LSAC Data User Guide 2103 - Growing Up in Australia: The

LSAC Data User Guide 2103 - Growing Up in Australia: The

The Longitudinal Study of Australian Children:

An Australian Government initiative

Data User Guide

November 2013

CONTENTS ........................................................................... ERROR! BOOKMARK NOT DEFINED.

1 ABBREVIATIONS .......................................................................................................................... 4

2 ACKNOWLEDGEMENTS AND CITATION .............................................................................. 6

3 INTRODUCTION ........................................................................................................................... 7

4 WHAT IS LSAC? ............................................................................................................................ 8

4.1

O

BJECTIVES

................................................................................................................................ 8

4.2

W

HO IS INVOLVED

? .................................................................................................................... 8

4.3

T

IMELINES

.................................................................................................................................. 8

4.4

S

TUDY INFORMANTS

................................................................................................................... 9

4.5

M

OTHER

/F

ATHER DATA

.............................................................................................................. 9

5 INSTRUMENTS ............................................................................................................................ 10

5.1

C

HILD ASSESSMENTS

................................................................................................................ 13

5.2

R

ESPONSE RATES

...................................................................................................................... 16

6 THE LSAC DATA RELEASE ..................................................................................................... 20

7 FILE STRUCTURE ...................................................................................................................... 21

7.1

M

AIN DATASET

......................................................................................................................... 21

7.2

S

UPPLEMENTARY FILES

............................................................................................................. 24

7.3

Q

UESTIONNAIRE VARIABLES

..................................................................................................... 33

7.4

D

ERIVED VARIABLES

................................................................................................................ 37

7.5

S

TUDY

C

HILD

H

OUSEHOLD COMPOSITION VARIABLES

............................................................. 38

7.6

PLE H

OUSEHOLD COMPOSITION VARIABLES

............................................................................ 41

7.7

A

GE INVARIANT INDICATOR VARIABLES

................................................................................... 42

7.8

I

NDICATOR VARIABLES

............................................................................................................. 42

7.9

V

ARIABLE LABELLING CONVENTION

......................................................................................... 43

7.10

M

ISSING VALUE CONVENTIONS

............................................................................................... 44

8 DOCUMENTATION..................................................................................................................... 45

8.1

M

ARKED

-

UP INSTRUMENTS

....................................................................................................... 45

8.2

F

REQUENCIES

........................................................................................................................... 46

8.3

D

ATA

D

ICTIONARY

................................................................................................................... 46

9 DATA TRANSFORMATIONS .................................................................................................... 50

9.1

T

RANSFORMATIONS TO ENSURE CONSISTENCY

......................................................................... 50

9.2

T

RANSFORMATIONS TO UPDATE INFORMATION

........................................................................ 50

9.3

S

UMMARY MEASURES FOR SCALES

............................................................................................ 50

9.4

O

UTCOME

I

NDEX MEASURES

..................................................................................................... 51

10 CONFIDENTIALISATION ....................................................................................................... 52

10.1

I

N

-

CONFIDENCE DATA

............................................................................................................. 52

10.2

G

ENERAL RELEASE DATA

........................................................................................................ 52

11 DATA IMPUTATION ................................................................................................................. 53

12 SURVEY METHODOLOGY ..................................................................................................... 55

12.1

S

AMPLE DESIGN

...................................................................................................................... 55

12.2

D

EVELOPMENT AND TESTING OF SURVEY INSTRUMENTS

........................................................ 56

12.3

D

ATA COLLECTION

................................................................................................................. 57

12.4

F

IELDWORK RESPONSE

............................................................................................................ 62

13 IMPORTANT ISSUES FOR DATA ANALYSIS ..................................................................... 64

13.1

W

EIGHTING AND EXTERNAL VALIDITY

................................................................................... 64

13.2

U

NIT OF ANALYSIS

.................................................................................................................. 71

13.3

A

GE AT INTERVIEW

................................................................................................................. 71

13.4

T

IME BETWEEN INTERVIEWS

................................................................................................... 73

13.5

C

ROSS

-

COHORT COMPARISONS

............................................................................................... 73

13.6

S

AMPLE CHARACTERISTICS

..................................................................................................... 76

LSAC Data User Guide – November 2013

2

14 USER SUPPORT AND TRAINING .......................................................................................... 78

14.1

O

NLINE ASSISTANCE

............................................................................................................... 78

14.2

G

ETTING MORE INFORMATION

................................................................................................ 78

15 REFERENCES ............................................................................................................................ 79

16 BIBLIOGRAPHY ........................................................................................................................ 80

LSAC Data User Guide – November 2013

3

1 Abbreviations

ABS – Australian Bureau of Statistics

ACARA – Australian Curriculum, Assessment and Reporting Authority

ACIR – Australian Childhood Immunisation Register

AEDI – Australian Early Development Index

AIFS – Australian Institute of Family Studies

ANU4 – Australian Nation University ranking of occupational prestige, 4 th

edition

ASCL – Australian Standard Classification of Languages

ASCO – Australian Standard Classification of Occupations

ASGC - Australian Standard Geographic Classification

ATSI – Aboriginal and Torres Strait Islander

BMI – Body Mass Index

CA – Carer Allowance

CAI – Computer Assisted Interview

CAPI – Computer Assisted Personal Interview

CATI – Computer Assisted Telephone Interview

CBC – Centre-Based Carer

CCB – Child Care Benefit

CSR – Child Self Report

DFRDB - Defence Forces Retirement and Death Benefits Scheme

DSP – Disability Support Pension

DSS – Department of Social Services

DVA – Australian Government Department of Veterans’ Affairs

F2F – Parent 1 Face-to-Face Interview

FCF – Family Contact Form

FDC – Family Day Care

FDCQA – Family Day Care Quality Assurance

FTB – Family Tax Benefit

FTBA – Family Tax Benefit A

FTBB – Family Tax Benefit B

GPS – Global Positioning System

HBC – Home-Based Carer

IVF – In-Vitro Fertilisation

LDC – Long Day Care

LOTE – Language Other Than English

LSAC Data User Guide – November 2013

4

LSAC – Longitudinal Study of Australian Children

MBS – Medicare Benefit Scheme

MSN – Medicare Safety Net

MR – Matrix Reasoning test

NCAC – National Childcare Accreditation Council

NILF – Not In the Labour Force

NSA – Newstart Allowance

OMR – Optical Mark Recognition

OSHCQA – Outside School Hours Care Quality Accreditation

P1D – Parent 1 During Interview Questionnaire

P1L – Parent 1 Leave-Behind Questionnaire

P1SC – Parent 1 Self-Complete Questionnaire

P2L – Parent 2 Self-Complete Questionnaire

PBS – Pharmaceutical Benefit Scheme

PLE – Parent Living Elsewhere

PPVT – Peabody Picture Vocabulary Test

PPVT-III – Peabody Picture Vocabulary Test, 3 rd

Edition

QIAS – Quality Improvement and Accreditation System (for Long Day Care centres)

RSE – Relative Standard Error

SACC – Standard Australian Classification of Countries

SEIFA – Socio-Economic Indexes For Areas

SRS – Simple Random Sample

TIS – Telephone Interpreter Service

TUD – Time Use Diary

WAI – Who Am I?

WISC – Wechsler Intelligence Scale for Children

LSAC Data User Guide – November 2013

5

2 Acknowledgements and citation

The current version of the LSAC Data User Guide has been updated by AIFS

The Wave 5 data files were prepared by the ABS and AIFS

Readers wishing to cite this document should use the following form of words:

Australian Institute of Family Studies (2013) Longitudinal Study of Australian

Children Data User Guide – November 2013, Melbourne.

LSAC Data User Guide – November 2013

6

3 Introduction

This Data User Guide is designed as a reference tool for the users of the Growing Up

in Australia: the Longitudinal Study of Australian Children (LSAC) datasets.

It aims to cover all of the things you need to know to use the LSAC data – such as an overview of the survey methodology, an outline of the file structure, variable naming conventions, and issues data analysts need to be aware of.

The following documentation is also useful to data users and is available on the study website, http://www.growingupinaustralia.gov.au/index.html:

• Questionnaires and interview specifications marked with variable names

(including mock questionnaires for Computer Assisted Interview (CAI) instruments)

• Data Dictionary

• Technical Papers on weighting, non-response and other issues

• Data issues papers.

Feedback from data users suggests that browsing the marked questionnaires and interview specifications is often the best way to understand the breadth of information available in the study and find sections relevant to the proposed research topic.

Please read the ‘Important issues for data analysis’ section carefully. This section outlines aspects of the sample design that have important implications for interpreting analyses from the study.

Other information relevant to data users is contained in the ‘Data users information’ pages on the website.

We welcome any feedback you have about this Data User Guide. If there is something that you expected to find in the manual and didn’t, or if you had difficulty understanding any section, please let us know by emailing [email protected]

LSAC Data User Guide – November 2013

7

4 What is LSAC?

Growing Up in Australia: the Longitudinal Study of Australian Children (LSAC) aims to examine the impact of Australia’s unique social and cultural environment on the next generation. The study will further understanding of child development, inform social policy debate, and be used to identify opportunities for intervention and prevention strategies in policy areas concerning children and their families.

4.1 Objectives

LSAC explores family and social issues, and addresses a range of research questions about children’s development and wellbeing. Information is collected on the children’s physical and mental health, education, and social, cognitive and emotional development, from parents, child carers, pre-school and school teachers and the children themselves. Its longitudinal structure enables researchers to determine critical periods for the provision of services and welfare support and identify the long-term consequences of policy innovations (see LSAC Discussion Paper No.1, ‘Introducing the Longitudinal Study of Australian Children’ for more details).

LSAC aims to provide a database for a comprehensive understanding of children’s development in Australia’s current social, economic and cultural environment. LSAC is delivering the first ever comprehensive, national Australian data on children as they grow up.

4.2 Who is involved?

Growing Up in Australia: the Longitudinal Study of Australian Children is conducted in partnership between the Department of Social Services (DSS), the Australian

Institute of Family Studies (‘the Institute’) and the Australian Bureau of Statistics

(ABS), with advice provided by a consortium of leading researchers known as the

LSAC Consortium Advisory Group.

The Wave 1 data collection was undertaken for the Institute by Colmar-Brunton

Social Research and I-view/NCS Pearson, private social research companies. Data collection for Waves 2, 3, 4 and 5 was undertaken by the Australian Bureau of

Statistics.

4.3 Timelines

Development work for the study commenced in March 2002 with the testing phase continuing through 2003, involving over 500 families. Recruitment to the study of over 10,000 children and their families took place from March until November 2004.

From 2004, the families have been interviewed every two years. In addition, betweenwaves mail-out questionnaires were also sent to families in 2005, 2007 and 2009

Sample design

The focus of the study is on the developmental pathways of Australian children.

Therefore the child is the sampling unit of interest. A dual cohort cross-sequential design was employed, as shown below.

Cohort

B cohort

Wave 1

0-1 years

Wave 2 Wave 3

2-3 years 4-5 years

Wave 4 Wave 5

6-7 years 8-9 years

LSAC Data User Guide – November 2013

8

K cohort

4-5 years 6-7 years 8-9 years 10-11 years 12-13 years

The two cohorts of children were selected from children born in a 12-month period:

• B cohort (infant cohort): children born March 2003 - February 2004

• K cohort (child cohort): children born March 1999 - February 2000

Further information about the design of the sample is available in the ‘Survey

Methodology’ section of this guide, and in LSAC Technical Paper No. 1, ‘Sample

Design’ (available from the study website, http://www.growingupinaustralia.gov.au/pubs/technical/index.html).

4.4 Study informants

The study collects data from multiple informants:

• Parent 1 (P1) is defined as the parent who knows the Study Child best; in most cases this is the child’s biological mother.

• Parent 2 (P2) is Parent 1’s partner or another adult in the home with a parental relationship to the Study Child; in most cases this is the biological father, but step-fathers are also common.

• The Study Child themselves.

• Parent Living Elsewhere (PLE) is a parent who does not live with the Study

Child; this is most commonly the biological father after separating from the biological mother. This collection was started in Wave 2.

• Teachers and childcare workers.

In addition, data are linked to the file from the National Childcare Accreditation

Council, Medicare Australia, the Australian Bureau of Statistics and the National

Assessment Program – Literacy and Numeracy (NAPLAN).

4.5 Mother/Father data

While Parent 1 is usually the mother and Parent 2 is usually the father, this is not always the case. However, many users prefer to analyse the data by parent gender,

(i.e. Mother and Father rather than Parent 1 and Parent 2). Therefore all the variables collected for both Parent 1 and Parent 2 are presented as Mother and Father variables as well. It should be noted that Parent 1 and Parent 2 may be the guardians of the child and not the child’s biological parent, so in this context Mother should be taken to mean ‘female parent/guardian’. Sometimes Parent 1 (and/or Parent 2) might change between waves. For instance, Parent 1 may be female in both waves but different people.

If there are two female parents, Parent 1 is coded as Mother and Parent 2 is coded as

Father. This will be maintained if the parents swap who Parent 1 and Parent 2 are in subsequent waves. This means that there are a small number of female Fathers that analysts should be mindful of when working with these variables.

LSAC Data User Guide – November 2013

9

5 Instruments

The following table summaries the data collection instruments used in each wave.

Table 1. Data collection modes by wave.

Questionnaire Mode Completed by

Indicator

Variable

W1 W2 W3 W4 W5

Face-to-Face

Interview (F2F)

Face-to-Face

Interview (F2F)

Parent 1 during interview (P1D)

Parent 1 during interview (CASI)

Paper

Computer

Paper

Computer

Parent 1 Leave behind

(P1L)

Paper

Parent 2 Leave behind

(P2L)

Paper

Computer Child Self Report

(CSR)

Audio Computer

Assisted Interview

(ACASI)

Computer

Paper

Parent 1

Parent 1

Parent 1

Parent 1

Parent 1

Parent 2

N/A

N/A

[*]p1dd

[*]p1dd

[*]p1scd

[*]p2scd

Study Child [*]csrd

Study Child Need consent from:

P1 [*]id40e &

SC [*]id40f

Parent 1 N/A

BK -

-

BK BK BK -

-

BK BK BK -

BK BK BK BK BK

-

-

- -

BK BK BK BK

-

K

-

-

K

-

BK BK

B

K

BK BK BK -

-

-

-

BK

K

- Time Use Diary

(TUD)

Time Use Diary

(TUD)

- - - K K

Parent Living

Elsewhere (PLE)

Parent Living

Elsewhere (PLE

CATI)

Home-Based Carer

(HBC)

Centre-Based Carer

Computer Study Child Need consent from:

P1 [*]id40i &

SC [*]id40j

PLE [*]plescd Paper – mailed out

Computer/Te lephone

PLE [*]plescd

Paper

Paper

Carer

Carer

[*]hbccbc

[*]hbccbc

-

-

B

B

BK -

-

B

B

-

-

- -

BK BK BK

-

-

-

-

LSAC Data User Guide – November 2013

10

(CBC)

Teacher Questionnaire

(TQ)

Paper

Physical

Measurements (PM)

Computer

Teacher [*]tcd K K BK BK BK

Who Am I (WAI)

PPVT Assessment

(PPVT)

Matrix Reasoning

(MR)

Study Child Blood

Pressure (BP)

Computer

Computer

Computer

Study Child Need consent from:

P1 [*]id30d &

SC [*]id30e

BK BK BK BK BK

Study Child cid44a1

Study Child [*]ppvtd

K

K

-

K

B

BK

-

B

-

B

Study Child [*]id44a1 - K K BK B

Computer Study Child Need consent from:

P1 [*]id47a &

SC [*]id47b

-

Interviewer

- - K K

BK BK BK BK BK Interviewer

Observations (IOBS)

Computer

NB:

1. The indicator variable can be used to see if data is present or not for a particular instrument (for more information see sections 7.8 & 7.9).

2. The [*] in the indicator variable should be replaced by the age indicator (a,c,d,e,f or g) as discussed below.

The following methods are used to collect study data.

• The Face-to-Face Interview (F2F) is conducted with Parent 1 (although in Wave 1

Parent 2 could complete some sections if this was more convenient). This component is undertaken with all participating families at a wave. Some interviews might be completed over the telephone in full and refer to p.56 (remote areas).

• The Parent 1 During Interview Questionnaire (P1D) consists of self-complete items for which it was considered important to achieve high response rates. In

Wave 4 it became a Computer Assisted Self-Interview (CASI).

• The Parent 1 Leave-Behind Questionnaire (P1L) consists of lower priority selfcomplete items. Efforts are made to obtain this data from Parent 1 while the interviewer is in the home. This form became part of the CASI.

• The Parent 2 Leave-Behind Questionnaire (P2L) consists of self-complete items.

Efforts are made to obtain this data from Parent 2 while the interviewer is in the home. If this is not possible the questionnaire is left for completion at a later time.

• Child Self-Report Interview (CSR) consists of items answered by the Study Child

For children younger than 10 years old it is administered by an interviewer. For

LSAC Data User Guide – November 2013

11

children 10 years old and older it is administered via Audio Computer-Assisted

Self Interview (ACASI). As part of the interview, physical measurements are taken and other assessments (such as measures of cognition or achievement) administered to the Study Child.

• The Study Child completes an Audio Computer Assisted Self Interview (ACASI) by themselves. This method allows sensitive content to be answered by the child in total anonymity.

• The Time Use Diary (TUD) documents a 24-hour period of the child’s life. In

Waves 1, 2 and 3, the child’s family were asked to complete two TUDs, one for a weekday and one for a weekend day. A different procedure was implemented in

Wave 4. In Wave 4, the Study Child (K cohort only) was asked to complete one

TUD. A TUD form with instructions on how and when to fill it in was sent to the study child prior to the interview. The study child was asked to fill in the TUD form on the day before the interview date. The next day during the interview the interviewer asked the child to describe “yesterday” using the TUD form. The day the diary referred to could be any day of the week depending on when the interview was scheduled.

• The Parent Living Elsewhere Questionnaire (PLE) was first included in Wave 2 as a mail-back questionnaire. In Wave 3 it became a Computer Assisted Telephone

Interview (CATI).

• The Home-Based Carer Questionnaire (HBC) is for children aged 0-1 and 2-3 years who receive childcare in a home environment, most commonly from a grandparent.

• The Centre-Based Carer Questionnaire (CBC) is for children aged 0-1 and 2-3 years who receive childcare from long day care programs in centres, schools, occasional care programs, multi-purpose centres and other arrangements

• The Teacher Questionnaire (TQ) is for children aged 4-5 years and older who attend a school or, for some 4-5 years olds, a preschool or long day care centre.

• Interviewers make observations (IOBS) with permission of the respondent about the interview, state of the house (where the interview was conducted) and the neighbourhood characteristics of where the respondent lives.

• In Wave 1 the Australian Early Development Index (AEDI) was included as a nested study, which involved the AEDI questionnaire being sent with the K cohort

LSAC Teacher Questionnaire in Victoria, Queensland and Western Australia. The

AEDI is a community-level measure of young children's development based on a teacher-completed checklist. It consists of over 100 questions measuring five developmental domains: language and cognitive skills; emotional maturity; physical health and wellbeing; communication skills and general knowledge; and social competence. More information on the AEDI can be found on the following website: http://www.rch.org.au/australianedi/edi.cfm?doc_id=6211.

• The Family Contact Form (FCF) recorded information about any contact between the interviewer and the family of each of the selected children at the time of Wave

1, regardless of whether they agreed to participate in the study or not. The information was mainly used by the fieldwork agency, with the only information from the FCF available in the publicly released dataset being the information on

LSAC Data User Guide – November 2013

12

the family’s home and neighbourhood. In subsequent waves, this information was included as part of the Interviewer Observations of the Face-to-Face Interview.

• Between Waves Questionnaires (Wave 1.5, Wave 2.5 and Wave 3.5) are brief questionnaires sent to respondents to complete and return in the year between main waves of data collection. Between wave surveys enable maintaining contact with study families and collecting information about activities and development in the year between the main waves.

5.1 Child assessments

5.1.1 Physical measurements

5.1.1.1 Weight

For the B cohort in Wave 1, the child’s weight was obtained by calculating the difference between the weight of Parent 1 (or another adult) with the child and the weight of the parent/other adult on their own. For the B cohort at all subsequent waves, and the K cohort at all waves, the child’s weight was measured directly.

In Wave 1 the scales used were Salter Australia glass bathroom scale

(150kgsX50gms). In Waves 2 and 3, these scales were used along with HoMedics digital BMI bathroom scales (180kgsX100gms). In Waves 4 and 5, Tanita Body Fat scales were used.

5.1.1.2 Height

Height is measured for children aged 2 years and older. In Waves 1, 2 and 3 height was measured using an Invicta stadiometer, from Modern Teaching Aids. In Waves 4 and 5 a laser stadiometer was used. Two measurements were taken, and if the two measurements differ by 0.5cm or more a third measurement was taken. The average of the two closest measures is included on the data file.

5.1.1.3 Girth

This measurement is taken for children aged 2 years and older using a non-stretch dressmaker’s tape, positioning the tape horizontally over the navel. Two measurements were taken, and if these differed by 0.5cm or more, a third measurement was taken. The average of the two closest measures is recorded on the data file.

5.1.1.4 Body Fat

A body fat measurement was included in Waves 4 and 5, with the reading provided by the same scales used for weight (Tanita Body Fat scales). Issues with the body fat measurement are outlined in the Issue Paper series.

5.1.1.5 Head circumference

This measurement was only taken for the B cohort in Wave 1, using an Abbott head circumference tape. Two measurements were taken, and if these differed by 0.5cm or more, a third measurement was taken. The average of the two closest measures was included on the data file.

LSAC Data User Guide – November 2013

13

5.1.1.6 Blood pressure

This measurement was taken for the K cohort in Waves 4 and 5 using A&D Digital

Blood Pressure Monitor - Model UA-767. Two measurements were taken by the interviewer, with a one-minute interval between the measurements. Both of the readings are included in the data file.

5.1.2 Who am I? (WAI)

1

The ‘Who am I?’ is a direct child assessment measure that requires children to copy shapes (circle, triangle, cross, square, and diamond) and write numbers, letters, words and sentences.

For the LSAC testing, there was a change to Who Am I? Item 11

‘This is a picture of me’ was replaced with a sentence to be copied ‘John is big’. The ‘Who am I?’ assessment was used for the children at ages 4-5 years (Wave

1 K and Wave 3 B cohorts) to assess the general cognitive abilities needed for beginning school.

The study child was given his/her own answer booklet to draw and write in. What they wrote/drew was assessed by experienced researchers at Australian Council for

Educational Research (ACER). See the new Data Issues series for details of the Rasch

Modelling used to score the WAI.

5.1.3 Peabody Picture Vocabulary Test (PPVT)

2

A short form of the Peabody Picture Vocabulary Test (PPVT - III), a test designed to measure a child’s knowledge of the meaning of spoken words and his or her receptive vocabulary for Standard American English, was developed for use in the study. This adaptation is based on work done in the United States for the Head Start Impact

Study, with a number of changes made for use in Australia.

Different versions of the PPVT containing different, although overlapping, sets of items of appropriate difficulty were used for the children when aged 4-5 years, 6-7 years and 8-9 years. A book with 40 plates of display pictures was used. The child points to (or says the number of) a picture that best represents the meaning of the word read out by the interviewer.

Scores are created via Rasch Modelling so that changes in scores represent real changes in functioning, rather than just changes in position relative to peers. See the

Data Issues Paper No. 2 for more details (available on the study website, http://www.growingupinaustralia.gov.au/pubs/issues/index.html).

1

The ‘Who Am I?’ is copyrighted by: Australian Council for Educational Research, Melbourne, 1999.

2

The ‘Peabody Picture Vocabulary Test, Third Edition (PPVT-III) Form IIA’ is copyrighted by Lloyd

Dunn, Leota Dunn, Douglass Dunn. American Guidance Service, Inc, 1997 and published exclusively by AGS Publishing. Permission to adapt and create a short form for LSAC was granted by the publisher. The PPVT- III - LSAC Australian Short-form was developed by S. Rothman, Australian

Council for Educational Research (ACER), Melbourne, from the Peabody Picture Vocabulary Test,

Third Edition (PPVT-III) Form IIA, English edition.

LSAC Data User Guide – November 2013

14

5.1.4 Matrix Reasoning

3

Children completed the Matrix Reasoning (MR) test from the Wechsler Intelligence

Scale for Children, 4th edition (WISC-IV) at ages 6-7, 8-9 and 10-11 years. This test of non-verbal intelligence presents the child with an incomplete set of diagrams (an item) and requires them to select the picture that completes the set from five different options. The data file includes raw scores (number of correct responses) and scaled scores based on age norms given in the WISC-IV manual. The instrument comprises

35 items of increasing complexity. Children start on the item corresponding to their age-appropriate start point. If a child does not answer correctly on the first or second start-point items the examiner should ask two items prior to the age-appropriate startpoint (called “reverse administration”). Reverse administration was not implemented in the LSAC instrument. See the discussion of this issue in Data Issue Paper No. 8

(available from the study website, http://www.growingupinaustralia.gov.au/pubs/issues/index.html).

3

The ‘Wechsler Intelligence Scale for Children – Fourth Edition’ is copyrighted by Harcourt

Assessment, Inc., 2004.

LSAC Data User Guide – November 2013

15

5.2 Response rates

The number and percentages of survey instruments of each type that were completed at each wave is shown in Table 2. More detailed information on non-response can be found in the “Weighting and non-response” technical papers.

Table 2. Waves 1-5 instrument response.

Wave 1

Instrument (a)

B cohort K cohort

Eligible (b)

Actual (c)

%

Eligible (b) Actual (c) %

F2F

P1L

P2L

TUD 1

TUD 2

AI

PPVT

HBC

CBC

TQ

AEDI

W1.5

5107

5107

4630

5107

5107

N/A

N/A

788

436

N/A

N/A

5061

5107

4341

3696

4031

3751

N/A

N/A

342

233

N/A

N/A

3573

100

85

80

79

73

N/A

N/A

43

53

N/A

N/A

71

4983

4983

4286

4983

4983

4983

4983

N/A

N/A

4761

1366

4935

4983

4229

3388

3867

3582

4880

4382

N/A

N/A

3276

720

3594

100

85

79

78

72

98

88

N/A

N/A

69

53

73

Wave 2

Instrument (a)

B cohort K cohort

Eligible (b)

Actual (c)

%

Eligible (b) Actual (c) %

F2F (d)

P1D

P1L

P2L

TUD 1

TUD 2

PPVT

MR

PLE Mail-out

HBC

CBC

TQ

W2.5

5107

4606

4606

4099

4606

4606

N/A

N/A

400

791

1672

N/A

5107

4606

4504

3536

3128

3477

3459

N/A

N/A

96

533

1144

N/A

3246

90 4983

98 4464

77 4464

76 3804

75 4464

75

4464

N/A 4464

N/A 4464

24 612

67

68

N/A

N/A

N/A 4447

64 4983

4464

4358

3495

2949

3446

3460

4409

4402

199

N/A

N/A

3632

3252

99

99

33

N/A

90

98

78

78

77

78

N/A

82

65

Wave 3

F2F (d)

P1D

P2L

TUD1

TUD2

PPVT

Instrument (a)

WAI

MR

PLE CATI

TQ

Eligible

(b)

5107

4386

3900

4386

4386

4386

4386

N/A

346

4114

B cohort

Actual

(c)

4386

3831

2753

2959

2950

4266

4197

N/A

272

3395

%

86

87

71

67

67

97

96

N/A

77

83

Eligible

(b)

K cohort

Actual

(c)

4983

4331

3707

4331

4331

4331

N/A

4331

510

4275

4331

3807

2680

2961

%

87

88

72

68

99

79

85

2963 68

4273 99

N/A N/A

4270

403

3643

LSAC Data User Guide – November 2013

16

Wave 4

Instrument (a)

F2F (d)

CASI

P2L

CSR

ACASI

TUD

PPVT

MR

PLE CATI

TQ

Eligible

(b)

5107

4242

3706

4242

N/A

N/A

4242

4242

439

4143

B cohort

Actual (c) %

4242

4210

2677

4181

N/A

N/A

4185

4180

377

3427

Eligible

(b)

K cohort

Actual

(c)

82

99

4983

4164

72 3512

99 N/A

N/A 4169

*

N/A 4169

*

99

99

N/A

4169

*

86

83

572

4025

%

4164

4116

84

99

2645 75

N/A N/A

4094

3994

99

96

N/A N/A

4103 99

493

3352

86

83

Wave 5

Instrument (a)

F2F (d)

CASI

P2L

CSR

ACASI

TUD

PPVT

MR

PLE CATI

TQ

Eligible

(b)

5107

4077

3512

4026

*

N/A

N/A

4026

4027

537

4021

B cohort

Actual (c)

%

4085

4010

2444

4014

N/A

N/A

3977

3985

404

3490

80

98

70

100

N/A

N/A

99

99

75

87

Eligible

(b)

4983

3952

3277

3872

3873

*

3871

*

N/A

N/A

614

3857

K cohort

Actual

(c)

3956

3857

2333

3850

3844

3649

N/A

N/A

464

3225

%

79

98

71

99

99

94

N/A

N/A

76

84

*

Represents instances where a child interview was completed and the main interview with the parents was not. Specifically, in Wave 4 there were five cases (K cohort) and in Wave 5 there were eight cases for the K cohort and four cases for the B cohort.

N/A=Not administered

(a) Questionnaire acronyms are detailed in previous section.

(b) ‘Eligible’ means the number of LSAC children for whom a questionnaire was applicable (e.g. children are eligible for an HBC questionnaire if the child’s main care is attended for 8 hours or more per week and this is home based care)

(c) ‘Actual’ means the number of respondents for whom a form was returned.

(d) Response rates for Wave 2, Wave 3, Wave 4 or Wave 5 as a proportion of Wave 1 families.

5.2.1 Parent 1 Questionnaires

In Wave 1, interviewers encouraged the parents to complete the P1L and P2L forms while the interviewer was in the home. Interviewers were also able to pick up forms in some cases, when forms were left behind. Forms not given to interviewers were mailed back. Two reminders were made for forms that were not returned.

In Wave 2, Parent 1 had two forms to complete. Interviewers were instructed that the

P1D form ‘must’ be completed when they were in the home (resulting in a high response rate). The P1L was generally left behind for mail back, as there was not enough time for these to be completed as well. Interviewers were generally not required to pick-up the forms. Up to four reminders were made for forms that were not returned, however the P1L forms showed lower response rates in Wave 2

LSAC Data User Guide – November 2013

17

compared with Wave 1. This may be because P1 had already completed one form and also because interviewers did not generally pick up forms.

For Wave 3, there was only one Parent 1 self-complete form. Interviewers were instructed that this form must be completed while the interviewer was in the home.

However, only two thirds of parents were able to do so. Three reminders were given for forms not returned.

In Wave 4 Parent 1 was asked to complete a CASI, which resulted in a response rate of 99% of eligible respondents. This is higher than the response rate of 88% of eligible respondents achieved in Wave 3 using the self-complete form.

In Wave 5 response rates are very similar to response rates obtained in Wave 4. This is due to no mode changes and attrition tapering off.

5.2.2 Parent 2, TUD and Teacher forms

Response rates to the P2L and the TUD were broadly similar between waves (Wave

1, 2 and 3), while the carer and teacher questionnaire response rates were much improved in Wave 2, with similar response rates at Wave 3. In Wave 4 the TUD response rate was 96%. The higher response rate could be contributed to the change in the procedure and the informant. In Waves 4 and 5, the interviewer collected the TUD information from the child (not the parent) as part of the interview rather than leaving a diary which then had to be completed and mailed back by respondent families after the visit.

5.2.3 PLE response

The PLE questionnaire was introduced in Wave 2 and applies for children who see their “parent living elsewhere” at least once a year. There are three stages where nonresponse can occur: (1) obtaining contact details from Parent 1; (2) obtaining permission from Parent 1; and (3) receiving a response from the PLE.

In Wave 2, contact details were given for 69% of cases for the B cohort and 70% of cases for the K cohort, and responses received from 35% of PLEs sent a questionnaire for the B cohort and 47% for the K cohort.

Due to the relatively low response in Wave 2 to the mail-out questionnaire, a change in methodology was introduced in Wave 3. Where Parent 1 provided contact details,

PLEs were telephoned and asked to respond to a Computer Assisted Telephone

Interview (CATI). The response from PLEs who were approached was very positive.

Of the 856 PLEs that interviewers attempted to contact, interviews were achieved with 675 (79%), and only 53(6%) refused an interview. Most of the remaining nonresponse was due to not being able to contact the PLE.

In Wave 3, the Parent 1 was explicitly asked the permission to contact the PLE.

Therefore, it was easy for the Parent 1 to refuse to provide any information about the

PLE or refuse the PLE’s participation. This meant that no information was obtained for 260 (18%) PLEs.

It is worth noting that while there was no direct question asking the Parent 1 permission to contact the PLE, some Parent 1 refused the PLE’s participation.

Table 3 summarises the situation with regard to PLEs in Waves 3, 4 and 5.

LSAC Data User Guide – November 2013

18

Table 3. Waves 3, 4 and 5: Information obtained with regard to PLE.

PLE identified during P1 interview

Eligible

PLE*

B cohor t

578

346

Wave 3

K cohort

837 1415 674

510

Total B cohort

Wave 4

K cohort

856 439

Total B cohort

Wave 5

K cohort

878 1552 773

572 1011 537

Total

911 1684

614 1151

*The PLE is considered eligible when: (1) the PLE satisfies the parental requirements i.e. PLEs who see the Study Child at least once a year; (2) PLE‘s contact details are available; (3) Parent

1 did not explicitly refuse to contact the PLE.

LSAC Data User Guide – November 2013

19

6 The LSAC data release

Data users are required to read the manual for the access to and use of DSS longitudinal survey datasets, complete a dataset application form and sign a deed of license. Users must abide by strict security and confidentiality protocols. Instructions on how to access data can be found on the LSAC website: http://www.aifs.gov.au/growingup/data/index.html.

6.1.1 Data security requirements

The deed of licence stipulates numerous security requirements for the data, including:

• The LSAC CD-ROM MUST be kept secure in a locked filing cabinet or other secure container when not in use.

• The LSAC data (and any derivatives of the LSAC data) MUST be stored on a password protected computer or network.

• Your password MUST include a mixture of upper and lowercase characters, be at least 8 characters long, and include some non-alphanumeric characters such as #,

;, *, etc.

• Any printed unit record output MUST be stored in a locked filing cabinet or other secure container when not in use. Any printed unit record output MUST be shredded if no longer required.

• You MUST NOT provide the unit record data to any unauthorised individual.

• There MUST be a means of limiting access to the work area where the data are kept and there must be tamper evident barriers to access (i.e., if there were a break-in, it would be obvious from broken glass, damaged lock, etc).

• If you have an individual license and you change employers, you MUST inform

DSS prior to doing so. Data MAY be able to move with the individual, depending on the research to be undertaken and the new employer. You must NOT leave the data with your old employer if you move.

• If you change your research project you MUST seek permission to use the data for the new project from DSS.

6.1.2 How data files are provided

All data are provided in three formats, SAS, SPSS and STATA, however users can transfer the data to other formats if they wish. The CD-ROM and/or website also includes extensive data documentation, including this document, marked-up questionnaires and variable frequencies. The data files and the other documentation are discussed in detail in later sections of this document.

LSAC Data User Guide – November 2013

20

7 File structure

For the Wave 5 data release, the following datasets are available:

• Ten datasets comprising the main datasets for each wave and cohort (lsacgrb0

4

, lsacgrb2, lsacgrb4, lsacgrb6, lsacgrb8 lsacgrk4, lsacgrk6, lsacgrk8, lsacgrk10 and lsacgrk12)

• 20 Time Use Diary datasets: o

One cleaned datafile with problematic cases deleted for each cohort for

Waves 1, 2 and 3 (diaryb0, diaryb2,etc.) o

One datafile with the cases deleted from the above files after cleaning for each cohort for Waves 1, 2 and 3 (poortudsb0, poortudsb2, etc.) o

One datafile with all cases and no data cleaning performed on them for each cohort for Waves 1, 2 and 3 (ucdiaryb0, ucdiaryb2, etc.) o

One datafile for K cohort only for Wave 4 (tudk10) o

One datafile for K cohort only for Wave 5 (tudk12)

• Three Medicare Australia Datasets representing information from the 3 Medicare

Australia databases the information was drawn from (mbs, pbs and acir)

• Two Study Child household composition datasets, one for each cohort (hhgrb, hhgrk)

• Two PLE household composition datasets, one for each cohort (plehhgrb, plehhgrk)

• Two Wave 2.5 datasets, one for each cohort (lsacgrb3, lsacgrk7)

• Two Wave 3.5 datasets one for each cohort (lsacgrb5, lsacgrk9)

• LSAC NAPLAN dataset (lsacnaplan)

• LSAC MySchool dataset (lsacmyschool)

Note: Wave 1.5 datasets have been added to the Wave 1 datasets. This is possible because all respondents that responded to Wave 1.5 had to complete a Wave 1 interview. This is not the case with other between wave mail outs, respondents may have completed any prior combination of interviews.

This structure has been used to reduce the size of the main datasets and because some data are formatted using more than one record for each child.

7.1 Main dataset

The main dataset consists of the data from all questionnaires except the Time-Use

Diary, Wave 2.5, Wave 3.5, some household composition information and LSAC

NAPLAN data. Data from the instruments are presented in the following order.

FCF (Wave 1 files only)

4

File names in this section are for the general release datasets (see ‘confidentialisation’ section below), users of the in confidence data should substitute ‘ic’ for ‘gr’ in the file names

LSAC Data User Guide – November 2013

21

F2F

P1D (except Wave 1 files)

P1L (except Wave 3 and 4 files)

P2L

PLE (except Wave 1 files)

Teacher/Carer Questionnaire

5

Wave 1.5 data (Wave 1 files only)

A number of derived variables are included in the output dataset alongside the raw responses used in their derivation. Additionally the main datasets contain status variables (e.g. date of interview, whether each type of form was returned, etc.), ABS

Population Census and NCAC data, and weights.

7.1.1 Australian Bureau of Statistics Census of Population and Housing data

Public data from the Australian Bureau of Statistics Census of Population and

Housing have been added to the file to enhance the range of neighbourhood characteristics available for analysis with the LSAC data. Census data is available for the child’s residence at Waves 1, 1.5, 2, 2.5, 3, 3.5, 4 and 5.

The items currently included are:

SEIFA – rounded off to the nearest 10 for on the general release file

Remoteness Area Classification

Percentage of persons aged under 5, 10 and 18 years

Percentage of persons born in Australia

Percentage of persons speaking English only at home

Percentage of persons with ATSI origins

Percentage of persons completed year 12

Percentage of persons above median income category

Percentage of persons working

Percentage of households with internet capacity (in 2006 Census only)

Percentage of households with broadband (in 2006 Census only).

Census data is either linked at the Statistical Local Area (SLA) level or, where this wasn’t available, the child’s postcode. One estimate is provided for each time point representing a linear interpolation of the data at the censuses either side of the time period. For example if a SLA had 4.2% of people with ATSI origins in 2001 and

6.5% with ATSI origins in 2006 then the estimate for the proportion in 2004 would be:

,

5

Since the CBC or HBC forms were only dispatched to the child’s main care type each child could only have one of these completed for them. Hence for Waves 1 and 2 HBC and CBC data are merged into a single set of variables where possible. This data is given in the order of the HBC questionnaire, with questions appearing only in the CBC form given at the end.

LSAC Data User Guide – November 2013

22

If data is only available for one of the Censuses then no interpolation is performed. A

‘link type’ variable is included to tell data users whether the linkage was performed using SLA or postcode and whether the 2001 census, 2006 census, 2011 census or all were used.

7.1.2 National Childcare Accreditation Council data

A key research question in LSAC relates to the impact of child care on children’s developmental outcomes over time. While LSAC collected parent-report information on children’s child care histories and carer reports on the child care environment, relatively little systematic information was collected on quality of child care.

The National Childcare Accreditation Council Inc. (NCAC) has quality assurance data on every Long Day Care (LDC) centre, some Family Day Care (FDC) schemes and some Before and After School Care providers. The LSAC dataset includes linked

NCAC data for most children using LDC or FDC at Wave 1, where contact details of this care were obtained and matched with NCAC data. The match rate obtained during the linkage process was 78% for Wave 1, 82% for Wave 2, 84% for Wave 3 and 92% for Wave 4.

One complication in using the NCAC data is due to the change of accreditation systems for both FDC and LDC. In Wave 1, all cases had FDC assessed under the guidelines laid out in 2 nd

edition of the ‘FDCQA Quality Practices Guide’ (NCAC,

2004), while from Wave 2 and onwards all cases have been assessed under the 3 rd edition of this reference, which was introduced in July 2005. The revised guidelines contain the same Quality Areas, but have had the number of principles used to assess these areas reduced from 35 to 30. The old scheme has 10 Quality Areas assessed by

35 principles, while the new has 7 Quality Areas assessed by 30 principles.

For LDC, all Wave 1 centres were assessed under the ‘QIAS Validation Report, 2 nd

Edition’ (NCAC, 2003). From July 2006, accreditation decisions were made under the

‘QIAS Quality Practices Guide, 1 st

Edition’. As a consequence some of the Wave 2 and 3 accreditations were made under the new scheme, while some were made under the old scheme.

Before and after school care arrangements are assessed in the guidelines laid out in the ‘OSHCQA Quality Practices Guide, 1 st

Edition’ (NCAC, 2003). In Wave 2 and 3, accreditations were made under the new scheme, while some were made under the old scheme.

The variables included are:

Date of accreditation

Date of validation

Accreditation status

LDC v1 Quality area 1: Relationships with Children

LDC v1 Quality area 2: Respect for Children

LDC v1 Quality area 3: Partnerships with Families

LDC v1 Quality area 4: Staff Interactions

LDC v1 Quality area 5: Planning and Evaluation

LDC v1 Quality area 6: Learning and Development

LDC v1 Quality area 7: Protective Care

LDC v1 Quality area 8: Health

LDC v1 Quality area 9: Safety

LDC v1 Quality area 10: Managing to Support Quality

LSAC Data User Guide – November 2013

23

LDC v2 Quality area 1: Staff relationships with Children and Peers

LDC v2 Quality area 2: Partnerships with Families

LDC v2 Quality area 3: Programming and Evaluation

LDC v2 Quality area 4: Children’s Experiences and Learning

LDC v2 Quality area 5: Protective Care and Safety

LDC v2 Quality area 6: Health, Nutrition and Wellbeing

LDC v2 Quality area 7: Managing to Support Quality

FDC Quality area 1: Interactions

FDC Quality area 2: Physical Environment

FDC Quality area 3: Children's Experiences, Learning and Development

FDC Quality area 4: Health, Hygiene, Nutrition, Safety and Wellbeing

FDC Quality area 5: Carers and Coordination Unit Staff

FDC Quality area 6: Management and Administration

OHS Quality area 1: Respect for Children

OHS Quality area 2: Staff Interactions and Relationships with Children

OHS Quality area 3: Partnerships with Families and Community Links

OHS Quality area 4: Programming and Evaluation

OHS Quality area 5: Play and Development

OHS Quality area 6: Health, Nutrition and Wellbeing

OHS Quality area 7:Protective Care and Safety

OHS Quality area 8: Managing to Support Quality

Demographic data

The data used to develop the quality areas was collected from six sources:

• A self-study report prepared by centre management;

• A validation survey completed by the director;

• A validation survey completed by staff;

• A validation survey completed by families;

• A validation report completed by an independent peer; and

• A set of moderation ratings completed by independent moderators.

Data on 35 principles was collected. Each principle was related to one of the ten quality areas. Response categories for each principle were: ’Unsatisfactory’,

‘Satisfactory’, ‘Good Quality’ and ‘High Quality’. Proportionally-weighted factorscore regression coefficients for principle ratings were calculated to determine the extent to which each principle contributed to a Quality area. For further information, see Rowe (2006).

As no data about the child was obtained, no consent was required from parents to collect this data (although parents did need to give details of their carers to assist in the linking).

7.2 Supplementary files

7.2.1 Time Use Diary data

In Waves 1 to 3, responding families were given two Time Use Diaries (TUDs) to complete at each wave. Each record in the TUD data relates to a single diary, i.e., each child can have up to two records (one for each TUD).

The key component of the TUD data is to gather information on children’s activities and context for the 96 15-minute periods of each 24 hour block. In addition to these

LSAC Data User Guide – November 2013

24

variables, the TUD data includes the child’s unique identification number in order to allow linkage with the main dataset. It also includes the following general descriptors:

Date diary should be completed

Day of week diary should be completed

The diet of the study child on the day in question (Waves 2 and 3)

The relationship of the diary writer to the child

Over what duration the diary was completed

Actual day and date of completion

Hours of work done by respondent on day of completion (Waves 2 and 3)

What kind of day was described in the diary

Due to scanning problems in Wave 1, and other data quality issues that are likely to apply equally across waves, a number of imputations and corrections have been applied to the TUD data (see Data Issues paper on the study website for details). So researchers can determine the effect of these imputations/corrections to the data on any analysis. An uncorrected version of the TUD data is also provided, as well as files containing imputations/corrected versions of cases that were considered unsuitable for data analysis even after correction.

LSAC Technical Paper 4 includes a detailed discussion of issues that should be considered when using the time use data. The Technical Paper is available from www.aifs.gov.au/growingup/pubs/technical/index.html.

In Wave 4 a new methodological approach was undertaken. The study shifted away from the parent being the informant to the study child being the informant. In Waves

4 and 5 only the K cohort completed the TUD, which was substantially different from the TUDs that the parents completed in earlier waves. The TUD in Waves 4 and 5 had the form of an “ABS Activity Episode” diary. This data is stored as a long file as opposed to the wide files the previous diaries were stored as.

Example analysis

SAS

The following code gives the proportion of children eating or drinking while watching a TV, video, DVD or movie at any time of day for the B cohort at Wave 1. Statements

1 and 2 tell SAS to create a new dataset beginning with the data in the mtud.diary2 file (you will need to use your own library name). The third statement tells SAS to treat the time-use data as a multidimensional array (x) containing 96 rows of 40 columns each. The next statement tells SAS to set up a new array of 96 variables

(Tveat) into which the data for eating in front of the TV will be derived.

Statements 5 to 8 contain a do loop which runs across all 96 time periods. Statement 5 tells SAS to create a variable ‘i’ to keep track of which time period is being worked on, and to give it the values 1 to 96 in turn. Statement 6 tells SAS to allocate the value

100 at the position in the ‘Tveat’ array for the current time period if the child was eating or drinking (column 4 in the array ‘x’) and was watching a TV etc. (column 12 in ‘x’). Statement 7 says the value of 0 will be assigned if the child either wasn’t eating or drinking or wasn’t watching TV etc. and the diarist wasn’t unsure of the child’s activities for the time period. This means that cases where the diarist wasn’t sure, or didn’t fill any information in for activities in this time period, will have missing data. Statement 8 finishes to do loop, and statement 9 finishes the data step so

SAS runs the above statements.

LSAC Data User Guide – November 2013

25

Statements 10-13 produce the means of the variables in the ‘Tveat’ array (which SAS gives the names Tveat1 to Tveat96 by default). The mean here will be the percentage of children from whom an activity was known that ate or drank in front of the TV etc. at each time period. Line 12 uses the day weight variable ‘bweightd’ to ensure the proportion is representative of the population and represents each day of the week equally.

(1) data diary2;

(2) set mtud.diary2;

(3)

(4) array x [96,40] b2da0101--b2de0196; array Tveat [96];

(5)

(6)

(7) do i=1 to 96; if x[i,4]=1 and x[i,12]=1 then Tveat[i]=100; else if (x[i,4]=0 or x[i,12]=0) and x[i,1]^=1 then Tveat[i]=0; end; (8)

(9) run;

(10) proc means data=diary2;

(11)

(12)

(13) run; var Tveat1-Tveat96; weight bweightd;

This data can be used to produce a graph known as a tempogram.

Figure 1 shows the data produced by the example program along with the equivalent data for the K cohort at Waves 1 and 2. It shows that children did more of this as they got older, and that this activity was most common in the early mornings.

Figure 1. Tempogram of children watching TV, video, DVD or movies while eating or drinking by wave and cohort.

LSAC Data User Guide – November 2013

26

SPSS

The equivalent code to derive the tveat variable in SPSS would look like do repeat eat b2da0401 b2da0402 … b2da0496/ tv b2da1201 b2da1201 … b2da1296/ dk b2da0101 b2da0101 … b2da0196/ tve tveat1 to tveat96. if (eat=1 or tv=1) tve=1. if ((eat=0 or tv=0) and dk=0) tve=0. end repeat.

STATA

The equivalent code to derive the tveat variable in STATA would look like

7.2.2 foreach n of numlist 1/9 {

7.2.3 gen tveat`n'=1 if (b2da040`n'==1 & b2da120`n'==1)

7.2.4 replace tveat`n'=0 if ((b2da040`n'==0 | b2da120`n'==0) & b2da010`n'==0)

7.2.5 }

7.2.6 foreach n of numlist 10/96 {

7.2.7 gen tveat`n'=1 if (b2da04`n'==1 & b2da12`n'==1)

7.2.8 replace tveat`n'=0 if ((b2da04`n'==0 | b2da12`n'==0) & b2da01`n'==0)

7.2.9 }

7.2.10 Medicare Australia data

In Wave 1, 97% of parents of study children gave consent for their children’s data to be linked with Medicare Australia data for the duration of the study. This includes data from the Medicare Benefit Scheme (MBS), the Pharmaceutical Benefit Scheme

(PBS) and the Australian Childhood Immunisation Records (ACIR). Data from these sources provide an indication of usage history of MBS, PBS and ACIR services.

Linkage was successful for 93% of children (incomplete consent forms resulted in data not being released for about 400 children).

Since the child’s use of medical services is ongoing, the Medicare Australia data are not broken into waves, but are provided as three separate files:

• ACIR: Each record in the file represents an immunisation that the child has had.

• MBS: Each record on this file represents a benefit claim.

• PBS: Each record represents a benefit claim.

LSAC Data User Guide – November 2013

27

7.2.10.1 ACIR file

Records are currently available for payments received from birth to early 2013. The following variables are included on the file:

• Child identification number

• Vaccination code

• Vaccination name

• Scrambled provider ID

• Date of receipt of payment

• Date of immunisation

Some of the vaccination codes contain dose numbers, which indicate where a vaccine has been received in a series of doses. The sequence of doses for these has been included in the dataset (i.e. 1 st

, 2 nd

, etc.). If a dose is missing it means that it was either not reported to ACIR, or it was missed.

7.2.10.2 MBS file

Records are currently available for services between January 2002 (or birth for the B cohort) and early 2013. The following variables are included on this file:

• Child identification number

• Item number

• Item name

• Amount of benefit paid

• Hospital indicator

• Scrambled provider ID

• Date of payment

• Date of service

Some cases have very small or negative benefit amounts. In relation to negative benefits, this indicates that an adjustment has been made to the Medicare benefit records. There are several reasons why this may happen:

• sometimes this is a correction of a data entry made against the wrong individual reference number on a Medicare card, (i.e. service is initially incorrectly recorded against someone else on the same card);

• the provider has issued an amended account; or

• a new cheque has been issued to replace lost/stolen/un-presented cheques.

In relation to small benefits:

• there are a number of item numbers which have small benefits, e.g. many pathology related claims;

• there are also small amounts for things like bulk bill incentives (generally around

$5 - $6); or

• the claimant had reached the Medicare Safety Net (MSN) threshold. Once the threshold has been reached, the family's out-of-pocket expenses are tallied and a payment is calculated for a percentage of the substantiated amounts. In effect there can be two payments made for the same doctor's visit - one to the doctor for the service and one to the claimant for MSN purposes.

LSAC Data User Guide – November 2013

28

7.2.10.3 PBS file

The final of these datasets contains the PBS data. Again, each record represents a benefit claim. Records are available for medications supplied between May 2002 (or birth for the B cohort) and early 2013. The following information is included for each record:

• Child identification number

• Item code

• Item name

• Quantity

• Benefit paid

• Prescription type (original, repeat or unknown)

• Payment category

• Payment status

• Date of payment

• Date of supply

7.2.10.4 Example derivations

There are simple techniques in SAS, SPSS and STATA to summarise across multiple records to create derived items from the Medicare datasets. The following code samples create a variable (ben07) for the amount of PBS benefits paid for a child in

2007. Note that this variable will initially be missing for cases that had no PBS claims in 2007 as well as those for which data linkage was unsuccessful. The ‘match’ file can be used to distinguish between these cases and set ben07 to 0 for those with no claims. This file contains a variable called ‘medicare’ which is 1 if linkage is successful for a case and 0 otherwise.

SAS:

proc means data=m.pbs nway sum; class hicid; var benefit; where datesupp>=mdy(1,1,2007) and datesupp<=mdy(1,1,2008); output out=temp sum=ben07; run; data temp; merge temp m3.match; by hicid; if medicare=1 and ben07=. then ben07=0; run;

LSAC Data User Guide – November 2013

29

SPSS:

temp. select if (datesupp >= date.dmy(1,1,2007) & datesupp <= date.dmy(31,12,2007)). aggregate

/outfile='/temp.sav'

/break=hicid

/ben07=sum(benefit). get

file='/temp.sav'. match files /file=*

/file='/match.sav'

/by hicid. if (medicare=1 & missing(ben07)) ben07=0. execute.

STATA (note that the collapse command will delete all other data than hicid and ben07, make sure to save it to a new file): collapse (sum) ben07=benefit if (datesupp>=mdy(1,1,2007) & datesupp<=mdy(1,1,2008)), by(hicid) merge hicid using match replace ben07=0 if (medicare==1 & ben07==.) keep if ben07!=. sort hicid save temp, replace

7.2.11 Household composition data

At each wave of data collection, responding families are asked to give the details of the people currently residing in their household, as well as people who have come and gone between waves, but lived with the study child for at least three months.

This dataset contains one record for each study child, detailing the composition of their household since their recruitment to the study up to the most recent data collection.

Details collected about the study child, Parent 1 and Parent 2 are included in each main dataset, along with a number of derived variables on household composition.

LSAC Data User Guide – November 2013

30

7.2.12 LSAC NAPLAN data

In Wave 3, 81% of parents of K cohort children gave consent for their child’s data to be linked with NAPLAN data for the duration of the study. Linkage was successful for 96% of children. For 4% of children, the NAPLAN data were not found, either because these children had not sat NAPLAN tests yet or they sat the NAPLAN tests in 2008 or 2009, but a match was not found. Families who did not give consent or who did not participated at Wave 3 were asked again at Wave 4. Out of 964 families who were followed up in Wave 4, 847 gave consent to link NAPLAN results.

In Wave 4, 95.5% of parents of B cohort children gave consent to link

NAPLAN/AEDI results.

This percentage excludes 9 B cohort families where the study child was home schooled. Linkage for of NAPLAN data for children whose consent was received at

Wave 4 is scheduled to be released in April 2012.

The Wave 4 LSAC NAPLAN release includes K cohort NAPLAN results for years

2008 and 2009. The update of the LSAC NAPLAN file with NAPLAN results for years 2010 and 2011 is scheduled for April 2012. Starting from 2013, the LSAC

NAPLAN file will be updated with new NAPLAN results every two years and released along with the main wave release.

In Wave 5 LSAC NAPLAN release includes B & K cohort NAPLAN results for 2008 to 2012.

LSAC Technical Paper 8 includes a detailed discussion of data compendium and data issues that should be considered when using the LSAC NAPLAN data. The report is available from www.aifs.gov.au/growingup/pubs/technical/index.html.

7.2.13 Wave 2.5 data

The data from the Wave 2.5 mail out is included in two separate datasets. Unlike

Wave 1.5 in relation to Wave 1, families that responded to Wave 2.5 did not necessarily respond to Wave 2. Merging these with the Wave 2 datasets would have resulted in a number of largely blank cases on the data file.

The data in the Wave 2.5 file consists of questionnaire items, a small number of derived items, and linked census data based on postcode of responding families at the time of Wave 2.5. Unfortunately, formatting of the questionnaires resulted in some respondents skipping items they should have answered. Imputation has been performed on some items where it was possible to infer the data for these questions based on responses to other questions. See the Data Issues paper for further information.

7.2.14 Wave 3.5 data

The data from the Wave 3.5 mail out is included in a separate dataset, in the same way that data from Wave 2.5 was included.

The data in the Wave 3.5 file consists of questionnaire items, a small number of derived items, and linked census data based on the postcode of responding families at the time of Wave 3.5. Imputation has been performed on some items where it was

LSAC Data User Guide – November 2013

31

possible to infer the data for these questions based on responses to other questions.

See the Data Issues paper W3.5 for further information.

7.2.15 ACARA MySchool Data

Data has been obtained from ACARA. ACARA is responsible for collating NAPLAN data received from Australian schools, collecting school characteristics and managing the MySchool Web site. Some of the data ACARA collects and collates on Australian schools is publically available on the MySchool website. School data about the schools LSAC participants attend has been linked onto the LSAC survey datasets and is available to data users.

LSAC Data User Guide – November 2013

32

Variable naming conventions

The variable naming convention was developed so that variables have predictable names across waves and informants, and so that thematically linked variables have similar names wherever possible. A two-page ‘help sheet’ is included on the LSAC

Data CD to help users learn these conventions.

7.3 Questionnaire variables

Variable names follow the standard format in most cases. Exceptions to this naming convention (derived items and household composition variables) are explained in sections that follow.

Standard format: A tt xxxxx where:

A = child age indicator tt = topic indicator xxxxx = specific question identifier.

7.3.1 Child age indicator (alpha)

The child age indicator is an alpha symbol that indicates the child’s age, allowing for comparisons between the cohorts where data has been collected for both cohorts at that age. For instance:

a indicates the child is aged 0-1 years (which is the B cohort in Wave 1)

b indicates the child is aged 2-3 years (which is the B cohort in Wave 2)

c indicates the child is aged 4-5 years (which is the B cohort in Wave 3, and the K cohort in Wave 1)

d indicates the child is aged 6-7 years (which is the B cohort in Wave 4, and the K cohort in Wave 2)

e indicates the child is aged 8-9 years (which is the B cohort in Wave 5, and the K cohort in Wave 3)

f indicates the child is aged 10-11 (which is the K cohort in Wave 4), etc.

This is an example of how the child age indicator is used for the item ‘Parent 1 rating of parenting self-efficacy’:

• Wave 1 B cohort: apa01a

• Wave 2 B cohort: bpa01a

• Wave 3 B cohort: cpa01a

• Wave 1 K cohort: cpa01a

• Wave 2 K cohort: dpa01a

• Wave 3 K cohort: epa01a

Those items of information that do not change (e.g. details of birth, age began or stopped something, etc.) are given the age indicator z so that they have a consistent variable name across cohorts regardless of the age of the child when the information was obtained. For example, zhs03a indicates ‘birth weight of the study child’ regardless of whether the information was collected when the child was aged 0-1 years as for the B cohort, or aged 4-5 years as for the K cohort.

LSAC Data User Guide – November 2013

33

7.3.2 Topic indicator (alpha)

The topic indicator is taken from the topic field of the data dictionary. An effort was made to make abbreviations used meaningful (e.g. family demographics is fd).

A list of topics and their abbreviations is in Table 4.

Table 4. Topics used in LSAC datasets.

Abbreviation Topic

fd fn gd hb he ho hs id lc pa pc pe pl

Family

Demographics

Finances

General

Development

Health Behaviour and Risk Factors

Home Education

Environment

Housing

Health Status

Identifiers

Learning and

Cognition

Outcomes

Parenting

Program

Characteristics

Parent Living

Elsewhere

Parental Leave in

Australia

Scope

Demographic information relating to the family such as education, ethnicity and religion

Financial information such as income and use of government benefits

Scales which contain items from multiple domains of child development

Behaviours and other risk factors that potentially impinge upon the health of the Study Child or his/her family. Includes behaviours such as parental smoking and drinking as well as risk factors such as a parent experiencing diabetes during pregnancy

Information on factors likely to impinge on the child’s learning while at home, such as parental support for education, number of books in the home and TV use. Also contains information on parent interaction with teachers such as parent teacher interviews even when asked from the teacher’s perspective.

Information on housing such as number of bedrooms, tenure type and payments.

Information about the physical and mental health status of the study child or his/her family such as

Body Mass Index, diagnosis with conditions and number of hospital stays.

Questionnaire process variables such as sequence guides, consents, and details of proxy respondents.

Information on the child’s development in the areas of learning and cognition including language, literacy and numeracy

Information on parenting styles and other information effecting parenting such as selfefficacy

Characteristics of the educational or childcare program such as type of program, number of days or hours the child attends and staff satisfaction.

Details of the child’s PLE such as the relationship to study child, interactions with resident parents and child support.

Data from the ‘Parental Leave in Australia’

Nested Study

LSAC Data User Guide – November 2013

34

Abbreviation Topic

pw re sc se tp

Paid Work

Relationships

Social Capital

Social and

Emotional

Outcomes

Teaching

Practices

Scope

Information on work status such as employment, occupation and work/family interactions

Information on the quality of relationships primarily focused on the relationship between

Parent 1 and Parent 2, but also on broader family harmony

Information on social capital such as attitudes to neighbours, their neighbourhood and use of services

Information relevant to the social and emotional development of the child such as temperament, behaviour and emotional states

Practices employed by teachers and childcare workers in their work such as time use, use of resources, and general philosophies.

For example:

• apa01a (Parent 1 rating of self-efficacy) has ‘pa’ as the second and third letters as its topic is ‘Parenting’; and

• zhs03a (Birth weight of study child) has ‘hs’ as the second and third letter as its topic is ‘Health Status’.

7.3.3 Specific question identifier (alphanumeric)

The last 5 digits of a variable name make up the specific question identifier. These digits contain whatever information is necessary to uniquely identify each item. Each has an arbitrary two-digit question number, not related to the questionnaire positioning. Items of related content are grouped together as much as possible. For example:

• bhs12a is whether Parent 1 is concerned about the child’s weight

• bhs12b is whether Parent 1 considers the child to be ‘underweight’, ‘normal weight’, ‘somewhat overweight’ or ‘very overweight’

The 5th digit of the variable name can also be an informant or subject indicator where a question is asked of or about more than one person. The indicators used are:

a Parent 1

b Parent 2

c Study Child

m Mother

f Father (or family home for census data)

t Teacher/carer

i In-between waves respondent

LSAC Data User Guide – November 2013

35

For example:

• bhs13a is Parent 1’s rating of their own overall health status

• bhs13b is Parent 2’s rating of their own overall health status

• bhs13c is Parent 1’s rating of the Study Child’s overall health status

• bhs13p is the PLE’s rating of their own overall health status

• bhs13m is the Mother’s rating of their own overall health status

• bhs13f is Father’s rating of their own overall health status

An exception to the above rule is in the area of childcare and education (variables with topic indicators pc and tp). Here the prefixes a, b, c, d and e are used to mean different things at each wave depending on the options available to the child at that age (see Table 5).

Table 5. Subject indicators for education and childcare variables.

Age 0-1 Age 2-3 Age 4-5 Age 6-7 Age 8-9 Age 10-11 a b c d e o

1 st childcare

2 nd

3 rd childcare childcare

1 st childcare

2 nd childcare

3 rd childcare

Other childcare

Any extra care

Main educational program

1 st childcare

2 nd childcare

3 rd childcare

Any extra care

Main educational program

Before school care

After school care

Program child would attend if attending school

Any extra care

Main educational program

Before school care

After school care

Childcare at other times

Program child would attend if attending school

Main educational program

Before school care

After school care

Program child would attend if attending school

Any extra care

All items that form a scale have a single question number. Where applicable, the name of the item also indicates the relevant subscale or sub-subscale (please note that this is done only where it is possible to do so, due to the eight character limit for the name of an item).

An example of how this is applied is shown with the Conduct Problems and Peer

Problems subscales of the Strengths and Difficulties Questionnaire (see Table 6).

These are subscales that both Parent 1 and the teacher filled out in Waves 1 and 2 for the K cohort.

LSAC Data User Guide – November 2013

36

As shown:

• The 6 th

character in the variable name in this case represents an informant indicator: ‘a’ is for Parent 1, ‘t’ is for teacher.

• The 7 th

character indicates the subscale: 4 for Conduct, 5 for Peer. (Note: Also available as part of the SDQ are 1 for Prosocial, 2 for Hyperactivity and 3 for

Emotional.)

• The final character uniquely identifies each item. (Note that different items were used for the Conduct subscale in Waves 1 and 2 due to the change in the child’s age).

Table 6. Variable names of SDQ

1

conduct and peer problems subscales

Conduct Problems

Often loses temper

Generally well behaved, usually does what adults request

Wave 1

Parent 1

K cohort name

cse03a4a cse03a4b

Wave 1

Teacher

K cohort name

cse03t4a cse03t4b

Wave 2

Parent 1

K cohort name

dse03a4a dse03a4b

Wave 2

Teacher

K cohort name

dse03t4a dse03t4b

Often fights with other children or bullies them

cse03a4c cse03t4c

Often argumentative with adults

cse03a4d cse03t4d

Can be spiteful to others

cse03a4e cse03t4e

Often lies or cheats

Steals from home, school or elsewhere

na na na na dse03a4c na na dse03a4f dse03a4g dse03t4c na na dse03t4f dse03t4g

Peer Problems

Rather solitary, tends to play alone

cse03a5a cse03t5a dse03a5a dse03t5a

Has at least one good friend

cse03a5b cse03t5b dse03a5b dse03t5b

Generally liked by other children

cse03a5c cse03t5c dse03a5c dse03t5c

Picked on or bullied by other children

cse03a5d cse03t5d dse03a5d dse03t5d

Gets on better with adults than

cse03a5e cse03t5e dse03a5e dse03t5e

with other children

1

The SDQ is copyrighted by Robert Goodman, UK, 1999.

7.4 Derived variables

The derived items start with an age indicator as outlined in section 7.3.1, followed by

an informant or subject indicator and then a mnemonic that relates to the subject matter of the derived item. So for example, the Peer scale of the SDQ for the K cohort teacher in Wave 2 is dtpeer, where d=child aged 6-7 years, t=teacher, peer=Peer scale of SDQ.

LSAC Data User Guide – November 2013

37

7.5 Study Child Household composition variables

In order to keep the variable names under 8 characters, it was necessary to have a slightly different convention in the Wave 2 data release. Household composition variables have the following structure:

A f ##xmmm

Where:

A= Child age indicator f = f (for ‘family’)

## = Question number (numeric) x = Sub-question indicator (optional) mmm = person identifier

Note that:

• The age indicator above is as described in section 7.3.1.

• ‘f’ is a constant to indicate that it is the household composition that is being described.

• The question number and sub-question indicator indicate the question being responded to.

• The person identifier indicates the member number, or other identification information. For every household, the Study Child is Member 1, Wave 1

Parent 1 will be Member 2, and Wave 1 Parent 2 is Member 3 (or will be missing if there is no Parent 2 at Wave 1). Any additional people in the household at the time of Wave 1 are given Member numbers 4 through to whatever is required. Each household member retains the same member number throughout the study, even if they leave and re-enter the Study

Child’s home.

• Due to the requirements of the CAI instrument, some families have ‘gaps’ in member numbering, for example where someone is Member 5, but Member

4 has never been assigned.

• Member 1 is denoted by ‘m1’ in the above convention, Member 2 as ‘m2’ and so on as required.

• As families change, from Wave 2 on Parent 1, Parent 2, Mother or Father can have any member number apart from 1. For this reason an extra set of variables has been derived to give the details for the Parent 1, Parent 2,

Mother and Father at any age. This subscript is an age indicator and then either ‘p1’, ‘p2’, ‘m’, or ‘f’.

• A set of indicator variables tracks the household member number of Parent

1, Parent 2, Mother and Father at each wave. For example bp2mn tells you the household member number of Parent 2 when the child is aged 2-3, while cmmn gives the member number of the mother when the child is aged 4-5.

Some examples:

• zf02m1 - the gender of the study child (z=unchanging characteristic, f=‘Family’, 02=gender, m1 =Study child)

LSAC Data User Guide – November 2013

38

• bf01m2 - whether the Wave 1 Parent 1 is present in the household when the child is aged 2-3 (b=child aged 2/3, f=‘Family’, 01= present for wave, m2=Wave 1 Parent 1)

• cf01m3 - whether the Wave 1 Parent 2 is present when the child was aged 4-

5 (or whether there was a Parent 2 at all in Wave 1 for the K cohort) (c= child aged 4-5, f=‘family’, 01=present for wave, m3=Wave 1 Parent 2)

• af08am - Relationship of the Mother when the child was aged 0-1 to the

Study Child (a=ages0/1, f=‘family’, 08=relationship to study child, am=mother of child at age 0/1)

• df01cp1 - Whether the Parent 1 of the child when aged 4-5 is present in the household when the child is aged 6-7. (d=child aged 6-7, f=‘family’,

01=present for wave, cp1=child’s Parent 1 when child is aged 4-5)

• cf13dp2 - Whether the Parent 2 of the child when aged 6-7 had a medical condition or disability at the time the child was 4-5 (c= child aged 4-5, f=

‘family’, 13=whether person has a disability, dp2=Parent 2 when child is aged 6-7).

Table 7 shows the information that is available for each person.

LSAC Data User Guide – November 2013

39

Table 7. Question numbers used in variable names for household member characteristics.

##x Question

01 Present for wave

02 Gender

03 Age

04 DOB

05 Temporarily away from home (as per Wave 1 question)

06 Relationship to parent 1

07 Relationship to parent 2

08 Relationship to study child

09 Country of Birth

10 Year of first arrival in Australia

11 Language other than English spoken at home

12 ATSI status

13 Has a condition or disability for 6 months or more (as per Wave 1 question)

13a 1st specific condition

13b 2nd specific condition

14 Date stopped living with study child

15 Reason stopped living with study child

16 Temporarily away from home (as per Wave 2 question)

16o Temporarily away from home (other) (as per Wave 2 question)

17 Has a condition or disability for 6 months or more (as per Wave 2 question)

17a Has sight problems (as per Wave 2 question)

17b Has hearing problems (as per Wave 2 question)

17c Has speech problems (as per Wave 2 question)

17d Has blackouts etc (as per Wave 2 question)

17e Has difficulty learning (as per Wave 2 question)

17f Limited use of arms or fingers (as per Wave 2 question)

17g Difficulty gripping (as per Wave 2 question)

17h Limited use of legs and feet (as per Wave 2 question)

17i Other physical condition (as per Wave 2 question)

17j Other disfigurement (as per Wave 2 question)

17k None of the above conditions (as per Wave 2 question)

18 Restricted in everyday activities

18a Has difficulty breathing (as per Wave 2 question)

18b Has chronic pain (as per Wave 2 question)

18c Has nervous condition requiring treatment (as per Wave 2 question)

18d Has mental illness requiring supervision (as per Wave 2 question)

18e Has head injury (as per Wave 2 question)

18f Has other long-term condition (as per Wave 2 question)

18g Has other condition requiring treatment (as per Wave 2 question)

18h None of the above restrictions (as per Wave 2 question)

19 Date began living with the study child

20

Household member was in the household for at least 3 months, but moved in and left between current and previous wave

LSAC Data User Guide – November 2013

40

7.6 PLE Household composition variables

From Wave 4, the household information for the child’s parent living elsewhere

(PLE) has been collected. PLE household composition variables have a similar structure to that of the Study Child Household composition variables:

A f ##xple#

Where:

A= Child age indicator f = f (for ‘family’)

## = Question number (numeric) x = Sub-question indicator (optional) ple# = person identifier within PLE household with ple (for Parent Living

Elsewhere) and # member number

Note that:

• The age indicator as described in section 7.3.1.

• ‘f’ is a constant to indicate that it is the household composition that is being described.

• The question number and sub-question indicator indicate the question being responded to.

• The person identifier comprises the constant “ple” to indicate that it is PLE household and the member number. For every PLE household, the Study

Child is Member 1 (ple1) and PLE is Member 2 (ple2). For example, variable f02ple2 refers to a PLE gender when a Study Child is 10 to 11 years old. Any additional member in the household is assigned a PLE member number that remains the same throughout the study, even if they leave and re-enter the PLE’s home.

Table 8 shows the information that is available for each PLE.

Table 8. Question numbers used in variable names for PLE household member characteristics.

##x Question

01 Present for wave

02 Gender

03 Age

04 DOB

05 Temporarily away from home (as per Wave 1 question)

06a Relationship to PLE

08 Relationship to study child

09 Country of Birth

10 Year of first arrival in Australia

11 Main language spoken at home

12 ATSI status

LSAC Data User Guide – November 2013

41

PLE household file also includes the following variables (asterisk refers to child age indicator ):

• *datplec – date of PLE PLE CATI interview;

• *plepar – whether PLE has a partner;

• *pleparmn – PLE partner member number in PLE household;

• *dfd02p3 – date of recent PLE marriage;

• *dfd02p4 – date of PLE cohabitation.

7.7 Age invariant indicator variables

There are 5 variables at the start of each of the main data files which contain no age indicator. These are:

• hicid – unique identifier assigned when child was selected by Medicare Australia

• cohort

• wave

• stratum – stratum at the time of selection

• pcodes – postcode at the time of selection

Users wishing to create long datasets should note the presence of these variables when removing age indicators.

7.7.1 Study child unique identifier

Each study child has a single, unique identification variable to ensure matching and merging across instruments, files and waves. This number was allocated at the time of selection by Medicare Australia.

The first digit indicates which cohort the child is in (1-4 = Infant; 5-8 = Child) and what fieldwork phase (see “Methodology” section for more detail) the child was selected to be part of in Wave 1 (Phase 1 = 1 and 5, Phase 2 = 2 and 6, etc).

The second is the state the child was selected from (1 = NSW, 2 = Vic, etc).

The third indicates the part of state the child was selected from (1-2 = capital city; 3-4

= rest of state).

The remaining 5 digits are a random number allocated by Medicare Australia.

Note that the stratum for selection may differ from the location of the child at interview and that the fieldwork phase may change from wave to wave.

7.8 Indicator variables

There are indicator variables in the main data files that indicate which parts of interview were incomplete. These variables were created to flag to data users (through yes/no values) that no data, or only partial data exists for an instrument (for example the CASI) or an informant (for example parent 1). The data may be incomplete due a number of different reasons. There may be no data if a self-complete form was not returned; parent/child did not provide consent to obtain/provide the data; one of the informants refused to participate; or when the interview was partially completed.

For example, on the day of the interview the parent may consent to the child participating but refuse to participate themselves. In this example there would be data for the sections where the study child is the informant, however there would be no

LSAC Data User Guide – November 2013

42

data for the sections where parent 1 is the informant. To identify these cases a data user can use the following indicator variable nopar (* refers to the age indicator).

Another example is teacher’s responses. To identify cases where a teacher form was not returned, a data user can examine the variable *tcd. A data user can also examine the following indicator variables: *partresp to identify cases that were incomplete due to an interview stopping half way as opposed to just certain sections being refused, or *hhresp to identify cases where the household interview was completed.

There are a large number of indicator variables and data users are encouraged to investigate the reasons for data being incomplete through these variables. Note that the indicator variables do not follow the general variable naming conventions described above. Some indicator variables are listed in Table 1. Indicator variables can be found in the data dictionary under the topic ‘Identifiers’, along with other variables that fall under that topic. For more information refer to the data dictionary.

7.9 Variable labelling convention

The labels used for the variable dataset take the following general form:

(Age) – (Informant/Subject) – (Questionnaire Position) – Construct Label

Age is a label for the age indicator from the variable name, so:

• a= 0/1

• b=2/3

• c=4/5

• d=6/7.

If no age indicator is present in the variable name, or the age indicator is z, then this part of the variable label will not be included. E.g. label zf04m1 = "SC – DOB”, here no age is associated with the variable because it doesn’t change with time, hence no age indicator is included. label df03m1 = "6/7 - SC - Age", this variable is a variable that changes over time so the age indicator is required in order to establish when the question was answered.

Informant/subject gives the informant or subject of the question as contained in the variable name. For household composition variables involving Parent 1, Parent 2,

Mother or Father, the age of the study child at which the person’s status as parent is determined will also be indicated (e.g. [email protected]/1 is the Mother when the child is aged 0-

1 years old). If the information only exists for one subject or informant in the study this part of the variable label will not be included.

Questionnaire position indicates the location of the question the data was obtained from within the LSAC questionnaires (e.g. F2F H2 is question H2 of the Face-to-Face

Interview). This part of the variable label is left blank for derived items such as scales, and other non-input items, but included for Mother/Father variables where the location of both the P1 and the P2 variables are given.

Construct label provides a description of what information is actually contained in the variable (e.g. ‘Sex’, ‘Birthweight’, etc.). This part of the variable name will be consistent for each variable representing the same construct for a different subject/informant or wave.

LSAC Data User Guide – November 2013

43

For example:

• the Parent 1’s rating of their own health quality at Wave 1 for the B cohort

(ahs13a) has the variable label ‘0/1 – P1 – P1L D1 – Global Health Measure’.

(0/1 is the age indicator, P1 is the informant/subject indicator, P1L D1 indicates the variable comes from the first question of Section D of the Parent 1 Leave-

Behind questionnaire, ‘Global Health Measures’ is the construct label).

• total score for the Parent 1 parental warmth scale for the K cohort at Wave 2

(dbwarm) id ‘6/7 – P2 – Warm parenting’ (6/7 is the age indicator, P2 is the informant indicator, there is no questionnaire position as the variable is calculated from multiple questions, ‘Warm parenting’ is the construct label).

7.10 Missing value conventions

Missing data are coded as follows:

-1 Not applicable (when explicitly available as an option in the questionnaire)

-2 Don’t know

-3 Refused or not answered

-4 Section refused

-9 Not asked due to one of the following reasons:

(a) question skipped due to answer to a preceding question (e.g. if a child never repeated a grade, the following question regarding what grade the child repeated was not asked/skipped);

(b) a form was not returned or consent to participate was not given (e.g. if a teacher form was not returned then teacher’s responses for this hicid are set to -9. To identify cases for which a form was not returned/or consent was not provided a data user can use an indicator variable (see Table 1 for details));

(c) one of the informants refused to participate (e.g. if a parent refused to participate but not a child then parent’s responses are set to -9. To identify cases when the parent refused to participate, a data user can use *nopar indicator variable);

(d) a form was partially completed (e.g. parent 1 completed the interview over the phone (P1 CATI) but face-to-face component did not occur. To identify these cases, a data user can use *partresp indicator variable).

(see

7.8 for more detail)

-99 Negative income (loss)

. Missing data – data not collected where it might be expected (e.g. the respondent skipped a question they should have answered in a selfcomplete form), or made missing due to an unreliable value (e.g. weight of

Parent 1 recorded as 800kg).

LSAC Data User Guide – November 2013

44

8 Documentation

A number of tools can be used to navigate the LSAC dataset:

Marked-up instruments

Frequencies

Online LSAC Data Dictionary

Excel spreadsheets of the Data Dictionary (good for creating hardcopies)

Users should also consider which documents they want to print out and which they want to look at electronically. We have found that the marked-up questionnaires and interview specifications are best printed and provide the easiest method of browsing to familiarise yourself with the data available. The Data Dictionary is best used for searching for specific items and mapping items from wave to wave.

These tools are described in more detail below.

8.1 Marked-up instruments

The associated variable name has been added beside each question in the questionnaires and interview specifications. Derived variables are also included. See

Figure 2 for an example.

Figure 2. Example of the marked-up questionnaires.

A mock questionnaire (interview specifications) has also been generated for the CAI instrument used in Waves 2, 3, 4 and 5. Figure 3 is a sample of this.

LSAC Data User Guide – November 2013

45

Figure 3. Example of Wave 2 interview specification.

8.2 Frequencies

The frequencies are a listing of the response categories for each question and the number of cases in each category. Figure 4 provides an example of the listing.

Figure 4. Example of the weighted frequencies apw01a1

0/1 – P1 – F2F C1.1 – Main activity – FT work

Frequency

1.636675

Percent

0.03

Cumulative

Frequency

1.636675

Cumulative

Percent

0.03

-4

No

4763.971

341.3922

93.28

6.68

4765.608

5107

93.32

100.00

Yes

The frequencies are useful for simple queries related to particular questions (for example, how many births were a normal delivery, or what are the codes used for

Wave 1 question A15). Variables for which there were a wide variety of responses meaning unaltered frequencies would run for several pages (eg. Study Child weight) have been rounded off to enable grouping of responses.

8.3 Data Dictionary

This is available as both an ‘online’ version and in Excel. Both versions of the data dictionary are searchable and can be sorted. Each record describes a single variable and has the following fields: variable name variable name without age (useful for sorting)

LSAC Data User Guide – November 2013

46

topic number (allows derived items to be sorted in with the input variables they come from) question id (i.e. variable name without age or subject/informant, useful for sorting) position in file order file wave cohort position of question in questionnaires person label child’s age variable label briefly describing each data item; topic construct measure question as found in the survey instruments response categories population with data

SAS format notes field indicating other information users should know about the data item.

8.3.1 Excel Data Dictionary

The Excel data dictionary contains two spreadsheets, one with the complete detailed listing of variable attributes, another with a shorter listing in a print-ready format. The print-ready format contains the variable name, question, responses and population fields; however, it is not a difficult task for users to make their own printable versions if they prefer other fields.

The Excel version can be easily filtered using the drop-down menus in the first row of the spreadsheet. For example, to find all the items on teacher practices in the lsacgr6 file (K cohort at Wave 2) first click on the drop-down menu in the ‘File’ field as shown in Figure 5 and select ‘B2’. Next, repeat the process for the ‘Topic’ field selecting ‘Teaching practices’.

After the search is finished all variables can be displayed by either clicking the ‘show all’ option in each of the fields that have been filtered (see Figure 5), or by selecting

‘Data > Filter > Show All’ from the menus.

More advanced searches can be performed using the ‘Custom Filter’ option which produces a dialogue box to assist with your searching. For example, to find all the questions that contain the word ‘internet’, go to the ‘question’ column and open up the filter menu and click on ‘Custom filter’, in the dialogue box change ‘equals’ to

‘contains’ and type ‘internet’ next to this.

LSAC Data User Guide – November 2013

47

Figure 5. Example of filtering in Excel.

8.3.2 Using wildcards for filtering

A good understanding of the variable naming convention is valuable for using the

Data Dictionary. Both the on-line and Excel Data Dictionary can be searched and filtered using wildcards, which can be used to return thematically linked sets of variables. Two wildcard characters are used by both these programs:

* represents any combination of letters and characters

? represents any single character

Some examples of the use of these characters are as follows: apw23a* returns a range of variables apw23a1a through to apw23a4b. apw23a4? returns two variables apw23a4a and apw23a4b.

?pw23a4a shows if this variable exists over different waves apw23?4a shows if this variable exists for different people in the same wave.

?pw23?4a shows if this variable exists for different people in different waves.

LSAC Data User Guide – November 2013

48

8.3.3 Some useful tips navigating the Data Dictionary

• Only items currently on the main datasets are included in the data dictionary

6

.

The User Guide provides information on the composition of other datasets.

• Items on the data dictionary are in the same order as on the data files, but can easily be sorted into other orders, for example grouping topics.

• Searching the on-line data dictionary finds whole words (e.g. searching for ‘child’ won’t find ‘children’ as well). However, an asterisk will represent any combination of characters. So searching for ‘child*’ will find ‘child’, ‘children’,

‘childcare’ etc.

• The introduction page for the data dictionary contains a list of topics and constructs that can be used for finding the information you want.

• The ‘Question ID’ field gives the variable name without any wave or person indicators. Filtering by this field is the best way to tell which questions were asked of or about which people at which wave.

• The ‘Topic ID’ field gives the topic and associated two digit question number for each item where this is appropriate. It can be used to link derived items with their associated input items.

Please contact the LSAC Data Management team if you need any help with using the

Data Dictionaries.

6

The data dictionary reflects the variables that are included in the main datasets (i.e. lsacgrb0, lsacgrb2, lsacgrb4, lsacgrb6, lsacgrk4, lsacgrk6, lsacgrk8, lsacgrk10). Items from the study child household and the PLE household modules, the NAPLAN items and the Medicare items are not in the data dictionary.

LSAC Data User Guide – November 2013

49

9 Data Transformations

The data from many of the responses to questions have been transformed to assist data users.

9.1 Transformations to ensure consistency

LSAC contains a number of items that have been asked slightly differently in different waves. Where this is logically supportable, items are recoded to match the variables produced from other waves. These recoded versions are provided in addition to the original item response. Some examples of this are:

• Income is generally collected as a continuous variable; however for the PLE in

Wave 2 it was collected using five categories. To assist users in comparing the responses of different informants, an additional variable containing the continuous income information recoded into these five categories is added wherever income has been collected continuously.

• In Wave 1, respondents were asked if the child received any regular childcare from a grandparent. In Wave 2, respondents were given the option of this being a maternal or paternal grandparent. In addition to the two variables giving this information separately for maternal and paternal grandparents, an extra variable has been added for whether the child is being cared for by a grandparent.

9.2 Transformations to update information

From Wave 2 on, there are a number of places in the questionnaire where respondents are asked about something happening since the last interview (or in the last 2 years if the study child is living in a new household). For example, in Wave 1 Parent 1 was asked how many homes the study child had lived in since birth, while in subsequent waves Parent 1 was asked how many homes the study child had lived in since the last interview. The datasets for the subsequent waves contain variables on the number of homes since the last interview and a tally of all the home the study child has ever lived in.

9.3 Summary measures for scales

The appropriate summary measure for each scale is included, based on advice from the Consortium Advisory Group. Where it is possible to logically implement either a mean or a sum score for a psychological scale or subscale, the preference of the

Consortium Advisory Group was to provide the calculation of means, except in cases where convention would dictate another scoring system. This enabled the calculation of scale level derivations where data measuring a construct has multiple contributing data items and where some contributing items are missing. Using a sum calculation for these scales would have lead to the exclusion of cases with any missing data. All contributing data items to these scales are included on the datasets.

For scales where there are different sets of items for children at different ages or for different informants, multiple versions of the same scale are calculated based on just those items shared between two versions of the scale. For example, the parenting hostility scale began as a 5-item measure for 0-1 year olds, but had one item dropped for children aged 4-7 years, and a further item dropped for children aged 8-9 years.

LSAC Data User Guide – November 2013

50

On the file for 0-1 year olds three different versions of the scale are calculated: one using all 5 items, another just using the 4 items included for children aged 4-7 years, and another using just those 3 items used for children aged 8-9 years. As a general rule data users should select the variable containing the greatest number of contributing items that is appropriate for their purpose. So, for analyses just using the hostility scale at aged 0-1 years, or for those comparing the hostility scale at ages 0-1 and 2-3 years, analysts should use the 5-item version. For analysts comparing hostility between the ages of 0 and 7 years should use the 4-item version, and for analysts comparing hostility between the ages 0 to 9 years should use the 3-item version.

9.4 Outcome Index measures

A unique component of the derivation and analysis work was the development and derivation of the LSAC Outcome Index, a composite measure that indicates how children are developing. LSAC tracks the development of children across multiple domains, and the Outcome Index provides a means of summarising this complex information for policy makers, the media and the general public, as well as data users.

In contrast to some other indices, which focus on problems or negative outcomes, the

LSAC Outcome Index wherever possible incorporates both positive and negative outcomes, reflecting the fact that most children have good developmental outcomes.

Thus the Outcome Index has the ability to distinguish groups of children developing poorly from those developing satisfactorily.

The rationale and methodology used to develop the Outcome Index are described in the LSAC Technical Paper No. 2 ‘Summarising children’s wellbeing: the LSAC

Outcome Index’. Papers on the derivation of the Waves 2 and 3 Outcome Index are forthcoming. Any users planning to use the Outcome Index are strongly advised to read the technical papers as they contain important information about the correct use of the variable (www.aifs.gov.au/growingup/pubs/technical/index.html). From Wave

4 the Outcome Index is not calculated.

When undertaking longitudinal analysis involving the Outcome Index, analysts should be cautious with using outcome indices from different waves in a pooled data file as they can use different measures at different waves to create the sub domains.

LSAC Data User Guide – November 2013

51

10 Confidentialisation

Two types of data are available to data users:

In-confidence data

General release data

10.1 In-confidence data

The only information not included is name, address and other contact details for the child, family, childcare agency and teacher or carer. Access to the in-confidence datasets may be granted where data users are able to demonstrate a genuine need for the additional data and that they meet the necessary additional security requirements.

10.2 General release data

In addition to the information removed for the in-confidence file, some other items have also been removed, and some items have either been transformed, had response categories collapsed, or have been top-coded (i.e. recoding outlying values to a less extreme value).

The following items are removed:

Qualitative data provided by respondents;

Census and postcode data for the location of carers and schools.

The following items are transformed:

Postcode – postcodes are given an indicator so that all children selected in the same postcode can be identified;

Date left hospital after birth – number of days between birth and departure.

The following items have response categories collapsed (i.e. response categories combined to form an aggregate category):

Parents’ occupation – output at 2-digit Australian Standard Classification of

Occupations (ASCO) level, or rounded off to the nearest 5 if ANU 4 ratings of occupational prestige;

Occupation in previous job – output at 2-digit ASCO level;

Socio-Economic Index for Areas (SEIFA) variables - rounded to the nearest 10;

Country of birth (coded as 0 if fewer than five contributors);

Religion (coded as 0 if fewer than five contributors);

Language Other Than English (LOTE) (coded as 0 if fewer than five respondents).

The following data items are top-coded:

Income

Housing costs

Child support paid by Parent 2

Children and Parent’s current height, weight and waist circumference

Number of hours spent in childcare

LSAC Data User Guide – November 2013

52

11 Data imputation

Limited imputation of data is undertaken in LSAC. In general, imputation occurs only when there is clear contradiction between data items and there is a good reason to believe one item over the other.

Some basic principles are applied for this task

11.1.1 Virtual roll-forward

‘Roll-forward’ is the term in CAI design that refers to the use of data from a previous wave of data collection to determine the questions that need to be asked in a subsequent wave. For Wave 2 a limited set of data was rolled forward, largely to assist with the household composition module. Time and resource implications meant that roll-forward could not be used in some other parts of the questionnaire where it may have reduced respondent burden.

For example, in Wave 2 respondents were again asked about the age the child stopped being breastfed, in order to obtain the information from those cases where this had not yet happened at the time of Wave 1. In re-asking this question, some respondents gave different answers to their Wave 1 responses. Given that recollection of respondents is likely to be more accurate closer to the event (i.e. the cessation of breastfeeding), it was decided that in cases where Wave 1 data exists the Wave 1 value is taken as correct, and the Wave 2 value is ignored (i.e. as if the Wave 1 data had been rolled forward and the question never asked in Wave 2). This means a single variable is produced that represents the best estimate from the two waves of data. (Users are able to tell at which wave the timing data was collected by referring to the question from each wave asking if the child is still being breastfed.)

Note: From Wave 3 onwards there is a greater use of roll-forward, which reduced the number of situations where such conflicts could occur.

11.1.2 Longitudinal contradictions

Another possible contradiction in the data may occur where respondents report at a subsequent wave that an event occurred at a time before a previous wave, when the previous wave’s data indicated that this event hadn't happened yet.

In these cases the time of the previous wave is treated as the time of the event. For example, if a parent reported at Wave 2 that the child stopped being breastfed after two months, however at Wave 1 the child was three months old and was reported as still being breastfed, the age of breastfeeding cessation would be set to three months.

This strategy for fixing the time of an event is also used for:

• Date when new members joined the household;

• Length of attendance at a particular childcare facility;

• Date left the household for Wave 1 members and temporary members (bf14m1, bf14m2, etc.)

• Age stopped breastfeeding (zf05c)

• Age first had non-breast milk (zhb07)

• Age first had solid food (zhb10)

• Age entered child care arrangements (bpc11a, bpc11b, etc.)

• Age last lived with 2 biological parents (bpe23c)

LSAC Data User Guide – November 2013

53

11.1.3 Other imputations

On inspection of the data, problems were revealed in a small number of items that were solved using imputation:

Employment status: some assumptions are made to assist in coding the parent to employed, unemployed or not in the labour force where missing values were present.

Type of educational program (K cohort Wave 1): There appeared to be some confusion with parents and interviewers as to whether the child was in pre-school or pre-year 1 at school. The type of education program variable was amended based on the teacher data and other information provided in the questionnaire.

Parental income: Outlying values, particularly those with responses to other questions (e.g. categorical income, sources of income) that make the income value appear incorrect, were adjusted.

Parental height: It was found that there were some changes in height between waves for some parents of study children. While most were minor (most likely due to estimation error) some were more substantial, and called into question the reliability of differences in Body Mass Index recordings between waves.

Time Use Diary data: Responses were recorded by marking an oval which indicated whether an activity/situation occurred in each 15 minutes time period.

A number of “false positives” were discovered in the Wave 1 TUD data.

Imputation was used to reduce the number of false positives. A number of imputations were also performed to improve data quality in all three waves.

Further details of these imputations are given in the Data Issues papers.

LSAC Data User Guide – November 2013

54

12 Survey Methodology

LSAC employs a cross-sequential design that follows two cohorts of children, initially aged 0-1 years (B cohort) and 4-5 year olds (K cohort) in 2004.

Families are visited by interviewers every two years to collect data for the main waves of the study. In the “between” years, a mail-out survey was conducted, to help maintain contact with families and obtain some additional information, at Waves 1.5,

2.5, 3.5 and 4.5.

The key features of the initial sample design and methodology for each wave are included in this section. A full description of the sample design is given in LSAC

Technical Paper No. 1, and details of the weighting and non-response analysis are given in Technical Papers no. 3, 5 and 6

(www.aifs.gov.au/growingup/pubs/technical/index.html).

12.1 Sample design

A two-stage clustered sample design was employed, first selecting postcodes then children, with the clustered design allowing analysis of children within communities and producing cost savings for interviews.

Stratification was used to ensure proportional geographic representation for states/territories and capital city statistical division/rest of state areas. The sample was stratified by state, capital city statistical division/ balance of state and two strata based on the size of the target population in the postcode.

Postcodes were selected with probability proportional to size selection where possible, and with equal probability for small population postcodes. Children from both cohorts were selected from the same 311 postcodes. Some remote postcodes were excluded from the design, and the population estimates were adjusted accordingly.

Children were selected with approximately equal chance of selection for each child

(about one in 25).

Apart from some remote areas, the sample was selected to be representative of all

Australian children (citizens and permanent residents) in each of two selected age cohorts:

• children born March 2003-February 2004 (B cohort)

• children born March 1999-February 2000 (K cohort).

12.1.1 Sample selection and recruitment

The sample was selected from Medicare Australia’s enrolment database. Within the selected postcodes, the population was ordered by date of birth and then a random start and skip applied to select the children. The actual number of children selected depended on which stratum the postcode was in, but for most postcodes the aim was to recruit about 20 children per cohort per postcode.

The selection of children and corresponding Wave 1 fieldwork occurred in 4 phases, partly to reduce the age range of children at interview, and partly because some of the target population had not been born at the time of the first phase selection.

LSAC Data User Guide – November 2013

55

Families of 18,800 selected children received letters of invitation to take part in the study sent by Medicare Australia. Families could “opt-out” of the study by phoning a

1800 number or returning a reply paid slip. Medicare Australia 1800 staff were given training about the study and were able to answer queries and make note of other information (for example, telephone numbers).

After a 4 week opt-out period, Medicare Australia gave the contact names and addresses of remaining families to I-view, the Wave 1 data collection agency. I-view then sent another letter to families saying when an interviewer would be in their area.

I-view maintained a 1800 number for families selected in the study which was transferred to the Australian Bureau of Statistics (ABS) once ABS had responsibility for the data collection (from Wave 2 on).

12.2 Development and testing of survey instruments

12.2.1 Pretesting

Pre-testing of new material and processes is undertaken at each wave of the study, comprising small scale pre-tests and cognitive interviews. In Waves 1 and 2, more formal piloting was also undertaken. Small scale testing is also undertaken for the between-wave surveys.

Wave 1

• Development began in March 2002.

• Small scale pre-testing occurred in September-October 2002.

• Pilot test with about 50 families from each cohort was conducted in March-April

2003.

Wave 2

• Development began in July 2004.

• Small scale pre-testing occurred in September-October 2004.

• Pilot test with 86 families conducted in April 2005.

Wave 3

• Development began in March 2006.

• Pretesting occurred in a number of stages from mid 2006 to March 2007

• No pilot test was required.

Wave 4

• Development began in February 2008.

• Pretesting occurred in a number of stages from mid August 2008 to June 2009

• No pilot test was required.

Wave 5

Development began in February 2010.

Pretesting occurred in a number of stages from mid June 2009 to March 2010

No pilot test was required.

12.2.2 Dress Rehearsal

In Wave 1 a Dress Rehearsal (DR) sample of 526 families was recruited to test the content and processes intended for the main waves of the study. Over 1000 children were initially selected from 25 postcodes in Victoria, Sydney and rural/remote New

LSAC Data User Guide – November 2013

56

South Wales and Queensland. Postcodes in Victoria were selected at random, but the other postcodes were selected as areas that may provide challenges to the data collection process.

Wave 1 DR – August-November 2003 (526 families interviewed)

Wave 2 DR – September-November 2005 (423 families interviewed)

Wave 3 DR – July-October 2007 (420 families interviewed)

Wave 4 DR – July-October 2009 (387 families interviewed)

Wave 5 DR - July-August 2011 (451 families interviewed)

After each DR, both processes and content have been refined to increase efficiency and reduce the time in the home.

12.3 Data collection

12.3.1 Interview length

Details of the instruments administered each wave are given in the “Content of Each

Wave” section.

In Wave 1, an average of 126 minutes was allowed for time in the home by the interviewer. In-home data collection with the B cohort averaged about 1 ½ hours, while interviews for the K cohort averaged about 2 ½ hours.

In Wave 2, although an average of 90 minutes had been allowed for the time in the home, the actual time was shorter, averaging 66 minutes for the B cohort and 85 minutes for the K cohort.

In Wave 3, an average of 100 minutes in the home was allowed for time in the home; the actual time was 91 minutes for the B cohort and 98 minutes for the K cohort.

In Wave 4, an average of 110 minutes in the home was allowed for time in the home; the actual time was 102 minutes for the B cohort and 108 minutes for the K cohort.

In Wave 5, an average of 110 minutes in the home was allowed for time in the home; the actual time was 98 minutes for both cohorts

12.3.2 Interviewers

As part of standard ABS interviewer induction, ABS interviewers receive two weeks of intensive training across a range of standard procedures and practices.

All interviewers received 8 hours of home learning (Computer-Based Learning module,

Home Study Exercises, reading of Interviewer Instructions).

In Wave 1, 150 interviewers and field supervisors from I-view were trained during a series of 4-day sequential training courses conducted in Melbourne, Brisbane, Perth and Sydney during February to early March 2004. The principal trainers were the same for all courses ensuring consistency in training.

Psychologists conducted the training for the Who am I?, the PPVT and the interviewer observations. A large part of the training involved practice interviews, with one day devoted to interviews with parents and children.

For Wave 2, 147 interviewers from ABS were trained in a series of 3-day training courses in Sydney, Melbourne, Brisbane and Perth during March and April 2006.

LSAC Data User Guide – November 2013

57

Two training teams were used, comprising staff from both AIFS and ABS. This time,

AIFS staff undertook the direct assessment training, after receiving training from a child psychologist (the use of Computer Assisted Interviewing for the direct assessments helped ensure the consistent administration of these assessments).

For Wave 3, 176 interviewers from ABS were trained in a series of 2-day training courses in Brisbane, Melbourne, Sydney and Perth during March and April 2008.

Interviewers who had not worked on LSAC previously were given background training in LSAC before the 2-day course commenced. Two training teams were used, comprising staff from the ABS, AIFS and DSS. Again, AIFS staff undertook the direct assessment training.

For Wave 4, 181 interviewers from ABS were trained in a series of 3-day training courses in Brisbane, Melbourne, Sydney and Perth. Two training teams were used, comprising staff from the ABS, AIFS and DSS. As in previous waves, AIFS staff undertook the direct assessment training.

For Wave 5, 198 interviewers from ABS were trained in a series of 3-day training courses in Brisbane, Melbourne, Sydney, Adelaide and Perth. New to LSAC interviewers (defined as anyone who did not participate in Main Wave 4) attended the first day of classroom training where topics such as ‘Background to the Study,

‘Physical measurements”, ‘Direct Assessments’ and ‘Notebook security’ were covered. All Interviewers attended Days 2 and 3 where the P1, K and B child interview are covered in detail (apart from what was done on Day 1). New

Interviewers were teamed with an experienced Interviewer allowing for mentoring throughout the training course, and for the new Interviewers to be the Interviewer during practice sessions.

12.3.3 Fieldwork periods

Wave 1

Selected postcodes were divided into 2 groups for maximum field efficiency. The target population was also divided into 2 groups: children born March-August (older) in one group and children born September-February (younger) in the other.

The fieldwork was divided into 4 phases:

• Phase 1 started in mid March 2004 for the older children in the first group of postcodes;

• Phase 2 started at the end of April for the older children in the second group of postcodes;

• Phase 3 started in June for the younger children in the first group of postcodes; and

• Phase 4 started in late July for the younger children in the second group of postcodes.

Follow-up continued throughout 2004. The blue line in Figure 6 shows the distribution of interviews over time for Wave 1 fieldwork.

Wave 2

Again there were broadly 4 fieldwork periods, although the dates for these varied from state to state. Regional offices of the ABS were able to organise the work to suit

LSAC Data User Guide – November 2013

58

the availability of interviewers and other work. As far as possible, ABS tried to interview the children born in March-August in the first 2 periods, and children born in September-February in the later fieldwork periods. 84% of the interviews were conducted prior to September 2006.

Figure 6 shows the distribution of interviews over time for Wave 2 fieldwork.

Fieldwork started later than in Wave 1 due to the additional work required to prepare the CAI instrument.

Wave 3

Fieldwork was organised as per Wave 2. The green line in Figure 6 shows the distribution of interviews over time for Wave 3 fieldwork.

Wave 4

Fieldwork was organised as per Waves 2 and 3. The dark blue line in Figure 6 shows the distribution of interviews over time for Wave 4 fieldwork. However, as the children are getting older, the age differences within a cohort are less significant, and to assist the efficiency of work allocations to Interviewers, in Wave 4 not as much emphasis was given to following interviews within the set phases.

Wave 5

Fieldwork was organised as per Waves 2, 3 and 4. Figure 6 shows that the distribution of interviews for Wave 5 fieldwork was more spread out across the months than for previous Waves.

Figure 6. Month of interview for study families in Waves 1 to 5

LSAC Data User Guide – November 2013

59

12.3.4 Contact process

Wave 1

For most families, the interviewer only had the name and address of the Medicare cardholder, and which cohort the child was in. In a small number of cases, families who were keen to participate had contacted the 1800 numbers and supplied phone numbers and/or best times to call.

Interviewers were required to make up to 6 visits to the address, at different times of the day and on different days of the week. A major challenge was that 7% of addresses were post office box addresses, and although families with these addresses were specifically requested to make contact with the 1800 number to supply a residential address, only a small proportion did so. In addition, many of the residential addresses held by Medicare were found to be out-of-date by the time the interviewers visited. Interviewers made significant attempts to locate families for whom they did not have a current residential address, by referencing White Pages and electoral rolls and speaking with neighbours and other local contacts.

Between waves

Contact is maintained with study families between waves by sending birthday cards, annual calendars and newsletters and via the between-wave mail-out questionnaires in

Waves 1, 2 and 3. These processes have resulted in some families contacting the ABS to update their contact information, which helps when trying to arrange appointments for the main waves of interviewing.

Subsequent waves

Pre-interview letters plus a brochure outlining the processes for that wave were sent to all families who had not opted out of the study since the previous wave, unless it was confirmed that the address was out-of-date. Interviewers then followed up with a telephone call to make an appointment for an interview. If the contact information was out of date, the interviewers tried to contact secondary contacts of Parent 1 (these details were given by Parent 1 in Wave 1 and are updated each wave) to locate the family. One visit to the address was also made. If the family could not be located, the interviewer referred this back to the office for tracking.

After an appointment for interview was made, the interviewer confirmed the appointment the day before the appointment.

12.3.5 Foreign language interviews

Wave 1

As part of the Medicare Australia mail-out, a brochure was included with information about the study in nine languages. Medicare Australia staff made use of the Telephone

Interpreter Service (TIS) to assist with calls where required.

Apart from this brochure, no other study material was (or has been) translated into other languages, and instead interpreters were used. An interpreter was required in 3% of interviews, with over 50 languages involved. In most cases (138), a member of the family or friend was preferred as the interpreter. In 76 cases an I-view employee was able to act as interpreter, and in 96 cases, an interpreter was employed.

LSAC Data User Guide – November 2013

60

Wave 2

A total of 110 interviews (1%) were conducted in a language other than English, in 23 different languages. Family or friends assisted in 58 cases, ABS interpreters helped in

37 cases and a TIS interviewer was used for 15 families. An interpreter was arranged whenever requested or judged necessary by the interviewer. The reduction in use of interpreters between waves is presumably due to the increased confidence in English that has been gained by respondents in this time.

Wave 3

A total of 97 interviews needed an interpreter, in 24 languages. Family or friends assisted in 58 cases, ABS interpreters helped in 31 cases and a TIS interviewer was used for 8 families.

Wave 4

A total of 93 interviews needed an interpreter, in 26 languages. Family or friends assisted in 50 cases, ABS interpreters helped in 29 cases and a TIS interviewer was used for 14 families.

Wave 5

A total of 81 interviews needed an interpreter, in 18 languages. Family or friends assisted in 47 cases, ABS interpreters helped in 24 cases and a TIS interviewer was used for 10 families.

12.3.6 Indigenous communities

Although the sample selection process excluded 40% of areas classified as remote by the ABS (areas that typically have a high Indigenous population) there were still a number of postcodes selected that contained some remote Indigenous communities, hence strategies have been put in place to enumerate these communities.

Where feasible, communities were visited or telephoned, and personal contact made with a number of community organisations from whom assistance was gained to identify whether families were in residence and willing to be interviewed. Travel to remote communities was only undertaken if there was an appointment for an interview.

Aboriginal and Torres Strait Islander families are included in representative numbers in non-remote centres. However, there has been a higher rate of attrition from the study among these families. See the weighting and non-response technical papers for more details (www.aifs.gov.au/growingup/pubs/technical/index.html).

12.3.7 Remote areas

In the initial sample there were 12 postcodes selected in areas classified as “remote” by the ABS Australian Standard Geographic Classification (ASGC) Remoteness

Classification. Interviewers were either recruited from these areas or travelled to these areas when the field agency did not have a suitable interviewer in the locality.

Where visits were not possible, telephone interviews were conducted: 12 in Wave 1,

42 in Wave 2, 87 in Wave 3, 83 in Wave 4 and 73 in Wave 5. The increasing number is due to sample dispersion.

LSAC Data User Guide – November 2013

61

12.4 Fieldwork response

12.4.1 Wave 1 recruitment

The final response to the recruitment of children was 54 per cent of those families who were sent the initial letter by Medicare Australia. The response rate was higher for the B cohort with 57 per cent of families (5,107) agreeing to take part, compared with 50 per cent of K cohort families (4,983).

About 35% of families who were sent the initial letter refused to take part in the study. The main reasons given to interviewers for not participating in the study were: not interested/too busy (57%), not capable/moving/overseas (9%), husband refused

(9%), and illness/death (8%). The remaining 13% of families were not able to be contacted, despite intensive efforts from interviewers.

Non-response analysis was undertaken to determine how representative the sample is of all Australian children in the scope of this study, and adjustments have been made to the survey weights to allow for this. For further information on the weighting and non-response, see LSAC Technical paper no 3, “Wave 1 weighting and non-response analysis”. www.aifs.gov.au/growingup/pubs/technical/index.html

LSAC Data User Guide – November 2013

62

Response in later waves

Table 9 summarises the response from families in later waves, using the Wave 1 sample and “available” sample as the bases for comparisons.

Table 9. Sample size and response rate for each wave and cohort of LSAC.

B cohort K cohort Total

Main waves

Wave 1 original

Wave 2 available

1

Wave 2 responding

2

Wave 3 available

No.

5107

Resp. rate

Resp. of rate of

Wave 1

(%) available sample

(%)

No.

100

Resp.

Wave 1

Resp. rate of rate of available sample

No.

(%)

(%)

4983 100 10090

Wave 1

Resp.

Resp. rate of rate of available sample

(%)

(%)

100

5047

4606

98.8

90.2 91.2

4971 97.3

Wave 3 responding 4386 85.9 88.2

Wave 4 available 4929 96.5

4913

4464

4829

4332

4774

98.6

89.6

96.9

86.9

95.8

90.9

89.7

9960

9070

9800

8718

9703

98.7

89.9

97.1

86.4

96.2

91.1

89.0

Wave 4 responding 4241 83.0 86.0

Wave 5 available 4884 96.6

4164 83.5

4735 95.0

87.2 8405 83.3 86.6

9619 95.3

Wave 5 responding 4085 80.0 91.1

Between-waves

Wave 1.5 sent 5061 99.1

Wave 1.5 returned 3573 70.0 70.6

Wave 2.5 sent 4859 95.1

3956

4935

3584

4712

79.4

99.0

71.9

94.6

83.5

72.6

8041

9996

7157

9571

79.7

99.1

71.0

94.9

83.6

71.6

Wave 2.5 returned 3268 63.5 64.0

Wave 3.5 sent 4772 93.4

3287

4641

65.5

93.1

66.0 6555

9413

63.4

93.3

65.0

Wave 3.5 returned 3012 59.0 63.1 2972 59.6 64.0 5984 59.3 63.6

1 available sample excludes those who opted out of the study between waves. Some additional families also opted out permanently during the fieldwork process

2

those who had home visit

Table 10 details the reasons why interviews were not obtained in Waves 2, 3, 4 and 5.

Table 10. Response status and reasons for non-response by wave

Response status

Responding

Refusal

Non contact

Away-entire enumeration period

Death of study child

Total starting sample

No.

9070

Wave 2

284

540

61

5

%

91.1

2.8

5.4

0.6

0.1

No.

8718

436

552

93

1

Wave 3

%

89.0

4.4

5.6

1.0

0.01 0

Wave 4

No.

8405

637

526

135

%

86.6

6.6

5.4

1.4

0

Wave 5

No.

8041

774

715

88

1

%

83.6

8.0

7.4

0.9

0.01

9960 100.0 9800 100.0 9703 100.0 9619 100.0

LSAC Data User Guide – November 2013

63

13 Important issues for data analysis

The new Data Issues series has been initiated with a set of papers that had appeared as attachments to previous versions of the Data User Guide. These will be added to as other issues are addressed. The current set of papers includes:

• Issues Paper no.1. Cleaning of Time Use Diary Data

• Issues Paper no.2. Report on Adapted PPVT-III and Who Am I?

• Issues Paper no.3. Imputations to solve missing data problems in Wave 2.5

• Issues Paper no.4. Investigation of Educational program type (cpc06a4) in

Wave 1

• Issues Paper no.5. Cleaning of income data

• Issues Paper no.6. Height differences

• Issues Paper no.7. Data issues in Wave 3.5

• Issues Paper no.8. Data issues in Wave 4

• Issues Paper no.9 Data issues in Wave 5

Other important issues are addressed below.

13.1 Weighting and external validity

The LSAC study design, based on a complex probability sample, is specifically designed to produce valid estimates at the population level. Unlike clinically based or convenience samples, the LSAC sample is population-based by design. By properly accounting for the survey design when analysing the data it is possible not only to make inferences about the children and families participating in the study, but to make valid inferences about the entire population of children in the relevant age groups.

The LSAC sampling strategy has three important elements that distinguish it from a simple random sample (SRS):

stratification to ensure proportional representation of all states and both capital city and ex-metropolitan areas;

clustering by postcode to both reduce field enumeration costs and allow the study of community level effects on children’s development and wellbeing; and

weighting to adjust for potential non-response bias and to provide population estimates.

• It is the responsibility of data users to determine when and how each of these needs to be accounted for when developing their analyses.

13.1.1 Stratification

Stratification, by state and part of state, was employed to ensure that all geographic areas within Australia are represented in the sample in proportion to their population.

This produces a more even distribution of the sample across geographic areas than could be expected from a simple random sample.

The use of stratification can be expected to reduce standard errors compared with a simple random sample with no control over the geographic spread of the sample. As such, when trying to extrapolate to the population the stratification should be incorporated in the analysis of results from the survey in order to correctly calculate standard errors and confidence intervals.

LSAC Data User Guide – November 2013

64

13.1.2 Clustering

The use of clustering in the sample design has important consequences for the analysis of data from the study. Clustering is useful in reducing the field costs associated with the survey enumeration. Clustering also has the added benefit of making possible the analysis of community level effects, by ensuring that sufficient sample is selected from each postcode included in the survey.

However, the use of clustering violates the standard assumption of independence of the observations that is fundamental to many statistical routines in major statistical packages. When children or carers have more similar characteristics within a given postcode than children or carers selected purely at random, the responses within postcodes will be correlated. This correlation will lead to an increase in the standard errors and size of the confidence intervals. The extent of this increase is measured by the design effect, which is the ratio of the variance of an estimate from the survey to the variance that would have been achieved by a simple random sample of the same size.

Failure to account for clustering in the analysis can lead to under-estimating the size of standard errors and confidence intervals. In some circumstances this can result in misleading conclusions of statistical significance.

13.1.3 Weighting

The Wave 1 weights provided in the LSAC data files take into account both the probability of selecting each child in the study, and an adjustment for non-response.

An analysis of possible differences in the characteristics of respondents and nonrespondents was undertaken and identified two factors associated with the probability of participating in the survey - whether the mother speaks a language other than

English at home, and whether the mother has completed year 12. Both of these factors were incorporated into the Wave 1 survey weighting so that, to the best extent possible, the use of the sample weights offset the bias that may be introduced into the data due to differential non-response patterns.

At each subsequent wave of data collection weights have been adjusted to account for the differential probability of response as estimated by regression. The weights are then calibrated back to the stratum benchmarks and a small number of cases have their weights top or bottom coded to prevent any case having too great or small an effect on the data.

From Wave 3 onwards it is required to produce longitudinal as well as cross-sectional weights for the first time. Cross-sectional weights adjust the sample attained at current wave to be representative of the population at the time of selection, while longitudinal weights do the same for the sample that has responded to all waves of the survey.

More detailed information on the weighting variables can be found in LSAC

Technical Papers no. 3, 5, 6, 9 and 10.

(www.aifs.gov.au/growingup/pubs/technical/index.html).

Three types of weight are included in the LSAC datasets:

Child population weights – these weights are used to produce population estimates based on the LSAC data (e.g. based on LSAC data there are 22,464 children born in March 2003 to February 2004 in Australia that were never breastfed).

LSAC Data User Guide – November 2013

65

The sum of the responding B cohort child population weights is 243,026 and the sum of the K cohort child population weights is 253,202, which are the ABS estimated resident population counts of children aged 0 years and 4 years, respectively, at end March 2004, adjusted for the remote parts of Australia that were excluded from the study design.

Child sample weight – this is the child population weight rescaled such that the sum of the weights matches the number of children in the sample (e.g. 5,107 B cohort and 4,983 K cohort in Wave 1).

This weight is used in analyses that expect the weights to sum to the sample size rather than the population, particularly when tests of statistical significance are involved.

Time Use Data day weight (for Waves 1, 2 and 3 only)- this is the sample weight adjusted so that each day of the week receives equal weight in analyses of time use data.

Data files for Wave 1 and Wave 2 each have one population weight and one sample weight. Given that there are no cases that responded to Wave 2 that didn’t respond to

Wave 1, these weights can be used for both longitudinal and cross-sectional analyses.

At Wave 3, two sample weights and two population weights are necessary as this is the first time that respondents could return to the study after missing a wave. The first of these weights the full Wave 3 sample and should be used for cross-sectional analyses. The second weights the sample that has responded to all three waves, and should be used for longitudinal analyses.

A complete list of LSAC weighting variables is given in Table 11.

Table 11. Weighting variables

Variable name

aweight B aweights B aweightd B bweight B

Cohort Type Waves cases responded to

Population 1

Sample

Day

1

1

Population 1 & 2 bweights B bweightd B cweight B cweights B

Sample

Day

Population 1 & 3

Sample

1 & 2

1 & 2

1 & 3

Used for

Wave 1 cross-sectional analyses

Wave 1 cross-sectional analyses

Wave 1 cross-sectional analyses

Wave 2 cross-sectional analyses and longitudinal analyses involving Waves 1 & 2

Wave 2 cross-sectional analyses and longitudinal analyses involving Waves 1 & 2

Wave 2 cross-sectional analyses and longitudinal analyses involving Waves 1 & 2

Wave 3 cross-sectional analyses and longitudinal analyses involving Waves 1 & 3

Wave 3 cross-sectional analyses and longitudinal analyses involving Waves 1 & 3

LSAC Data User Guide – November 2013

66

Variable name

cweightd B bcwt bcwts bcwtd dweight dweights B eweight eweights B bdwt bdwts cdwt cdwts bcdwt bcdwts bcdewt bcdewts cweight

Cohort Type

B

B

B

B

B

B

B

B

B

B

B

B

B

K cweights K cweightd K dweight K

Waves cases responded to

Used for

Day 1 & 3 Wave 3 cross-sectional analyses and longitudinal analyses involving Waves 1 & 3

Population 1, 2 & 3 Longitudinal analyses involving

Waves 2 & 3, or Waves 1, 2 & 3

Sample

Day

1, 2 & 3 Longitudinal analyses involving

Waves 2 & 3, or Waves 1, 2 & 3

1, 2 & 3 Longitudinal analyses involving

Waves 2 & 3, or Waves 1, 2 & 3

Population 1 & 4 Wave 4 cross-sectional analyses and longitudinal analyses involving Waves 1 & 4

Sample

Population 1 & 5

Sample

1 & 4

1 & 5

Wave 4 cross-sectional analyses and longitudinal analyses involving Waves 1 & 4

Wave 5 cross-sectional analyses and longitudinal analyses involving Waves 1 & 5

Wave 5 cross-sectional analyses and longitudinal analyses involving Waves 1 & 5

Population 1, 2 & 4 Longitudinal analyses involving

Sample

Waves 2 & 4, or Waves 1, 2 & 4

1, 2 & 4 Longitudinal analyses involving

Waves 2 & 4, or Waves 1, 2 & 4

Population 1, 3 & 4 Longitudinal analyses involving

Waves 3 & 4, or Waves 1, 3 & 4

Sample 1, 3 & 4 Longitudinal analyses involving

Waves 3 & 4, or Waves 1, 3 & 4

Population 1, 2, 3 & 4 Longitudinal analyses involving

Waves 2, 3 & 4, or Waves 1, 2,

3 & 4

Sample 1, 2, 3 & 4 Longitudinal analyses involving

Waves 2, 3 & 4, or Waves 1, 2,

Population 1, 2, 3, 4

& 5

3 & 4

Longitudinal analyses involving

Waves 2, 3, 4, & 5 or Waves 1,

2, 3, 4 & 5

Sample 1, 2, 3, 4

& 5

Population 1

Longitudinal analyses involving

Waves 2, 3, 4, & 5 or Waves 1,

2, 3, 4 & 5

Wave 1 cross-sectional analyses

Sample

Day

1

1

Population 1 & 2

Wave 1 cross-sectional analyses

Wave 1 cross-sectional analyses

Wave 2 cross-sectional analyses and longitudinal analyses involving Waves 1 & 2

LSAC Data User Guide – November 2013

67

Variable name

dweights K dweightd K eweight eweights K eweightd K dewt dewts dewtd fweight fweights K dfwt dfwts efwt efwts defwt defwts gweight

Cohort Type

K

K

K

K

K

K

K

K

K

K

K

K

Waves cases responded to

Used for

Sample

Day

Population 1 & 3

Sample

1 & 2

1 & 2

1 & 3

Wave 2 cross-sectional analyses and longitudinal analyses involving Waves 1 & 2

Wave 2 cross-sectional analyses and longitudinal analyses involving Waves 1 & 2

Wave 3 cross-sectional analyses and longitudinal analyses involving Waves 1 & 3

Wave 3 cross-sectional analyses and longitudinal analyses involving Waves 1 & 3

Day 1 & 3 Wave 3 cross-sectional analyses and longitudinal analyses involving Waves 1 & 3

Population 1, 2 & 3 Longitudinal analyses involving

Waves 2 & 3, or Waves 1, 2 & 3

Sample

Day

1, 2 & 3 Longitudinal analyses involving

Waves 2 & 3, or Waves 1, 2 & 3

1, 2 & 3 Longitudinal analyses involving

Waves 2 & 3, or Waves 1, 2 & 3

Population 1 & 4 Wave 4 cross-sectional analyses and longitudinal analyses

Sample 1 & 4 involving Waves 1 & 4

Wave 4 cross-sectional analyses and longitudinal analyses involving Waves 1 & 4

Population 1, 2 & 4 Longitudinal analyses involving

Waves 2, 3 & 4, or Waves 1, 2

& 4

Sample 1, 2 & 4 Longitudinal analyses involving

Waves 2 & 4, or Waves 1, 2 & 4

Population 1, 3 & 4 Longitudinal analyses involving

Waves 3 & 4, or Waves1, 3 & 4

Sample 1, 3 & 4 Longitudinal analyses involving

Waves 3 & 4, or Waves 1, 3 & 4

Population 1, 2, 3 & 4 Longitudinal analyses involving

Waves 2, 3 & 4, or Waves 1, 2,

Sample

3 & 4

1, 2, 3 & 4 Longitudinal analyses involving

Waves 2, 3 & 4, or Waves 1, 2,

3 & 4

Population 1 & 5 Wave 5 cross-sectional analyses and longitudinal analyses involving Waves 1 & 5

LSAC Data User Guide – November 2013

68

Variable name

Cohort Type

gweights K defgwt defgwts

K

K

Waves cases responded to

1 & 5

Used for

Sample

Population 2, 3, 4, &

5

Sample 2, 3, 4, &

5

Wave 5 cross-sectional analyses and longitudinal analyses involving Waves 1 & 4

Longitudinal analyses involving

Waves 2, 3, 4 & 5 or Waves 1,

2, 3, 4, & 5

Longitudinal analyses involving

Waves 2, 3, 4 & 5 or Waves 1,

2, 3, 4, & 5

13.1.4 Survey estimation and analysis techniques

Survey estimation and analysis techniques are available that can take all three key features of the study design into account, and many of these techniques are now included in commercially available software. Incorporating the study design features into analyses of the study can produce externally valid results at the full population level. Estimates of means, proportions and totals incorporating the study design provide the best estimate of the true means, proportions and totals within the total population.

Analytic techniques, particularly modelling, aim at exploring relationships within the data, are able to estimate the best fitting model for the underlying population not just the best fitting model for the sample, when properly applied to account for the study design.

13.1.5 Useful references

An overview of population survey methods is given by Levy and Lemeshow (1999).

They discuss the use of stratification, weighting and clustering in survey design, and the impact it has on the analysis of sample survey data.

For a thorough discussion of the mathematical techniques used to analyse data from complex surveys, see Chambers and Skinner (2003).

13.1.6 Software

There is now a range of software available from a number of vendors that supports the analysis of data from complex survey designs incorporating stratification, clustering and weighting. These include SAS (using the SURVEYMEANS and SURVEYREG procedures), STATA (using the svy commands), and SPSS (through the SPSS

Complex Samples add-on module), as well as software packages specifically designed for the analysis of sample survey data such as WesVar and SUDAAN.

Use of the appropriate analytic techniques from one or more of these packages is recommended for researchers analysing the LSAC data. Results that properly account for the sample design features will have the greatest external validity and should be appropriate for drawing inferences about the total population of children from which the sample was drawn.

LSAC Data User Guide – November 2013

69

A template for using the SURVEYREG and SURVEYMEANS procedures in SAS is shown in Figure 7.

Figure 7. SURVEYREG and SURVEYMEANS procedures in SAS

proc surveyreg data=<filename> total=<stratumfile>;

stratum stratum;

cluster pcodes;

model <standard SAS model details>;

weight weights; run; proc surveymeans data=<filename> total=<stratumfile>;

stratum stratum;

cluster pcodes;

var <variable names>;

weight weights; run;

Where: stratum: is a variable you can calculate for lsac0 using the formula:

stratum=int(mod(hicid,10000000)/100000); pcodes: is the postcode of selection (already on the data file) weights: is the sample weight (preferred to the population weight for this analysis)

<stratumfile> is a file that contains the number of Primary Sampling Units (in this case postcode clusters) in each stratum. It is included on the data CD or can be set up using the following code.

52 86

53 32

54 103

61 28

63 38

71 9

73 3

74 1

81 23

; run; data stratum;

input stratum _total_; datalines;

11 295

13 168

14 160

21 202

22 58

23 95

24 316

31 116

33 121

34 108

41 110

43 34

44 131

51 82

LSAC Data User Guide – November 2013

70

13.2 Unit of analysis

The child is the unit of selection in LSAC and estimates produced from this survey are of children, not of parents or families. It is important this point is understood when producing population estimates from this survey.

Using the estimates to count families/parents will produce an over count of the number of families/parents, due to the multiple (or over) counting of children from multiple births. Although this will not make a huge difference to the actual numbers, it may be important in the interpretation of the information and in comparing data from other sources.

Although it is possible to produce ‘family’ weights, it is not considered a worthwhile use of resources given the small number of analyses this could possibly meaningfully affect

13.3 Age at interview

Different ages of children should be accounted for in any analyses focused on age dependent measures such as cognitive and motor development. Figures 8 and 9 show the age distribution of the two cohorts at each Wave. The figures show the age of the child as a base figure (ie, 0, 2, 4, 6 or 8 years) plus a number of months. For example, a B cohort child aged 3 years 1month at time of interview in Wave 2, is shown against

“13” on the x-axis on the red line.

LSAC Data User Guide – November 2013

71

Figure 8. Age distribution of B cohort sample at each wave.

Figure 9. Age distribution of K cohort sample at each wave.

LSAC Data User Guide – November 2013

72

13.4 Time between interviews

Effort is made to ensure that the time between interviews is close to two years, however in some cases this is not possible. Figure 10 shows the distribution of the intervals between waves.

Figure 10. Distribution of time between interviews.

B Cohort Wave 1 to 5

K Cohort Wave 1 to 5

13.5 Cross-cohort comparisons

It should be noted that the two cohorts of LSAC were selected and weighted to represent similar but different populations. For the B cohort the reference population

LSAC Data User Guide – November 2013

73

is ‘0 year old children in Australia in 2004 excluding those from certain remote postcodes’, while for the K cohort the reference population is ‘4-year old children in

Australia in 2004 excluding those from certain remote postcodes’. One implication of this is that the K cohort will have a greater number of children born overseas as there was more time for families to immigrate to Australia between the birth of their child and selection into the study. The 2001 census contained 4.4% of 4 year olds that were born overseas compared with 0.8% of 0 year olds. In comparison, the weighted percentages for these figures in LSAC at Wave 1 were 4.2% v 0.4%.

However, there are also other demographic differences between the populations that are reflected in the benchmarks used to weight the two cohorts. Figure 11 shows the population percentages in each state by part of state by gender stratum for the B cohort and K cohorts. The B and K cohort figures generally match closely, however the population from which the K cohort was selected was a little more likely to live in capital cities (66.5% v 63.6%). Figure 12 shows the population proportions for mothers having completed Year 12 by state and part of state for each cohort. The B cohort population was more likely to have completed Year 12 in every part of the country, with the ABS Census figures nationally being 56.6% for the B cohort against

48.3% for the K cohort. Figure 13 shows the populations proportions for mothers speaking a language other than English at home by state and part of state for each cohort. These proportions were more closely matched between the B and K cohorts.

The implication of this is that just because the two cohorts have been weighted using similar variables, it does not mean that the variables that they have been weighted on are not responsible for the differences observed between the two. For example, because the two cohorts have had non-response due to maternal education adjusted for, it does not mean they will have equal proportions of mothers who had completed

Year 12 when the weights are applied. Therefore different levels of maternal education could explain differences observed between the two samples in the educational attainment of children.

Figure 11. Cohort benchmarks by state, part of state and gender.

Note: there are no respondents from non metropolitan ACT.

LSAC Data User Guide – November 2013

74

Figure 12. Proportion of mothers who completed Year 12: Cohort benchmarks by state and part of state.

Note: there are no respondents from non metropolitan ACT.

Figure 13. Proportion of mothers who speak a language other than English at home: Cohort benchmarks by state and part of state.

Note: there are no respondents from non metropolitan ACT.

LSAC Data User Guide – November 2013

75

13.6 Sample characteristics

To assist in the assessment of the representativeness of the Wave 1 sample, selected characteristics were compared with ABS estimates: gender, state and region were compared with the ABS September 2004 Estimated Resident Population figures; the other characteristics were compared with (previously unpublished) population data from the ABS 2001 Census of Population and Housing (see Table 12).

Table 12. Wave 1 sample characteristics compared with ABS data.

B cohort

LSAC ABS

% %

K cohort

LSAC ABS

% %

Gender*

Male

Female

Family type

Two resident parents/guardians

One resident parent/guardian

Siblings

Only child

One sibling

Two or more siblings

Ethnicity

Study child Indigenous

51.2 51.3 50.9 51.3

48.8 48.7 49.1 48.7

90.7 88.1 86.0 82.0

9.3 11.9 14.0 18.0

39.5 36.2 11.5 12.1

36.8 35.6 48.4 45.9

23.7 28.2 40.1 42.0

4.5 4.3 3.8 4.3

Mother speaks a language other than

English at home 14.5 16.8 15.7 17.6

Educational status

Mother completed Year 12

Father completed Year 12

66.9

58.5

56.6

50.2

58.6

52.7

48.3

45.3

State*

New South Wales

Victoria

Queensland

South Australia

Western Australia

31.6

24.5

10.4

34.1

24.6

9.9

31.6

25.0

10.2

33.7

23.8

20.6 19.3 19.8 19.7

6.8 6.8 6.8 7.2

10.1

Tasmania

Northern Territory

Australian Capital Territory

Region

Capital City Statistical Division

Balance of state

2.2

1.7

2.1

2.3 2.7

1.4 1.7

1.7 2.3

2.5

1.6

1.3

62.5 63.7 62.1 62.1

37.5 26.3 37.9 37.9

Total 5047 4983

Note: ABS= 2001 Census for families for 0 and 4 year olds, except where * based on September 2004 Estimated

Resident Population for families of 0 and 4 year olds

.

For most characteristics, the Wave 1 sample is only marginally different to the ABS data. The largest difference is in the educational status of the parents. Children with mothers who have completed Year 12 are over-represented in the sample, with proportions 10 per cent higher than in the 2001 Census.

LSAC Data User Guide – November 2013

76

Other differences include:

• children in lone-parent families are under-represented;

• children with two or more siblings are under-represented and only children are over-represented in the infant cohort, particularly for the B cohort at Wave 1;

• children from an ATSI background, although not for the B cohort at Wave 1;

• children with mothers who speak a language other than English at home are underrepresented; and

• children in New South Wales are under-represented.

Table 13 shows the number of children in the Wave 1 sample with selected characteristics, and gives the Waves 2, 3 and 4 response rates for children with these characteristics. As can be seen in the table, the greatest sample loss has been from

Indigenous families and families where Parent 1 speaks a language other than English at home.

Table 13. Response rates at Waves 2, 3, 4 & 5 by selected sample characteristics

Wave 1

N

% responding to Wave 2

B K B

% responding to Wave 3

K

% responding to Wave 4

B K

% responding to Wave 5

B K Characteristics B K

Full sample

Study child male

Study child female

Study child

Indigenous

Mother speaks language other than English

Mother did not complete Yr 12

Father did not complete Yr 12

5107 4983 90.2 89.6 85.9

2610 2537 90.0 89.8 86.3

2497 2446 90.4 89.4 85.5

230

740

187

778

78.3 81.8 64.8

83.9 83.8 75.0

1688 2044 84.8 86.6 78.8

1890 2016 90.0 90.0 85.9

86.9

87.2

86.6

66.3

76.6

81.7

87.0

83.1

83.8

82.3

63.0

72.3

73.2

77.0

83.6

84.0

83.2

51.3

75.5

77.1

80.5

80.0

80.3

79.7

60.0

58.2

70.4

70.3

79.4

79.5

79.1

59.9

58.4

73.4

72.2

New South

Wales

Victoria

Queensland

1615 1573 90.3 90.2 84.4 86.3 79.6 80.3 77.6 76.9

1251 1245 88.4 86.3 85.1 86.0 83.9 81.4 78.1 74.4

1054 988 91.4 90.8 88.0 87.2 87.1 88.5 86.6 86.2

South Australia 347 339 91.1 89.4 88.2 86.7 83.3 84.4 79.5 80.2

Western 533 507 89.7 91.5 83.9 87.6 80.7 85.2 77.7 80.5

Australia

Tasmania

Northern

Territory

Australian

Capital

Territory

113

87

107

136

82

113

90.3 94.1 92.0

90.8 89.0 83.9

97.2 94.7 95.3

91.2 101.8

87.8

94.7

56.3

99.1

95.6

73.2

93.8

94.7

51.7

92.5

94.1

67.1

92.9

Capital City

Statistical

Division

3194 3095 90.6 89.3 86.2 86.8 80.4 80.9 76.4 77.0

LSAC Data User Guide – November 2013

77

Wave 1

N

% responding to Wave 2

% responding to Wave 3

% responding to Wave 4

% responding to Wave 5

Characteristics B K B K B K B K B K

Full sample 5107 4983 90.2 89.6 85.9 86.9 83.1 83.6 80.0 79.4

Balance of state 1913 1888 89.6 90.0 85.4 87.2 86.9 87.7 85.7 83.2

14 User support and training

User training sessions are offered by AIFS to further develop the information provided in the user manual and to allow users to interact with the LSAC Data

Management team and benefit from their knowledge and experience with the data.

These sessions consist of an introduction to LSAC and the newly released datasets including:

• study methodology

• introduction to the datasets

• issues for data analysts (e.g. weighting, clustering, confidentialisation)

• variable naming

• user resources (eg data dictionary, labeled questionnaires).

See the LSAC website for details on when training sessions are being offered.

14.1 Online assistance

An email alert list is used to convey key information and updates to users. Important information distributed via the email alert list is also stored in the data access area of the Growing Up in Australia website. This area contains: all reference material made available to users (in downloadable form)

Excel Data Dictionary critical updates and alerts as distributed through the email alert list updates on data user workshops

14.2 Getting more information

More information on Growing Up in Australia and its progress can be found on the

LSAC website: http:// http://www.growingupinaustralia.gov.au/index.html

Further enquiries can be directed to [email protected] or by contacting:

LSAC Data Manager

Australian Institute of Family Studies

Level 20/485 La Trobe Street

Melbourne VIC 3000

Tel: +61 3 9214 7879

Fax: + 61 3 9214 7839

LSAC Data User Guide – November 2013

78

15 References

Baxter, J. (2007). Children’s time use in the Longitudinal Study of Australian

Children: Data quality and analytical issues in the 4-year-old cohort, LSAC

Technical Paper No 4. Australian Institute of Family Studies, Melbourne.

Chambers, R. L., & Skinner, C. J., (Eds.). (2003). Analysis of Survey Data.

Chichester: Wiley.

Cusack, B., & Defina, R. (2013). Wave 5 Weighting and non response, LSAC

Technical Paper No 10. Australian Bureau of Statistics, Canberra.

Daraganova, G., & Sipthorp, M. (2011). Wave 4 Weights, LSAC Technical Paper No

9. Australian Institute of Family Studies, Melbourne.

Levy, P. S., & Lemeshow, S. (1999). Sampling of populations. Methods and

applications. 3rd Edition. New York: Wiley.

Misson, S., & Sipthorp, M. (2007). Wave 2 weighting and non-response, LSAC

Technical Paper No 5. Australian Institute of Family Studies, Melbourne.

National Childcare Accreditation Council (2003). OSHCQA Quality Practices Guide,

1 st

Edition. NCAC, Sydney, Australia.

National Childcare Accreditation Council (2003). QIAS Validation Report, 2 nd

Edition. NCAC, Sydney, Australia.

National Childcare Accreditation Council (2004). FDCQA Quality Practices Guide.

2 nd

Edition. NCAC, Sydney, Australia.

National Childcare Accreditation Council (2005). FDCQA Quality Practices Guide.

3 rd

Edition. NCAC, Sydney, Australia.

National Childcare Accreditation Council (2006). QIAS Quality Practices Guide, 1 st

Edition. NCAC, Sydney, Australia.

Rowe, K. (2006). The measurement of composite variables from multiple indicators:

Applications in Quality Assurance and Accreditation Systems – Childcare.

Background paper prepared for the National Childcare Accreditation Centre.

Sanson, A., Misson, S. & The LSAC Outcome Index Working Group (2006).

Summarising children’s well-being: the LSAC Outcome Index, LSAC Technical Paper

No 2. Australian Institute of Family Studies, Melbourne.

Sanson, A., Nicholson, J., Ungerer, J., Zubrick, S., Wilson, K., Ainley, J., . . . . . . . . .

Wake, M. (2002). Introducing the Longitudinal Study of Australian Children, LSAC

Discussion Paper No 1. Australian Institute of Family Studies, Melbourne.

Sipthorp, M., & Misson, S (2009). Wave 3 Weighting and non-response, LSAC

Technical Paper No 6. Australian Institute of Family Studies, Melbourne.

Soloff, C., Lawrence, D., & Johnstone, R. (2005). Sample Design, LSAC Technical

Paper No 1. Australian Institute of Family Studies, Melbourne.

Soloff, C., Lawrence, D., Misson, S., & Johnstone, R. (2005). Wave 1 weighting and

non-response, LSAC Technical Paper No 3. Australian Institute of Family Studies,

Melbourne.

LSAC Data User Guide – November 2013

79

16 Bibliography

The following publications provide more information on techniques for analysis longitudinal and survey data:

Australian Bureau of Statistics. (1996). Women’s Safety Australia: User Guide.

Chambers, R. L., & Skinner, C. J., (Eds.). (2003). Analysis of Survey Data.

Chichester: Wiley.

Deville, J. C., & Särndal, C. E. (1992). Calibration estimators in survey sampling.

Journal of the American Statistical Association, 87, 376-382.

Deville, J. C., Särndal C. E., & Sautory, O. (1993). Generalised raking procedures in survey sampling. Journal of the American Statistical Association, 88, 1013-1020.

Kalton, G. (1983). Compensating for missing survey data. Research report series,

Institute for Social Research, University of Michigan.

Lepkowski, J. M. (1989). Treatment of wave nonresponse in panel surveys. In D.

Kasprzyk, G. Duncan, G Kalton & M. P. Singh (Eds.), Panel Surveys (348-374), New

York: Wiley.

Levy, P. S., & Lemeshow, S. (1999). Sampling of populations. Methods and

applications (3rd ed.). New York: Wiley.

Pfeffermann, D., Skinner, C. J., Holmes, D. J., Goldstein, H., & Rasbash, J. (1998).

Weighting for unequal selection probabilities in multilevel models. Journal of the

Royal Statistical Society Series B, 60, 23-40.

Skinner, C. J., & Holmes, D. J. (2003). Random effects models for longitudinal survey data. In R. L. Chambers & C. J. Skinner (Eds.), Analysis of Survey Data (205-

218). Chichester: Wiley.

Tabachnick, B. G., & Fidell, L. S. (1989). Using Multivariate Statistics (2nd ed.).

New York: Harper and Row.

Wolter, K. (1984). Introduction to Variance Estimation. New York: Springer.

Wright, B. D., & Masters, G. N. (1982). Rating scale analysis. Chicago: MESA Press.

Information on related studies:

Fergusson, D. M., Horwood, L. J., Shannon, F. T., & Lawton, J. M. (1989). The

Christchurch Child Development Study: A review of epidemiological findings.

Pediatric and Perinatal Epidemiology, 3, 278-303.

Freidin, S., Watson, N., & Wooden, M. (2002). HILDA Survey Coding Framework:

Confidentialised Data, HILDA project technical paper series. The Melbourne Institute of Applied Economic and Social Research, University of Melbourne.

Frick, J. R., & Haisken-DeNew, J. P. (2001). Structuring the HILDA Panel:

Considerations and Suggestions, HILDA project discussion paper series. The

Melbourne Institute of Applied Economic and Social Research, University of

Melbourne.

Henstridge, J. (2001). The Household Income and Labour Dynamics in Australia

(HILDA) Survey: Weighting and Imputation, HILDA project discussion paper series.

LSAC Data User Guide – November 2013

80

The Melbourne Institute of Applied Economic and Social Research, University of

Melbourne.

National Longitudinal Survey of Children and Youth (NLSCY). (1999). Overview of

survey instruments for 1998-99 data collection cycle 3, Catalogue no. 89FOO78XPE, no. 3, Canada: Statistics Canada.

Watson, N., & Fry, T. R. L. (2002). The Household Income and Labour Dynamics in

Australia (HILDA) Survey: Wave 1 Weighting, HILDA project technical paper series.

The Melbourne Institute of Applied Economic and Social Research, University of

Melbourne.

Watson, N., & Wooden, M. (2002). The Household Income and Labour Dynamics in

Australia (HILDA) Survey: Wave 1 Survey Methodology, HILDA project technical paper series. The Melbourne Institute of Applied Economic and Social Research,

University of Melbourne.

Willms, D. (Eds.) (2002). Vulnerable Children, University of Alberta Press,

Edmonton.

Wooden, M. (2001). Design and Management of a Household Panel Survey: Lessons

from the International Experience, HILDA project discussion paper series. The

Melbourne Institute of Applied Economic and Social Research, University of

Melbourne.

Wooden, M., & Watson, N. (2000). The Household Income and Labour Dynamics in

Australia (HILDA) Survey: An Introduction to the Proposed Survey Design and Plan,

HILDA project technical paper series. The Melbourne Institute of Applied Economic and Social Research, University of Melbourne.

LSAC Data User Guide – November 2013

81

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