ECONOMETRIC MODELS OF DOMESTIC WATER CONSUMPTION IN THE TUCSON METROPOLITAN AREA by

ECONOMETRIC MODELS OF DOMESTIC WATER CONSUMPTION IN THE TUCSON METROPOLITAN AREA by
ECONOMETRIC MODELS OF DOMESTIC WATER CONSUMPTION
IN THE TUCSON METROPOLITAN AREA
by
Leon Nicholas Ray
A Thesis Submitted to the Faculty of the
DEPARTMENT OF HYDROLOGY AND
WATER RESOURCES
In Partial Fulfillment of the Requirements
For the Degree of
MASTER OF SCIENCE
WITH A MAJOR IN WATER RESOURCES ADMINISTRATION
In the Graduate College
THE UNIVERSITY OF ARIZONA
1972
STATEMENT BY AUTHOR
This thesis has been submitted in partial fulfillment of requirements for an advanced degrei e at The University of Arizona and is
deposited in the University Library to be made available to borrowers
under rules of the Library.
Brief quotations from this thesis are allowable without special
permission, provided that accurate acknowledgment of source is made.
Requests for permission for extended quotation from or reproduction of
this manuscript in whole or in part may be granted by the head of the
major department or the Dean of the Graduate College when in his
judgment the proposed use of the material is in the interests of scholarship. In all other instances, however, permission must be obtained
from the author.
SIGNE
APPROVAL BY THESIS DIRECTOR
This thesis has been approved on the date shown below:
•-n
Hasan . Qashu
Associate Professor of Hydr ogy
and Water Resources
7
/ 9 :47_
Date
ACKNOWLEDGMENTS
First, I would like to thank members of The University of Arizona
faculty, particularly Dr. Hasan K. Qashu, Dr. Jerome Wright, and Dr.
Russel Gum, members of my Committee, for their advice and assistance,
and Dr. George Leaming of the Bureau of Business and Economic Research
for assistance in designing the questionnaire.
Next, I would like to thank Mr. Frank Brooks, Head of the Tucson
City Department of Water and Sewers and Mr. A. B. Hobbs of that Department, and Mr. Pat Dwyer, Head of the Data Processing Department and the
members of his staff, Richard Cronnican, Ray Ponikvar, and Albert Ybarra,
for their help in getting water consumption data from the City of Tucson's
computer files and in mailing the questionnaires.
I am grateful for the services of the University Computing Center
and the Data Processing Department of the City of Tucson, which were
absolutely essential to the retrieving and handling of the data in this
research.
TABLE OF CONTENTS
Page
LIST OF ILLUSTRATIONS
vii
LIST OF TABLES
viii
ABSTRACT 1.
INTRODUCTION, FUNDAMENTAL ECONOMIC
CONCEPTS AND URBAN WATER DEMAND
Importance of Water Consumption Research
The Classic Price-Demand Model
A General Supply Function
An Empirical Water Supply Function:
Jacobs' Wells
The General Demand Function
Elasticity
• Marginal Revenue
The Flat Rate and "Requirements" Approach
Commercial and Industrial Water Demand
Demand by Type of User
Objectives
2.
1
1
4
7
10
11
14
14
15
19
20
20
EMPIRICAL WATER DEMAND MODELS
21
Models of Howe, Linaweaver and Colleagues
Bruner's Models
The Strange Case of Buckeye
Zeizel's Water Consumption Model
22
26
28
31
3. DATA SOURCES AND DATA ACQUISITION
33
33
33
35
36
36
36
Sources of Data
Tucson Department of Water and Sewers
Pima County Tax Assessor
Census Bureau
Questionnaires
A Trial Procedure
iv
TABLE OF CONTENTS--Continued
Page
Implementing the Trial Procedure The Questionnaire Method Lot Size Estimating Property Value Questionnaire Design Information Desired Wording of Questions and Layout of
Questionnaire Specific Discussion of Questions
The Covering Letter 4.
37
39
40
40
43
43
47
48
50
. RESULTS Responses to Questionnaires Retrieving and Processing Water Consumption
Retrieval of Data from Water Consumption
History File Water Consumption Collation Data Selection and Tabulation Program Regression Analyses Correlation Matrix Discussion Regression Equation Water Rate Structure Rate 1 Subset of the Basic Observation Set Correlation Matrix Regression Equation Rate 2 Subset of the Basic Observation Set Correlation Matrix Regression Equation Deciding Which Observations to Include Regression Using Subset 3a Correlation Matrix Regression Equation 51
• •
51
53
53
54
54
54
55
56
60
60
66
66
69
69
69
74
75
80
80
112
5.
DISCUSSION AND CONCLUSIONS 113
6.
SUMMARY AND RECOMMENDATIONS 116
APPENDIX 1: FACSIMILES OF COVERING LETTER
AND QUESTIONNAIRE 122
vi
TABLE OF CONTENTS--Continued
Page
APPENDIX 2: SAMPLE SELECTION AND ADMINISTRATION
OF THE QUESTIONNAIRE
125
APPENDIX 3: CODING OF THE DATA
129
LIST OF REFERENCES
134
LIST OF ILLUSTRATIONS
Figure
Page
1
General Solution to Supply-Demand Model 6
2
Supply-Demand Model with No Solution 6
3
General Monopoly Supply-Demand Model
8
4
Supply Function for Single Firm, Short Run 8
5
Jacobs' Supply Function for Water, Tucson
Ba sin 12
vii
LIST OF TABLES
Table
1.
Page
Summary Questionnaires Mailed, Returned, and
Coded 53
2.
Correlation Matrix for Basic Set, All Rates 57
3.
Regression Coefficients and Related Statistics
Basic Set, All Rates 61
4.
Subsets of Data for Which Regressions Were Run .
. . .
63
5.
City of Tucson Water Department Rate Schedules .
. • •
64
6.
Correlation Matrix for Basic Set, Rate 1 67
7.
Regression Coefficients and Related Statistics
Basic Set, Rate 1 70
8.
Correlation Matrix for Basic Data Set, Rate 2
9.
Regression Coefficients and Related Statistics
Basic Set, Rate 2 76
10.
Correlation Matrix for Subset 1, All Rates 81
11.
Regression Coefficients and Related Statistics,
Subset 1, All Rates 82
12.
Correlation Matrix for Subset 1, Rate 1 84
13.
Regression Coefficients and Related Statistics,
Subset 1, Rate 1 85
14.
Correlation Matrix for Subset 1, Rate 2 87
15.
Regression Coefficients and Related Statistics,
Subset 1, Rate 2 88
viii
72
ix
LIST OF TABLES--Continued
Table
Page
16.
Correlation Matrix for Subset 2, All Rates 90
17.
Regression Coefficients and Related Statistics,
Subset 2, Ail Rates 91
18.
Correlation Matrix for Subset 2, Rate 1 93
19.
Regression Coefficients and Related Statistics,
Subset 2, Rate 1 94
20.
Correlation Matrix for Subset 2, Rate 2 96
21.
Regression Coefficients and Related Statistics,
Subset 2, Rate 2 97
22.
Correlation Matrix for Subset 3, All Rates 99
23.
Regression Coefficients and Related Statistics,
Subset 3, All Rates 100
24.
Correlation Matrix for Subset 3, Rate 1 102
25.
Regression Coefficients and Related Statistics,
Subset 3, Rate 1 103
26.
Correlation Matrix for Subset 3, Rate 2 105
27.
Regression Coefficients and Related Statistics,
Subset 3, Rate 2 106
28.
Correlation Matrix for Subset 3e 108
29.
Regression Coefficients and Related Statistics,
Subset 3a 109
30.
Summary of Results from Thirteen Regression Runs. 31.
Percentage Decrease in Mean and Standard Deviation of Consun lotion, Rate Area 1 to Rate
Area2
. . .
111
-
Oe
e
0000
4
f1
600•0000
•
114
ABSTRACT
Data by household on water consumption and twenty-seven
economic and cultural variables assumed to be related to residential water
consumption in Tucson were collected. Consumption models were constructed by a linear regression computer program. To develop empirical
water demand models, the classic economic demand function, C f(p),
is expanded to the function, C = f(p, x 2 , x 3 , ..., xn ), where p is Price,
and x 2 , x 3 , •.., xn are other variables. Both mean household consumption and variability among households decrease as water price increases.
In the lowest water price area water consumption was correlated significantly to the property value and pipe diameter at the meter, and slightly
less significantly correlated to number of bathrooms and number of trees.
In the second highest price area consumption was strongly correlated to
property value, pipe diameter, number of bathrooms, number of trees,
having a dishwasher, automatic clothes washer, garbage disposal, lawn,
lawn area, sprinkling system, whether property value is more than
$40,000, and whether the lawn is watered in the summer; it is moderately
correlated to the number of people per house and negatively correlated to
having an evaporative cooler.
CHAPTER 1
INTRODUCTION, FUNDAMENTAL ECONOMIC CONCEPTS
AND URBAN WATER DEMAND
There are, of course, many variables which are related to water
consumption. One problem is to determine which of these are relevant
and feasibly measurable--that is, which are highly correlated to water
consumption and which of these variables are controllable. Those variables which are the intersection of these three sets are of interest in
water resource planning.
Importance of Water Consumption Research
Projected water consumption is a very important factor in planning
the general development of urban areas, especially in arid regions where
costs of providing water may be a constraining factor. Thus it is important to study the variables which determine water consumption, or, as
economists call it, the "demand function" for water (which, together
with the "supply function," determines use). In Clausen's study of
Optimal Operation of Water-Supply Systems (1970), a water demand function plays a vital part.
In the conclusion of his report, Zeizel (1968b) stated that
"...there is a relationship between median family income and water
1
2
consumption for the City of Tucson... (but)) further investigation is
needed to determine the exact relationship." On the other hand, he also
said, "The broader question of which socio-economic parameters
(besides income) would best predict future water demand for the City of
Tucson should be...answered before further investigation into the effect
of the income variable is conducted."
In an article called "A Data-Collecting Network for the Sociosphere," Kenneth Boulding (1968, p. 98) said, "Decisions are always
made in the light of the decision-maker's image of reality," rather than
reality itself. Thus, government leaders are making decisions based on
bad or non-existent information, that is, unreal images of the social and
economic conditions of the society on whose behalf they are making
decisions. In order to make intelligent decisions and avoid costly, disastrous errors, it is necessary to collect pertinent, meaningful information.
We presume that this is as important in the area of water resources
systems as it is in the international systems to which Boulding was making
reference.
Boulding adds a discordant note, however, by saying that the
...value orientations of the decision- makers7..are frequently...strong,
and these...filter to exclude information which might challenge their existing images of the system" (p. 99). What he seems to suggest here is
that sometimes the decision-makers are seeking an image of reality which
3
will support a preconceived conclusion or decision, ex post facto, rather
than an image which supports a different one, even if it be closer to reality. This is indeed a discouraging note for the researcher, who would
like to think that all he has to do is to collect pertinent, meaningful information, publish it, and then let intelligent decisions happen. It is clear
that excellent research in water resources has produced pertinent and
meaningful information and it is also clear that often intelligent decisions
have not followed merely because this research was done. Although this
is an important problem, it is outside the scope of this thesis.
Howe and Linaweaver (1967, p. 15) wrote that " ...forecasting
(the) demands that will be placed on a (water) system is difficult at
best." He criticized two particular methods for forecasting water demand
because they " ...can lead to substantial over- or under-design of systems" (p..15). One of these methods is that "...of estimating population,
multiplying by an average daily per capita use, and then applying peakto-average ratios based on entire cities to estimate peak demands"
(p. 16). The other method is the "revised standards" of the Federal
Housing Administration (July, 1965, pp. 51-53). These standards, in
part, promulgate forecasting based on the assumption of four persons per
dwelling unit and 100 gallons per capita per day for average demand. The
FHA suggests estimating maximum daily demand at 200% of average daily
demand and peak hourly demand at 500% of average daily demand or 700%
of average daily demand if "extensive lawn irrigation is practiced."
4
lawn irrigation is practiced." Among the faults of these two methods is
that " ...they contain no allowance for climate , economic level, price
and other effects" (Howe, 1967).
Since Tucson is in an arid area, the consequences of poor water
resources planning decisions may be more costly than where water is less
scarce. Therefore the criticisms which Howe has leveled at these two
methods and the need for more accurate water consumption models may be
very important for Tucson. In dealing with the water demand function for
which Clausen cites a need, this thesis will be concerned with the socioeconomic parameters, price, and other effects which are mentioned by
Zeizel and Howe.
The objective of the remainder of this chapter is to review the
classic economic price-demand model, realign its constraints and definition so that it relates meaningfully to the urban water demand situation,
describe the previous empirical studies and evaluate them, and present
the framework for the present study.
The Classic Price-Demand Model
The classic supply-demand model describes how, in a free market
situation, the price and quantity of a particular commodity offered for exchange on that market are determined--that is, how the "forces" of supply
and demand (Adam Smith's invisible hand) operate to determine the two
5
variables: price and quantity. This model comprises two independent
functions,
q
=f(p)
(1)
and
=
g(P)
(2)
which are, respectively, the demand and supply functions, where q is
the quantity of water and p is the price per unit of water. The simultaneous solution of these equations (which define the two functions) gives
the transaction price (p) and the quantity (q) of the commodity which will
be exchanged. Figure 1 shows a graphic solution with hypothetical demand
and supply functions. The ripples are in the curves to suggest that real
supply and demand functions may not be as regular as theoretical ones
used by most economists (after Lancaster, 1969, p. 16). An important
assumption which is made about supply and demand equations is that they
will yield one solution. This may not be true always. Figure 2 shows a
situation where there is no solution--the sellers are asking more than any
buyer is willing to pay, resulting in no sale, no price, no market. This
situation might have occurred in Southern Arizona to force the Indians to
abandon their irrigation system. A long or severe drought requires that
the canal system must be extended to the next nearest source of water,
and the Indians may not have been able or willing to do this. In terms of
supply and demand functions, this means that the supply function has
6
Figure 1. General Solution to Supply-Demand Model
Supply
Demand
with
7
shifted upward to the position shown in Figure 2, while there was no
upward shift in the demand function. When this happens with a vital
commodity such as water, the response is to emigrate, perish, or make
a drastic change in culture--such as shifting from irrigation farming to
hunting and food gathering.
Within their market areas, water utilities are monopolies and the
perfect competition models above must be modified in order to describe
this monopoly situation. The supply function for water in the community
is renamed the marginal cost curve for the single firm which is the private
company or municipal department which is the only supplier which the
customers can patronize. The quantity of water which the utility will produce is determined by the simultaneous solution of the marginal cost and
marginal revenue equations, as illustrated in Figure 3, which is a shortrun model. The price is then derived from the demand equation. This
process will be discussed in more detail later.
The next sections will look briefly at a general supply function and
a specific water supply function for the City of Tucson; sections after that
will examine general demand functions, and then, in Chapter 2, the main
topic of the paper, specific demand functions for water, will be discussed.
A General Supply Function
The general supply curves in Figures 1 and 2 illustrate the
economic principle that as the price offered for a commodity increases,
8
Marginal Cost
J)
o
-'-4
a,
Pc
Demand
Marginal Revenue.
qm
" qc
Quantity
Figure 3. General Monoply Supply-Demand Model
Total Cost
•
Marginal Cost
Average Cost
Quantity
9
the quantity of that commodity which will be supplied (i.e., offered for
sale) also increases. Present producers will increase production and
other firms will enter into production as they become aware of the price
increase. Conversely, a price decrease causes a drop in the quantity
supplied.
The supply function for a single firm is its marginal cost function,
which represents the change in total cost which results from one incremental unit of production. Important relationships exist between the
marginal cost function and other functions which describe the economic
anatomy of the firm, as shown in Equations 3, 4, and 5.
MC = g(q)
(3)
TC = SMC + C = _(g(q)+C= G(q)
(4)
AC
TC
=—
G(q)
(5)
Total production cost (TC) is the integral of the marginal cost function
(MC). The average cost per unit produced (AC) is the total cost divided
by the amount produced (q). The integration constant (C) is the amortized cost of capital investment. The firm's marginal cost function is not
monotonically increasing like the supply function for a whole industry;
however, the meaningful area for operation of the model is along that
portion of the curve which is monotonically increasing. Figure 4 (after
Samuelson, 1969, p. 516) shows these curves and their interrelationships.
10
An Empirical Water Supply Function: Jacobs' Wells
In the past, the greater portion of research in water resources has
been in the area of supply, rather than that of demand. The nature of this
work is primarily technological: engineering, hydrology, geology, meterol-
ogy, statistics, etc. Such research usually does not get much deeper
into economics than cost accounting and computation of benefit/cost
ratios. Even then, the estimates of benefits and indirect costs are determined sometimes rather arbitrarily, or they are based upon political
instead of economic considerations.
James Jacobs (1968) developed empirical water supply functions
by dividing the potential well fields near Tucson, Arizona, into eighteen
units, calculating the costs of producing water from these eighteen areas
and delivering it to the existing 22nd Street reservoir, and ranking these
areas according to increasing cost. These costs included the pumping
costs, the amortized drilling and construction costs, land acquisition
costs, etc., and, as indirect costs, the loss to the community caused by
the termination of farming activity on the land which overlies the well
fields. The sizes of these eighteen areas, which Jacobs called "diversion units" were defined such that each area (except two) could produce
ten million gallons of water per day at the rate of pumping currently in
practice on that land. Jacobs had decided that ten million gallons per day
was an optimum increment for orderly expansion of this type of water-producing system. (For hydrological reasons, two of these areas would
produce less than ten million gallons per day.) When these costs are
ranked in ascending order and plotted, they form the stepped function
shown in Figure 5. The water unit shown on this supply curve is acrefeet per year, where 9,200 acre-feet per year is equivalent to 10,000,000
gallons per day, assuming the wells are operated 360 days per year.
Because Jacobs treated all the costs as fully amortized and variable for each diversion unit, this marginal cost curve looks more like the
supply curve for a competitive industry rather than like the short-run
marginal cost curves in Figures 3 and 4.
The General Demand Function
As long as all the other parameters and variables which could
affect the consumer decision-making process are held constant, the
amount which a customer is willing to purchase, of any commodity or
service, will vary inversely with the price of that item. The rationale for
this statement is based partly on observations of society but primarily on
logical conclusions drawn from fundamental assumptions. As mentioned
supra, the function is monotonically decreasing, but need not be regular.
Generally, goods and services are scarce, that is, there is not
enough of everything for everybody. In this scarcity situation, the goal
of the consumer is to choose a combination of amounts of each commodity
and service which maximizes his "satisfaction" (admittedly a subjective
concept); each item has a price and this forms the basis for a calculus of
12
o
o
‘ntn
C, 7
I
CL)
0
./.:Xt? .2 0
U
.
L
rn
-
o
13
choice: the quantity of each item to be acquired is adjusted so that the
satisfaction derived from incremental amounts of each item (in money
terms) is the same.
General consumer demand theory states that as the quantity of a
good which a consumer has increases, the satisfaction or value ("marginal
utility") to him of each successive incremental unit declines; it follows
from this principle that also the price which he is willing to pay for incremental units (the marginal price) declines along with the marginal utility.
The demand function (Equation 1) in Figures 1, 2, or 3 intersects
the Quantity axis when the price is zero. This is the point at which the
consumer has so much of a commodity that he will not accept any more of
it even if the price becomes zero. Although this situation is merely a
hypothetical curiosity with most commodities, it becomes important in the
study of water resources because it is precisely what happens when water
is sold at a flat rate--that is, where the consumer is offered "all you can
drink for $X."
As mentioned at the beginning of this section, the demand function of Equation 1 and Figures 1, 2, or 3 are based on the traditional assumption that all the other relevant factors, variables, or parameters which
could affect consumption remain constant. Thus, when there is a change
in one of these other factors, the result is a new demand function, which
is often called a shift in the demand curve (Samuel son, 1961, p. 434;
14,
Lancaster, p. 17; Alchian, 1967). The demand function,
C =f(p)
(6)
then becomes the family of functions
C = fi(p)
C = f2 ( p )
•
•
•
•
•
•
(7)
•
•
C = f(p)
p)
where C is consumption and f n are different functions resulting from changes
in non-price factors.
Elasticity
The total revenue (R) is the product of the quantity consumed (C)
and the price (P).
R = P * C
(8)
The elasticity is defined as
e —
dC/dP
dC/C
C/P
dP/P
(9)
Marginal Revenue
Marginal revenue is the derivative of the total revenue function
with respect to price, or with respect to consumption:
R = P * C
(10)
Since C = f(P), R =P * f(P)
(
MR =
(il)
PC
.)
(12)
And since one may also say
P =f(C),
(13)
MR= dR _ d Cf(C)
dC
dC
(14)
The unregulated monopolist determines the amount of water he is going to
produce by the simultaneous solution of his marginal cost and marginal
revenue functions. But the price is established from the demand function.
Figure 3 (after Samuelson, 1961, p. 531) shows graphically how this is
determined. The price (P c ) and quantity (q 0 ) are what the customers
would have if competition existed. The unregulated utility will raise
price to P m and customers will buy q m . The important thing to the customer, however, is that he is paying a higher price and consuming less
water than he would if free competition were to be possible.
The application of these concepts to a real situation depends on
whether the data and mathematical techniques to construct these functions
are available and whether the management of the utility pursues the goal
of maximizing profit.
The Flat Rate and "Requirements" Approach
Flat rate billing means that the basis for charging the customer is
something other than the amount of water used, such as lot size, frontage
feet, property value (ad valorem tax), number of bathrooms in the house,
number of faucets, water using appliances, fixed or minimum service
charges, and other criteria which could be used.
16
When public utilities first began selling water, they used flat
rates because reliable metering hardware was not then developed, and
historical tradition helped maintain the flat rate to present times.
Another factor which supports the use of the flat rate is that it is
expensive to install meters on all services, read the meters periodically,
and compute bills on the amount consumed. It is often argued that this
expense would increase the total cost of providing service, and that
people needed most of the water that they used.
This argument has been refuted by the experience of the City of
Boulder, Colorado, as reported by Hanke and Flack ( 1968 ) . They found
that "...as a result of complete metering, Boulder had the capacity to
serve 11,000 more people with the same water supply." This meant that
the City could defer costly expansion of their production and treatment
facilities by metering their customers. There was a saving in variable
costs because the variable costs involving larger treatment facilities
were not incurred, less water had to be treated in the existing facility,
and a long run saving was made because design parameters could be reduced. Hanke and Flack also asserted that there is a saving to be made
by deferring capital investment expenses. This is true in a non-inflating
economy, where the present value of construction performed now is higher
than the present value of construction done some years in the future. In
an inflating economy the validity of this assertion depends on subtleties
17
of inflation rates, discount rates, present versus future costs of acquiring
land or other resources which might be available now but not in the future
or vice versa, and technological developments which could affect costs.
Another saving to the community to be considered is the reduction
of sewage treatment costs which naturally follow a reduction in indoor
water consumption.
Hanke and Flack (1968) and Bruner (1969), as well as other
authors, make reference to the "requirements" approach to water pricing,
comparing it (unfavorably) with the "price-demand" approach, though
they do not define the former term explicitly. An implied definition of
the "requirements" approach appears to be based on the following considerations: an adequate supply of potable water is an absolute necessity
for the maintenance of life, health, sanitation, and a decent standard of
living. Though the slogan, "Cleanliness is next to godliness," never
appears explicitly in technical literature, it is implied by the "requirements" approach. It is assumed, therefore, to be the duty of the water
utility to provide everyone (except the poorest, perhaps) with all the water
they need at a price which they can afford to pay--so they may be clean
and godly.
The experience of the City of Tucson, where only about 50% of
the water sold by the Water Department is collected in the sewers, and
literature which will be cited later, show conclusively that residential
18
water is used for many other purposes than mere maintenance of health
and sanitation.
In 1960 the City of Boulder, Colorado, 50% of the water customers
were metered and annual use was 243 gallons per capita per day; in 1965
all customers were metered and annual use was 149 gallons per capita
per day, which is a 40% drop. At the same time, no drop in health or
sanitation standards was reported, indicating that this water was not
essential to maintaining health and sanitation.
When water is billed at a flat rate, the per unit price of water is
zero, and consumption expands to the zero-price quantity which is illustrated in Figures 1, 2, and 3. Consumption stops at this point because
any more water is perceived as a nuisance, that is, it has negative
marginal utility. In physical terms, this means that the lawn may be
damaged by more water, or that a leaky faucet has become annoying. This
is the point at which a customer would have to be paid to accept more
water, or conversely, it is the point at which he would pay to get rid of
the "excess" water by drainage works or flood protection. This concept
is shown graphically by extrapolating the demand functions in Figures 1,
2, and 3 into the negative price quadrant.
The main argument against flat rate billing is that it encourages
waste because there is no economic reward for careful usage. Plumbing
repairs and other waste reducing rneasurcs cost money and other effort,
and with a flat rate, the customer's incentive to stop waste comes from
the annoyance of the dripping faucet and from guilt about waste, which
might be reinforced a bit by preachments from the utility's publicity department.
The water consumption prediction model which follows from the
requirements or flat rate approach is to multiply the per capita "requirement" by the population projection and a safety factor.
Commercial and Industrial Water Demand
Conceptually, water consumption by business and industrial
establishments can be put into two categories: (1) water used personally
by the workers and customers of these establishments while they are on
the premises, and (2) water used in producing the service or product of
the business, directly or indirectly. For example, water used by air
conditioning systems, washing floors, or maintaining a decorative fountain to attract customers may be considered as indirectly related to the
production process. Businesses which obviously use water directly in
their productive processes are laundries, car washes, buildings with
large air conditioning plants, and restaurants. These categories are not
hard and fast. Pay restrooms produce revenue directly, even if the fee is
designed to prevent use by non-customers or vandalism. The restrooms
in a gas station are there specifically to draw customers, while those in
other places are for the people who are already there.
20
Demand by Type of User
At present, the Tucson Water Department does not break down
water sales by types of users--that is, consumption by single residenbes,
apartments, hotels, laundries, car washes, office buildings, etc. By
reading the names and addresses of the recipients of water bills, rough
inferences can be made as to whether the customer is a private person or
a business. These inferences are subject to error because some water
bills for businesses are sent to individuals and some water bills for residences are sent to business films, particularly real estate firms.
Objectives
The purpose of this thesis is to determine which of the several
variables which might be related to residential water consumption are the
most significant in water demand functions. To do this, we first determine what data are available and their sources, and then we develop a
procedure or procedures for using this data to develop and test water consumption models.
CHAPTER 2
EMPIRICAL WATER DEMAND MODELS
Up to now, we have been dealing with traditional economic
demand theory, which begins by making theoretical analyses of the
behavior of consumption in response to price changes, assuming that all
other parameters which could affect consumption remain constant, which
is expressed by the equation,
C = f( p ) ,
(15)
where C is consumption and f(p) is some function of price. Changes in
any of the non-price parameters cause a shift in the demand curve, thus
producing the group of functions,
C = fi(p), i = 1, 2,
n.
(16)
This expansion of the demand function into a family of n functions is
adequate for purposes of developing economic theory. When we get to the
area of econometrics, however, and the goal is to define a particular
demand function for a particular commodity, from observations of empirical
data, it seems better to define a single consumption function, such as
C =
x2, x3 , •••, x n )(17)
where C is consumption, p is price, and x 2 , x 3 , ..., x n are the (independent?) variables other than price, The process of building a model then
21
22
becomes a problem of identifying these other variables, discovering the
character of these variables (linear, exponential, etc.), and determining
which ones are really important—that is, ranking them by importance.
The development of electronic computers and multiple regression programs
makes it feasible to gather data on parameters which might be correlated
to water consumption and fit them to a consumption model. Some of the
more relevant empirical studies which have been done by this technique
are discussed below.
Models of Howe, Linaweaver and Colleagues
One of the best known and most thorough studies of residential
water demand are those done by Howe and Linaweaver (1967), Howe
(1968), Linaweaver (1966, 1965) and others who participated in the
Residential Water Use Research Project at John Hopkins University. This
residential Water Use Project gathered consumption and price data from
39 study areas which consisted of 34 to 2373 dwelling units each. These
were divided into five categories: (1) metered with public sewer in the
western United States, (2) metered with public sewer in the eastern
United States, (3) metered with septic tanks in the eastern United States,
(4) flat rate with public sewer, and (5) apartments where buildings but
not individual tenants were metered.
Each of the selected areas was homogeneous with respect to the
value, lot size, age of subdivision, water price, and climate. Master
23
meters were attached to the water mains serving each area; these meters
recorded consumption in fifteen-minute intervals. Thus, data were
obtained to compute peak hourly flow, peak day flow, average day flow,
and seasonal peaks. Howe made the assumption that indoor ("domestic")
consumption was fairly constant with respect to seasons and that outdoor
("sprinkling") use was practically zero during winter or the rainy season.
Based on this assumption, average daily sprinkling uses were calculated
by subtracting average daily consumption during the non-sprinkling season
from average daily consumption during the sprinkling season.
The season of peak water consumption coincides with the periods
when lawns and shrubs need artificial water, so it is logical to conclude
that most of the excess consumption is used for "sprinkling" --most, but
not all. Some of this excess consumption is used to operate evaporative
coolers, for increased washing of paved areas, and perhaps more laundry
may be done.
The data used in these regression models were averages for each
of the thirty-nine areas in the study rather than observations of individual
dwellings. As it was specifically pointed out by the authors, "There are
potential statistical difficulties involved in using averages in regression
analyses, since the variance of each average is inversely proportional to
the product of the number of time periods and dwelling units included in
each average "
24
The total residential demand is the sum of two components,
domestic (indoor) demand and sprinkling (outdoor) demand:
C = CD + CS
(18)
where CD is domestic demand and CS is sprinkling demand. The domestic
demand model for metered areas with public sewer, east and west United
States is
CD = 206 + 3.47*V - 130*P
(19)
where V is property value in thousands of dollars and P is the unit price
for water some unit price per thousand gallons. (The authors' price unit
is not too clear.) "East" comprises the cities of Des Moines, Fort Worth,
Little Rock, Washington, Baltimore, and Philadelphia; "west" comprises
Oakland, Los Angeles, and San Diego. The domestic demand model for
apartments and flat rate customers is
CD= 28.9 + 4.39*V + 33.6*D
(20)
where V is property value and D is population density in persons per
dwelling unit. The domestic (indoor) demand model for metered areas
with spetic tanks is
CD = 30.2 + 39 . 5*D
(21)
where D is population density in persons per dwelling unit. It is interesting that price is not a significant variable in metered areas with septic
tanks. Apparently concern for overflowing the septic tank is a more
limiting factor for indoor consumption than price. In Howe's study areas,
25
the mean price per unit of water was higher in the metered-with-septictank areas; this fact makes it even more surprising that the unit price for
water is not significant.
The equations for outdoor (summer sprinkling) are in exponential
form. In the ten western areas in the study the average summer precipitation is .15 inches; in the eleven eastern areas the average summer precipitation is 11 inches. The authors concluded that fitting the data
separately for eastern and western states gave the best results. The
equation for metered customers with public sewer in the western United
States (that is, with .15 inches average summer rain) is:
CS = 1130 * p**(-703) * V**( .429) (22)
For metered customers with public sewer in the eastern United States
(11 inches average summer rain) is:
CS = .164 * B**( -.793) * (W-.6*R)**2.93 * V**1.45 (23)
where B is irrigable area in acres, W is potential evaporation in inches,
and R is rainfall in inches. For flat rate customers, in the east and west,
the only significant variable is property value, and the equation is
CS = 100 - V**.783
(24)
The authors were surprised to find that irrigable area was not significant
with flat rate customers or with metered customers in the west. Their
conclusion was that " ...the apparent instability of the (irrigable area)
exponent may be partly due to the small range of variation in (irrigable
area) within each group of study areas..." Part of the trouble, too,
be that irrigable area was estimated, not measured.
26
Bruner's Models
In a dissertation on water demand in the Phoenix metropolitan area,
Bruner (1969) developed residential water consumption models based on
four variables: price of water, "income," lot size, and climate. His goal
was to show that on an annual basis "...residential water consumption
was a function of price, income, and lot size" (p. 9). A second point
was that " ...seasonal changes in residential, commercial and public
water consumption (in the aggregate) were a function of climate" (p. 9).
The most substantive consumption models which Bruner developed
followed very closely the methods of Howe and Linaweaver (1967), who
fitted regression lines to data from thirty-nine geographic areas with
different rate structures. For his data, Bruner selected six areas in the
Phoenix metropolitan area which had different rate structures and took a
sample of twenty single family residences from each. The form of the
consumption model is
C = f(Pf, Pi, Y, LS)
(25)
where C is consumption, Pf is the fixed part of the water charge, Pi is the
incremental water charge, Y is "income," and LS is the lot size. The
water consumption and billing rates were obtained from the utilities serving
each area, the lot size and full cash value were obtained from the Maricopa County Tax Assessor's Office.
The empirical data which were thus gathered were fitted to the
multiple linear regression model of the form
27
C = b-u + b 1 *Pf + b 2 *Pi + b 3 *Y + b *LS E
4
(26)
where the lots are coefficients, E is an error term, and the other variables
are as described supra. C is average monthly consumption, that is 1/12
of annual consumption, probably in gallons. (The author did not define
the unit explicitly.) When the coefficients are computed using a linear
regression program, the following equation was developed:
C = 105 - 1.45*Pf = 218*Pi ± .015*Y - .001*LS. (27)
Bruner also fitted his data to a curvilinear or exponential equation of the
form:
C = b + b logPf + b logPI + b logY + b lOgLS + E
2
0
1
3
4
(28)
The values which he found for b through b were, respectively, -1866.78,
0
4
-10.63, -134.98, 491.53, and -.46. In comparing these two models, he
concluded that " ...based on the computed coefficients of determination,
there is little to choose between the linear and curvilinear models with
respect to their ability to explain changes in residential water consumption."
In the introductory chapter of Bruner's dissertation, income was
discussed (pp. 9, 11) as one of the variables affecting residential water
consumption. In the following pages, however, there was no definition
of income--the reader was left to wonder whether this meant personal
income, gross income, net income after taxes, family income, etc. The
lack of a precise definition for income fades from concern, however, when
28
Bruner discusses the actual data which are used for the models, for he
says that the Maricopa County Tax Assessor's "...full cash value will be
used in the model as a measure of home value and as a proxy variable for
income" (p. 128). Thus, the actual variable which is used is property
value and not income at all. The concept of using property value as a
"surrogate for income" or a "proxy for income" presumes that there is a
definable functional relationship between a person's income and the value
of the house in which he lives. Although it is true that people with high
incomes have the means to live in more expensive homes than those who
have low incomes, it seems that more evidence than a common sense presumption should be cited to support the contention that people with high
incomes actually do elect to live in more expensive homes than those with
low incomes before we can assume that property value is a legitimate
substitute for income.
The negative coefficient in Equation 27 for lot size is surprising;
it suggests that people with large lots might use less water than those
with small lots; or it indicates that there might be co-variance difficulties
affecting the regression analysis.
The Strange Case of Buckeye
When the findings of Bruner (1969) concerning the experience of
the City of Buckeye, Arizona, with rate changes are compared with the
finding of an author not cited by I31 uner (Haney and Hamann, 1965), an
-
29
Prior to the installation of an electrodialysis plant in September,
1962, the water delivered by the City of Buckeye had a "...total dissolved salt content... (of) about 2,000 to 2,500 parts per million." At
this time the price for 7,000 gallons monthly was $3.70. Upon installation of the treatment plant, rates for the same amount of water increased
to $10.15, and total consumption dropped sharply the very next month.
During the next two years, rates were increased twice more, accompanied
by further reductions in consumption. Thus far, traditional economic
theory is confirmed. But in the next two years, water consumption declined slightly while the rate remained constant. Bruner's regression
equation for this water demand function is
log(C) = 2.4960 - .8715 *log (p)
(29)
In this study Haney and Hamann (1965) reported the following:
From a survey of customers in Buckeye, Ariz., the following
per family "hidden costs" of using water with 2,200 ppm total
solids were estimated: home water softeners, $4.90; water conditioning agents, $2.94; bottled water, $6.42; repair and replacement, $4.10; and excess soap and detergent use for people not
using home water softeners and water conditioning agents, $2.33.
It was concluded that the citizens of Buckeye paid approximately $40,000 in water bills for the fiscal year ending June
1960. The hidden costs listed above totaled an additional
$80,000 for the year. Thus, the average monthly water bill per
customer was $5, and hidden costs, in excess of the water bill,
amounted to more than $10 per month (p. 1076).
One of the key assumptions in price-demand theory is that the
units of the commodity under consideration be identical in quality. The
demand function which Bruner developed for the City of Buckeye was
30
developed from data representing two qualities of water, which is like
comparing two separate, albeit closely similar, commodities. The technique of mixing data from two different qualities of a commodity to create
one demand function needs careful justification whenever it is attempted.
The five dollar average water bill cited by Haney and Hamann
apparently assumes an average monthly consumption exceeding 7,000
gallons. The important point, however, is that before the electrodialysis
plant began operation the total water bill, including the hidden costs was
$15 per month. After the installation of the electrodialysis plant, the
total cost of getting softened, treated water into the home dropped. With
a decrease in price, traditional economic theory predicts an increase in
consumption; the facts here seem to indicate that the opposite has happened. Part of the explanation may be that when the customer was making
separate payments for bottled water, water softening service, excess soap
and detergent, and excess plumbing repairs, he didn't realize how much he
was paying for water. If all of these hidden charges had been gathered
together and billed as a unit by the water utility, as they in effect had
been done when the utility began delivering treated water, consumption
might not have been as high as it was. Water used outside the house
and for flushing toilets does not need treatment, and the price for this
water was raised too, of course. Bruner (1969, p. 138) found that
seasonal peaks and troughs in Buckeye's water sales were considerably
31
flattened out after the electrodialysis plant began operation. Factors
tending to reduce the seasonal peak are reduction of sprinkling and the
installation of recirculating pumps in evaporative coolers. This suggests
that much of the cut-back in consumption was in the water that did not
need desalinizing.
Zeizel's Water Consumption Model
Eugene Zeizel (1968a, 1968b) made a study of the relationship
between annual residential water consumption and median family income
in Tucson, from which he computed the following equation:
C 63,194 ± 17.732*X
(30)
where C is annual water consumption per customer (not identical to household) and x is median family income in dollars. The R
2
is .72, which
means that 72% of the variation in consumption is explained by income.
The unit of study from which the data were obtained was not individual households or residences, but totals from four selected geographic
areas within the city called "books." A book is the area covered by one
meter reader in one day. These areas remain constant although the number
of customers varies, owing to turn-ons and turn-offs. The basis for the
selection of the four books (areas) was (1) that each book had to be included within one census tract, (2) that the books contain as few
commercial customers as possible, and (3) that the books represent two
different levels of income, The average consumption per customer was
32
computed by dividing the total consumption for the book by the number of
customers. This was done by the month for the four books for six years,
producing 24 observations of average annual consumption. Zeizel indicated that errors as high as ten percent in water consumption are possible.
Median family income data for the four census tracts which
covered the four book-areas for the year 1959 came from the U. S. Census.
City-wide annual percentage changes in income were derived from the
"Effective Buying Income" data which is compiled and published by Sales
Management Magazine. These percentage changes were applied to the
1959 census data to generate 24 observations of median income. It was
assumed that city-wide changes in income had the same proportional
effect on all levels of income.
CHAPTER 3
DATA SOURCES AND DATA ACQUISITION
This chapter is concerned with developing a methodology for
getting the data necessary to achieve the objectives of the thesis. The
first step in accomplishing this is to determine what are the sources of
data which are available; then procedures are developed to obtain the data.
The final method used to obtain household information was a questionnaire; sample selection of the questionnaire is described in Appendix 2.
Sources of Data
This section describes the sources of data and the information
which each source is able to provide.
Tucson Department of Water and Sewers
Subsequent to Zeizel's study, the Water Department and Data
Processing Department began keeping records of monthly water consumption by individual customers on a magnetic disk file. This file contains
the accumulations of monthly consumptions and the numbers of years'
observations for each customer--not the actual monthly consumption for
the most recent twelve months. That is, for a certain customer the
history file might contain the total January consumption for 1, 2, or 3
33
34
years, the total February consumption for a different number of years,
etc. The number of years (times 30) is also in the history file, so the
average consumption for each month may be computed by dividing total
consumption by the number of years (divided by 30). A change in the
method of keeping the consumption history file is contemplated by the
Data Processing Department, and in the future the monthly consumption
for the most recent twelve months may be kept, rather than accumulated
consumption for several years.
The computer used by the City is an IBM 360/40 with magnetic
disk storage. A special program was written to select specified customers
who constitute the random sample from the Consumption History Disk File
and extract the consumption data and punch it on cards for further processing by the University's CDC 6400 computer.
The Department also has printed records of the billing for each
customer account for each month. At least three lines of computer printout are required for each customer for each month. There are about
70,000 active accounts, so the bound volumes of billing records for one
year create an eighteen-foot stack of computer printout. This means that
the time required to do manual searching for a large sample of customers
is prohibitively long. The records are stored in a utility tunnel in the
basement of the City Hall.
35
Pima County Tax Assessor
The Pima County Tax Assessor's Office has data on land value,
improvement value, total property value, lot size in hundredths of an acre,
the number of rooms, bathrooms, and plumbing fixtures, the existence of
a swimming pool or a private well, school district, land use classification, and several other items of data of lesser interest to this econometric
study. The information is available in two forms: (1) 9" by 11" cards,
and (2) a magnetic tape file.
The cards contain all of the information concerning the parcels of
property. There is one card for each parcel of land in Pima County. This
is the card on which the assessors record the information when (if) they
visit the property; thus it is considered to be the original or basic source
document on land parcels. These cards are filed by a number called
"state code number" or "parcel number."
The magnetic tape files are under the supervision of Dare Griffith
in the Computer Section of the Finance Division of the County. The tape
files contain a subset of the information on the file cards: lot size, land
value, improvement value, and land use code are the items of interest
to the water consumption study. There is also other information on the
tapes which might be of interest as search keys. The records are in sort
by the parcel number mentioned above, and the individual land parcel is
the logical unit about which ihe files are organized
36
Census Bureau
The United States Census Bureau has income, property value, and
plumbing fixture data at the level of census tracts. Data concerning
individual households are not made public. This was part of Zeizel's
source of information concerning median family income.
Questionnaires
The individual householders have a wealth of information concerning the things which cause them to use water, such as the number of
water-using appliances they use, the people in the household, landscape
style or feature, the amount of time they spend at home, etc. They might
be induced to reveal some of this information on a questionnaire, perhaps,
maybe, we hope.
A Trial Procedure
It seemed most desirable to gather data on individual customers,
by type of customer. The data which Zeizel (1968b) and Howe and
Linaweaver (1967) and their colleagues worked with were the means of
groups of households. When the only data which are used are the means
of groups, no information about variations within the groups enter the
study; it seemed desirable to find out how large such variations were,
and whether they were significant. The customer or household seems to
be a natural unit of investigation because this is the unit at which water
consumption is measured and billed. This is the level at which water
37
consumption is controlled, and if the person who controls or supervises
water consumption is also the billpayer, price is a potential variable in
determining consumption. Of course, it is desirable to study water consumption by all customers--residential, commercial, institutional, and
industrial. However, information on the type of customer was not readily
available, as was mentioned in the previous chapter. It was decided to
restrict the study to single family residences, as these constitute the
largest group of customers, and it could be determined whether the work
of previous researchers could be replicated in Tucson.
Since the Tax Assessor's card file contains the lot size to the
hundredth of an acre (and dimensions in hundredths of feet also), property
value, the existence of a private well and/or swimming pool, type of
dwelling, number of rooms, bathrooms and plumbing fixtures, it was
decided to choose a random sample of Tucson Water Department customers,
match them with the Tax Assessor's card file, and record these observations. This would permit a study which could create a model similar to
those of Howe and Linaweaver (1967) but using individual residences as
the data base instead of means of groups of them, while adding some new
variables.
Implementing the Trial Procedure
The first step was to select a simple random sample of customers
of the Tucson Water Department who were single family residences. It
38
was decided to do this by selecting the sample from the computer printout
of the January, 1968, billing list. Meter reading and billing are done on
a cycle basis, with 21 meter-reading days per month. The computer
printout for one day's billing is called a block or folio; each folio comprises several "books," which are the number of customers which one
meter reader covers in one day. Since Tucson is a winter resort community, January is probably the month of maximum housing occupancy, which
should mean that this is the month which will have the highest number of
active accounts. A systematic-random sampling technique was used
instead of a pure random sampling technique. The system was to choose
the fourth account from the top of every fourteenth page (every seventh
even page). This system would pick up about 350 accounts when the
twenty-one blocks were covered. If the fourth account name is that of a
business, a forward scan was made to the nearest account which was a
person-name. The customer number, name, address, minimum rate, and
January consumption were copied onto a keypunch form.
The next step was to get the state parcel number for each water
customer in the sample. The Tax Assessor's office does not maintain a
cross reference listing of parcel numbers by the street address of the
parcel or the name of the owner. (The County Treasurer's Office has an
alphabetized list of the person or mortgagor who pays the taxes.) With
the street address of the land parcel, it is possible, in theory, to find
39
the parcel number by referring to a sequence of three or more maps. The
first of these is a large, grimy wall map which contains most of the
streets. The street and block containing the parcel must be located on this
map; then the name of the tract or subdivision is read from this map. The
subdivision name is the key to the detailed maps. There are over 100
tract names in Tucson. The tract detail maps are bound in about 80 large
books. Two or more levels of detail maps must be referenced before a
parcel number can be found. A dry run of this procedure determined that
ten to fifteen minutes are required for each piece of land. The assistance
of one of the employees in the office, John Beard, was vital in finding the
parcel numbers. Once the parcel numbers were obtained, finding the
cards was relatively easy. The dry run was misleading; the actual time
to get one parcel number was 30 to 60 minutes, and about ten percent of
the parcel numbers could not be found at all. Thus, it was decided that
a new procedure needed to be worked out.
The Questionnaire Method
The new procedure was to use a mailed questionnaire which would
be enclosed with the water bills. In anticipation of a 20 percent response,
or less, the sample size was increased to 3,000; this would produce
about 600 responses. Generally, returns from questionnaires usually
introduce an educational and cultural bias to the responses, and probably
an income bias as well. One great advantage which the questionnaire
40
has is that many more aspects concerning possible water-related variables
may be investigated than would be possible if information sources were
limited to that which is maintained in the Tax Assessor's files. But
respondents' answers to questions about lot size and property value may
not be as accurate as the Tax Assessor's data.
Lot Size
Householders usually will give lot dimensions to the nearest five
feet; on a "normal" lot, this is an error of about 300 square feet. The Tax
Assessor's magnetic tape data are accurate to the nearest hundredth of an
acre, which implies a potential error of 200 square feet, while the manual
card file data has both acreage and length-by-width dimensions to the
nearest hundredth of a foot, which implies less than one square foot of
error.
Estimating Property Value
There are several methods for estimating the value of residential
real estate: (1) actual sales price, (2) loan appraisal value, (3) legal
appraisal, (4) insurance appraisal, (5) revenue stamps, (6) Tax
Assessor's "full cash value," (7) discounted rent, and (8) self estimate
of the owner. Each of these estimates may be different and each has its
own purpose; none can be said to be "wrong."
If the house has been sold recently, the actual selling price is a
good indication of its value at the time of the transaction. Because of
41
the effects of inflation, depreciation, and changing nature of neighborhoods
the selling price loses validity as a measure of current value within a few
years. The selling price may not be a valid measure of current fair
market value if the house was sold to another member of the same family,
if the owner had to sell on short notice, if the buyer is new to the community, unsophisticated, or has to buy quickly, or if the sale is forced by
probate or foreclosure. Real estate does not come in the standardized
units of identical quality about which traditional supply-demand theory is
constructed. Nearly all houses are bought with a mortgage, and the
interest rate which is charged affects the true price--especially if it is
a fraction above or below the long-term prevailing rate. The practice of
giving or taking "points" significantly affects the true price.
The loan appraisal value tends to be conservative, low. It is the
lending agency's estimate of what they feel certain they can sell the
house for if they foreclose.
Legal appraisals are ordered by courts when necessary for probates, litigations, or other reasons.
The purpose of an insurance appraisal is to estimate the cost to
replace a structure in the event of total loss. The basis of this estimate
is current construction costs or projected future construction costs.
Federal revenue stamps are purchased by the buyer of the
property and applied to the deed to make it valid, The number of stamps
42
is supposed to indicate the price paid for the property, and the value of
the stamps is on file with the County Recorder (of deeds). This is a poor
indicator of the true value of the property because the number of stamps
purchased may overstate or understate the value of the property according
to whether the buyer of the real estate wishes to get a high loan or reduce his property taxes.
The Tax Assessor's full cash value is that office's best estimate
of the fair market value. Usually it is a little low; this reduces the
number of complaints about over-valuation.
If a house is rented, it is an income producing asset. As such,
the (discounted) present value of the income stream which it is capable
of producing is the value.
The self-valuation of a house by the owners or occupants should
be accurate to within about $5,000. They are aware of the present
physical condition of the house, the purchase price, and effects of inflation. Wishful thinking will tend to bias their estimates upward.
If all of these measures of property value are highly correlated
(and they should be since they are estimators of the same statistic),
then any one of them should work in an econometric model of water consumption.
43
Questionnaire Design
In designing the questionnaire several factors were considered:
the information to be obtained, phrasing the questions, the method of
administering the questionnaire and the constraints on the layout and
printing which this entails, and lastly the method of coding and evaluating the responses. The most important goal in designing a questionnaire
was to induce the respondents to return it, with most of the questions
answered. Other design features, such as convenience in processing the
responses and simplicity of administration were secondary.
Information Desired
Stated most broadly, all of the variables which could possibly
have an effect on residential water consumption are the information
desired. With a questionnaire, the only information which can be obtained is that which is easily known to the respondent, and which he is
willing to give. Also, the questions must be short, easy to understand,
easy to respond to, and few in number. If not, the rate of returns will be
low. Therefore the desired information was limited to these variables:
property value, number of persons living in the residence, number of
bathrooms, number of water-using appliances, landscape style, lot size,
lawn size (if any), pool size and type (if any), and certain control information.
44
Control
Information. Control information is that which is not
pertinent to the subject matter of the survey but is necessary for meaningful evaluation or processing of the returns.
The questionnaire is directed to single family residences, but
some Water Department customers who are not single family residences
may receive questionnaires in spite of reasonable efforts to avoid this
occurrence. The Tucson Water Department presently does not code its
billing records by type of customer (single family dwelling,,aparLment,
store, office, etc.), although there are plans to do this in the future. In
the process of selecting the random sample, an inference as to whether
the customer is a residence or a business is available by looking at the
name to which the bill is addressed; but this is not infallible. Therefore
a question asking whether the enclosed bill applies to a single family
residence is necessary.
Some houses are rented, and it is necessary to know whether the
tenant or the landlord pays for the water. If the tenant, who controls
the water usage (at least partly), does not pay the bill, he is a flat rate
customer. If the tenant pays the bill he is not. This affects waterusing habits, and it is necessary to ask a question which will distinguish
these two types of tenant.
Some older houses have private wells which were the water supply
before utility service was extende,c1 to the area, and some of these wells
45
could be in current use as secondary (alternate) water sources. A question
about whether city water is supplemented by a private well is necessary
so that responses from such houses can be eliminated.
Property Value. The studies of Howe and others, and Bruner (1968)
considered property value as an important variable. In Bruner's study
(1968), though it claims to use income as an independent variable (p. 9,
p. 11), actually uses property value as a substitute for income. Therefore it was decided that property value would be an important variable for
the study. The specific information desired is the current market value of
the house and lot. To get this information to the nearest $5,000 would be
adequate.
Number of Persons and Bathrooms. Of course, it is people who
cause water to be used, not bathrooms per se. However, house occupants
may use more water if they are not constrained to sharing the same bathroom. Therefore it seemed that both the number of persons living in the
house and the number of bathrooms at their disposal are both important
variables.
Water Using Appliances. The following appliances were included
on the main questionnaire: laundry tub, wringer-type washing machine,
automatic washing machine, dishwasher, garbage disposal, evaporative
cooler, and refrigeration cooling which uses water. The pre-test form had
all of these items except the garbage disposal. It should be noted that
46
the heating capacity of the water heater affects the ability to make
concurrent use of some of these appliances and showers. However, it
was decided that a question about heating capacity would not be fruitful.
It would lengthen the questionnaire, lower the return ratio, and might not
produce accurate answers.
It should not be assumed that the change in water consumption
caused by the acquisition of an appliance is the same as the amount of
water consumed by the appliance itself. "Eating out" and owning an
automatic dishwasher are both signs of affluence, and they affect water
consumption in opposite directions.
Landscape Questions. These questions concern landscape style
(natural desert, gravel, lawn, or "other"), the lawn size if there is one,
whether it is watered in the summer and winter, whether there is a sprinkling system, and the number of trees and shrubs which get watered. It
also includes questions to determine if there is a permanent or portable
swimming pool and its size.
Rent. Some of the houses may be rented. The monthly rent is
more familiar to the renters than the value of the property, and they could
be more reasonably expected to answer this question than estimate the
value of the property. Along with this question, the respondent must be
asked whether the tenant or the landlord pays the bill, as this is an
important factor affecting water consumption.
47
Wording of Questions and Layout of Questionnaire
The way that the questions are worded and how they are arranged
on the page are very important, as they will affect the response ratio and
the quality of the answers. The assistance of Dr. George Learning of the
Division of Business and Economic Research was invaluable in the phrasing of the questions and the layout of the form.
It was decided that the questionnaire must be kept to a single page,
because response rate falls off sharply for forms which exceed one page.
To fit the Tucson City Water Department's standard billing envelope, the
width would have to be less than 7-1/2 inches. With a length of 13-1/2
inches, the sheet will fit the envelope when folded twice. It is also
important that the form appears attractive and uncrowded. These were the
physical constraints which the questionnaire had to meet.
It was decided to put the easiest questions first and to put the
money-oriented questions, which might be sensitive, last. The questions
were numbered because this gives a logical grouping to the questions and
it makes the number of questions look fewer than there really are. The
pre-test has nine numbered questions although it calls for 27 separate
responses, and the final form calls for 29 separate responses with eleven
numbered questions. For the sake of appearance, the form needs a title,
even though it doesn't supply information. The title chosen was: "University of Arizona Residential Water Use Survey."
48
Specific Discussion of Questions
At the top of the questionnaire (Appendix 1) is an instruction to
"check here" and return the questionnaire if the accompanying bill is not
for a single family residence. The purpose of this instruction is to identify and eliminate businesses and apartments which might receive the
form, but which do not belong in the survey.
Question 1. The expected response to this question is a number
which is to be entered in the space provided. Most people will interpret
the word "persons" to include boarders and guests as well as regular
members of the family, it is hoped.
Question 2. With this type of question, best results have been
obtained by letting the respondent specify whatever units he wishes to
use. It was expected that most respondents would give the length and
width in feet, with only a few choosing true acres or commercial acres.
Question 3. The expected response here is for the respondent to
fill in the blanks with numbers. It seemed necessary to define each of
these three terms (full bath, 2/3 bath, 1/2 bath) in order to get a
meaningful response. The popular term "half bath" doesn't have a universally understood definition, so if it were used on the questionnaire, it
would have to be defined also.
Question 4. The expected response was a check mark in the
appropriate boxes. The phrase "connected and in use" was added so that
49
respondents would not count appliances which were in storage or out of
service. The pre-test did not have the item "garbage disposal" or the
underlining of the phrase "using water."
Question 5. Respondents were expected to check one of the four
boxes, and if they checked the "other" box, they were expected to explain this in the space provided. Some people checked more than one
box, indicating that they had a combination of landscape styles.
Question 6. Like Question 2, this is a size estimate question,
and the units were left open so that the respondent could specify his preferred units. Three pairs of yes-no sub-questions were included in
Question 6. On the pre-test, the inquiries shown here as Questions 5 and
6 were combined under a single numbered question heading, and some of
the responses indicated that the combined question was confusing.
Question 7. The expected response was a number. This question
was not on the pre-test.
Question 8. Three questions were combined under one heading.
The respondent was free to specify his own units for the size estimation
part of this question.
Question 9. This is a control question whose purpose is to permit
the elimination of houses which might have an auxiliary water source.
Question 10. According to Dr. Learning, the particular wording
selected for this question brings better results than asking for the purchase
50
price, what the house is "worth," assessed valuation, etc. After
consideration of several ways of presenting this question, it was decided
to use multiple-choice categories whose end-points would be multiples
of $5,000.
Question 11. This question was provided for renters who might
not know what the house would sell for. It was presumed that respondents
in owner-occupied houses would leave the question blank. The purpose of
the second part of this question is to separate renters who pay for their
own water from those who have water included in the rent.
The Covering Letter
A facsimile of the letter which accompanied the questionnaire is
shown in Appendix 1. The purpose of this letter is to induce the respondent to return the questionnaire by explaining that there will be benefits
to him for doing so and that there is value to his efforts. The tone of the
letter should be courteous, inviting. The respondent was assured that
the information provided will be kept confidential.
CHAPTER 4
RESULTS
Assessing the results included coding processing, and evaluating
the questionnaire responses, getting the data into a form in which regression analyses could be run, and construction of models from the results
of the regression programs.
Responses to Questionnaires
Of the 2,496 questionnaires mailed out, some response was
received from 1,338 (54 percent return), which is an unusually high rate
of response for mailed surveys. The high response ratio in this survey
might be attributed to the fact that the questionnaire was included with
the monthly water bill, that water research projects arouse popular
interest in Tucson, or that the covering letter was persuasive. The response ration for the pre-test was about 50 percent. Because of this high
response ratio, no follow-up procedures with the non-responders was
deemed to be necessary.
About a dozen people included their water bill payments with the
questionnaires; two people used the return envelope to send in their payments without returning the questionnaire; and two people sent hostile
note. All of these incidents were much fewer than expected.
51
52
Only twelve responses were received from the customers with
Rate Code 3, which is the lowest response ratio of all the rate groups:
29 percent. The area covered by Rate Code 3 is the Town of South Tucson,
which is generally regarded as a low income area and inhabited predominantly by ethnic minorities. This suggests that the people in other
rate code areas who have the same ethnic, economic, and cultural background as the residents of South Tucson did not answer the questionnaire
either. The backgrounds of some of these minorities includes a language/
literacy barrier. The results of the survey are probably biased to the
extent that the culture groups which comprise South Tucson is underrepresented. The data from this survey cannot be used to make inferences
about the proportion of the population which have appliance x. It is valid,
however, to say that of the households which have appliance x, the mean
water consumption is y.
Eleven people made the effort to answer the questions on the form
and then cut off, erased, or blotted out their customer numbers. These
people apparently were fearful of a large agency building up a data bank
on them and yet also they felt that the survey was important enough to
them to answer the questions and then expend additional effort to deleting their identification.
Many respondents amplified their answers to some of the questions with comments and volunteered informatioi, and 42 of them signed
their names and/or supplied their addresses.
53
Table 1 summarizes the mailing, response, and coding of
questionnaires by Rate Code.
TABLE 1. SUMMARY QUESTIONNAIRES MAILED,
RETURNED, AND CODED
Coded, punched
and collated
Number Percent
Mailed
Number Percent
Returned
Number Percent
1
2,249
90.3
1,186
88.8
52.8
1,186
90.0
2
179
7.1
102
7.7
57.0
102
7.7
3
41
1.6
12
.9
29.2
12
.9
4
25
1.0
18
1.3
72.0
18
1.4
2
.0
18
1.3
Total
2,496
100.0
1,338
100.0
(1)
(2)
(3)
(4)
(5)
Rate
Code
Unknown
Percent
Returned
2
53.7
(6)
1,320
100.0
(7)
(8)
Retrieving and Processing Water Consumption
This section describes the process of retrieving the water consumption data and processing it.
Retrieval of Data from Water Consumption History File
Monthly water consumption by individual customer is stored on a
magnetic disk file by Tucson's Data Processing Department. A special
program was written in the Basic Assembler Language for the Department's
IBM 360 computer to search this file, select the customers of interest,
and punch their monthly consumption on cards for further processing.
Two cards per customer were required, with six months' consumption per
card.
54
Water Consumption
A program was written for the University's CDC 6400 to read the
monthly consumption history cards as punched by the IBM 360, compute
the annual consumptions by customer and punch and print this information.
One of the side effects of this program is that it revealed that the Water
Consumption History File showed zero consumption for all customers for
November and December. Thus, the period of observation is ten months:
May or June 1969 to October 1969 and January 1970 to April or May 1970.
Collation
An IBM collating machine in the University Registrar's Office was
used to "match-merge, no-match-select" the cards containing the data
from the responses to the questionnaires and the cards containing the
annual water consumption. The merged data from these pairs (matched)
cards were combined into a single card for each respondent by another
program which was written for the CDC 6400. This new deck of cards is
the input for the data selection and tabulation program and for Charles
Huszar's (undated) Stepwise Multiple Linear Regression program; and
they are the basic set of observations from which subsets were selected
for analysis.
Data Selection and Tabulation Program
This program was v,,Titton in FORTRAN and is designe d to be used
on the University's CDC 6400 computer. Its purpose is to read the full
55
basic set of observations and to select subsets from the basic set
according to specified criteria and then to print these observations in
tabular form. As an example, the selection criteria may be to select all
observations with annual water consumption greater than 70 hundred cubic
feet. Another example is to select all observations with more than two
bathrooms and a portable or permanent swimming pool. Almost any combination of criteria may be linked together with logical "and" and "or"
operators. As an option, this program also performs transformations on
some of the variables and recodes some of the variables for zero-one
analysis and punches these data on cards for later entry into a modified
version of R. L. Gum's (undated) multiple linear regression program.
Regression Analyses
Some initial regression analyses of the data were done with the
Stepwise Multiple Linear Regression Program written by Charles Huszar
(undated). This program follows the procedure described by Efroymson
(1960). The ability to perform certain elementary mathematical transformations in the input data was built into the front end of this program.
These transformations are called by punching a sequence of transformation commands on control cards. The structure of the transformation
commands is similar to the machine-language used on some of the early
three-address computers. These transformations were used to convert the
lot, lawn, and swimming pool dimensions into square feet.
56
The results from these analyses indicated that zero-one
transformations of the property value codes should be made and that the
answers to the yes-no questions should be recoded so that "no" and "noresponse" would be coded as zero and "yes" would remain coded as one.
The transformation vocabulary which was built into Huszar's program does
not contain the operators to perform these tasks. Therefore, it was necessary to modify the data selection and listing program to perform these and
other transformations and punch the transformed data on cards.
Since the data were now fully transformed and repunched, it was
decided to use a different regression program--the one written by R. L.
Gum (undated).
The basic set of observations is all the questionnaires which were
punched and collated with water consumption except those which were not
single family residences or which had private wells to supplement the
City supply.
Correlation Matrix Discussion
Refer to Table 2. Only correlation coefficients which are greater
than .29 are shown in this table. This indicates that the only variables
which are correlated to water consumption are (1) the existence of a
lawn, (2) the fact that the lawn is watered in the summer, (3) the number
of trees and large shrubs, and (4) property value, and (5) the diameter
of the pipe going through the meter. It should be noted that the water
57
z epoo en[/
Je;OUIPTOE edT cl
TUGJ
OrtIPA A.118d0Jd
iequmN
SOOJI, 30
Cf)
UOTI.P5TIJI JGUIUMS
1-1
1-1
CO
PO„TV UMPrj
CO
gz4
IIMPq
Csl •
CO
co CO
0")
CO
•q,
TOAPJO
1.JOSOC[
r4
0
r.L4
CO
TeinveN
(Y)
resodsTG ebpc[xes
JOI4S2MLISKI
5'141000
UOTI.PJO5TJje
(
e)
co
JOLISPM 105UTIM
(24
124
stp.pg
0
OUnd ; 0 JequinN
PalV
1.ori
Cs.)
1.1)
(Y)
58
z epoo onTpA
ial.eureTG odTd
CO
i.uoE
enteA Aq_ledoid
seal", 3 0 qumN
uoTve5Tni
JO MUMS
azy umpa
IIMPq
TGAP.,19
1.1GS
Ta11. -eN
iesodsm efrectree
iagsemlisTG
buTT000
u 0 Tw 1 a 5 TJJ 0 21
za uPAA.Ta6uTim
-
stpes
Und Jo legamN
spaN
CO
If)
*q.
r-1
CO CO U) CO
•
•
59
rate code, the number of people living in the house, the number of
bathrooms, and water using appliances are not significantly correlated
with water consumption. Outdoor features such as lot size, lawn size,
sprinkling systems are not correlated with water consumption. Having a
swimming pool doesn't seem to be important either. Of all the variables
which could have been in the table, only twenty-one have correlations
which are noteworthy (that is, greater than .29). In general, these
empirical results confirm common-sense hypotheses. The correlation
between lot area and rate code means that the lots outside the city limits
(where the water rate is higher) are larger than the lots inside the city
limits. Higher valued homes tend to have more full bathrooms, dishwashers, and garbage disposal machines. While none of these individual
features is correlated to increased water consumption, their aggregate,
as expressed by overall property value, is correlated to water consumption. Houses with dishwashers tend to have garbage disposal machines
also. The correlation between having a lawn and the practice of watering it in the summer is very high (.83) 1 while the correlation between
having a lawn and watering it in the winter is weaker (.32). There is a
positive correlation between the size of the lawn and the tendency to give
it water in the summer. No residences had both a wringer washer and an
automatic washer; hence a very high negative correlation would be expected. The reason that this did not happen is that many houses had no
washing machine at all.
60
Regression Equation
The form of the regression equation is
Consumption 7- b o + b i x i + b 2 x 2 +
-
+ b 38 x 38 + Error
(31)
The descriptive names of the x terms and the values of the coefficients
bi are shown in Table 3, together with the correlation coefficients,
standard error, t-test, and other statistics which apply to each variable.
Since the correlation coefficient for the rate code and consumption
was only -.18, while the regression equation coefficient was -.46.3, it
was decided to separate the observations by rate code and run the regression program on each of these subsets of data. Table 4 shows all of the
subsets of data for which the regression program was run; this will be
discussed in detail later. The next section discusses the rate structure
in detail.
Water
Rate Structure
Table 5 shows the water rates charged by the City of Tucson
Department of Water and Sewers to the customers covered by the survey.
These rates consist of a fixed monthly minimum charge which includes
a minimum amount of "free" (that is, flat rate) water plus a variable
charge which is a function of the volume of water which is consumed
above this minimum. Four different rates are charged and these rates
are determined by the geographic area of the customer: (1) inside Tucson,
(2) outside Tucson and South Tucson, (3) inside South Tucson, and
61
CY)
CV
CV
r-I Cs]
r--I t_() r--I
r-I r-I
a)
r--I
r-I
0)
CD
CD
(ll
cc
(D
0 CD 0 0 0 0 0 0 0 0 0 0 0 0 0 0 CD 0 0 0 (0 0 0 0
0
co
s-,
•
a:5
0
•
co co .--1 co
co -11 NI co
•
•
•
•
o
r--1
71 .1
co (NI di
0
co
di' co 0 o co Lo 0 0 .--1
NI cri CO d -i up d-, .1-i ,,r, <zii
•
•
•
.
•
•
•
•
•
Lo
co
.çr, c) •qt d-I .1. 4 c5) c.1 c-,3
•
•
•
•
(9
r-I
•
CO
CV
cq
(--) o
•
CD
5
0
'
SO
(CI 0 .1-1
It)
4-,-1--.
U)
•
LO0")
c, co ,---1 -,3- 1.0 .zzt Lo c-q (9 N. LO NI C0 in r--. 0 0 In 0) 0 0
,
,
.
(0 00 NJ 0 Hi 0 N. Hi NI N. NI NI NJ N. ,---1 CV CD C --- NI Na 0) 0 0
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
.
•
•
•
•
•
•
•
r--I
CS)
CO
LO
N.
r-t
(0
(i)
X
1--I
(I)
(v)
0.1 cc r--I
Q)LU 0.1 r-I
•
00
o 0
rci
rrjr-4
0
NI
r-I
•
•
•
cs3
co co
0 00
•
•
•
•
•
cc N- CO C.1 CO 1-1 NI 0") (D (D N. NI CY) CO 0
CO N. (.0 CO N3 Hi •Çl •-zri cz) Ln CO 0 Cr) LU c0
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
CO I
r -I
r--1
I CV Lo NI
I r-I
00 LC) 00
-,
•
0 csz 0 0) 0 0) 0 (.0 0 0 0 '<r i 0 0 0 0) 'sr Hi 0 0
0 0 (9 (9 Cr)
•
•
0
•
•
cc 1_0 0 c-. )
• • •
•
•
•
CO 0) 0 in in (9 c0 (.0 •<1,
•
•
•
•
•
•
•
0
•
00 N. 00
CDCOLOO)N.00000N.0)r-lr-lLOCOCO
r-i
co
(0
r--I
u) D.)
r-I
r-1
0-1
0=1
E-1
,
Cr) CO N. r-1 LO CD CO CO '71"' LO CV NI CS) CO
CO 1-1
r-I CO 0 0 CV 00 CV NI NI
CV C) 0 0 D r-I D 1---1 C0 Cr
-1 CO LO
'1
0
•
•
•••••••••••••••••••••
•
4-,
CD
--1
0
-t-1
14-1
LI--1
0
..Q
co CD 0 0 1-4 01 C0
NI CD 00 r-I N. Ns LO 'q' •çt, ,q, cp
0 0 0 o
•
•
.
•
•
•
•
•
•
•
.
C:1
CD
CD OD CD 0 CDCD CD (\IC° 0 CD
co N. co cc co NI cp
•
.
•
•
rs,
I (NI 1-1
0r-I
r-1
r-1
En
.__
u) u)
._. ..
,4-1 LH LH
L1-1
o
4-1
o
o
5
,..., o
o o o
ai
X
0
0 a)
a) _b
,c1 rc45 .-Q n .-0. . (-) (g
r-4,:-.]
cc;
-16 -1
- - `4
i-1 `4
r,
"j
-r-,
i--i
cd
•
•
•
71
0
--I
>
It >
t) (,) -4-i
ti)ai
-H
3
-r-I
r?1) - r-1
..r
"
0
a)
-1-,
,-Q
°2 E
ci c0 s_, co lii 0
,(C C n W 0
cr) 0 ô
•-• ° 0
rc5
.-1
-cs
--t
0 u)
0
u)
° 71, --,®
Cs
rci 2
:5) '--- 0 (CS qp...,
U)
•
TS
› ;E3
-C
-1 )
-r--t D U) -I--'
rd S-1 S.-r S. ›, " 4:2
al
U) ._.
N.. 0) co NI co
r
s
•
LO N NI r-1 LO
N. r--1
CO I
a)
Q.
-1
-1-1
•
I
$_ °
0 .--]
0
ai tri ai
0 4.::.
ai
o0
.-C1 ,-Q ,-Q
--, ,.
7
:11
04 u)
T: IL
a)
u)
0 0 t
0
-.1 co co
(.0
,_
aD
0 0_, a)
t-i)
0
\ \
u)
4-, c-.1 1---1 ,-Q ro
s-_,
Q.
-H 0 a) U) (i)
4-1
•
I-I(Y)
I CO i 1-1 I
I
0
•
•
Lo .-i r--I CO CO LO N. CV NI 0) `<r CO N. CV
E.
Sa,
0) L" -I--'
0
0
.--1 4-4
.--
- -I
2 (CS
-
-'-'4
-0
Cf) '4 a_,
0 r---1 0 0 F-1 C/2
(
▪
62
r-1 '11r-1
LO
r-1
r-1
wsr
r-1
0 0 CI 0 CD 0 0
cv
• •
co
(Y)
Lo
•
co
•
•
CO t's,LO
•
•sr
•
LO
•
•
r-i
co tr)
cV co
.sr, •q1 co cv cv
• . • • • • • •
Lo
cN
CD
•
•
CO
Lo
j1
cr.>
r -1 0 0 0 0
•
•
•
•
•
•
•
,
0.) 0 0 CN3 (.0 0-) („0 r-1 CO CO CC) N. LC)
r--1
C) OD C)
CV CO
•
ar
•
•
•
•
•
•
•
•
•
•
CO
CD
CN3
(Y)
1-1 I
I
I
r--1
CS) LO t•-• 0.) d'
1-1
C) 0 OD LO
•
•
is)
co
•
•
(NI
.
•
•
Cs.]
•
•
co cs) co Lo
r-1 CO r-1
•
•
•
•
0') CO CO C`,3
LO
LC) LO
•
'n:14
•
•
•
•
•
0 CNI CO LO CO
C ) Lo co c_o
•
•
•
•
•
CO 0) CO CO rn
Cq CO CV
r-1
Lo
C)
di
cy)
r-1
co
•
L.0 CV
r-1 0 0 r-1 Cs]
CV CV Cy)
•
•
•
•
•
•
0
h, LO cip n--1 LO C.] 1-1
N/ LO OE) 0")
LO
•
•
•
•
•
•
•
I cr) co N. CO CD N. I-0 r-1 (-0 CO (X)
1-1
C
n
NJ CO
1r4
at
=1
I
I
Q)
Q)
._. (1.) c
a) Ti --i
>, 0 _r_cii ,—i cv co ,71-, up
-1-,
(1:5
a)
5-4
RI
(D L', CO
0 0 0 0 0 0 0 0
.-81 -d
i Q-n '--'
+
O
(I) - ,-- 1 (-0 ---1 rn.1
,---i
r-- I n-1 .----1 r--- -1 r---I r
—I
(7,j c:i c,.n 0 (G (,ï ccJ
n. r-L1 al 1:4 0-n > .--- > > > > >>
--
-
63
TABLE 4
SUBSETS OF DATA FOR WHICH REGRESSIONS WERE RUN
Rate 1
Data Subset
All Rates
Inside City
Rate 2
Outside City
Mo Cons ?, 0
Basic
Ann Cons
0
Mo Cons
0
Subset 1
1218
1092
98
1092
1015
63
1086
1006
72
1032
967
57
Ann Cons>0
Mo Cons
1
Subset 2
Ann Cons ?'-10
Mo Cons
Subset 3
Ann Cons>70
Lot Size
1
625
Subset 3a
Lawn size'l
-
- o
-1
64
ta
e-1
+..)
O
0 0 0 0 Ln Li"
u-) LO LI) C)
LO
CO
0
ai
0 0
(0 S:) s...,
(2 cf)
• • • •
L.-. co o) c.)
0
0
•
U) 1-0 1-0
1-0 OP LO
•
•
•
71-1'Sr
4
.
.
o o 0 o o 0
UP UP Ln 0
'1-1
• • • •
r•-• co cr) cq
0 10 U, U)
CD LI) OP 1-0CO
•
•
•
•
•
Lo
.q‘
CS)
CV
if)
•q,
DDDD
0 0 CD 0
CV
CV
)
CSD 1-4 CO
0 (9
•
•
•
•
•
•
• • • •
UP LO CO CD
u" 1-0 CD CD
'Sr
g:4
CD 1-0 0 0
1 CO CO
•
•
CO
D
•
CV
•
•
CO CO CO '71.1
0
a
0 CD 0 UP CO
'711 CO CO
•
•
•
•
CO CO 'çri
+-,
LH
-- --I
Q),-H
•
a a a a
›.
.
.,
-- ri-,
-H
a a a a
• o .
r* OP 0 CV
(73
,
n---1 r-I
0
OC!)
a)
o a ,
a a a a a a
D
a .,a a a a
,--i
iD •
r", C"-- N. L'n
,--,
n
--t -1-1
0
›-: ‹ 0
o
_Q)
Q)
• .-o
41
0
° o
..-4 ,_Q--,
'ci
.
Q)
-1-,LO
a)
.
• o
a
--1
D .7
N•• 0
•
rd
.7
7-
0
1-0 0
• • •
,-, ,--1c.)
0 ti--1
›
1-1
,-,
.--,
0
,-, 0
,-,-4
(ts a
7
7
7
UP CD - 0
C--- 0 -Lo CD
•
•
•
1-1
1-1 CV
a
40
_. co
;-)
0
aj
o ,--t
.--.
0
0
o ,-0
a
-,-,
u) ,--, ,.4o
cq
0.,-,
-1-4
a
a
.--C
CO 0_.,
.-,
co Hi ri CV
o
u) —
W 4-1
›
0
,Q 0
(d
.p
D-H
a
LI-1 c)
Cr) 4-,
n-• 4-4
0 >-4
0 rd -I
.--,
c) .-
0
0
o cp
c..2,
u) >,
CO
65
(4) the special remote areas known as Skyline Belaire, Rudasil, North
Northridge Estates, and Coronado Foothills. Within each of these geographic areas, there is a further subdivision of the fixed portion of the
rate which is determined by the pipe diameter at the meter.
On February 1, 1970 a rate adjustment was made. This adjustment consisted of (1) lowering the flat rate minimum allowances to 700
cubic feet for all pipe diameters and (2) a change in the usage rate for
volumes of water above the 700 cubic foot minimum in the amounts of
$.05, $.07, $.07, and $-.15 per hundred cubic feet, respectively, for
Rate Areas 1 through 4. Customers who use less than 700 cubic feet per
month are not affected by these changes; these are functioning flat rate
customers. Since one full year of observation for each customer before
and after the rate change is not available, there is no opportunity to study
the effect of a rate change on consumption. With the limited consumption
data which is available, the effect of the price change is masked by
seasonal fluctuation in water consumption.
It is significant to note, perhaps, that the rate for water is higher
in South Tucson, which is an area generally regarded as low income, than
in Tucson, which has a broad spectrum of incomes.
The rates described here do not include city, state, and utility
sales taxes, which total seven percent.
66
Rate 1 Subset of the Basic Observation Set
This subset consists of all the observations in the basic set which
are within the City of Tucson. There are 1092 observations in this subset.
This is the lowest rate charged by the Water Department.
Correlation Matrix
Refer to Table 6. Only correlations which are greater than .29 are
shown in this table for clarity. Two of these variables have correlation
coefficients which are just barely .29. Only four correlation' s with water
consumption meet the criterium for appearing in this table: these are
( 1 ) the fact that the lawn is reported to be watered, (2) the number of
trees reported, (3) the property value, and (4) the pipe diameter. The
correlation between water consumption and the existence of a lawn
(whether watered or not) was .30 for the basic set of observations (Table
2), but this has dropped to .24 with the Rate 1 subset, and hence it is
not shown in Table 6. Also, the lawn size and winter irrigation variables,
which were present when all rates were considered together (Table 2),
are not significant from Rate 1 customers, and they are missing from Table
6. Since all the rate codes in this run are "Rate 1" this is no longer a
"variable" and thus no longer part of the correlation matrix. The remainder
of the correlations are similar to those previously discussed.
67
9
(
enTuA
Ailedoid
C.1
CO
seau jo JequinN
E21.
uonebTni Ieunung
CI)
uarv umpq
E7)
1
umpq
o4
0
TOA PIS
LO
al X
SOCI TP.IMPN
(D
iaqs anal s Tc{
•q'
0
61111009
LIOTI.P.105T1jad
(24
0
C.)
101.1 S PM JObLITIM
st2S
Trnd 30 iequmN
c\1
CO
0)
Cs1
68
CO
z apoo anteA
Cs.)
CO
CO
•
•
dr
CO
CS)
lel.GuIPTC1 Gd Id
wad
nTpA
Apadoid
saau 30 JociumN
uo p.p5T.ui [email protected] TIMMS
-
28.1V UM2a
LIMPrl
TGAP15
S OG TPI1I12N
101.1S emtis
-
ru
buTT000
u0p.p..To5Tijeu
Jetisem Jearrim
smps
iTna Jo siaquInN
N. LC) 0
CO CO
(Y)
CO
•
CD
CO
69
Regression Equation
The form of the regression equation is the same as previously stated
(Equation 30), except that there is one less term because the Rate Code is
no longer a variable:
Consumption = b o + B 1x 1 + b 2 x 2 + . • • + b 37 x 37 + Error
(32)
where consumption is measured in hundreds of cubic feet ( c f) and the
names of the x terms and the values of the coefficients b. are shown in
.
Table 7. The values of other statistics are shown in this table also.
Rate 2 Subset of the Basic Observation Set
This subset contains all of the observations outside the city
limits of Tucson or South Tucson, with the exception of the special remote
subdivisions. This is the third highest rate charged by the Water Department. There are 98 observations in this subset.
Correlation Matrix
The correlation matrix is shown in Table 8; for sake of clarity,
only those correlation coefficients which are greater than .30 in absolute
value are shown.
This table differs from the two previous tables in that there are
many more high correlations. Water consumption has high correlations to
14 variables: (1) the number of people living in the residence, (2) the
number of full baths, (3) possession of an automatic clothes washer,
•
▪
70
i
.
-I-I
X
O
."n:ti (Z) h.. CO CV
NI r-i NI
r--I CD
(0
0
0
NJ ri LO r-1 r-I r-1 r-I r-i r-1 CD r-1 r-1 r-I OD 1-1
0
OD
0
cO
co
rri
(24
1
-.-4
s-,
CCS
rrj
1
'-.
>
__,
3-,
N __, •,--1
0
0 0 CD CD 0 CD 0 0
0 0 0 0) CD CD CD 0 0 0 p CD 0 0 0
0 0 •,:t1 CO 0 CO 00 CO
L`... 0 LO d-' CNI CO ,---I di
0.1 •sr 0 CO 0-) C \I ,--I N. 0 0 n-n UD 1-1
• •
,.._, co
•
•
•
•
•
co --1
co 71-' up dt. co d -w co d-n co d- d- d di
•
•
•
•
•
,
•
•
•
•
N3
co
CO
0 - --1
CO Q to'
CV
rc3
(1)
•
,
•
,
•
•
•
cD
-1-1OD
ni
,
•
r-1
cs.3
co
,---1
0 N. .--1 ‘Zr Lo co N.
r-1 0 CO CV CD r-1 CD N.
•
•
•
•
•
•
•
.
CO C) r-i
I
-0
N
.
,--t co CO 1.0 CS) (''D 0 .) N. r-1 CD OD r-1 1--I CV V
I-1 CV N... CV r-i CV N. r-i CV CV N. CO CV CO CD
•
•
•
m
•
•
•
•
•
•
0
•
•
a
•
•
OD
N.
CD
N.
ri
r-1
LID CV CO
N.
CI) ( D CO CV 1.1")
El
I
cp co eH c3,dr
•
•
•
•
•
•
CV LID LO CD co N3
CO ‘1"' Lo N. .31-, N3 N.
r-1 CO CD CD CO I-0 CO CO CO 0) I-0 0 I-0 CO 0-)
•
•
•
0
•
•
0
•
•
•
•
eH
CV I COeH
CO r--I r-iI1-1
•
Lo
•
•
•
H
.-0
L-.
(0 COD cc)•zr
LO
crD
, •st' If) uD c0 CO c-.1 0 CO
, zr N... 0 cD (N
co 0-) CO
C 0 CO 0
CO CO
NID Lo c.] N. N. CV LID
.-,
•
•
•
•
•
•
•
•
•
•
•
•
•
•
0
0 -.1 • • • • • 4. • •
(C3
‘7I4 cc) 0")
0,1 0-) CO 0-) 0,3 C.1 t•-. 0-)
N. OD N. Cr) CD OD
-I--i 5-1 CV
r-1
U) Fi.:1
r-1
- 1-1
r-i
CV
r-4
cp i-i
CO CO CV th CD CD
CV 0 0 CV CO CV c‘i
eH r-4 CV 0 0 CD 0 1-4D r-1 CD CD CD
•
•
•
•
•
•
•
•
•
•
•
•
•
•
• • 0 • • •
•
•
r-1
CO CO CO CD CV LO r-1
•
4-,
0
..--1
O
OD CO 'q N. LO CV 1-0 C''') CV CD N. 0 ', 7' 0 C.0
LID 0 <zr N. CV CD 00 CO
0 (....„ co N... up [...., •q1ri Cr) •s:r. r-1 CNI LO CD CO CO 0 Lc) (..c) co CD CO
471 '4 OD
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
14-1
•
•
CV CO LO cnI gq'
CV CV
CD 'n:Ii LO .-4 CO 'q'
0
r-1 I 1-1 (X) CO
CO
CD
C
O
1
-1
r-1
i
*Zt
.
i
r-1
-I
CO
C
')
C
I CV r-I
r-i O
I
[I)
0
Q)
th co
Q)ri
÷-,
rd fci
O
W
Q_,
a
4.4 v-i ‘4-1
Li-i
0 0 0
__,
4_,
, -,-I (Ci
e,
ça,
c)
a)- .-r-d
Q) ..i
--. ,..
u)
-Q -Q -Q
W f.n0
ri
...._. CO CO
.--
(Ci
,,..... ,...,
U)
,-, r---, .-Q cci
0
.75.1
-.-1
I
0
-8
ef)in
"0
i-,
•
0
0 Q_,
6). 0 CO u)
,
r-
. „4› n
0
n
(0
F_I
7:5
E
o „ a) ui
---4
1.." .1-.
4-, •,..4 U) W 7.,4
O °
4-,
›
t5) ai
C/D
E---I
CI,
t
i)
.,_,
4-1 4-a,
-i) 0
ait1--t
W 4-,o
0o
- .-, _..
ro ▪ s_. s. >.., S-I 4--,$-n S-n , ro a.) (ri .-7!
s--. r.5)
0 a:I 0 0
Q)
0
,_s
0 0 0 ....
---1
0 a
0
(Ci
--1
al
S--1
W
0
t3)
4-'
,C) ,"- ,Q ,C) ,_Q . (-) 0)
.4-! -11 'CI
> 5, ni rC1
--, ro .._ (1, ,p
-,
f.-,
C'n (°hFC1 (CI r C10 -I-,
- r-I
0
.--_ , 111 mi
H Q-I
____,
6)
75 0 :/ . -5 .-) rill
Qr::10'7,U,-:100,-qu),>u).Z, a4
,... 12G
F
Z 1.4 4
,_, Z ,._1
O
)
-
71
0 0
r--I
LO
N. 70
(CI
1_4
'41
E
E
D
r
n-•
s. (.0
NI $-1 CD CV
•
•
•
•
CV 'SP
L.(-)
n
Q)
N. 0) up 714 (.0 0) 0) CV CO
CO CV
$-1 1-i
CV NI 71'
•
•
•
•
•
•
•
•
•
LE)
4J
r-" )
71" CO CV N.
ai 0 CO ..J4 C)
•
•
•
•
0
LO I!)
-1--,
U) CO S. CO
O
H
I
0)
Cq 'CV CO CO
•
•
•
CO 0) 0) CD Lo o.) 71 1 CV CV
0 0 CV CV /-1 CD CD c) cp
•
a
•
•
•
•
•
•
•
r-- 4
Cq 7r1 (.0 N. CO ( 0 If)
N. 0) C.I I-0 ri ',I' CO CO CV
•
•
i CO 1- 40)
•
•
•
•
•
•
I r4 I rl
•
•
CO CO
CV CY) CO CO LO D M D 0)
D r
0. ) 0 1-1 $-4$--1 CO LO 0) CO 0) NI CV CO
•
•
•
•
•
•
•
•
•
•
•
•
0 CO CO 0 r-4 CO LO
r-i • r-1 1--1 CV CV NI CO CO
r-
•
CO 4-0
0 0
•
•
•
00
CO r-4 $-1 0
•
•
•
•
I
I
Cq 0) 00
CY) CO N. (.9
C`') CO 4-0
0 0
•
•
•
•
Q.)
o
•
.-r-1 0 0
•
•
•
•
a
)
Q.)
I (0
co
•
Lo
•
ri
CO
0
o
co co 0
•
•
cp
ri
•
<0 0
.
•
•
•
(::5-)
ri
CD
0 1-1 " CO 'g-' I-0 CO N. CO
Z
i
X
+-,
0 0 0 0 0 0 0 0 Q)
0
-0 -0 -0 -0 - -0 -0 -0 -0
$_, o
a, (d
a)-0,0 00000000
a) o
›-,0--,00000000
al (..) Q
',-EZ
a) o a) ou a) a) a) o
a
0
.:i
-.--J
.
o
o
IT!, '-â : II; 4-i
r7C 5
(;
r (il 7d ' --'
H
cs
'
7d
7
I
3
:
1
F1
7
---i
rcI
.-i
'
0 0 0
a., a, a 12ci Ix a > > > > > > > >
4--'
,
72
1.d)
= onFA
E
= anieA
c.)
ro
odeuspuei
uonvumwoo
ialaulEqa adId
0. 3
•cr
.50
anieA Apadoid
CO
CO
0
0
V.
CC
".
•re
o
CO
V
•
V.
1
seau Jo laquinN
UlOgSAS BUTTNLITAS
V.
co
•
uone5p_ii latuums
CO
tn
CO
E-i
CO
paJv
•cr
•
n3
UMP1
co
0
P-1
<1
F-1
,
cc'
cc
•
0.-4
.
•
v•
•
•
vesou icumeN
iesods
<4
co
a
0.1
a50qle.9
•llT
•-•
CO
HQ
LO
latoop anneiodnA2
I
•
I
•
latismqsyj
CO
•
laqsem opptuolny
LO
CO
•
co
to
O
•
CO
C•1
I
cr
to
CO
•
•
•
o
CO
0
qnj, A_rpunvi
t2
0
0
sqln'a
tn
CO
LO
•
CO
0
jnj Jo 20qwnN
UO
CO
:3
CO
'D
LO
CO
ealy 10-1
CO
aidood jo laquinN
, r.
,
ro
3
a)
a
c
„,,
...
0
o c
-..0
..co
,-.;-,,, 0. t01/7
':
,
-,S),
-
.-&'
Cll
-3'
"L'';'
''''''''
;2.
'C''n
9
0. L',. ,
.. 0,
1'
V,
(4
n3
'
0,
....
' --.
7....
C
•.,," 2 ,F,,
'..', ,,
c.- c.)
1--1
E
to
a)
..-+
u3
0,
..-n
L ' 1
E.
8
.
'a'Cl ,0
,O
(5
>
'ra
,I.,
; ;>.`', K *4=-a. -,'a.
a.
t:
-4
,...
-__'
2' ‘.6
--.49
C
r'.:2
' -.'
.--.1
CO
-.5. ''q
c
0
E-4Cl
'''
Cl
Cf)
c.)
v
>.
,0 —,C),
t.,
•••.,
4.,11
C
,2
E —
O
(C
0
CO
0
.a...
C
011 ',. ,
=.,L.L., .A..r. Ug Tl> --->:
;E'.
'D
7,1.0
., '0 '01
--,1 ,..,
> > > ..,
73
(4) possession of a dishwasher, (5) non-possession of an evaporative
cooler, (6) ownership of a garbage disposal machine, (7) having a lawn,
(8) reporting the lawn area, (9) having a sprinkling system, (10) watering the lawn in the summer, (11) the number of trees, (12) property value,
(13) pipe diameter, and (14) having a house valued at more than $40,000.
Of particular note is the negative correlation between water consumption
and having an evaporative cooler. This can be explained by the negative
correlations between having an evaporative cooler and property value
(-.27 ) and having a garbage disposal (-.28). This means that people
who tend to have higher valued homes and/or garbage disposal machines
tend not to have evaporative coolers. Thus, even though the evaporative
cooler does consume water, houses with them consume less water than
houses without them for other reasons--particularly, they do not have
garbage disposals and they are generally lower valued than houses which
lack evaporative coolers. There is a correlation between the number of
people in the house and having a lawn and watering it in the summer, as
well as water consumption. Large lots are associated with natural desert
style landscaping and property value; and large lot sizes, per se, do not
affect water consumption. In addition to being correlated to water consumption, the number of full bathrooms is positively correlated with having
a dishwasher, garbage disposal, sprinkling system, a large pool area,
a high property value, a larger pipe diameter, and a landscape style which
74
combines a lawn with natural desert or graveled areas. Houses which
have laundry tubs tend also to have wringer-style washing machines and
larger permanent swimming pools. Neither of these features seems to
affect water consumption; it should be noted that very few homes had
either of these features. People with wringer-type washing machines don't
have the automatic type, obviously, but they do tend to have permanent
swimming pools. Houses with automatic dishwashers tend not to have
evaporative coolers, but they do tend to have garbage disposal machines,
lawns, larger lawn areas, sprinkling systems, several trees and shrubs,
a high property value, a landscape style which combines a lawn with
desert style or gravel. A garbage disposal unit accompanies high property
value and high water consumption. People with natural desert landscaping
tend to combine it with a lawn, but they tend not to water the lawn in the
summer. There is a correlation between having a lawn, answering the
lawn size question, watering the lawn in summer and in the winter, having
a sprinkling system, and combined lawn and desert or gravel landscape
style. The number of trees is correlated with property value as well as
water consumption. Property value is also correlated with the size of the
swimming pool.
Regression Equation
Again, the form of the regression equation is similar to Equations
31 and 32, except that there is one less term. than Equation 32:
75
Consumption bo + bp(' + b 2 x2 +
+ b 36 x36 + E
(33)
where consumption is measured in hundreds of cubic feet (ccf) and the
names of the x terms and the values of the coefficients bi are shown in
Table 9. The values of other statistics are shown in this table also. The
reason that there is one fewer independent variables in this equation than
for the one for Rate Code 1 is that there were no observations for houses
between the values of $35,000 and $40,000 in the area of Rate Code 2
(outside the city limits).
Deciding Which Observations to Include
The first three regression runs which were just discussed used the
basic observation set. This includes all those questionnaires which were
coded and matched with water consumption history and which had no private well and which were single family residences, duplexes or small
apartment groups (5 or fewer units ). There is a question as to whether
all of these questionnaires should be included in the observation set.
Many of these customers use less than the monthly minimum of 700 cubic
feet per month. These customers are paying effectively a flat rate and do
not have an economic incentive to conserve water any more than they are
presently doing because it will not reduce their bills. But then, such
consumption may be low because those customers are fearful of going
over the minimum. Therefore, it seemed useful to select from the basic
set only those customers whose consumption was more than 7,000 cubic
feet for the ten-month period of observation, and re-run the regression
• -
76
CO C) LO C
CD
r--I r--I
(N] r4 1-41-14
r---1
cc
(D t--I r4 r.-4
or)
CD
CD)
CT)
Cs]
QQQQQQ
QQ QQQQQQQQQQQQQQQ
.
7:23
Q
N-- ,---1 0 CV 0 00 0 ,711 v) 0 o ce) N.) cr cr
CO 'qt LO •q( v) •=z1" LO si, 00 (.0 v) 00 00 L--- r--1
cz) L.0 (0 0 •qi (0 CO
LO CD 1--- 'sr 0,3 CO CV 'q'
•
•
•
•
•
•
•
(CI
•
•
•
•
•
•
•
•
•
•
-LI
7i
4
-1
C
O
›
0
ai (1) 4 4-4
4-34-1
CO
CD
CV
CO
N.
(NI
Lo
<zr,
co, o .t, ,Kr co N. co
Q rts
UD
•
N-
•
•
•
•
4-,
Cf)
2)
H
H
•
•
o
s--,
0
LO
N
.•
-1-J
Nil
•
4-1
•
•
•
o
•
•
•
•
•
•
•
•
•
•
0
.--1
CD
c-1
1-1
•
•
•
•
7:3
S-4
ai
•
'sr'
Lc) (-NI ,sr, 00 1-0 N- 'V 0 )) CV n-n ( 0 c‘.1
.---1 (NI h. CV ',1 CV '14 CV n--- 4 CC) '":1' I---I r-i CV CD
co •, 1-1 co ,-i cr) 0 (.0 c...) co (0 LC) r- 4 LO CO C)
N.- 0 cY) C-- 0,1 0,3 N- 0,) CO 0 0 0) 1-0 Cq ,---I
'Sr 0 LO CC) r--I CO (0 (AD
N- d' CO 0) 0 'I'l CO 00
1
•
(0 cp
•
CO
1
•
•
•
•
r-I r-I CV
•
•
•
I
I
•
•
I ,-I
•
•
•
•
•
•
•
•
r-I
I CV I-1 r-1 I
I
3-4
Sri
(D N. 4-4 CD r-I
s.. c0 0,1 0 d"
•
•
•
0
•
,--1 r--4 C'-] CO 0
CO CO CV c‘i 00
cn,1 0 4-1 Sr CV 0 CV '71''
1-a) 14.0 140 N-- 0
o
•
•
•
•
CC') d-' CO
.
•
•
N.
•
LO
CV CV 'qi CO , 1' C.,)
(f) LO N. 0-) C) c0 .-..i 11) in 1--I
( 0 N.- 0) CNI 0 0,1 Cs) LO N- cs.3
•
.
•
•
• • • • • •
N.- 0 NI ,11• CO cp LID
) ce) 00
c.) 1-0 CV ,71-,v
r---1
CO
C/) fil
Lo 0,4
,
,
•
0
I-I
•
co
•
CO LSD r---1
C)
•
0 0 00 C‘1 0 ,---1 0 (.0
("C)
a)
.--1
r---1
•
•
•
•
LO 00
CO 00 10 r LO 0 (.0 c0
CO CO r---1 r--1
r-1
C) CV v) LO
cr) cr) Q
z:11
•
•
•
•
•
•
•
•
•
•
•
•
•
.
•
14 0)
cr)
•
•
•
c..)
•
0
-e---I
4 1
cr) 0\1 cs1 N. 0) LO (D (0 LO .---I csa f--- LO .-i c)
Lo
(NI ,---1 0 00 CO n--4 'Kl" ,---1 CD 0 cN ,--1 up
CO CD 't' ,--4 CO LO •=zr c0
O
Q
-
W
O
cNi 0 N. n-t (-0 (.0 0) .---1
•
•
•
•
•
•
0
•
•
•
•
•
t:»
C)
71:1
ra
U)
0 X
a)
HZZ
s--. s-, s-,
... (Ci
0) (D (D (D
a) ,,,
,1:1
t
1
_4 Z, Z
I ;
•
•
0
0)
CO
I
4-4 -)-' $-r
S--1 a.) cc5 @
V SD., 0
CO
"
(1)
-I--,
_,
(1:3
CO
'CS
a)
k..)
(0
--,(1) —1
-.-1
-.t
:::, co .-:-: :_s o
,5 Kt; Øf,
-I-I
0
s•-n
co
(I.)
-r--1
$, --,
a) ,-- r'Q
w (0
E
-1
E
.
„
;,:4 ç_2 0
---, > co (D `,1-1 .4_, 0 ro
t:Y) ari s o ro .., 7)1
0
Ti 4-, 4-4
-1-1 0 -I-,(I) .---1
-I-J - ri (0
0
0
(Cf 4'
(0 >, S--I 0
t]] D CO E--1 Ci,
-r-1
r--I
Cf) -1-'
(0
,
0 )-, 0 cn
CO
Ti
s-, a) ---, to
a) > q Q
0
r- -4-1
- o
--,
...c1 _Q ,Q -cs cy)
PI
•
a.
, 0 t
,0
(0
0 t,'
4 4-n 1U
0 0 0 EPI
481
•
LO CO CS)
Lo c.) 1-1
TS
0
'61 En
(
$-n .__[ f)
,(1) CO D
4-1 4-1 4-1
-r i
W
CD —I
a:s
" —1 ,-Q ,_
>
ro
rci
(1) fn. rO
,--I
r-I CC CC
''. \ . '''' \
0
0
ai
-1-1
--i
4-1
(-0 ai
•
(i)
Cf) U) U)
a)
•
I
I
I
o
•
•
CO ,..--I CO 0) LO CO 00 v) d-'
C')I
I r-4 I CO CO I os)
CO .KI' ‘qi LO ,zr, 0)
...--1 cs) I to co LO
I-0
ri)
al t-W
(0 s-, .--,
›
ai
" '-Q4-,
,a
'
.
n [ILI (L) ,'.-in (D 1--4 0 0 1-i Cf)
:
4
CI) -ZI la.
▪ ▪
77
'7ri 0 0
'7r1
S-1
5,
•q+ Lf) •qi
`714 rl
• • •
LI) ri CO
`711
CN
•
.1
o
•
•
•
(C) `n34 CV CV
CV CV CV CV
•
•
•
•
r-e- CV
CV
cr) (:)
LI)
•
co co
CV
•
(cf
a)
r D
o
•
• •
0-) co
•
"<14 LO N. CD L-0
o CV CV CV 0 0 0 0
•
•
•
•
•
•
•
Co
(A
CS) LO
0
ri
H
1
LO
I
H
0
‘1' CV CS)
Lo .1, co o) (.0 co N cr) co
•
•
•
•
•
•
•
9
•
•
•
NI r-i r-1 r..-I I 1
I
1
N. co
Lo
LO
NI ri ri ri ri
0 0 C) 0 0 0 0 0 0 0 0
7:5
Cri
r-1
CV
•
•
•
•
•
•
•
CO
0
szzt' Lo
•
0
•
•
•
co
•
ri
V,-
LO
`q' CV CV CS)
•
Lto (NI cs) opoo LID
•
•
CS) C)
CV CO CO
Dr-H
co
CO CS) CO OD CV 0 0
'n,1' 0 CO CS)
•11
`71i
0) 1-0 CO sq.
0 CV 0 0
•
•
•
1
co
oo
(0
CO
CO
•
I
N.
DL"---
•
•
•
CO
d CO
•
•
•
•
•
L-r)
CO 0 I-0
'7" (.9 C) N.
• • •
'- D
-.) co 1.0
•
1-0
•
•
Co
a)
•
LO
U")
t--n
0 ri CV CO 'Sr i LO LO CO
-1-'
0 0 0 0 0 0 0 0
..Z.E)
0
CLI
TS 7:3 0 Ti Ti Ti -ci
$., a)
a) '(i rzi'' 0 0 0 0 0 0 0
ai
a) a)
›,0--10000000
:al° 0-. 0 n a) a) a) a) a) a) a)
ai 731 -I-,a) a)
--1 , Q, .---I r--I ri .--, ri ---1 r---1
-
-
tf,
,---, a__
0 0 0 ---i (0c (ii ai ai (Ci (Ci ai (Li
> > > > > > >
cr;
Lit OA r-r,'
.
(IQ
a,
78
analyses with these customers. It should be pointed out, of course,
that using more than 7000 cubic feet in a ten-month period does not mean
that the 700 cubic foot minimum was exceeded each month. However,
this does assure the selection of customers who have exceeded the minimum occasionally, and therefore they have an economic incentive (even
if slight) to reduce consumption. This group of customers is identified
as Subset 1 in Tables 4 and 31.
By checking manually the monthly water consumption, it was found
that many customers showed zero consumption for months other than
November and December of 1969. There are several possible explanations
for these zero consumptions, and they indicate that such customers should
not be included on the observation set: (1) the house was vacant and
there actually was zero consumption for a billing month, (2) it is a new
house which did not exist during the early history of the study period,
(3) there are errors in the Consumption History File. The latter come from
several sources. An "over-read" means that the meter reader has recorded
a higher amount than the meter shows. If this error is undetected, the
customer's record shows that he consumed a higher amount of water than
he actually used for the current month, and a lower amount of water than
he actually used for the following month. An "under-read" occurs when
the meter reader records a lower number than the meter actually shows.
If this error is undetected, the customer's record shows that he consumed
a lower amount of water than he actually used for the current month, and
79
a higher amount of water than he actually used for the following month.
If the amount of this error is high it may result in the computation of a zero
or negative consumption. A negative (or zero) consumption is entered in
the Consumption History File as zero consumption for that month. Errors
are detected when the customer complains, or by spot checks. Even
though the water bill and the billing record may be corrected, these corrections are not transferred to the Consumption History File. Thus, it was
decided worthwhile to eliminate the customers which had zero consumption
in any month except November or December (where everybody's consumption history showed zero consumption) from the observations and re-run
the regression program. This group of customers is identified as Subset 2
in Table 4.
Subset 3 of Table 4 contains the customers who have no zero consumption in any month and a total consumption of more than 7000 cubic
feet for the ten-month observation period. This is the logical "and" of
Subsets 1 and 2.
The last subset in Table 4, Subset 3a, contains the customers
from Subset 3 who gave a non-zero (positive) response to the lot size,
lawn size, and property value questions. This subset was developed for
studying the lot size and lawn size variables with respect to water consumption.
80
Tables 10 through 29 show the correlation matrices and
regression coefficients with related statistics for the regression runs for
these subsets. The results are summarized in Table 30.
Régre s sion Using Subset 3a
Subset 3a contains all of the observations in which each monthly
consumption was equal or greater than 100 cubic feet, and annual consumption (for ten months) which was greater than 70 hundred cubic feet,
and positive response to the lot size and lawn size questions. Only in
Rate Area 1 were there enough responses meeting these conditions to do a
regression run.
Correlation Matrix
This is shown in Table 28. The surprising thing here is that when
only observations which have lot area and lawn area greater than zero are
used, these variables are not significantly correlated to water consumption or any of the independent variables either. However, in regression
runs of the other observation sets, in Rate Area 1, there was no correlation between water consumption and lot area or lawn area, either. It is
only with the Rate Area 2 observation sets where lot area and lawn area
are significantly correlated to water consumption; these observation sets
included many observations which had zero as the recorded areas. In
Subset 3a, water consumption is significantly correlated to the number of
•
•
81
CO
CO
0001ST-OT
z onTPA
CY)
CO
[email protected] 1_0 1.112 TC[ dT
enfeA
Apedoid
tr) LO
00 00
CO
•
•
CO
to
•
0,3
co
•
CO
CO
edpospu -eq
uoppuTquroo
1
C
O
r-I
•
1-1
u") c•I
•
•
1COD
c0 00
-
unA.Pri
•
•
CO
TPsodsTQ aEceq..125
PMLIS TC[
Ty000
uo p.p.le5 pjau
-
-
ieis py 105UTJAk
st_n.Pg
-und jo JecitunN
CO
LO
0
0
0
0
0
0
...
n
0
0
r-1
.E/).. 0 0 0 0 -Cf}
CD 0 0 0
• C) CD 0 0 5-n
t-5)
---1 .-r-d
S-4
0
.-
(0
01 u)
--1
4--.
0 0 t
_
( )
,_[ LO CD 0 Li) ,-
0
-
4-1 r-I CV CO CO 4-'
I
cf.) i
a)
D
J).
.--1
--1
cq co 0 0
• 0 ---f
u) 11),
rd .-1
O › 121 0
O -v)- -u)- -u)- <n 2 ---+
----I
rd - El > s. w
up u) cp $'
O ,-_ ---1
C:
O'—'0 (0 4.--, —
a
V a 0
rid
od
;
-
•I--' 0 (0 w
,„ ---1
ai
HO
.-.
Lf)
$_, u , 0
0 0:1 s-, Ril
a 4)
›
10
5-1 -I-,
(0
-t,
__, >-+ !.,..
.- . I-I CV CO to
a) t (cs
a) 0_4 CD 0)
a
::::
(s)
co H
0 0 0 0 En
Cl
.--g 0 .1-4
-'-' --1 > ro rd _s:-_
, a:3 ra
70'• `78 7.(11 7'ci 761 (0 o
,(Qr_LI0-,Q)}-1,--Icno,12,Q>>>>>>0
82
•sti C) N. c0 NT Ni —1 Ni
-1 c)
1
CD
Ni
CD
CD
CD
CD
Cd
r2
0 0 0 CD CD 0 CD 0
0 CD 0 CD C) CD 0 C) CD CD 0 0 CD CD 0
--I 0 CD '4' ri CO 0) C-1
,.. 0 LO *K1 4 cq c0 r---1 ,7-'
CV Cr 0 CO 0 r-i 0 Op
•
•
r-1 cc
CO
LO
L
----
CO
•
•
•
•
•
,---1 0 cD LC) r—I '7P r—I
c0 d'' cD 's1 -' •sr •11 Ni Ni
co •srt LO •V
o
•
•
•
•
•
•
•
•
•
CO
•
•
•
ri
Ni
LO •q' co
0 c0 Ni cD r-H
•
•
•
•
•
•
c0 LO CD
CD N.. r—i CD 0
NI CO LO
NI 0 CO op
C)
1---1 NiNiNi C'] CO ,H1
•
•
•
•
•
•
•
•
•
•
•
•
•
CS)
U)
•
ri
v1-1
Ni
•
•
ri
Cd
•
CD 0 M
Ni
•
•
CD
CO
.-4
4-,
LC, CO
`3'
OD
0
•
ON]
CO
c0 4 LO
•
•
•
•
•
(r) (r)
c0
0
•
'-r
•
cY)
NI 0
•
0
NI
•
NiN,
CO
•
oo CD
CCU)Lo
•
• •
•
•
•
•
•
•
Ni
•
0)
•
•
•
•
•
e
r-I I-0 CO 0,3
CO
4-1 4 ,
000
co
'EL
09
Q)
---,
,_,
P-1
0-,
•
•
•
LC)
C) CO CC)
r-I
uD cD
n-1 CD
0
0
•
r
eq
•
•
CO
r-4
•1-1 u-)
LO (D 0) LO
cD Ni Ni NI r—I Ni r—I
•
•
•
•
•
•
•
•
CO UP *ç:r r-I (NI CV
•
0
•
•
•
•
•
•
•
4 0")
CO
•
•
•
Cq
Cs]
Cd
--I
0
CI)
0
09 m
..-1 0 .4-, 0
-i--, --4 .
-aj -'-0', ›, E —6)
(31,
ni
H
ti
,
CO
H P-4
i--.1
s-1 --I
1--t
"8 -1
"
0
---1
s--,
•--0
co co
C-'La)) a 0
•"--... "---,..
,--4 ..Q '-1cf
I-) ( (cit1.)-•i) 1
r.cy
-1
°
1
0 ro 7' cr---1
Q)
rn
a) a) a)
4
_, ,-Q ,.(--/ ,-0. .(-) t:51CI
°
O
•
0
5_4 $_. ›, $.4 4-, -
w
c0 cq
c0 LO
a
,
a)
•
Ni (JD
o0
c» co cr •,:zt",c•I Lr) co
.1-1 co co Lo c) Lo
LK) (D •Çzr •1-' 0)
LO
UP CD
LH LH LI--I
:
0
.. 0 0
(
CI
5-4
•
u)
12Q rû OS1
0
co
cr)
m., (la
•
CN1 C']N.
r-I
co to
CD
•
N. 0 •q, L.C) 00 00
CO •qi •q' Cs)
•
• •
•
•
•
•
•
7.
c0
tr)
I
CS) CS) rl CO
r-1 C5)
CD 0
r
•
szr
OD 'V (D 0 c0 CO 0) 00 --I
•••••••
•
•
••••••
-4U)(D Ni
cD
<=,
•
r-I I Ni r-I rl I C.1 LO
ri
ri
•
0
N. CO cr)
CO LO OE) C.5)
uD
U)
cq
NI N. Lo Lc) NI co r4
I
cD
I
(Y) ,—I
(Y)
c0
'11
N
o
LO rH rl r—I r—I ri ri 0 rH rH r—I
CO
ty
,
I
—I
q
1w
0
_ o'
Q)
g
--I
wi
0 •--
a
: : C71
:
Q-,
°
--1 .-.2
(1)rTtl
>)
CÛ4 'n'Cl) -1-"' '7-1 .-C1
W :
.1-(C!"
'.
rd
°
S•-,
- : : 4::
o._ ) ' - " " ' - -m- i > ( 0 c 0 - 7 RI 4-' L'' (CS ;-1 k-71 çal
0)
'71
1---:1 r'--?1 4
1
r'>11 1-1 ,----
CZ,
4"-'
-'
n ri-i C) '74 C.) I---1 0 0 I-1 Cr) ..-' U) Z
Q-,
83
(Z)
-,-4
X
is) c)
r-i
71"
•11
r-I r-1
1-1 r-I r-I r-I r-1
(1)
rcl
I
0 CD 0 0
•r-I
40
0 CD 0 0 0 0 0
cr) co
u) co co 0) u-) c0 cs) 0) 0 CV CO
r-i
CO CV CO CV cV • t4 •q' cf) cV
•
•
•
•
•
•
•
•
•
•
•
•
•
•
rts
-o
4-i
(I)
o
CI CO
CD •<-'
4J
1.1)
D
•
(f)
(1)
H
0
rci
0-) CO Lc) Lr) cr) 71-1 CV
0 0 0 0 cV cv n-4
•
•
•
•
•
•
•
•
•
•
co
•
‘1-1 CV
CO N] N. N., (.0
LO CT) 0 N. CS)
NI d CO N. CO (0
cr) 0) CO IT) r--4 CO
•
0
•
•
•
•
•
•
•
0
•
•
•
•
CO CV
1 (O CV
Sr C)
1
CV
-0
•
•
0 "1-'
cr) r-I
C) CO C.1
CD r-4
•
•
•
•
4-3
r -I
r-q
C5) 0 CV CO 0 71-' CO CO
(0 co 0.-) co r-1
•
•
•
•
•
cs) co c) co co a)
r-I
r- 4
•
•
•
•
"q' CY)
CV CV CO CV
0
M CV N- c0
00 c0 LO
r--1
r-I 0 0 DD
rd
•
D
(
•
•
•
•
I
•
•
•
1
•
LO CO
cV
•
CV
0 0 r
•
•
•
•
1
0
C.)
-I-1
CV CO r--I 0 CO 1-4 CO CO 0 0 CV Sr N..
Lo Lc) cN
CO 0 CV CO 0 1-0 Sr 0 Lc) cp cr-1
Q)
-I-I
C)
;-{1
'II
0
14 Sr
a)
•
0
I
I
•
•
•
•
•
•
•
•
•
•
•
c0 CV CO Sr C--. 0) 0) CO cV CO Sr
I ,--1
—1 c:, co
r--I Sr c.o Na NI
1411
0
—1
o
Q)
Z
o
-.--1
4.-r
$_,
10
(d
>
Li-I
0
a.)
71
X
-,--,
$_, o
a, rci
0) 7:5 co
>, 0 ---1 .---4 cs.1 co Sr LID CD C--.. CO
a) (1)
C.)
4-,
(1.3
o 0
0 0
0-4
a,
-1-1
0
CD
8-I
(2) 00000000
0, ,--4 r-I
4_, o
0 0 0 0 0
7-1
(L1
> >
> > > > >
84
0001ST-OT
Z enTPA
CO
ialeurem edTc1
GniPACo
Aziedald
co
r•-.
•
I
CY)d4
drCO
co co co
i1
•
•
•
•
• •
I
umpa
j @A ale
'XOS O{
co
Tpirtl.eN
•
TesodsTc abegips
‘71-1
IGI1S PMIT sTa
131:Gm/An
buTt000 uopreiabT-TPU
Jeqs em JebuTJAA
-
.
Stil
Tua Jo loquirtN
t:S)
.7:3
0
0
.--(f)
rci 0
0
0 ..-4
-.--1 4-3
+-,
CO
(0
0 CD 0 0 0 __,
0 0 0 0 0 ,--.
c) o o co o to
I
$-., -.--1
Cll 'al
0
,--1
4-, 0
14
0
4-1
(ci
t3) .--1
u)
>
.--1
- r-1 (0
(0
C.)
(D ,,z5--, 0 >
t
a ça,
(
-,-t
E 0r
.Q 8
(0
,--I-,
,
0
,
L!) 0 0 Lo c) 4-'
,-1 CV CO CO d' CD
III
CO
0
CD
C)
I-1
I
i
(N
U) Co
°
CD LCO
z•-n
0
1---I
ii
)
c) cV CO Lo co N-. co p E
0 •• a) a) a) a) a) a) ..
o Q.
71 c)
(1)
w E g 4a' ,_, °
n--1
4_, ro
d
› cd 0(c3
CI) (c5
704 70I 701 7dI 7-ci ) 7010
<4 ri-ii-i (_)
14 co a, r4 >
> > > > > > CD
-,
a
▪
85
i
rr-.I
X
rc3
a) ._,
`Zli 0 N CO N N ri NI NI ri
1-1 C)
tr)
ri r--I r-I ri r--1 1-1 C) 1-1 ri rl cr) r-1 r-I
0
C5)
c)(::)
•
D
cp
oc)
Cl)
i
(cs
..-1 • 0 0 0 0 0 0 0 0 D CD 0 0 0 0 0 0 D 0 CD 0 C) 0 0 0
73
$.-i
(CI
Ti 1
..,
s_t
_, > 0
N.,1 p uP •q , 0 (ID co N NI szi" OP CO CO .--1 OP N. ri 0 CS) N N OD ri 0
N. p CD `q' N CO r-i St' CO d" d" d' CO 'Sr CO CO Sr LO CO di d' CD CV NI
•
••••••••••••••
•••
••••
.
)c)
—1 00u
c,)
I 0 0 ''-i
-I-I
CO CI
r-1
LO
.1.'
CO
DCV
(to'
1-1
CO CO r-1 N. OP r CO CV Niri D ri
•
Cs1 0 CO - *d"
•
N. 0 CO NI 0 ri 0 N. r-1 N
(1)•
•
•
•
•
•
•
•
•
•
N
•
(NI NJ
r-1 LO
- N CO ri N CO OD CY) NI N 0 0
•
•
•
•
•
•
•
•
•
•
•
•
r-1
CO
ri
r-i
r-I
i-JODCOLOCOLOOSPI-01-1 ri CO ri LO Sr 'Sri N. N LO ri 0") 0 'S1' 0 CO
(f) r-I LO 0 O.I Na LO CO CO r-1 rl ri OD LI) r-I'Sr NICOO)SODSPCON.CD
0 • • • • • • • • • e • • • 0 • • • • • •
H CO r-I N 1
1...1
i
CO r...1 1...-1 .
I ri 1-1
i
r-1
Isi LO C \I
1.1) ri 1
i
i
I
I
H
75
s.,
(c3 c..) c)
73
co co .q, cq co uo co c) co --I .z:31 co (:)-) co N-ONCOODUDC01-0
S.-, CO CD UP CO .--I CD .--i OP Q0 OP ‘zr CV CV OD CD c) cr) c) <NI .--1 (N (Y) a) v4
• 0 • 0 • • • • • • 0 • • • • 0 • • • • Lr; cc <7; • • •
:
CO $..1
.}..3 5, (NI
r4
Lc)
t.... O) CO C) ,-10PNICIPODOCOCOCO 0
✓ i r-i N
ri
rl r-I ri rl ri r--I
co
N. OD
ri
ri
r-1
o
▪
LO LO CS) 'Sr ri 'Sr
N 0 CD CD CD
•
•
•
•
•
o
•
n-1 d"
•
•
OP CV CV N. Sr Sri N. CO 0 OD OD N. r-i
r-I 0 0 0 r-i
•
•
•
•
•
CD ri
•
•
•
•
•
N
•
•
•
ri
•
•
s,
0
o
t'
Q)
-
,--;
O
N. CD CO CD t-n d''
cl" 0 00 LO (s) 0 cs-, r-ICO 1-1 CDLONCDDNICOOLOSI 1 LOCOM (J
4=1; „r) • • • • • LID CO 'qt CO 0 r-I di CO UP 0 N. .---1 0 up 1...0 cr , No up up
r7
4-4
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
up
uP cV ,;t"
0.)
,---.1 I
I
I co co n-i r-i rl a) N. CO C.0 CV UP
CV OP d' CV 0 I
r-I
Ni r-i
O
1--1 i Sr N rl
co
r--I
I
(3
trI
a)
ri
r-Q
CO
-r-I
S--1
rcS
>
0
U) U)
4-' -1--1 4J
a)
a) rt)
I2C) M 0)
$. ,..[
r-I
r---1 CO CO
92 CI) '->
'-'-' (1:3
N \
U) I--, ti)
r.L1 N ri r-C-1 ai ,5..
rcs
o
75
V)
LI)
0 o t
_
L.) a 0
.-1
10
n
--1 0 CD
> -1-1
ei ,---
a
L-1
0 r-r-i
o
3.-1 -1--I
aj rci (c)
'-a-rf
0
0
co
II
w
075
-0
•
11j
14
0
---I o
a) cf)
4-1 0 r--I
+-I -r-I U) w 0
Z +j >, L-i 0
t)) (0
.r-I ti) up El I2-'
rci
—s1
ty) LH
-,=.,
o
o
Cf.) .-
- rl
0 0 0
0 0 1--1 s_.,
0
"61
(fl 4
Q)
Q
I-1 ..--1 s--, 0 Q)
-4-1
s, .4--.
(1)
-,r0
s, rc3 S--1 -.1 S-.4 1-.-...
ro
0
X
..,
t3)
7
(3
1
•
o ,,,-. 7._ -' w
o
0 a) o.) a) ..b
r,.
t:D ..
0 (c5 ... T1)' f-'
,Q ,c) -O. .0 (it)» rci
_Q
.-Q
CD
-n .'n
-Q al
,-- :.,' ç7,
:
__ 0 ïl .C.. Q-, ,c1
CO ,--
>
,
;
:.
,
-
_._,
2 r,. '
'-:•',
=33 - , 4_, ‘-__I (_,- r( »sj 6.1--n -,,c),E
n ,-
r-1
`,_-5 :3: CD' , -17,' :---:' n.1) '-'''
(5 co
b
=I
I I 17.;
‘5_,'
cj
'4 1-4 r-Z .Z. ':;-',-, t-21
0 i—ji C7)
C/) Z a, ra,
(24
i I '4'
f:11 CD
(-5 1---:
0
„__,
LI-I
,
o
.
..
)
86
0 0
LOcD
1-1•qi
1-1 r4 r r-1 e-I r-1
(2
0 0 0
(0
co up
• •
co co up di up oP 01 CO (..D
CV CV ""4 -' 'Cr CO CV I-1 r-1
• • • • • • • • •
711 0 CV
•
0 0 0 0 0 0 0 0
4
Cn1
1"--U ")
c..1
C5") 0 0 0 CD CV CV
•
0
•
•
•
•
CD CS) '11 CV CY)
0 0 0 CD
•
•
•
•
•
LO
r-I
.
4-,
CO N. 1-1 Lo OD N. LO N. up
N. c)
c 0 co N. c (''')
•
•
•
•
•
•
•
•
•
q1 '41 0
oP
•
•
n_O
•
co
o") I
co CO
I
1
E-1
co o) d D 0 co co d' co
cn
oz)
r•-.
1-0 (NI
•
•
G
•
•
•
•
•
•
0")
ts, in CV
Lo
•
(D)
LO
CO
•
•
•
•
r-I CV CV CV CO CO
r-4
CO co CD
CS) LC) CO
a) CD CO CO
CO r-I r-ID r-I 0 CD
•
•
•
•
•
0
0
1.0 N. CV
CV 0 CY)
C3) CO 0") CV CO CO
•
0 CV 0
• • • • • • •
•
•
•
(.()
a)
OE) CS) D
la')
1-4
1 (Ni
1
I
•
•
CO a)
0 .)
•
•
OE) CO
n-1
12)
4-1
(1)
rci
Q.)
'Cf
>I
r°
CIA
°
CV CO
(:)
Q.)
Q.,
•
CO
CO
.7:1'
cD
,„,
> > >
1", CO
a)
0 0
(-J (C5 (,:;
87
17)
LO
01
000.0P$ ueql oJoul
8 = enleA
•
01 0
C.•
el
CO
C')
ralatuvla adid
•
Lo
leAed
anieA Apadold
01
•
CO CDC'
LO
•
•
•
•
pod amvvod
LO
01
CO
cualsAs buTpluPdS
uoTw 6 TailluTAA
uon2bwi
E
V) Li)
01 CO
CO
• •
CO
•
el
01
01
.19LULLITIS
•
01
IV 11/V1E1
aIA1S adeospueri
•
-''-no
C•1
CO
r1V VI
V. 01 CO CO
•
•
•
•
utnel
liasact reinlpiq
•
CD OD
LO
TesodsTa abegipp
to
buTrooD ani123od ,?na
01
0) CO
01 0
•
•
I
I
0
CO
LO
V V
•
•
•
0, 0
v 01
6uTi000 ttoT;e 1 a6 P;e21
latispm Dpeuloin
•
qnj, Azpunei
paxv30-1
eIdood Jo .lacitunN
•
01
Jailsem iabuph
stneE{ 1jn Jo laquinm
V)
01
V. VD
v. 01
JatispmxisTa
04
v
.
C.)
0
0
OD •--, 01
0 0 CO
.
•
•
I
.-I
Nr
CO
CO
•
,r)
01
LO LO
•
•
V.
CO
0
LO
tn
04 ..4
V CO
1.
•
0 0,
0 V.
.-1
cr
88
LI) Cs.] r
rH ri ri Hi ( n) ri rH g-1 g-1 1-1
C) r- 1rH
(3) 1-1 rH
Cr)
r -1
0 0 0 CD 0 0 CD 0 0 CD CD CD CD 0 0 0 CD 0 0 0 C)
0 c.) 0 co •sr t---. co N. co Lo d-, 0) N- 0) 1-0 cV 0 CO t---- N. CO CO CO
(D 0 00 1-0 ,--1 CO cV •71 -, CO •ql (0 •7r, '71' CO .ZZI' ‘qi .qt CD 'ql CO CO 00 .-1 ,—I
--1
•
•
•
•
•
•
•
•
0
,--1 Lo
I0
Ti ---1
PLI
H
g
•
•
•
•
•
•
•
•
•
0
•
•
•
•
•
•
N.
rH
tf)
1.0
rd1--1
(D "-I
r CO
Lf)
CO Q
4...
•
N.
CO
V)
rH
H
(D CD
F.4
(0
CI CO CO CO LO C.0 0) -1 LO 0 (0 0 N. CV 0 LI) LD CSD .LD. co co
'q' r-1 (0 cV (-V (0 (D (--1 (--I 03 CD CD
.--4 CD r`, r-1 CNI L',-
(0 'st 0 'sr co 0
(D
CO
(20
•
•
•
,-i .--(
•
•
•
•
•
•
•
•
•
•
•
•
•
•
CV
•
•
•
.
•
.--(
CY)
(D,-1
N.
•
•
0
•7r,1---1
CO
CO
s
CO
,-- 1 cs) (D (D N (D ,--1 c0 dr , a) N d
- -, co c0 0) ,z14 d-'
0 1..0 Lo u) p N. (r) co c0 0) N (D CO CO N- (D r--1 CO 0.)
'qi CO r-1 (D Ç')
(f) (Y) LO CO 1-0 'i-t N. N
Ig
COEi
•
•
.
•
f
C')
I
I
•
f
•
•
•
I-1 g---1 Hi
H
i H
•
•
•
•
0
•
•
•
I
1
I
El
Cr)
.--C3
.--1
(21
pa
o
.-d
.
•
.
C')
I
I
•
•
•
•
•
•
1
1-1 1-1 I
•
1-1
I
CD ,zr 0 '14 Co 0) (D 1_0 CO 0 N (D d -' LO 0 (D ,-- 1 co dr, di-1 N. 0-) (D
,
0-) 0, ,--1 .1-1 . t' (3) C) c\1 O)C) OE) (0
C°•
•
•
•
•
•
•
•
•
•
•
•
•
•
e
•
•
•
•
•
• ':1-'• co
0-) co 0 ri 'qi C3 CO C) CO t',. Hi LN
L---.. CO
C)
__. 0 r-1
CO co a) (9 (9 d' LO d', c0 d-, (0 (D (0 CO (D
t.' sq," 'P
L-f)r_I
(D
C) O)
5-4
H
o., c-_, D ,r .--I c.] N. 0) 0) E.--. 1-0
CO W
r.0
1=4
C)
0
- ri
4 -' 0,1 "Zr a) c0 N- .--1 CO 0 CO 1-0 CY) CO 00 Co 01 CV Cs) 1-0 .--1 c0 (D 0) Cv
(CicV 0 d" 0 0 (-V 0 c0 N dr , c0 00 ,---1 0 cc (--1 0 Lo ,q, 0 co r-1 Ç') .--i
(---1
'C
(D•
CO
H
•
•
Z
•
•
•
I
I
•
•
•
•
•
•
I
I
•
•
•
•
•
•
•
I
a
•
•
•
•
I
I
0
0
L9
CD
11
ri-,
4-,
0
-r-1
.
.
if) 0-.) N N 0-) u) d-, co 0 t_r) 0) co co 0 1-0 .--1 CS) C) Cr) CO 'Zr 0-)
LID 0 (D CO (D Cq CD 1-0 (0 (D (-1 —1 t--- CO d-, dt, (D 0 00 LO .-1 .-1 `n1' CD
a)
r4
,--1 eD
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
.
---1
(-1-1 ..., 1.10 I 0 (--A (.0 (D co c0 CO CD cV 0) 01 '714 C) 1-0 4q,(-0 CD N. I Co N
çzi
-1
-,
4-1 .-&-Z
cv
co
Lr")
1
CO
V
...
C'l
CO
(C) 0.1
(D
(3)
(-1
LO
LO
0.1 I
Z
0
r---1
.---1
I
I
I
I (--1
I
I
0I.
I
1C-:?(
10
CO
r-)
ij
cc
ril
Ii
(24
(.)
P.1
I
a)
a)
124
...Ei
c7).4. -
-
0
(Ti
-,--1
$--(
0
>
0
0-1
(r) u) cf)
'--' -"-, --,
ra o ai
Po a:, r,a
a)
a
s•-n
s..,
a)0) ,--1
5-4 4-1
0
0
E
0..) .
o u,
u)
0
,.. ...,
0
4_,
0 ,___i
0
a) .
0 0 t
(D 0
r.. ro z-,0 a 0(cs 4-, (f)
>•(
s--, 0 7. .,1
ci)
(Ci
U) (()
a) 0
. cc cc
4., 01 (--1 ,Q (ci
LH 4-1 LH
0 0 0 El
7.
:1 0
-,--1 0 0>--f
0 (r) ,_
-H
u)
(I)
q ,
l-i
"H
d 2, --I
0 s-, s.. $-.. >, --, 4-.)
0 . 0 4rs.., ,..,. Œs .--4
0 >-4.--1
.-1
1:p a) °
.--
g
Ci)
0 0 1_, 0
,Q °
.-Q°
,-C1
,C1
4-1
0
461
.
ç'‘
_
0 --I ...0 CD, -0.
'.>
.7
1-4
'1:i
a) 0
- ..-1
$--. ,Q
a) f---.1
t5.) (d
(:5) CO F -1 0-, `.-)
.,--1
0
o ,..., ',_.
'
$_.n
0
ri)
4-4 4L:
CI-I
0
0
Q 0
-,--1 $_,
:-_-1 0.76
1
ai
'-
(Ci
IZ p-.1 4
1Z
'
1-.- J- -3-' <4
i2i (11 (.) ',--=', 0 ,--4 0 L.) a u) ;_------ cf) ._,---, 0_4 (3_4
.
▪
89
CV r-I
r—I
0
r-I r-I
dr
741
Cs]
r24
C)
C5")
0
•
LO CO
•
•
CD 0 CI 0
C)
CI.)
LO LSD 0 CD
C`,3 CO CO 'di CO Or) N CN1
•
•
•
•
•
•
•
•
•
CO
"qt
0 CO
CO
1-77.;
c0
(3)
di co C)
•
•
•
r--I r—I
n-n
<31 cs) 0 0 c0 CO
cN1
r--1 0 CD
•
•
•
cr) LU C) LO
CO CO Up V) 0
co
LO 0"-) 0.) cs)
LO
0
•
•
•
•
•
•
•
•
•
1.0
0
(I)
•
•
•
•
•
(Or-4
I
Cs3
EH
N
I
I-0
C.C)
r-I
n-40")
•
•
•
CY)
CO cY.) LO c‘1
co cNi
•d' c oz)
•
•
•
•
•
•
•
•
N.. LO
Lo CO
LO
CD 0.) 0") Cs) 0) CD Cs)
CO CO 0")
ri
o
co
I
II
I
'61
CNI
•
cD
(I) (O CS) r-I
•
•
1
1
•
•
1
1", CNI 1-0
C)
•
•
CID Cs1 C)
1
1
•
1
C)
•
1
0
o
CIO c-1 CO 1-1 CO N. 1-1 CI)
(-0 0 CO (Z) CO CD I-0 CD
•
•
•
•
0
•
•
•
CO
U')
1-0
Cr)
L
U(..0 0) 'n:V CO CS) Cr/ h. CO
11111
CO
0)
0 0 0)
•
•
•
LU(OC)
']
o a) a) a) a) a)
-
7-;
C-1
ri
> > > > :>
( 7:1
90
z enTeA
-10 1.GuIPTG
edTcl
CO
ZUGE
LO
riT
PA
ODD
CO
Alledald
,
:34CO up
•
•
•
uopebTali Jet-MIMS
L!)
PalV umpq
op
LOC']D
co co c )
•
UM
S
•
CO
CO
CI relnyem
Lc)
,
•
0
CO
CO
0,3
-101_18 PmT_Ispci
L.r)
.sr
5 uTT 00 0
uonala 5 T-TleU
stgeg
TTnj jo JegumN
CO
•
0
0
0
0
0
Q) C)
0
(-CI
0
(1)
$-.
W
7-1
0
D
0
0 0 C)
61 0)
0-1
W
(1)
-.--I
Q.) ›
0
Q
•
0 Mi t3)
Q)
(C3
$„1
;7,1 › ,: ,L.! CO (0
r_i_l
U
..
}-1
,--1
-
I—I
•
0 (Ci
ID
.11
'S ._(. ct cD cp cp rci
4-3
0
( 4-L
1-4
H
U) CD
C\I C9 (I)
U
)
(t) -I-'
i
I
I
01 rri
ca c:, Lo Lr) C) 0
o
9
--1 0.) -r-cji
s-, -H > --, - 0
H ,--1 .-1 ,--1 c\I
-I
.1-.I
,
0
S-1
cp
0
-, 0
4 -H
0
0
Q
.
)
W
LO Cs/
•
5-1
a)
Q
• ri
•
u)
Q.)
78
H C \I cl) if) CCI
CD P-A .-C/ 0 0 0 CD 0
a
fal
ci
)
,...,
0 7; T ,0
1
2 :71 ,o 70-1 70' 7cs
t›- n-,
V > >> > > V
-
-
-t
-
91
N. CO cs1 NI 1-4 rI NI Hi
0
HI
▪
•
• I
▪
CD
r-1 H-1 HI HI n--1 1-1 1-1
r-I
,Z14
(7)
0)
CO
00
CD 0 CD 0 0 0 0 CD 0 0 0 0 0 0 0 0 CD 0 CD 0 0 CD CD 0
0
LO 'Zr CD N. CO NI CO di c0 • ,4-1 • NI NI CO 0 0 cnI C.0 r--(
NJ Cr)
N.C)(D - rNICc r-4Cc
CcNI NI NI
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
r-1
1-1
LCD
•n4'
NI
cp œ
•
Er) co co
0 CO • C)
•
•
•
•
•
crs CO
C)
▪
NI
•
•
•
N, cc) Lo • Cr) LO r-I N
•
•
•
•
•
0,3
•
•
•
0 0 CO CD
1-0 ‘7,11
NI
N c0 N./ CO CD 0
•
•
•
•
•
•
•
•
CO
CO
r-1
CO
NI
r-4
r-4
4-, 0) N. Hl Cs.] CD ',14 CD CO CO CD r-1 Lf) IS) 0 ,3 rH CO LC)
(i) "V CO NI 0 CO CO G)
CN1 c.)
c.Y) cr)
•
•
•
.
•
•
•
•
•
•
I
CV)
•
0 NI
r-1
•
•
•
•
•
•
0) C5) N cr)
(D (C) CO NI 0) 0
•
•
i•
•
•
•
CD CS)
(c) c‘)
•
•
sr) CD 0) Cr)
c) sr) 1-0
•
NI (--1 n—n I NI CO Cs1
cy) (2) N. CO N C5)
00 CS) CO
0
•
•
•
•
•
•
•
•
I
CD
CO
(-0 CD CO cD CS) CO 0 'el
0) CO CO 0 LO
0") 00 0) 0
•
•
•
•
•
•
•
•
•
•
N. CO (D 01 CD CS) H4 0) CO 0) N NI CO CS) '`;14COO)(D LI)
Hi
r-1
rH Hi
r-s1
0
CO CD CD (0 C) 1-0 r"-1C)
r--1
•
•
CO 0 CO 0 C)C)
•
•
•
•
•
•
•
1 -0
•
CD tn
C) H4 0
•
•
•
(C) S14 CO (NI
•
•
C)
•
•
Csq CN1 CO tn
C.1 CO CNI N NI Hi
•
•
•
•
•
•
•
0
o
-la
j
(1)
...-i
0
VI
(4-1
C:0 CD N. (D 00 r -I N c \I Cr) N. CO CO (D 0 NI ( --I 1.0 0 HI H, C) N CO 0
N. 0 NI HI 0 N N LO CD NI C-..... S. N Cc CO CD CO 0 CY) r--, 'qf H4 CO 0
•
•
,C) rn
•
•
(..,0 f
G)
•
•
LI) CO
t
.--.1
•
•
•
•
•
n-•-i 1 ,-i
i
0
u) u) u)
s-,
(1.)
0
7.(51
0
(-0
s-,
-51..
-H
0
(ri
a.
▪
al cd RI
120 CQ Ca
$..
0
.--1
,--1 0') CO
0
r_s.
\
r..
\
u)
.--1 .-C2 rd
>
4-1 LH LH
4-1
0 0 0 H
4-1
0
0fd
HI H S-4
H
X
0
(I.) 0 C.)
a)
„cl r.Q ecz
,-0
(:.)
E.
Z
1
f
:i
•
co rn LO HI r-1 (S) '1
I ,--1
u)
-1-4
H -I-.
(1) all
-0 0)
:7-5174.
•
•
•
•
•
•
•
0 CS) CS) CNI LO Cg
0) HI
N
I
0
0.
--1
(cl
a) ,
•
0 0 1-4 HI
Cf) 4-‘
(1:1 0
. fli S-4 tO) Hi
(cs r-4
ta)
0
RI
Q)
fl
-1-1
s__, r.Q
cll
0 +' CL'
-H __, G) G)
-., .._
P4 (1)
4_,
0.)
• --4
,-Q (cs
,
•
Hrd
S-
i -I.--, 0 0 ::i
(Ti 0 s _ 0
n :- 03 0
W OZ (...̀..) 1-1 0 0 1-4 C/) ,--- CO l''. CA-I fa,
CO n-4,:',
4 `,--__,
' Z, 'Z F-1 ,I> ,----t; Cc'.
'-,
Ht
, 0
74 15 .11:1
4-4 (>
'O
:-j
•
•
•
•
CO N 0 i
C)
7.-), c-cf
t')
0
0 u)
u
-,-. 0-' o r-4
0 (n
7:5
o -si
(0 4--, ci) ri) o
(..) S:), ,cij
•
ta)
›.., s., 0 -,7,1
0
cn
ul
CI rCi
t:» u) H a. ,-,
• s-, (D -.-1
I-1
-H
0
(1) ._.
'CS (r3 s, .., ta) 4-4
-H
a) a1
(i.,
•
.4
I Cc NI (-I
92
0 0
Lf)CD
,;11 r—I
r—I
r—i r—I
rH
0 CD 0 r-I
C) 0 0 0 C) CD 0
•sr in co 0')
c0 0) cq 00
cn1
•
•
•
•
•
•
•Kti
•
•
cycDcc0
co csi cq
•
•
•
•
•
r-1
11)
rcs
if) N CO (3) CO OD CO L0 1-0
d' NJ CO
N. 0 CD 0 0 NI NI 0 CD C)
•
•
•
•
•
•
•
•
•
•
•
•
•
LO
o
•
co
cr)
(1)
co
c, ․) 0.1 c.) CS) (0 'Kr
•
•
•
•
•
•
I
Cr)
0
CD
•
•
OP
•
CO LO
CS) d' CO CD CO
•
•
•
•
•
co (NI
'-I
I
I
CO 00 <3') (.0 sql CO 'Kr Cr) CN
CO
cO CS) r-i •q,NI N.
OD NI N.
•
•
•
0
•
•
•
•
•
•
•
CS)
r-1 CO CO 0 r-1 up
NJ NI NI (OC')
C"--
o
CO
CO
C)
07.1
00 0 0 CN CO CY)
r—I
C)
•
I
•
•
•
I
•
I
I
''Zji
N.
(.0
CO
r—i
•
•
•
•
•
•
•
•
I
I
o
NI
I
r—I
Cq
NJ
0 (NI cc)
•
CO
•
•
0 CY) co
CO CD 0 OD 'Sr r-i
•
CO
I •qt
•
•
•
•
•
co co
•
•
LO CO
•
•
CO
OD CD O) '-CO
NI NI r—I r—I r—I
0 I'.
N NI CO
1,-1
1
I
I
40)
''
@
--,
a)
Ord
0 'C5 (T5
1--1 CNI CO 'Kr 1 0 (.0
aj
F''
a,
0
Q
a)
.---,4—@Q_,
r—I
0
C' - .-4
R31
0 4-j
Crl
rir
,
(ci
S"
00
a) a) a) a) o a) o a)
l -RI 7ii
-,-1 7:51 7-d 771 7-c-I:i a_
o_, > > >
'-ri
>>>>
93
000' ST
00010TS 'On[
OLLIPTO
d'id
CO
JeAed TUE
enipA
Aq_iscioJd
LO
•
•
•
•
seau 3o JectumN
•
CO
eaav unnuri
umpq
ja/Ve.1 0
,
1.19Sau Tpanl_pi\T
•71-,
JseM14 S TG
vq,
•
e 1:2
5uT s
•q'
.5UT1OOo 1.10T1Pie TjJ
sl-T1Pg
TInj Jo JectumN
CO
o
-I-
ro
ni
0o
-
,—,
d u) u) 00
0 0 -0 v,
---
O
u
)
$
.-,
a)
a
a)
a.
0
--I
ai
r, Q)
,
ty, --1
1--1 ---n (0
0.) › n
0 0 0 0 0
cp c) cp c) c)
D 0 D D 0 cp D
D " " . D
(2, Lo 0 D in 0 0
,
.---1 C'.]
U) 4-'. 0
(
C)
-,'
a)
D 0)
• ,-4
(a
(:3 (cl
.-c)
ro
E
>,
t.._,
Q.) .-,
,
)
{/}
„
'‘.,
4-, 4-, 4-,4-,
KO-
CD CD CD CD CD
al
0 0 0 0 0 A
o c) cp c)I
cpn
0
Cl
2 9
Q 7cii 2 `._?, `Cf,' CO
cp co
u' —
rd 0
E o
0 '--'
CO CO `4'
c 0
cD00000,,r, ---1
,
rd
,
(7)
CO CO CO •1-' CO
0
a
--
1 n r-Li (3F-..] 0 Co I2L, r-r'', > -cf)- -Cf>
-cf). {/} .0)- > 0
-
CD N. CO CV C3
-
.1 4
(N)
4
1
•
•
4 I-I
r-I r-
r-I
ri
CY)
CS)
CD
cc
cc
CO
CO
•74-'
C:) CO 0 0 CD CD 0 CD CD CD 0 CD CD CD C) C:D 0 CD 0 0 0 CD 0
es] CD
CD N. 00 r-I C \I . 1' 00 CO CS) C‘.1 CD N-• r-I CD 0 (..0 CN] CO CV 0
00 .1' 'Sr 'Sr CO 'Sr •q' CO 'r' cN1 .1-n .1-, .q,
N. CD (.0 '[' C \I CO
•
•
•
•
•
r- 4 L1)
N7
N.
•
•
•
•
•
•
•
•
•
•
Cv)
•
(_0 0.4 c‘,1
•
•
•
0
rl
CO
es]
•q, cp co r---1 •t-, up co co
•
0 CO CNI
• 0 f"-,
•
•
st-'
CI)
o
•
œ if) cy) CO 0 N.
V) 'Sr
C‘i • CN1 r-I CV CO r---1 Cs.1
- (.9 N. CO
- C
•
e
•
•
•
•
e
•
•
•
•
•
C)
•
•
0 0) (-0 CD (.9 CO if) csD CO CO a)
00 0) CO 1-r) CS) (-0
r-100C.3cDCOr-ic.ICOCOr
•••
•••
•
431411•
•••••••
••
1-1 I r-
I CO r-I r--1
Lf) Cn1
•
•
r-i
•
•
•
•
•
•
•
r-I
cc
cc
cc
r-I
4-1
CO (-0 CNI
•
•
CO r-I
H
c0 CV
CN1 up CN1 0") (SD
r---1
CD "q'
Lo
CC) 0 CO
c)-)
cocDoorcs->moDc\IN.0)r---. ,--ic:DLococDo)c, incococsDN.
•
•114•••
••41
•••••••••••••••
C\1
N.00N.0).-40C•10)00CDCOCOODOI-0 LO000).-IN.00
CD
r-1
r-i r-i
r--1
r--1
C) r---1 0 0 CD
CNI
CT) LO
in
▪
TL)I
•
•
0-)
'Sr 0
CV 0 CD CD CD
•
•
r-i
ir) (9 CT) co C CD
0")
•••••••••••
C)
C cNa
cp
•••••••••
0
o
COG)
Cr) c CD 0)(-DCD
CY) I-1
CO
CO
'QCO 0,1 C‘I 0 NI CD OD
•
N.
•
•
•
•
•
•
•
•
•
•
00
0)
•
•
•
•
OCDC00)1-0(.0C00)
CD CY) 00
CY-) CO
•
•
•
•
•
•
•
•
N CY) CO CN1 C"] CO OD 1-r) (-0 N.c‘a
N. Lo cNi cp
",1 es] rl
'Q
I
CS) r4CO
cc
rl
-1
CO
r-1
o
0
U)
o
o
S•
oa_ t
r0
ro
171
Cc) i2Q
..,
Cl.) 0
0
5 (j)
,--,
,___, co co
0
- .-I
r. L . '---.,'-----,
--,
_ t , cLrin0
s_. 0_° u) co
'
.
1a lc .)_,>°
(Ci
j_a :s, h
a
o
-t :ly , e_,_4:
. _(3_:, , _ _ .,
14-i 0 4-1
>
1---1
0 0 0 0
..-4o ill Q) -,-,
4-4
0
•
0) (L) W CD W CD '- g E
nj n-) S 0 0 0
7!
(
---'
'
- ..c., r• , ,o, ,J
r:-(1
0
u)
-,--, o
-,-, •--1
0
1-1
--, ta)
I-!--I
-
5-1
E E , E E E :1 -- -- i ,,__,
,
-
0
-0
o 0
i3) fli
,--4
s--,
0 --.
.1--,
I Z ,--- '4 '4 '4 Hi, ,j -- f< cr :', Q f-Li C.) `4 C) I-21 0 (...) i.1 cn ;:->"
-,
ci
▪
- r
95
'Sr
r-i r--1
ci
CD 0 CD 0 CD CD CD
LO ‘7/4 (.0 0) 0) (Y0 (.0
r
c0 c
•
c
o
Q.()
Lc)
c
s)
ci
LI)
—I
(c3(
•
•
CD
0
•
•
csD
(-0
CD .0 c
•
•
•
•
•
•
C\1
0,1
•
•
•
9
0 0 CD
•
•
0
CO
N.
N- L.0
•
•
•
r-1 I
(Y)
0) c..1 (.0 CO CD
LO cN)
C) C') c 0) (..0
•
•
•
•
•
•
•
0) I r
CO CO
I r-4
0
CD
---
•
1
•
•
CO
(
•
. ) 0) CO CD
'q" CO c (0
•
•
0
L0 C.1 CS) CD
--1 CV ri
•
c0
r
Cs)
•
•
cq
0
•
CO N. I-0 C, 3
CV CV CV CO OD
cl
0
4-, (0 (0(0
CO
(Ci
Cv)
r-I
•
I
•
•
I
(7) (N 00 CO CV Lip OD (-0 CO
C)
ri CV
•
•
I
•
•
•
•
•
9
•
1
I
0
o
4-,
ci
•••4
(4-1
o
LSD N. 0 CO 0")
L.0
o
co co LU
C))
cp 1.0
N.
C')
c‘.3
CD N (.0
c's1
•
,-Q
I
I
•
I
•
CY)
•
•
•
r—I
r---1
o
CV
co
0
0
czt,
If)
(-0
N-•
0)
0 0
C)
(D
C)
72,
::-- :,-----.
71.J
',.-:, > , ,-- —
96
000‘0VS OrgoA
JalatuRia adTd
JaA 2 d
e-1
v.
0
ma
artjeA Avadoid
C4 •cl
0,N
tO
co
I°
vaiy Tood
I.
•7.
40 e)
00
CO
•
Tood luaueutiad
G-1
H
• •
01
01
•
0
CO
wa4sAs buypiupds
•
C0 ••nV
ir Nr
r••
CO
• •
uollebTin lawm
tior4P6p1i ieununS
PaiV tiN1P1
CO
01
C0
°I
0
adeospuel paumuloo
• .
umpi
, CD
0' CO CO
I
•
CO
• • •
ED CN1
V` V1
00
ipsodsTa a6ecpeo
CO c-I
01 01
ocl
CO
0
CO
buTi000 oATI.P.10dPAa
II
I
0 0
C.0 0
0) el.
1,
•
latispAngsTa
CO VI
0 c-, CO 0. 00
co
co co .4. c•-,
10 ( 0
,r .0.
•
•
in
co
I
CO
0
0.
.q.
611 Hoo 3 uonpiabpJau
0/
.
C/
•
I
1
logspm onewolny
.tailsem iabuTJAA
0
CO
0
l'•••
0 CO
•
0
01
0
CO
• Cl•
0.
• c:›
qnj, irupunei
staeg itru
Jo laquinN
•T CO
CO
0-1 NI'
2aNlcri
n-tQ
01
V>
COr
•-•1
01
•
•
Woad Jo loquinN
CO
CO
•-I
0, 01
0 0401
Or
0 CO
•
•
•
00
•n1•
•
97
CO (7) Lf)
r-1 r-1 I-1 r-1
r-1 IN r1 r-1 r-1
I-1 0 n-1
r-1
CS)
0")
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 CD 0 0 0 CD 0 0 0
CO C) CO
V) 0
•
•,‹1", co , z14 0)
CO (SD CSD
•
•
•
CO IN ‘7:1'
•
•
•
•
•
1 0 di co
•
•
•
co
•
cD Lo CO 0 CD CV NI CO
•
•
r-1
•
•
•
•
•
•
•
•
•
•
•
•
•
r-r-1
CO r--
•
•
•
CO
CO
•
•
•
(..0 co c)
c) cv
•
'Kt' co
•
•
r-1
cv
•
co
co
cv
cs)
•
(v)
CO 00 CD rl
•
•
•
•
•
0")
e
•
•
•
•
0
•
•
•
•
•
(Y)
Cr)
CO N. O. ) 0 C7)
I
0
LO
CO
CO
CO
CO
•t
i
CO
LO
CO
d
`
CO CO LO
•
•
•
r--1 CD CO CO
0) 0 0
•
•
•
r-1
c0 00 Lo
•
•
•
•
I-1 r-1 r-1
•
Qfp
cf)
•
ri
CO r-1 CO 0 CO
•
r-1
•
cv
I-0 OD C)
•
•
•
0
•
•
•
r4Cg
r-iI
CO CO IN N. LO LO CO CO N CO
LO
t--1 cq
•1-, co N
0 CO cp
N
CO
ri
•
N. co Lo Lo co
•
•
Lf)
CS)
IN
NI CO N
0
•
CO CD 0-) LI)
tr) N,
CiD 00 CD 0 'q"' CS) CO
(v)
CO IN 0 ri 0 N. IN N CiD IN '<ti IN 1 IN ri (73 10
S.
0
LO
(r)
•q,
ri5
•
r-1
if)
0
c)
CO
LID
N
1-1
CO
0)
1-0 0 0-) N cr)
•
•
0
•
•
•
•
10(V)
CO CO CO
N
71' if)
CO Cri 0 0 `4 CD •q' C0 CO CO
NI LSD r-1
'<14 CS) LO
CO CO ri
N r-1
LO I-1 rl 0 CD CO ri •srl cr) CO 1-1 ri 4,11 0 N3 C.C) LO
CO
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
e
•
•
CV
0
0
.1-1 c-;) co 00 ‘1-, co up (..0 (NI LO , 4-' LO CO 'Ci"' N NI CS) 0 CO ,--1 0 LO
•
•
•
.
.
•
•
•
•
•
•
•
•
•
•
•
•
1---1
-) .-i IN CO `1-' IN CO (0 N. 0) 0 CO LO
CO
0
OO
N.
t
N.
CC)
0
LI-I
r-1
N. LO co
(D 0•1
rl
4-1
IN 'r'' (D N. ri if) r—i CO
,--1
I
I
I
I
I
a)
o
a)
o
0
u)
(f)
co
a)
45 '-_, -'
U)
ro ai rd
a r° PQ CO
... +_,
..-1 ,_,
.---1
0
,--1 CO CO
0
i:5)
\ \
10-_,
ri. cv r-4 .-Q
0
LH LH LH
--1
0 0 0
ai
-.4
S-I
H
$.-4
4..,
5.
ro
.-1
C)
co
ai
Q
75
CD -H (1
0)14
(I) ,_u) -,--, 0a
(:)
)
4-,.
a) CD a) a) a) 'd 0-)
-9_, ,-..Y.,,---Q r-cl -Q r_,,b
(.0
(71
7 , ...
„,
al
s, L.) ,
o --4
0 rzi s_, (1)
> çz''
c), ..C1
:7; —
2 0 -,--_;
, 11C:; .-r--I P. -4* 0 Z,
;.-:-1 ,,4
1----1 '-7 ' 7
I I--
u)
TS
0 0 t
0 0_, 0.)
'0
WS
O
..--t
,-,
ai
0
a)
7. r-(11
in
W u) ^
.--
a)
Q,
$_,
cl.) al
—
LH
-.,
s, ro
(1)
0
0
-,-. 0 -1-,
al 4-'
rn
i:5)
E >"
.-i u.
s-n --I
cd s--, s, tj)
I-, s,
0
-,-,
s,
cll $--, ,--1
,..
a)
-,-,
1-.2i :,.) U 1--i V.) ,--- CO
I-0 (-.0
•
•
ri CO
cr) (
,---1
0
98
0 0 f-1
r-I
r-I r-I r-I
di di
C
a)
I
0
0 0 0 0 CD CD 0
cf) co Cs.]
CO N. ,---1
•
•
•
/
rd
r. d
,-,
-4 r--1
- '.1rn
s_..
C..1
(
Y)
(C) a) -r-1
c) uP of) da CD (..0 CO CD
CN1 CO s14 '<1-, CO C \I CN1 CV
•
•
•
•
•
•
•
•
CD
7=5
S--,
a)
44
cn-1-'
Q rri_
CO
r-I 02)
0
(T3
•
•
c)
CD
•
•
r-4 CV CV
•
•
0 N. N. N.
0 0 0
•
•
•
•
r--I
1-1
CO (.0 CO NI CO
CO
co
•
co
V)
co
E
•
r
•
aP
•
•
•
cs)
•
•
co
C,Q cvD CO CD
0
•
•
•
‘<t-,
uP
(.0
co
•
•
cq
•
•
•
co co a)
a) op
•
•
•
LO
uP
(.0 CO
NI N. N. 0 0
`z:14
C‘I 0 CD
•
•
•
I
co
•
•
•
•
c)
c) CD up
•
E
I
-1-'
(1)
-.--t
0
-J-I
4-.1
a)
Cs) d' d' CO CD CV CO Lf)
CD Cn1 ,--1
•
•
•
•
•
•
•
•
0) GO d' dl N. CO 0) 0)
CO
CY) L-0 N. CV I-0 1
1
I
i
I
,--1
•
•
•
0 i 0
'ql
0
o
a)-,
a)
-1
• r-1
X
al
a)
a)
0 7=5 (0
c) --4 r.-1
a c
rd '
Q
c„, CO .14 co co co
0 0 0 0 a) Ci) 0
:-.3 -1 :7) :-)-' :1 =J
1 ---; r.-4 --, --; --n
- ,---t 0 0
-5 .1-j
(r) rci ,-,•:', c:i :.;,i c .)
7 -O 0 7' -'0 --:41' -(i
O4 (4 0-C11 (4 12-4 > ,---- >
,':
99
O 000
1
000'ST
0 .1$ i onT 2 A
CSD
CO
-1e4eurem adTd
LO LC) CO CO
11.TPA
ATI G d al d
CO CO Lf) CO
•
•
•
•
C
PON LIM"
CO
•
UMPI
TGAPJS
1...TosaG Tp.myeN
Tpsodsra eEyeq.ree
v)
lagspAALIsTG
.101- 21A buTsfl
burT000 uo -p -2105Tijad
sweg
JagtunN
and Jo
C's1
Lfp
CO CO
•
•
-
0
CT)
a
-,--1
0 CD 0
0 0 0 0
0 0 C) CD CD
CD
0
CP
...,
CO 0
O
. . . . C:'
•---•
(D
CD Lr) c) cp i-r) CD
'-' (10
'
' r-1 CNI CO CO
(3.3 t»
0
i... 4_11.---1 0 c) 0 0 'c:)l4
'1 ' 1
,--, -I,-
•
(...)
Q)
.._,4-- CD
,
z
I-,
r,
O
•
•
41
,
:CI
1_4 „,.. .,_,
, >,
a)
>,
0 t rci
-I-) 4-
.2
-1-- (n- -5
,
\ / 0 0 0 0 , \
0 0 0 0 ."
4
,,
,_,
0 fl, 0 c c: c c o u!)
--)
).
)
)
; c3
ro 0 j $., -71 as —1.--1 (NI co al 0
,_.. 0 cf) ra., na > .cr)- -cr> {n- -cn- > (...)
1 00
'r CD N. CO CV
r-I
GD
r-I r--1
r--1 r-i r-i 1-4
r-i C)
GD
CD
GD
co
•sr,
CD 0 0 0 0 0 0 CD 0 0 0 0 0 0 CD 0 0 0 o 0 0 0 CD
LO
CD
•
c) (ID cc)
•
LO
•
•
cN c0
•
•
cN Lo co d"
c0
d'
di
•
•
•
•
•
•
•
CD CO
CD CD L.C. CN
d" 00 di 0) di di d"
•
•
•
C7)
CO
•
•
•
•
CV CD
(Y.)
•
c
•
NI
•
r-1
N
NI
co cq
cp
c0
Q)
•
•
•
(.0
Lo c0 cs) csi c0
0-) 0 0-) CV
•
•
•
CD
•
•
•
•
N.
•
•
cN
C N CO
•
•
•
e
cp (2)
LO Lc) di
CV 0) CO CO N LO CD 0
•
•
•
•
•
•
•
co
r-4
NI
CO
N..
(0
<1-t GD Lo (N.) Lo op cs) co 0) OD CD OD CS) N 1-0
NI
LC, (3) CO CD 0)
CO CO (D CO r-1
CN7
(Y)
ri
(V 'Zti
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Lo
NI r4 CV 1 I 1
rl 1 r4 I CV r1 r4 j NI Lo
0-)
•
•
0
I
GDGDNNIGD'r
c.) cz)
•
NI
r-I
Lr)
(3)
•
•
•
•
N. 0-) N CY)
•
GDGDGDGD')C'N GD
r-4 CO CV CD
0)
c‘I Lo
'r
•
•
•
•
•
•
•
•
C:) NI CS) CO CS) CV CO C.--
•
•
•
•
•
•
•
di co 0.)
Lo
•
•
N- CO
r-I
0
731
L.().<:14 (1) r--,
CO P GD C) CD
r-i
•
•
•
•
•
•
•
0-)
•<:)1
0 r-I CD
•
•
•
•
•
CD
•
1..0 CD CD
C)
CV CD 0 N CV CV N
•
•
•
•
•
•
(;)
•
•
•
•
1-4
0
o
"1-'
0
•ri
0
4-1
0
0
(r) dr, cp co r-1 0.3 co c.,1 ,---i
r----- 0 Lo co LO r---. 0 co C---- co r.---. 0) c.1 (--. d.,
0 CD 0 LO CT) N (..0 P in .1-, (.0 CO r-I LO `n1' CS) LO r---. LO OE) CS) LO CO
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
cs) a) cm r-1 Lo Lip
C., N-3 LID CO r-I OD I-i N r -I N. CO CNI LO .--1 c.)
)
eD
NI I
.-1 1
1
CO .-1
r4 1 ,--1 1 CO CV ri 1
1
I
1 NI
o
(1) U) U)
4-i
0
714
0
Q)
ci,
L1-1
o
se.
0
4-, 4-,$-4 4-,0 ,--4
(0 0 ro
1°) CO CO
0 0
s.-n ,_.
0.) En ,>
—1
—4 co c'')
,C (0
tpl
rEl) u)
al
,0., 0
''..
`4
7 3 '2' :4 ' r-4'
-
-
':::
-::. ,-ji;
n:%
l ) an
1a---1 F4-1 CD
0
Ts
u)
a) -,--1
r.,.., c- '''.--1 ,c) (ani$_,
-,-1 (1) >
(21
L1-1 LI--f 4-4 H
o u) ..u) --1
o o o
-1-' 0
4-,.. S-I 4-4,-i 0
.-- • 0 0
.0. s...4
9
• - s__, t3)
CDs--, 0 s-.4
0 $....
0
0 0
:.• (7
._Q ,---,Ti (7
_C)
r
-,
r!
, ---,,
- , -1--,0
:_73
()
'4 1-4
0
(1.
(C3
0
(1)
--r
(Z1
(i)
-.-4 i:» cf") E-4 11,
71 (0
0 (1)
s--, -0
0)
a)
ai
F--1
0
› 5..
E
4-4 Q) ri
÷, (I) a) 0
c..)) (c) ›'n /--1 0
..--1
..... -,•-n
._, s-, t:Y) "
1-4
..1
F-4
0
"-'
4-'
Q.)
C
1 (1 ) co
\4
0 :--Q
.-I--,
,-)
, -,- -1' Fi
..,-'
0 1--1 0 U
Cf) Z a,
•
101
1
CD I'D r—I CV `'14 r--1 r—I ri 3-1 ri rl ri rl
-,--1
X
Lc) o
1) (C/s.
E
ri
(24
ri 0 0 0 0 0 0 0 0
12)
co cr)
csl 0 cq CO CV CV
•
•
•
•
•
•
(.0 0 0 00 co
Cr) CO CV
•
•
•
•
•
•
CO 0) CO CO CO 0) CO 1-r) 1-0 D
, 1
•
I-0 I-0 D D 0 0 c..1
a,
•
•
•
•
•
•
•
•
CO CV
Lf)
•
•
•
z1 NI co
0 0 0
•
•
•
CO CO
Lf)r-i
(I)
cc
E-1
c ,--,
D co co gd, <NI co
•
•
LO
C)
•
ri
03 LO
•
cq
•
(C) LO CV C.0 CO 1-1
CV
LC) CO C:-.) CO CO CD CD CY) CO 0)
r-i
•
.
•
•
cNI
•
ri
ri D 0 0") 0 CO 113 CO CO
•
0
•
•
I
a
0
•
•
•
•
•
•
•
•
•
CC) CD tr) CV CS) 0 r -4 CV C.0 'Kr' C)
0,3 N N1 CV (Y) CO
ri
ri
Cc) CID CO N. 0 CN1 N. CO CS) CO h. LO CO
r
--1
•
-
la'
0
-t-i
0
I
1
1-1
(1)
0
CD
•
•
C)
•
•
•
ri
•
•
0 0 0
9
•
•
•
CV
•
N- N- c) LO (_0 (0 ri CS) 03 CO CO
CO CV CV
0 CV lr) N. CO 0 CS) CO 0
•
o
•
•
•
•
•
•
•
•
•
•
•
03 (3) 03 if) 0 N. f-... 0 0 (-0
N.
ri
0
ri
ri
'3' "nI' CN1 CV
r-1
I
I
I
i ri
c-.1 c...3
o
X
-,--)
O)
._,
(cia)$-, (1)
'0 (t)
a)
›., o-
co
up (s) N. co
al 0 C-)
0 0 0 C) 0 C) 0 0
0-1
,--I .!__,
r2; ?_.1
'-i
i G) 0
r -1
O cL1 4_, a -H—I
as) ro co a) (-o co co (Cl
0 - rf a:3
O
.---,
.1-, CL: CC)
CL, > > > > > > > >
102
000191
000`0TS i @nIPA
iaqaTuPTG adTd
an -reA
AT,TadoJd
Parsi
N-
C)
r
C) CO CO
•
•
•
1f)
(y)
•
IIMPrI
CO 0)
CO N.
umpa
•
•
Tan 2.15
-
pasee Teinl -eN
Jags1mITSTC1
JelPM buTsn
bullooD uop_e_TabT.Ija
Jag s em o ne wol ny
-
n--r
st.p. -es
CO
DD
C)
'Und Jo iBquinN
0
Q4
S-4
0 r—i
0 0 (D 0 CD
0 0 CD 0 CD
0
CD CD CD CD CD CD
0
„ „ . CD
CD
''-I
4-1 el) CD Ln c:D D if) c) cl)
M
,e72, 0 Q-.
c;
n--I
c...] CD CO 'C'
'
co
w ci
ti) .--1
CD 0
-,--1 ro
--I 1-1
M
r-I0
0 0 0 0 O'çrl -,-1
›
M
•--i (-0 0
}-., 0 w U)
O 0 0 '-C
(:=.-) res co >
O.,
a)
>1 ›, , / CD CD cp CD CD A E
C)
0
CD
CD
c)
a) ._, c ) _ ,-.4 w t
(C
1..)ta., WOCDCDOC) 0 (7)
DO (C .11,,
a)
I
... (ci
w G.)
- --)
/IN
,
-
• P
'
-4 .-Q
0.-n
e.I
p...,
ri
••
••
••
••
.11 CD LO U) CD
.• •--t
Lo
...
p ro
71 aj —1 —4 (N ro m rd ,.(,
> PD. (..) L.) ,--- cf) 0, m > <n- -cn -ÇA u) .up > U
>
0
,
,
-
- -
-
-
•
•
-
• N
•
103
:
.r c)
,
—I
X
Ri
E
ri (D
0
N. CO CV CV r--i rt Oj ri <7;ti ri 1-1 ri 1-1 r-I r--1 CD r-i ri
1-1
0
01) r-I ri
(7)
CD
0
CD
(1)
CO
<14
CO
(0
I
t-L4 -'.•-i CD CD 0 0 CD CD 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
LI)
0,714(...0 00 r-i
0
•
,4
•
•
co
•
r-,CO
•
•
•
I-0 co cn C5)
*di
dP co
•
•
•
CO r•-•
•
•
co co d'
•
•
•
CD 0)
•
CO
00
co
•
C.1 N. CV CD
•
•
•
•
r-i
CO
co o co ,-1 dr, Lo co (3-5 r 4 N. op LO co c.i CV N. r- 4 C3 Cs] cv cr) (.0 Lo •sr ,
-I 0 CO CV 0 rH 0 N. n--1 CV N. CV e-I CV CO r--i Csi Lo co co c'.1 co c) c)
(0
r
0•
•
•
•
.
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
CO CO
.--1
`
•
N
C,
.qi
1. 1
D
CO
NI
ri
1--1
1---I
4-, CO LC) 'Si' CO LO LO CO CO LO LID ri C') N. N. c3-) di d4 cy) co N. Lo .t-, co <zr
U) Lo Lc) co co co co o 03 r-4 CO O.) CD. C'') (7) LC)C''') CD LC) r—I LC) C) (-0 CD
•
•
•
(I)
H C') i--1 I-1 I
I
H
•
I
•
0
•
•
•
•
0
•
•
n-n I CV I CO ,-4 ri
I r-,I
I
I
I
•
•
•
N IS) CV
•
0
CD r-I I
(NI
N co 0 cr ,cv co •‹:1-, h,
cy)
CO CD CD di 0 CV CO
CUQLU
CU LUQr--1QN- C")CUQQ'CUQ N.
•
0
•
CV
•
•
a
•
•
•
O.]
ri
CD LC)
r
1-1-)
•
•
o
CD
Cr)
•
CD
•
0
•
-4
•
•
cn co co
O
al
---5
--r $-,
ai
•
•
•
CO
•
Cr)
•
714
•
•
•
•
CO C) ÇO
•
•
•
•
LO CO C5)
•
•
CO CO
r- 4ri r-1
r-i r--i
CO Up 0) CV CV c0 1.1) Lr) N. , zr, cr) Q N. 0
r-4
,--I CD CD CD ri ri 0 0 ri CV CV n-4
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
I
I
CO 0 CV CO N. .---n Lc.) up co di o d-' LO CO (0 N. CO CO n-n ,z11 CO r-i cp (0
.--i 0 (.0 r--1 CU LO crs N. --t co co c) as C') co (.0 LO 0 LO C') ,--i cv cr N.
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
0 CO (.0 CV 0) I
(.0
,71 CO LO CO CV CO CV CO CV r-i cri co r-i NI 1..0
CV
Cr) r--1
Ie-IINI`liCY)r-I
I C, ]
I
I
.--I
I
I
I
'--4-, ,-
✓
r-4
•
ri ri
i- 4
CD Q
•
•
CV CD 0) 0 CO
N. CS) CO 0 CV
0
714
ril
s...
.-
(1)
-I-, 4-1
al al
‘321 PQ
CC-1
5-n 4--,
5-,
(D (0
44.
W 0) ,''-'
---1
___, co CO
r.
rci t:rs
--.....,. --...„.
cn
ro
fir
P-t 0\1 - --1 .-Cj
0
-
O
CD
(D
ç:)
S.
CO
6)
E
C)
U)0
(I) r.r-i
E0
(l
(D u)
--I n
- -1-n ,a) .--1
,C4 O'-05
.4 . • r4 (f)
S. n
(
D 0
4-,
l.J a. 0
al (.0 >-n
s-, 0
-
-
Cn u)
I-I a)
Ca
I-1
Z15)
.1-1
t:5) Cf) E-I lar
ri
0
0
5-4 --I
› n
-0 (cl 5..n z_. ti) 4-4 4:_,'
- Cl,
0 u) -.0
0 E-I
0
(1) 0 1.--1 s--,
--i
.1-I
(f) :11
E-i
n
1-1 .1-I
i-H
0 0
4-,fti
>,
5-,
rci
ri
(1)
S-4
o
<5-4 S-I L-I
CY)
I. 5—I
0 aj • (7:3
76
:0
...-1
a5
n0.) $--n r-1 (D
a) 0 a) w w
'.-) 5 0 al c:}-4
cs al
0i,,(1)
:, >1,r ; -, , ,_,, (:_i 2, , , ,Q
.7 _,
,c : 5- _0-).,
Il»_, t-n'
0
—
rQ , r) ,-.4 ,-Q —(/ t
n, !
(:)._, _Q
!-_,
--'
,
_E. :5
L.;
- •
--,
1I
>
4-I
`4
z
4-I
0
0 4-1
i-H
O
,;..
.1- ', Z ,
f_4 .--.
.,1--1,;-.--
. (--.),
<C-4nr,:10
14o1-.:10U1-1U),5(i)O40-1
-
•
•
104
C)
* 1-1IO
X
0
(1)
(ti
rp
f
;D 0 0 0 r-1 0 0 0 0 0 0 0 0
.--i CV dl CO CO ln 'Kr' CO 0 CS) 0) <0
ro
-
•Sli r-I rH ri 1-1 r-I ri 6-1 ri
(15
•1-1
ro
d
0
.r4 (._0
cq
•
0.1 c-.1 .q -7:t ce) ce ) —1 r--1 ,--1
LO
C)C'tD
CD CO CS) L-0 CO 0
(V CO
0 CD CV
0 CD 0
C
• • • • • • • •
1--I
•
,
•
•
.
,
,
•
•
-
•
Lo cNi
co •çt'
-.,!-I
•
•
•
•
0
rt:$ 0 ---1 Ln
-I-, 4-
o
cn Q
00
Q)
•
•
0
0 (Y)
LI) r--1
•
0
H
CD NI
•
•
co
•
I
LO CO NI CS) N.. 1-1 N. 0)
(J) c0 ri NI CO CO
o
•
•
•
•
•
cs-) I
co co
f,-4
I
HI
rC5
7:1
0
--1
,(D)
0)
r-I CY)
•
•
Cr)
Cf)
•
41
r--1.q10),ZPOOCNO")1-01-1
r-i co a) CD NI C)
•
•
•
•
•
•
•
•
LO CY) 0 ri Na CO CO CD CO
ri Na C V
CYD CO
0
•
00 CO CO
(3) (-0 CO
Or) CO 0
rc,
DDDC' I
DD
•
•
•
I
•
•
•
I
•
f
•
•
•
•
•
•
f
o
-I-,
()
-1-I
U
-
4 4
0)0CSICT)C.01-1.)CY)CI)
NI 00 NI
ri NI.1iI '1-' ri LO CO 00 NI CO M
(D
(I)
•
•
I Ln
I
O
•
•
•
•
•
•
•
•
•
COLOOONICOOCO . 11 ri
'q' ri ri 1 N1 NI NI r-iii 0
r-I ri
r-1I
0
C)
..fl
--1
. rd--,
s_.
,—,a)
—
a)
ro
>
4-.1•- ,--1
O X
s._, a)
ni
a)
a) -d ai
>, o ---{ ,--4 (N co •71 t_ro
0 0 Q
a) a) a) C.)
-1 ::,' 7-1 -i
, a) a)
E-8
r----,
.-.-.-, r" - -1 7" -I
--
Q.,
--1.
0
9
(1
,
C)
---, ro - H CO Cri (C(. r(1:1 Fi
(21-1 IZ 12I21
Z
a-, > > > > >
-,
C)a,
-,--
,
)
(D
r.--. co
a) a) a)
71 r.-1
,D"
,-; .---i
(7:1 ru ro
> > >
105
0 0 0'oy$ o n m
000'0ZS
01 00'SIS onivA
vo'rto
J aAed Mg
enjeA Avodold
vary too,'
tood amepod
tuo;s/Cs buTptupds
•
C•1
uonefq.ul Jatuurns
01
01
P Ty IIMPq
UDto
C•1
CO
0, 01 •1•11'
`,111. CO.Cr
•
• •
adeospuvl paumwoo
C,1
U1
apCis adeospuel rocilo
Co CO
•
LIME, /
•
sr co
V In
Co
C1
IesodsTa a6vcpeD
1/1,
n
•
t
to
c n
latispmcpra
CO
CO
to
•
..)
•
to
C.
cv
c•-)
•cr
.
,..7,
e.. . .
04
01
111
CO
CO
.
•
aagsvm opouloinV
co co
t
..-I
OD CO
o UD
V 0.1,vo 04
CO V' CO
•
•
•
buTiooD uopeiabpja)3
CS CO
U1 CO
• •
Q
n
•
Co CO
to co
bullooD aAl3PIOd2A2
a
to
toLo,
•so
co
uonv6T1II - 10 1LITM
Co
V'
CO V)
CO
3a4au1v1a odid
I
CO
.-I
L.f3
lagsvm la6upm
•sr
goy JC.ipunv-1
.-1 Lf)
Lf)
0)
•
vo
stpvig c/z Jo loquinN
O
U1
a, 01 0
0 CO st.
•-•1
CO
stileg rind Jo aoqtuoN
•
• •
CO
St. O
V. 0
CO
CD
•
•
P0.0/ 4oi
D
c
',34,
at
'il
.,,
g
40
0:.
'S
t
c'
(1)
a
at
'0
'tS
E
0000
o
C1 C) CD 0
, c.., ... s
()
2 .v, 0:,-,' '.,1 _0 .
- •,
"oto
..a.OE,4, >,,.(24
.-',
,
-
Q
,0,---;,,,
,
>‘,3
?, ,
1 'OE1 -F:
,5 fe. 0 .0 .--, " C) , 0
:G• ,t,-; ',":ci ,:;- '7„' -,,n :
E .> ,--. vt,, t• ,1
E S 1::' f,'
Jo
O
.J. 3 ,'t 6 ,,
-
CD
"
"
C'''-r-6
.8
2 2 '73 .E:
C'
01 0 CO
C
CD 0 0 0 s, -2
E a' 4.-. — •-• .- ‘,.,- -,,,, 0 >
1—g? Q
p,
72,
',II
S ',,-; _
c,
c
o
:8 , - -','. • '3' ,J. .-L, d
K _9 „.; .,-:.,.; ,_: i.
•,:' '•- t'. -8 g
t:
3 ,',1.c', ,.`2 c`,-: c''
73
'21, 6 (S
.
.
v.
6: Z> ',-,-; ,:.> ',',', >'
-
▪
▪
• 1
106
CO
I
•••-1
X
CD
CO
al
Cll
CV
1-1 r-1 r-I r-I r-I r-I
-
c.1
a' E
..-n
r- i
Ti
CD
CD
r-I r-I r-I
a
a)
0)
r-i ,--I
0)
r-I
CV
O
r-I
CD CD CD CD CD CD CD CD CD CD CD CD CD rD CD CD CD CD CD <D CD C) CD
co N. a) CO (0 c..] CO N. V) uD 0 ) co (D It) .---1 cD cs.) It) Co ,-1 cr, cr)
Lo O co d-' .--i CO cV .q co .q Lf) 'Sr q cz) Lo .1 -1 .q, D 'Kr co co c.] ,--1 ,--1
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
• •
• •
.
r-I r,
a)
CO
0) c)
s-,
ai
Ti
C) LO ,--I r-I r--I 1-1 ,--1 ,--I .--I
,
.1
•5
0
,
,
,
cD
a1 0 ..-I
-1-J
,
r-t
Lf)
(0•V
4--.
cy)
CO 1:-) (t)
co
0) cp N. cq -q, CV N. N. CO CV N. cc) 0) co Co co r-I CD CO V CV a) `,31 ,44
(ci Cr) CD 'ql Co CD ,-1 CD N. .--i Co c.0 NI Co 1---I LO CV CV CV CO r-I .-I N. <D CD
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
a)
r-`1
Co G) ,--I
CO
r-i
CO
CD
It)
N.
CO
CO
U)
co
r--1
r
c) 10 0.)
c3-.)
co
1-1
00 N. r-1
CO CD Lo N. LO LO CS) c.C)
cD Co CO NI N. n- 4
(1) cDcOMNOOd'C0001-0CVG) • 1_0 CO (S:) (.0 cs)
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
1-1 r-I r4 r-I I
CV
0,1
I
I
E-1
E-iIIl
Cy)
C) CNI
c.0 c0
(.0
co cc) cy)
• CO C) Ni N. CO CS)
•
•
•
•
•
.
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
1-1 V)
S. CO CO CO C`,
CO a) v--1
Lf)
N.
LO
It)
Co
(
0 szr
CD OD
CO co co 1_0 ID •,:34 ,714 CO CO •1-1c0
c)
CD
C)
•
-
CO
co
-, (D up
NI Co NI CO NI (c)
NI CD 'q CD CD NI CD c0 CV •rçr CO cc)
•
•
•
•
•
•
•
•
•
•
•
•
(1)
LO LOCD
•
a) (0 CO G) CO
cr)
cD
CD ce) r--1
•
•
•
•
•
•
•
•
•
•
•
1
S-n
O
0
-I-'
0
o I
'il
(7) cr)
0', (D
CO CD cV cD 0) CD 0-) Co Co LO r-1 ,--I 09
`r' CD 1-0 1-0 N. (0 r-I CS) 'q' 'q' 1-0 1-1
•
•
- -
0
r0
41
44
e
•
I
0
C.)
I
•
/)
a)
0
2
o
-614
‘,. -Q1
RI
(cf
>
4--I
-,-I
.
a.)
O X
r-I
I
i
.---1
I
•
1' N-7 c0
,
co cr) co
r--1
I
00 ,--1
(0
(2) cD
(.0 .1..
0-.) ,_1
0`,1 Co
LO co
CO
c.„3 r_i
"
L0
c.] r-1
C..)
•
•
•
•
•
•
cV ,--1 ‘qt
(C) Co
,---1
r-i
r-4
r-I
S-t
O
.-.
Q.
S.
-1
a
)
ou
ai
c.)
S--, 4-,0
al Ri ai
i ca ca
---,
,-4 Co Co
5.,
(1) (cf
a) '-u)
C )0
-04
-1.--'
7:5
a) u)
-,-4 0 4-, 0 ,--1
-1-' .'-1 (r) 0 0
0
.-
'-'
0 Q_, Q)
(0 tj)
(ci ›, s.4
rd
CD
®'.'"`... -.' '''`.
Cf)
0) 0.1 C/D E-I
ri
s-, Q) - ri (1)
G-, (.1 r---1 ,CI (ci
(n-1
1
-,9 a)
> 121
4-1 41 4-1 4-1 ei
-d ro - r-I
0 Cf)
Q
tz 0
0 0
O
,--1 •
-1-1
cf)
(1) 7_, 1-1 -,-, ,_„
l4
›..,
r-1
0
rci --, C) Ri
s-, 0 s-, s..., s_.
W $. W
'7, .1.-1
".
(1) a) (i) (1) ._b CI) c°
.., ll,
0 ai
..Q ,C1 ..C1 ' 0 t=5)t3)
,Q
$-' .-Q4--)
›
'-Q
0 4_4
4 ,1 .-i ) Q-1 ,-4
47,.
a) ,_
o
Z ,_
I
I
CL,
a)
•
0
CD CO
.-.
CD
E
4
_,
f
(ii
•
•
•
t
L
•
$-1
•
c0
00 r-4 CO Cq 't-' co
Co NI 1-0 CO
I
.-
--4
•
•
•
•
CO 0) 0 CO Ln CO (0 LO •q' 1-0 CD (D
(
.-
e---1
O
Co
CD
6
Cr)
r--I
4
o
(.0
CD C)
Cs
1-4 1-1 LC)
g .!
A
.,5, Tl
rd rC1 J.,_i ai 4- , 0 fo D
Q.,
0 -, ›
,( r24 fl p_i 0 Z (.) '40 0 ,-4 cn
rd ,... D
Z ,1
o s, 4-, rd
S
0 7.!
O. `-'
0
,
a) a)
rc9 :a'
(Ti
t
P.,
ci) o
co Z a a,
CD
107
1-1 r-I
r-I r-I
CD CD CD CD CD CD CD
"KJ
cs) r--. CO
CV ,--I
Cdr-i
'0 I
5
ai
a)
•
•
'Kr U)
0 CD Cv)
-•-I
n
t.r) CO CD
Cr)
•
CY)
•
,--i
di
,zts
•
`14
•
0)
CY)
CO
NI
•
•
CN
•
CO
4---)-I-,
Cf)
•
as
CD ,--1 C5) G)
NI .-1 .---s c,D
•
•
r-I
•
•
CO
N..
'11
Od 00
Od
•
•
1-1
•
1
1- 1
•
0
•
0
•
0d dl (.0
NI 0 ud N
•
•
Ir-
•
I
•
I
•
TS
5-4
N.
al
(CI
.-i
40
I-I
0 `çii 'q'
.
.-n
0
.
0 N. CO CD CO CS) 'Zr
a) ud I----. r-I N. (.0 N
.
(.0
•
•
•
•
•
•
•
dl ud di c:. .-i r-I C)
0 oo co co co co oo
r-i r-i
ud o'd CO 'ST 'CD 0) .14
(D N 0 0 r---1
s....
0
r-1
NI
Cd
NI
•
•
•
•
•
•
$.4
•
•
•
•
III
o
C.)
4-,
0
-.-1
0
cs.) cq o
-)
co C.,
CO LI) r--1 NI
NI N. .-I CO
•
•
•
•
CY)
NI
(3)
00•
'T
0
r-1
OD
I -I
a)
N.
•
•
I I-0
CO
LI)
CY)
ITil r.Q
LI-I
,--1
't '
0 'ti NI
• • •
CDr-1
0
0
N
00
1111
I
-.,
0
+0
a)
s--, 0
Cd0
'0
0
X
731
! 0
1 fa4
al
›, 0 --I r-t N ced •st 4 ud (.0
(TS 0
C-) 0 a) 0) (1) 0) (2 )
a_i
Cd.----i
,__., C.) 0 :="
,---1
,-- -I •-n
1' •---n
CD ..--n aj r-- r,j '-- , _'J C-j C.:'
C24 i2
121-1 > > > > > >
108
CO
000'0T7
<anTeA
CO
000' ST
0 1 000'0T$ enT 2 A
cr
Jazeureta ed fd
-
enreA
Al_Jodald
(1)
(1)
LO
-1 GAPd
CO
CO
LO
co
<zr,
C)
Lo
•
CO CO
•
•
Cs4 *q..1
Cr)
•
0
CO
CO
cr.
cr.
CO
CD
CO
LO
Co
CO
•
0
Q-4
s..
o
o .--, 0
--1 a:t
0 n w
CO
0 0 Ti
(..)
r._
0 0 C) 0 CD
c) c) c) c) c,
(:)
, c, 0 0 0 0 cp
(
0
,---, . . . . . ,
LO CD cp
'-'
r-I (N CO CO
-..
cp ir) cp c)
a ,. ,.,
2ooooo2-2
Ira
43 cD> ncj)-1 3:3—
0>
o .._ ..-,
—
VI 0 (,1) 0 ›,
t
(CS " L." t:n "
"j
I-1
'
.
0 (CS "A
CO
•
(
\ / CD 0 0 0 0 {I} C:1'
v c, cp CD CD C, i'.
0 CD 0 C) CD 0 a; ,,
(i)
‘_- ,,
.---1 0 In LO CD LO ,—.I
o
w ai s " 0
4-•
_., Q) rcj ,----1 ,---1 C\I CO cO (0 0
•
-1.-I › (Ci 0 ID
,
--,
,,c n r.4 0 u ti, a_,
af.
>
"Cf3- -(1)- <r)" -Cr)- <f)" > O
109
c:, c)
I
.e--1
r-
0
ai
rH C)
-
c)
cl)
di N rH rH rH ri Cs.1 r-1 di .---I ,---1 rH rH rH r-1 CD rH rH r-i 0")
ri ri
Q
0)
•st
•Kr,
,
co
r-1
rH
(Cil
" i-I
i 0 0 0
CD
d'
0 0 0 0 CDOOCD0000000000000
1
co
Ti
ri Cr r-1 Cs] N., N. (..0 cr) cr) Lo
co Lo OD 1 0 d' d' 0 CO OD d, ,---I CO C.]
S--I
CO 0 CO di ,---4 co ,--i CO N di 'd' •qi co co ..---1 co . :1-, o. ) --1 di •q' cio N N
ai
•
d
•
•
•
•
•
•
•
•
•
_
•
•
•
•
•
•
•
0
•
•
•
•
.r-'Ci l 4 H ri CS)
Lo
0
'-' > 0
c..)
r--1
rH
iti Q) "r -1
-H
Cf)
n
QC')
CO
CV
r-1
,d
4-
'Sr 0 0 rl CO (C) CO rH 0) C--- 0 'Sr `Sri C''', CO 'Sr CO 0 N. 0 N. . ---1 CO 1-0
rH CD CO CD CNI c0 (*V r-1 1-1 Cr) ri CV 0) 0") 'Sr CV LO 0 0
)
e 0 syl 0,3 cp
(CS
0•
0
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Cr) C) r-1
'Srl
r--1
•
•
•
•
•
N,
0
r-1
0
L'..
CV
•
•
ri
(D
LO 0 ri 'Sr (3
r) CO 0 CS) Q OD LO CV CO ri 0 CV
LO
1-0 N.- d' LiD
cp
Co
cp co
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
r-1
rH 1
I c..]J
rH rHI
(
co
H
ri
Lo
•
cs.i —4
CV N,
•
•
LO
•
•
E-1
Li
al
1.:1
li
s_.4
rH c) (D (D (15--) c.-) CO 00 -1-1 N., c:, .---, cp ,---i NI N. Lo CV 1-0 1`, CO 00 CC) CO
CO CD 1-1 • 1-1 LI) LO LO CO r-1 C) 'SV N, CD 'Sri 'Sri '"1 'Sri 0 LO CV LO 'Sr1 r-I CV
•
OCO
•
•
1-1
4-1 H
•
•
•
•
•
0
•
•
•
•
•
•
•
.
C)
CO CO O'D C \I . q.' CO ÇO r-4 (V) 1-0 t', 1-0
C..,1 ri CO rH r-I ri rH ri CV CV CO ri CO
•
•
•
•
•
•
•
C..... C) 1-.-I C, Lo
CV ri rH
1,0 0, 1
ri) ri-1
O
c.1
Cr) rH CD 0 LO ri 1-1 CO 1-4 1-1 CO 1-0 <V '14 CO 0 CV 00 LO
N r-1 r-1 CD C) r-I CD CD rH (D 1-1 r--1 CO CL) 0
CO 0 0 0 0
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
o
•
I
-1
r
O
o
a'(3)
cr, CO
CV 0 0.) CO CO rH 1-1 CV CO CO CO CD CO CO CO C')
N. CD rH CC) OD (C) oD 1-0 CO 1-0 1-0 0 1-0 (...0 CO N... CO LO
•
•
•
•
•
•
•
y
•
•
•
•
e
•
•
e
•
•
•
•
0
•
•
•
•
'Sr' LO
ri LO C) ri ';Iiri CO Ci.) CV 'Sri LO
CO CD ICON-
1:1:11 ,-Q 'Si iLO
N.
CO ri
CV I CV ri rH CS) rH CV
rH
14-1
rH ri ri
I
III
Q)
I
0
4
—, cp ,t, Lo Lo c,
si-1
c::, (...0 .q co cr,
,
o
U) Cf) cn
a)
a)
71
-f:-51_,
-1-, +-n -1-1SH -1-i
0 al
a)
s-,
i
>
-ri
o x
fl-,
ô
L1-1 4-1 LH
0 0 0
s_.
a)
0
0
G),-Q
E
(D
14
\
,---I ,_Q
0.-n
(
LH
\
.--
H
ci)
(-) 0- 4 !C-13
al ti)
(.0
(r)
0 - -1
-r-i
n-,
0
0
0 -1-)
0 (f)
. if) r""
(1 )
,--4 CO CO
CO
Q) r-1
-8 ai
H ,_
CC) CC)
r-1
0
Sp,
s-,
OE:1
0 rc3 al
0
0
..,.,
s_,
0
ca
0) > n ,-,
(1)
Cf) H
r1-4 - !H
1-1
Z F_1
(ri
1-1
Q) U)
0
-1-1
0 4-1 (1) ri
4-' • ,-.1 Ca 0 0
(ci
ro
D
-.--1
›, $_, 0
ty) Ui H 0_,
..-1
--(1 ai 141__.
Ln 4-,
a) 0
(D e--1
s_., $_. $-, 1-,...
, , S-1 -1-1(CI
• ro
D al .--1
a) a) a) ._ w rCi-•,..-1
_-,
H
1-1
(J) '44-)
0
.___, S-r ,-,
-•-n '..s
W )-n
0
ti) E tp s o co ,..., CL) ,_, s, ,0
-(.-1 (Cç
(1)
,_Q . Q ,Q . o
4-3
Fil ( . . i n
' • ' "i C:1
H
L-n
E :.--: E 5i :"=»--, .,c_,-; ;) ,T4r1
lij
(3'
0 --I ',->
0
,77t- ,::'
73 .-r
1-41 Z F---1 '‘,.--, ‹: C-4 f---1- p--1 0 'K:- , (D 1--=.1 0 0 ,_1 u) >-- u) z', a
-
4,
0
CO
7.:i
-
r
:)
-'
--
-i
110
CD CD
CT) (Y)
0 CD 0
0 0 0 0 0 0 0 0
-Ci
00
LO C
'0
.5
(Ci(1)
•
ri
CO 0 (0 '3° N. CNQ 0 -Kr
•
•
•
•
•
•
•
•
•
•
•
or)
osi
CO
LC)
•
1- 1
r-I
ci
(ci
•
r:D r-1
W
"
c0
•
CO I-0
-
(ll Cq
r-1
•
c‘l 1_0
•
•
I
•
•
.
•
•
•
CO cq
I
CO 0
I
I
I
E-1
C (SD
CY) (0 N.
'Sr CO CO 0
c.)
°
ri 'Sr c 'Kr
0 0 cc)
r-1 0 0 0 c‘l
• • • • • • • • • •
CO
<0 CO 0 LO
c0 c‘a Q
.
I
I
rj
—1 0-)
•
0
•
•
•
cp (o
(o cr, co
•
•
•
a) C.)
LI) ri
•
•
•
•
c) co
c)
r
NI CO NI
co CN1 (NI
0) 0 r-f
CO
o
- cCi
a)
I
•
•
0
°
"ql
•
0 N. , zzr
C
•
I
•
•
rl
•
0.")
•
•
•
•
II
o
')r'
C
0
I
Lr)
0
nI CO
• • •
00 00 I-0 0")
• • • •
I co
N. 0')
rQ
CO 'Sr
I
i
r-I 03
r-i
CD CD
I
o
L.r) CO CO00
r--IC)
• •
0) CO
(JD
I
CO
I
o
0)
0
0
0)
7(51
(Ci
-1-1
U
CD
03
>
LI-1
0 ---I
CD
rd
CD
0
0 .-0 ro
›, 0 -(-4 __,
0 o n
X
ra,
r
co d-,
I-0 LO N.
a)
(0
>
4_,.-.
t..
C.) 0 0 0 (1) 0 OL) 0 CD
a) a)
, --, , n
7,4j --, ,
,=--_,, 4_,
;'
co rci rd Trid 0 (-0 _,_-,
0 a) --,-I (r) -,=1 ri
a_, o4 cci a > > > > > > > > la
7::,
----. -1--'
E
it,O F.',
6
'4
a)
$--r
)
-I-'
-.
,
P age Missing
in Original
Volume
112
baths, the number of trees, property value code, pipe diameter, and
whether or not property value exceeds $40,000.
Regression Equation
The form of the regression equation is similar to Equation 33,
where consumption is measured in hundreds of cubic feet (ccf) and the
names of the x terms and the values of the coefficients b1 are shown in
Table 30.
CHAPTER 5
DISCUSSION AND CONCLUSIONS
Perhaps the most notable finding of this study is the difference in
water consumption patterns between the residents of Rate Area 1 (inside
Tucson City limits) and Rate Area 2 (outside Tucson and South Tucson).
In Rate Area 1, the fixed cost of water service is $3.00 for the first 700
cubic feet per month plus $.20 ($.25 after February 1, .1970) per 100
cubic feet thereafter. In Rate Area 2, the fixed cost is $5.00 for the first
700 cubic feet per month plus $.29 ($.36 after February 1, 1970) per 100
cubic feet thereafter. Refer to Table 5. Thus, the fixed rate in Rate
Area 2 is 67 percent higher than in Rate Area 1, and the unit consumption
rate in Rate Area 2 is 35 percent higher than in Rate Area 1 (44 percent
higher after February 1, 1970). The percentage decreases in mean annual
consumption and the standard deviations from Rate Area 1 to Rate Area 2
are shown in Table 31. Notice that there is not only a decrease in consumption as the water price increases, but there is a decrease in the
variability in consumption (standard deviation) among the households as
in individual
the price increases. This suggests that there is a decrease
variation in water consumption habits as the price increases. The
correlation coefficients between the independent variables in this thesis
113
114
TABLE 31
PERCENTAGE DECREASE IN MEAN AND STANDARD DEVIATION
OF CONSUMPTION, RATE AREA 1 TO RATE AREA 2
Mean
Consumption
Standard
Deviation
Basic Data Set
-67%
- 8.6%
Subset 1
-23%
Subset 2
-36%
-10.2%
7 .5%
Subset 3
-18%
- 7.6%
and water consumption (and between the independent variables themselves)
are much higher for the Rate Area 2 customers than for the Rate Area 1
customers, as shown by all of the correlation matrices. In particular,
see Tables 6, 8; 12, 14; 18, 20; 24, 26. At the higher price for water,
more of the variation in water consumption was explained by the independent variables which were investigated than at the lower price. Thus,
as water rates increase, the accuracy of water consumption models which
are made by the techniques used in this thesis will increase, and so will
their usefulness as predictors of water consumption.
In Rate Area 1, water consumption was correlated significantly to
the property value and to the diameter of the pipe at the meter, and it was
slightly less significantly correlated to the number of bathrooms and the
number of trees. In Rate Area 2, water consumption was strongly correlated to the same variables mentioned for Rate Area 1; and in addition, it
was correlated to having an automatic clothes washer, dishwasher,
115
garbage disposal, lawn, the area of the lawn, a sprinkling system,
whether the lawn is watered in the summer, and whether the property is
worth more than $40,000. To a lesser degree, water consumption is
correlated to the number of people living in the house. Water consumption is negatively correlated to having an evaporative cooler.
CHAPTER 6
SUMMARY AND RECOMMENDATIONS
Projecting water consumption is vital to general urban planning,
especially in arid areas; some methods used in the past are inadequate;
and further investigation of water consumption in the Tucson area is
needed.
The classic supply-demand economic model shows that, in a
competitive market situation, the price and quantity of a good which will
be exchanged are determined by the simultaneous solution of two equations: a supply function and a demand function. This model changes for
the unregulated monopolist who chooses to maximize his profit (or minimize loss ): marginal cost and marginal revenue functions enter the model
and the quantity which will be produced and offered for sale is determined
by the simultaneous solution of these two functions and then the price at
which this quantity can be sold is determined from the demand function.
Most of the authors on water consumption stress the concept that
water is the same as other economic goods in that a demand function
exists for it, and that customers should be metered and billed according
to the amount they use, and that flat rate billing leads to waste. On the
other hand, flat rate, subsidized, or free water is justified in communities
116
117
or nations where sanitation and health are poor because the people do
not have ("can't afford") a potable water supply and adequate sewage
treatment. Water used for maintaining health, cleanliness, and sanitation flows through the residence--that it is used as a vehicle to carry
away bacteria, dirt, wastes, soap, and related chemicals which facilitate washing. Hence water supply and sewage treatment are inseparable
parts of one system.
Besides being used to maintain health and sanitation, residential
water customers also use water for lawns, gardens, swimming pools, and
washing paved areas and cars.
The ability to classify the consumption history of commercial
customers by type of business or industry is very important for developing
commercial water demand models; the City of Tucson Water Department
intends to do this in the future.
In order to develop empirical water demand models, the traditional economic demand function,
C = f(p)
(5)
is expanded to the function,
C = f(p, x 2 , x 3 , • • • x n )
(7)
where C is consumption, p is price, and x 2 , x 3 , ... , x n are non-price
variables. When the demand model is expressed in this form, and the
non - price variables are defined or identified, and data are gathered relating to observations of these variables, then multiple regression
118
techniques and electronic computers can be used to determine which of
the variables are most highly correlated to water consumption and then
these variables can be fitted into a consumption model. This was the
approach of the authors cited and of this thesis.
An attempt to gather data from existing sources was tried, but
this data could not be collated with water consumption. Then a questionnaire was sent to a simple random sample of customers of the Tucson
Department of Water and Sewers with the water bills; this resulted in a
set of observations of independent variables related to water consumption
which were collated with individual water consumption histories of the
customers in the sample.
Responses were received from about 54 percent of the sample,
were coded,
which is high for a mailed questionnaire. Of these, 1,320
in Rate
punched, and collated with water consumption. It was found that
Water Department) water consumpArea 1 (the lowest rate charged by the
diation was correlated significantly to property value and to the pipe
meter at the meter; water consumption was slightly less significantly
number of trees. In Rate
correlated to the number of bathrooms and the
the Water Department) water
Area 2 (the second highest rate charged by
just mentioned and
consumption was strongly correlated to the variables
automatic clothes washer, dishwasher,
also correlated to having an
sprinkling system,
garbage disposal, lawn, the area of the lawn, a
119
whether the lawn is watered in the summer, and whether the property
value is greater than $40,000. To a lesser degree, the water consumption is correlated to the number of people living in the house, and it is
negatively correlated to ownership of an evaporative cooler. It was found
that houses with evaporative coolers were in the low-value ranges and
usually lacked other water-using appliances and features, specifically
dishwashers, garbage disposal, multiple bathrooms, sprinkling systems,
and large lawn areas. The assumption that the number of water-using
appliances and features increases with property value is supported. As
the price for water increases, the mean water consumption per household
decreases, and so does the variability between households. Thus, as
water rates increase, it may be possible to make more accurate water
consumption models.
This study covered single family residences only. Supplementary
studies should cover water consumption by townhouses, apartments,
condominiums, businesses, industries, and public uses (parks, medians,
schools, hospitals). In agriculture, water productivity ratios in terms
of dollar value of the crop per unit of water used for each type of crop
are commonly made. Similar ratios for the businesses and industries in
Tucson should be computed.
There might be some cultural variables which affect water consumption significantly. Examples are: the percentage of time which the
120
residents spend at home, whether they do a lot of entertaining at home,
whether they eat many meals out, whether the family contains young
children, and attitudes or aesthetic values with respect to keeping a
landscape style suited to a humid climate. These should be investigated.
The water consumption data should be verified. This entails
locating the monthly billing printouts (the original block folios) and
checking the monthly consumption. Special attention should be paid to
the handling of turn-ons, turn-offs, transfers of ownership, over-reads,
and under-reads. Such errors that are located should be corrected. At
the same time, the consumption for November and December should be
picked up and added to the calculations.
When the November and December water consumption is obtained,
there is a full year's data on which seasonal studies can be made. Of
special interest should be the effect of seasonal variation on consumption
with respect to landscaping variables: style, lawn area, trees and shrubs,
lot size, etc.
Get property value and lot size data from the Pima County Tax
Assessor's magnetic tape file for a sample of the questionnaire respondents. Also, get plumbing data and swimming pool information from the
manual card files for the same sample of respondents. Check the Tax
Assessor's data with the questionnaire responses. Determine whether
there is a discrepancy; if there is a consistent bias, apply the bias as
121
a factor
to the questionnaire data and re-run the regression program. If
there is no consistent bias, get the property value and lot areas, etc.,
for
all the responses, make appropriate corrections to the data, and re-
run the regression program.
APPENDIX 1
FACSIMILES OF COVERING LETTER AND QUESTIONNAIRE
122
123
THE UNIVERSITY OF ARIZONA
Tucson, Arizona 85721
HYDROLOGY AND WATER RESOURCES OFFICE
In order to provide you with better water service in the future,
your Water Department and the University of Arizona are cooperating in
making a study of residential water usage in Tucson. Your participation
in this study will be both valuable and appreciated.
We ask that you take a few minutes to answer the questions on
the following sheet. Then return it to the University in the envelope
provided.
Of course, the information you provide will be kept strictly confidential. Information from individual questionnaires will not be released.
If, as a recipient of this letter, you are a landlord instead of a
tenant, please answer the questions as though you were the resident of
the household which received the water on the enclosed bill.
Many thanks for your kind cooperation.
Sincerely,
Leon N. Ray
Graduate Student in Water
Resources Administration
Facsimile of Covering Letter
124
UNIVERSITY OF ARIZONA - RESIDENTIAL WATER USE SURVE_
If the enclosed bill is not for a single family residence, check here (
and return.
1.
How many persons are living in the household?
2.
Please estimate the size of the lot?
3.
Please indicate the number of full and partial bathrooms:
Full baths (lavatory, toilet, and tub or shower)
2/3 baths (any two of the above)
1/3 baths (any one of the above)
4. Which of the following are in the house, connected and in use?
( ) Laundry tub
( ) Dishwasher
( ) Wringer washer
( ) Evaporative cooler
( ) Automatic washer ( ) Garbage disposal
( ) Refrigeration cooling usi_n_swater
5.
Please indicate the type of landscaping:
( ) Natural desert (cactus, desert shrubs, etc)
( ) Gravel
( ) Lawn
( ) Other 6.
If there is a lawn, please estimate the size.
Is the lawn watered in the summer? ( ) Yes ( ) No
Is the lawn watered in the winter? ( ) Yes
( ) No
Does the yard have a sprinkling system? ( ) Yes
( ) No
7.
How many watered trees and large shrubs are there?
(Don't count cactus or unwatered plants.)
8.
Is there a permanent swimming pool on the lot? ( ) Yes ( ) No
Is there a portable swimming pool on the lot?
( ) Yes ( ) No
If so, please estimate the size.
9.
Is the water supply supplemented by a private well?
10.
( ) Yes ( ) No
Please estimate what the house and lot would sell for today:
( ) less than $10,000
( ) $25,000 to $30,000
( ) $10,000 to $15,000
( ) $30,000 to $35,000
( ) $15,000 to $20,000
( ) $35,000 to $40,000
) $20,000 to $25,000
( ) more than $40,000
11. If the house is rented, indicate the monthly rent.
And who pays the water bill? ( ) Landlord ( ) Tenant
Thank You.
Facsimile of Questionnaire
APPENDIX 2
SAMPLE SELECTION AND ADMINISTRATION
OF THE QUESTIONNAIRE
This appendix describes some of the procedural arrangements and
mechanical steps concerning the printing, selecting the sample, mailing,
and handling the returns.
Printing
Printing and folding were done by the University's
Multilith and
Mimeo Bureau. Since tinted paper brings a higher response ratio than
white paper, a light yellow paper was selected for the questionnaire. The
covering letter and return envelope were white. The paper stock had to be
trimmed so that, after folding, it would fit the City's window-type billing
envelope.
Arrangements with the Cityof Tucson
A draft of the questionnaire form and the covering letter were
presented to Mr. Frank Brooks, Head of the Department of Water and
Sewers of the City of Tucson. He approved of the questions and covering
letter and gave permission to select a random sample from the Department's customers and to enclose the letter, questionnaire, and return
envelope with the water bills.
125
126
The water bills are in the form of an IBM card, which is produced
by the City's Data Processing Department. These cards are designed to
be mailed as a post card or to be enclosed in a window envelope with
other enclosures. Arrangements were made to intercept these cards as
they came out of the computer room but before they went to the Mailing
Department. Arrangements were made for work space.
Selection of the Sample and Mailing
Water bills are mailed to each customer monthly on a' cycle system
with twenty-one billing days in each cycle. To cover the Water Department's complete service area each of the twenty-one billing days had to
be sampled.
A typical day's billing consisted of a little more than two boxes
of IBM cards. Compressing the cards as tightly as possible, marks were
made every quarter inch across the top edges of the cards. Then a card
was drawn at random between each mark. If the name to whom the bill
was addressed was that of a business, the next card in sequence which
was a person's name was selected. The customer numbers on the main
survey were checked against the list of numbers in the pre-test; those in
the pre-test were eliminated from the main survey. (The name of the
mayor, which came up in the random drawing, was also skipped.)
The selected cards were taken to the computer room and reproduced; the reproduced cards were interpreted. These cards then became
127
a record of the questionnaires mailed and the input into the Water
Consumption History File retrieval program. This program was written
with consultation of one of the City's programmers.
The customer number from each billing card was copied onto the
questionnaire so that the responses could be collated with the water consumption when the responses were returned. The envelopes were stuffed
with the bills, the questionnaires, covering letters, and return envelopes
and then taken to the mailing room.
The frame from which the sample was drawn was the group of
Water Department customers whose bill mailing dates were from May 27,
1970 to June 24, 1970 (meter-reading dates from May 18, 1970 to June 17,
1970), and whose accounts were current. Customers whose accounts
were in arrears did not receive questionnaires. It was felt that they
would produce a very low response; also, their bills received special
handling and were sent out with dunning notices. About ten per cent of
the customers are in arrears. The mailing dates for the frame for the pretest were May 8, 1970 and May 9, 1970 (meter reading dates May 1,
1970 and May 2, 1970).
Handling the Returns
In opening the envelopes, special care was taken to check for
payments which might have been enclosed, so they could be taken to the
Water Department. Surprisingly few payments were enclosed, and only
128
two people sent payments without the questionnaire. Responses with
marginal notes were trimmed to fit three-ring binders, into which they
were put before coding.
APPENDIX 3
CODING OF THE DATA
A coding form was designed onto which the responses to the
questions were transcribed from the survey forms. The format which was
designed for the cards was, in Fortran notation (F8., 1X, Fi., 2F2.,
2F4., 14F1., 2F3., 3F1., F2., 2F1., F2., F3., 2F1., F3.., 2F1.), which
comprises 61 columns. All of the information from one questionnaire fits
onto one line of the coding form and onto one card. As the information
from the questionnaire was transcribed, those forms which had conflicting or questionable responses, marginal notes or letters, or lot size
given in acres were flagged on the coding forms (column 61) and these
questionnaires were separated from the rest of the questionnaires. Later,
these flagged questionnaires were sorted by customer number for future
reference.
Since most people were expected (correctly) to respond to lot,
lawn, and pool sizes with dimensions (length by width) instead of acres
or square feet, the coding form was designed so that two dimensions for
each of these items could be entered. When responses in acres or
square feet were encountered, they were factored into pseudo-dimensions
easily
.
129
130
General Coding Procedures
Generally, the responses were coded as literally as possible to
the way they were answered, rather than by making interpretations as to
what the respondent really meant to say. If the actual responses did not
seem to be what the respondent seemed to have intended, the coding sheet
was flagged with a "2" in column 61. A "no-response" to any question
was coded as zero (left blank). A "yes" was coded as "1" and a "no"
was coded as "2." Thus, the yes-no questions could have three possible
responses: 0, 1, or 2. When respondents wrote in "sometimes" to such
yes-no questions, it was coded as "1" (yes). For the appliance and
landscape style questions, where the appropriate response was a check
mark, a mark was coded as "1" and no mark as "0" (blank). A few
respondents indicated that they had two or more evaporative coolers or
laundry tubs; in these cases, the count was entered on the coding sheets.
Lot Size: The Commercial Acre Problem
Several respondents gave their lot size in acres, as was anticipated. Since there are 43,560 square feet in an acre, the number of
acres was converted into pseudo-dimensions by factoring by ten, multiplying the number of acres by ten and entering these factors on the keypunch data sheet.
Some respondents, however, specified the unit "commercial
acre" on their questionnaire, implying a unit which is less thi In 43,560
-
131
square feet. In Pima County there is no legal definition of "commercial
acre" and the County Recorder's Office does not give any official recognition of this term, and therefore there exists no official opinion as to how
many square feet a commercial acre should contain. Some people in the
office do not regard the use of the term "commercial acre" by real estate
salesmen as being an ethical practice. In the Pima County Tax Assessor's
records, the sizes of all lots are recorded officially as true acres to the
nearest hundredth of an acre and also in length and width to the nearest
hundredth of a foot.
Real estate salesmen and land developers define a commercial
acre as a true acre minus easements for streets, alleys, utilities, sidewalks, and perhaps other things. Thus, the actual size of a commercial
acre is as variable as the ethics of the individual real estate salesmen
and the lots which they are currently selling. It is likely that these
easements could range from fifteen to twenty per cent of a true acre.
In coding lot sizes it was decided to assume a hypothetical size
of 38,000 square feet, which is about 87 per cent of a true acre, as the
definition of "acre" for all respondents.
There were 1,076 respondents who answered the lot size question,
which was 81.5 per cent of the total respondents. Of this number, 50
specified
or 4.6 per cent specified "acre" as a unit and 8 or .7 per cent
"commercial acre" as a unit.
132
Lawn Size
There was no difficulty in coding lawn size, as all those who
gave a specific size used length and width in feet. Many people checked
the box indicating that they had a lawn, but did not give a size. Very
few people gave a descriptive response, such as "small." The size for
such responses was coded as zero, with a flag-code of 11 2."
Pool Size
For people with permanent pools, the responses and coding thereof was the same as for lawn size. One person gave a volumetric answer;
a surface area was estimated by assuming an average depth of five feet,
and a flag-code of "2" was assigned.
There were 76 respondents who reported having a permanent pool,
which was .57 per cent of the total respondents. Of these, 54 or 71 per
cent reported a numeric size.
Because of their construction techniques, portable pools are
circular. Therefore, it was assumed that the dimensions provided by the
respondents were diameter and depth. In all cases the larger of these
dimensions appeared to be a reasonable diameter, so it was used to
compute an area, which was factored and entered on the coding sheets
as a pair of pseudo-dimensions.
There were 54 respondents who reported having a portable pool,
which was .41 per cent. Of these, 48 or 89 per cent reported a numeric
size.
133
Rented Property
The two questions which were specifically concerned with rented
property were conditional questions; that is, "no response" was expected
for both of these questions if the property was not rented.
These questions were coded independently as answered. A "1"
indicates the landlord pays the water bill and a "2" indicates that the
tenant pays the bill.
There were 148 respondents who indicated non-zero rent, which
is 1.12 per cent of the total respondents.
LIST OF REFERENCES
Alchian, Armen Albert and Allen, W. R. 1967. University Economics:
Belmont, California, Wadsworth Pub. Co.
Boulding, Kenneth. 1968. A Data-Collecting Network for the Sociosphere: Impact of Science on Society, Vol. 18, No. 2.
Bruner, John Marston. 1969. An Analysis of Municipal Water Demand
in the Phoenix Metropolitan Area: Ph.D. Dissertation, Arizona
State University, Tempe, Arizona.
Clausen, George Samuel. 1970. Optimal Operation of Water-Supply
Systems: Ph.D. Dissertation, The University of Arizona, Tucson,
Arizona.
Efroymson, M.A. 1960. Multiple Regression Analysis: Mathematical
Methods for Digital Computers, Anthony Ralston and Herbert S.
Wilf, eds. , New York: John Wiley & Sons.
Federal Housing Administration. 1965. Minimum Design Standards for
Community Water Supply Systems: Federal Housing Administration
Report No. 751, Washington, D.C.
Gum, Russell (undated) Computer program to perform multiple regression analysis: Department of Hydrology and Water Resources,
The University of Arizona, Tucson, Arizona.
Haney, Paul D. and Hamann, Carl L. 1965. Dual Water Systems:
J. American Water Works Assoc., Vol. 57, No. 9.
Hanke, Steve H. and Flack, J. Ernest. 1968. Effects of Metering on
Urban Water: American Water Works Association J., Vol. 60,
No. 12.
Howe, Charles W. and Linaweaver, F. P., Jr. 1967. The Impact of
Price on Residential Water Demand and Its Relation to System
Design and Price Structure: Water Resources Research, Vol. 3,
No. 1. Reprinted by Resources for the Future, Inc., Washington,
D.C.
134
135
Huszar, Charles K. (undated) A Generalized Stepwise Multiple
Regression Analysis: University Computer Center, The University
of Arizona, Tucson, Arizona.
Jacobs, James Jerome. 1968. An Economic Supply Function for the Diversion of Irrigation Water to Tucson: M.S. Thesis, The University
of Arizona, Tucson, Arizona.
Lancaster, Kelvin. 1969. Introduction to Modern Microeconomics: Rand
McNally and Co., Chicago, Illinois.
Linaweaver, F. P., Jr., Geyer, J. C., and Wolff, J. B. 1966. Final
and Summary Report on Phase 2: Department of Environmental
Engineering Science, Johns Hopkins University, Baltimore, Md.
Reid, G. W. 1965. Projection of Future Municipal Water Requirements:
Southwest Water Works Journal, 46:18.
Samuelson, Paul A. 1961. Economics, An Introductory Analysis: McGrawHill Book Co., New York.
Zeizel, Eugene P. July, 1968. Progress Report on Econometric Model of
Water Consumption in the Tucson Metropolitan Area Memorandum
4 9, Operations Research Study Group, Hydrology and Water Resources Office, The University of Arizona, Tucson, Arizona.
. September, 1968. An Econometric Model of Future Water
Consumption for the City of Tucson, Arizona: Memorandum #10,
Operations Research Study Group, Hydrology and Water Resources
Office, The University of Arizona, Tucson, Arizona.
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement