Hilda News - 2014
1 December 2014
New HILDA Discussion Paper is now available online.DP 3/14 – Wooden - The Measurement of Physical Activity in Wave 13 of the HILDA Survey
Please find the abstract and download link below.
HILDA DP No. 3/14
Mark Wooden
The Measurement of Physical Activity in Wave 13 of the HILDA Survey
Abstract:
Every wave of the HILDA Survey conducted to date has included one simple question intended to measure the extent an individual engages in physical activity. Included each year in the self-completion questionnaire, it measures the frequency per week of participation in moderate or intensive physical activity. The question reads as follows:
In general, how often do you participate in moderate or intensive physical activity for at least 30 minutes?
Moderate level physical activity will cause a slight increase in breathing and heart rate, such as brisk walking.
Six pre-coded response categories are provided: “Not at all”, “Less than once a week”, “One to two times a week”, “Three times a week”, “More than 3 times a week (but not every day)”, and “Every day”.
The information collected from this question is very coarse. It does not measure the amount of time individuals actually spend in physical activity; it treats all types of activity (subject to satisfying the criteria that it must at least cause a slight increase in breathing and heart rates) the same, regardless of the intensity of that activity; and it only permits responses within bands.
Given the prominence of obesity as a public health issue, it has been suggested that the HILDA Survey should give greater priority to the collection of data that would measure both the extent of obesity within the Australian population and factors that contribute to obesity. As a result, data on (self-reported) height and weight began to be collected in wave 6 (and data on waist measurement in wave 13). In wave 9, a dedicated health module was designed that included, among other things, questions about diet. More detailed questions on physical activity, based on questions included in the Active Australia Survey (see AIHW 2003), were also trialled as part of the dress rehearsal for that survey wave. Interviewer feedback about both the quality of responses being provided and the relatively large imposition on respondent time, however, led us not to continue with these questions for the wave 9 main survey.1
In wave 13, with the re-inclusion of the health module (which is scheduled to be included every 4 years), we again gave consideration to the inclusion of questions that would enable the construction of more reliable and informative measures of physical activity. For this wave we were drawn to the International Physical Activity Questionnaires (IPAQ), which over the last decade or so have increasingly been used for population surveillance of physical activity among adult populations (though not in Australia, where the Active Australia Survey is dominant, and implemented in large part in the 2011-12 Australian Health Survey conducted by the Australian Bureau of Statistics [ABS])
This paper: introduces the IPAQ and the types of summary measures it generates; describes its implementation in wave 13 of the HILDA Survey; and reports brief summary statistics from the wave 13 data describing the distribution of responses on the key outcome variables and associations with other health-related variables.
- Journeys Home Waves 5 and 6
The microdata for Waves 5 and 6 of Journeys Home: A longitudinal study of factors affecting housing stability are now available for researchers use.
Journeys Home is an important national survey about the living and housing challenges that people may be facing. The study allows for analysis of the diverse social, economic and personal factors that are related to housing stability. More information on the study is provided at: http://melbourneinstitute.com/journeys_home/index.html.
For full details of how to access the Journeys Home data please see: http://melbourneinstitute.com/journeys_home/research/dataaccess.html.
25 November 2014
New HILDA Technical Paper is now available online.
TP 2/14 – Watson and Summerfield - Outcomes from Matching the HILDA Survey Sample to the Death Register
Please find the abstract and download link below.
- Call for HILDA Data Publications
- Journal Articles
- Books / Book Chapters
- Other Publications
- Working / Discussion / Research Papers
- Conference Papers / Public Lectures
- Student Essays and Dissertations
- Title of Article / Paper
- Journal / Publication / Conference name
- Date of Publication / Conference
- Author Name/s
- Indicate if published – add volume no., issue no., and page nos.
- Indicate if forthcoming
- HILDA Release 13 pre-orders
- Introduction to the HILDA Survey – Melbourne (1 December 2014), Canberra (2 December 2014), Sydney (3 December 2014) and Brisbane (4 December 2014). This 1 hour (or 3 hour) session aims to give you an overview of the HILDA Survey (and its datasets and documentation). This course is organised by the Melbourne Institute. See below for more details or go to www.melbourneinstitute.com/hilda/training
- Getting Started: Analysing HILDA with Stata – Melbourne (25-27 February 2015). This 3-day hands-on training course is designed for people interested in using the HILDA Survey data but have not yet done so or have only used the data a little. This course is organised by the Melbourne Institute. See below for more details or go to www.melbourneinstitute.com/hilda/training
- Fundamentals of Longitudinal Research and Design – Melbourne (2-6 February 2015). This 5-day course is designed to give participants an overview of the use of longitudinal methods in social research and is aimed at researchers new to using longitudinal surveys and those involved in designing longitudinal surveys. This course is part of the ACSPRI Summer Program. For more details see www.acspri.org.au/node/1428 - Please note: This course has now been cancelled.
- Longitudinal Data Analysis Course – Brisbane (24-28 November 2014). This 5-day intensive course is designed to deepen specialist knowledge and enhance the quality and meaning of the data used in decision making. This course uses both HILDA Survey data and LSAC data and is run by the Institute for Social Science Research at the University of Queensland. For details see www.issr.uq.edu.au/filething/get/31681/C5_LDA2.pdf
- Applied Longitudinal Data Analysis – Melbourne (9-13 February 2015). This 5-day intensive course aims to provide participants with a range of skills for the analysis of longitudinal data. This course is part of the ACSPRI Summer Program. For more details see www.acspri.org.au/node/1434
- Introduction to the HILDA Survey – Melbourne (1 December 2014), Canberra (2 December 2014), Sydney (3 December 2014) and Brisbane (4 December 2014). This 1 hour (or 3 hour) session aims to give you an overview of the HILDA Survey (and its datasets and documentation). This course is organised by the Melbourne Institute. See below for more details or go to www.melbourneinstitute.com/hilda/training
- Getting Started: Analysing HILDA with Stata – Melbourne (25-27 February 2015). This 3-day hands-on training course is designed for people interested in using the HILDA Survey data but have not yet done so or have only used the data a little. This course is organised by the Melbourne Institute. See below for more details or go to www.melbourneinstitute.com/hilda/training
- Fundamentals of Longitudinal Research and Design – Melbourne (2-6 February 2015). This 5-day course is designed to give participants an overview of the use of longitudinal methods in social research and is aimed at researchers new to using longitudinal surveys and those involved in designing longitudinal surveys. This course is part of the ACSPRI Summer Program. For more details see www.acspri.org.au/node/1428
- Longitudinal Data Analysis Course – Brisbane (24-28 November 2014). This 5-day intensive course is designed to deepen specialist knowledge and enhance the quality and meaning of the data used in decision making. This course uses both HILDA Survey data and LSAC data and is run by the Institute for Social Science Research at the University of Queensland. For details see www.issr.uq.edu.au/filething/get/31681/C5_LDA2.pdf
- Applied Longitudinal Data Analysis – Melbourne (9-13 February 2015). This 5-day intensive course aims to provide participants with a range of skills for the analysis of longitudinal data. This course is part of the ACSPRI Summer Program. For more details see www.acspri.org.au/node/1434
- Sample design, fieldwork and following rules
- Content
- Response rates
- Datasets and conventions
- Using the documentation
- Creating your own datasets and identifying key variables
- Examples of merging datasets in Stata
- Melbourne - Monday 1 December, University of Melbourne
- Canberra - Tuesday 2 December, Department of Social Services
- Sydney - Wednesday 3 December, University of Sydney
- Brisbane - Thursday 4 December, University of Queensland
- Accessing the data
- Finding the right data
- Visualising the data
- Merging household and individual data
- Dealing with missing values
- Using imputed values
- Creating analysis variables from HILDA data
- Using weights
- Generating descriptive statistics in the panel context
- Understanding data patterns in the panel context
- Answering the question
- Accessing the data
- Finding the right data
- Visualising the data
- Merging household and individual data
- Dealing with missing values
- Using imputed values
- Creating analysis variables from HILDA data
- Using weights
- Generating descriptive statistics in the panel context
- Understanding data patterns in the panel context
- Answering the question
- describe the genesis, construction and administration of the sexual identity question included in wave 12;
- show how the data collected can be used to construct population estimates of the size of the lesbian, gay and bisexual (LGB) community in Australia, and then compare these estimates with estimates from other sources and for other countries;
- examine and report on item non-response, the main weakness with self-reported data on sexual identity; and
- present preliminary evidence on associations between sexual identity and measures of both mental health and labour earnings, two outcomes where previous research has demonstrated clear and large differences between heterosexual and LBG sub-populations.
- Canberra
- Brisbane
- Melbourne
- Other location (if we have the numbers and an organisation to host it)
- Stata
- SAS
HILDA TP No. 2/14
Nicole Watson and Michelle Summerfield
Outcomes from Matching the HILDA Survey Sample to the Death Register
Introduction:
Beginning in 2001, the Household, Income and Labour Dynamics in Australia (HILDA) Survey has become a major resource in the social sciences for understanding the dynamics of the economic, social and subjective well-being of Australians. The health-related questions facilitate the examination of changes in self-reported health status, long-term disabilities, impairments and health conditions, behavioural risk factors (such as smoking, exercise, and diet), and use of health care services. In 2014, the HILDA Survey sample was matched to the National Death Index (NDI) so that details of the date and cause of death could be added to the data files. This offers researchers a valuable opportunity to study the changes in life circumstances leading up to death. It also helps us to more accurately identify the eligibility of sample members over the course of the panel, which will assist in the production of response rates and population weights.
This report details the matching process, the outcomes, and the variables available.14 November 2013
We are in the process of updating the HILDA bibliography, preparatory to turning over to 2015 and we would like to include details of all publications that make use of the HILDA Survey data. We therefore request that you send details of any publications or papers that use the HILDA Survey that are not already included in the HILDA Survey bibliography.
The bibliography can be accessed from our webpage or via the above link and includes:
Please check the website carefully and if your paper / journal article etc. is not listed, could you please forward any information regarding any of your publications so that I can update the website pages accordingly. Please include:
Finally, a reminder to all data users that a requirement for using the HILDA data is the inclusion of the appropriate acknowledgement and disclaimer in any publications that cite / use the HILDA data.
5 November 2013
HILDA Release 13 pre-orders, Introduction to the HILDA Survey - December 2014 and Getting started: Analysing HILDA with Stata - February 2015
Release 13 of the HILDA data (which includes waves 1 to 13) is now available for pre-ordering. The release DVD includes Stata, SAS and SPSS datasets with extensive documentation.
For further information on pre-ordering and the new fact sheets, please see our website : http://www.melbourneinstitute.com/hilda/data/
Introductory level
Advanced level
More information about the two training courses organised by the Melbourne Institute, including how to register, are provided on our website or see news item below.
25 September 2014
Current and potential HILDA users may be interested in the following training opportunities. They range from a 1-hour overview to intensive 5-day hands-on workshops.
Introductory level
Advanced level
More information about the two training courses organised by the Melbourne Institute are provided below.
Introduction to the HILDA Survey - December 2014
This short training session aims to give you an overview of the HILDA Survey and its datasets and documentation. The first part of this training (9.30am – 10.30am) will provide an overview of the following:
The second part of this training (11.00am – 12.30pm) will cover:
Participants can sign up for the first hour or for all three hours.
The training will be presented by Nicole Watson from the HILDA Survey team at the Melbourne Institute. It will be in a lecture style format and there will be plenty of opportunities to ask questions.
Location and dates |
|
Catering |
Morning tea will be provided |
Fee |
$25 |
To register, go to www.melbourneinstitute.com/hilda/training
Getting started: Analysing HILDA with Stata - February 2015
A hands-on introductory HILDA Survey training course will be held on 25-27 February 2015 at the University of Melbourne.
This 3-day training course is designed for people who are interested in using the HILDA Survey data but have not yet done so. It will take a descriptive research problem such as “What happens to single mothers who receive Parenting Payment Single? Do they stay on the program from one year to the next? Do they move to another income support program? Do they leave income support?” and go through all the steps necessary to answer this question. We begin with the HILDA DVD and assume participants have relatively little experience in working with large data sets.
The training will then cover the following topics:
There will be scope within the course to deal with individual questions which participants might have about data extraction and structure in panel data contexts.
The course will be presented by Prof Robert Breunig (Crawford School of Public Policy, Australian National University) and Nicole Watson (Melbourne Institute of Applied Economic and Social Research, University of Melbourne).
Location |
University of Melbourne |
Date |
Wednesday 25 February to Friday 27 February 2015 |
Time |
9.15am – 4.00pm |
Catering |
Lunch, morning and afternoon tea will be provided |
Fee |
$850 ($580 PhD and Masters students) |
16 June 2014
2013 Annual Report
The HILDA Survey Annual Report for 2013 is now available, and can be downloaded from:
http://melbourneinstitute.com/hilda/Reports/annual_report.html
2014 Statistical Report Volume 9
The HILDA Survey Statistical Report for 2014 is also available, and can be downloaded from:
http://www.melbourneinstitute.com/hilda/Reports/statreport.html
If you would like a printed copy you should contact Victoria Lane (HILDA Survey Administration Assistant)
12 June 2014
New HILDA Discussion Paper is now available online.
DP 2/14 – Uhrig, Watson - The Impact of Measurement Error on Wage Decompositions: Evidence from the BHPS and HILDA Survey
Please find the abstract and download link below.
HILDA DP No. 2/14
Noah Uhrig and Nicole Watson
The Impact of Measurement Error on Wage Decompositions: Evidence from the BHPS and HILDA Survey
Abstract:
While measurement error can be studied in various ways, little is known about longitudinal patterns of reliability for most types of survey measures. Reliability assessments using test-retest methods rarely investigate whether reliability itself is stable or the extent to which change in reliability affects findings in substantive models. Research in the social sciences often recognises that measurement error could influence results, yet established methods of error correction are rarely applied. Indeed in some types of analysis, measurement error is expressly confounded with the main conclusions of the research. For example, Jones and Kelly (1984) demonstrate how discrimination assessments can conflate discrimination with omitted variables and random measurement error present in variables included in the models. Wage discrimination monitoring exercises routinely add new variables to core human capital arguments about wage setting, or shift the nature of analyses to demonstrate how disparities remain in certain segments of the labour market or sections of the income distribution (Cobb-Clark and Tan 2011; Joshi, Dex, and McCran 1996; Kee 2006; Marini and Fan 1997; Miller 2005). To our knowledge, established measurement error correction techniques have never been incorporated into these approaches, although some effort has been made to examine the extent to which estimates are bounded within certain ranges (Bollinger 2001; Bollinger 2003). Focusing on gender wage inequality, we address two problems in the wage discrimination literature. First, we examine the reliability of employment experience measures and explore the extent to which measurement reliability varies over time, across genders and across measurement protocols. Secondly, we examine whether correcting for measurement error in employment experience influences substantive conclusions about the amount of gender wage inequality. We analyse panel data from Britain and Australia. Both nations adopted wage equality legislation in the 1970s. Both have household panel studies of comparable designs affording direct comparisons. However, measurement protocols differ between the studies thus providing an instructive comparison in how measurement can influence substantive conclusions about discrimination.
22 May 2014
You may be interested to know there is a Longitudinal Data Analysis course to be held in Brisbane on 3-4 June, presented by staff from the Institute for Social Science Research at the University of Queensland.
For more details, please see the information flyer or visit http://www.issr.uq.edu.au/mfsas-stream-c19 March 2014
New HILDA Technical Paper is now available online.
TP 1/14 – Wilkins - Derived Income Variables in the HILDA Survey Data: The HILDA Survey ‘income model’
Please find the abstract and download link below.
HILDA TP No. 1/14
Roger Wilkins
Derived Income Variables in the HILDA Survey Data: The HILDA Survey ‘income model’
Abstract:
Each wave, the HILDA Survey collects detailed information from each respondent on annual income received from each of a number of sources. While studies may of course make use of these individual income components, the primary purpose of the collection of this information is to allow estimation of the total personal and household annual income of each sample member. However, to achieve these estimates, steps must be taken to account for limitations of the collected data. First, some income components are not collected, most important of which are some government benefits, which were deliberately not collected on the basis that estimates based on eligibility criteria were likely to be more accurate than respondent recollections of these components. Second, non-response for income components, and indeed the presence of non-respondents in partially-responding households, needs to be accounted for. Finally, respondents mostly report gross or pre-tax income amounts, whereas primary interest is in the post-tax and transfer, or disposable, income of individuals. The HILDA Survey data managers therefore estimate income components not collected and impute missing values of collected income components, and then aggregate the income components to produce derived total personal and household income variables. Furthermore, income tax is estimated for each sample member to produce disposable income estimates at both the personal and household level.
This technical paper describes the methods by which the derived annual income variables are constructed as of Release 12 of the HILDA Survey, which contains data from Waves 1 to 12 (2001 to 2012). In particular, it explains how income components are aggregated and certain government benefits are estimated to produce total income measures, and how taxes are estimated to produce post-tax (disposable) income measures.1 The methods used to impute missing values, and the extent of missing values, are not discussed in this paper; these are described in Hayes and Watson (2009) and Summerfield et al. (2013).
The methods described in this paper apply to all 12 waves of the HILDA Survey, but not to all of the 12 releases of the HILDA Survey data up to Wave 12. Over time, the sophistication of the methods has improved, such that values of derived income variables in a given wave have changed for at least some individuals from release to release. An overview of the changes over Releases 1 to 12, and their implications, is provided in the Appendix. There is, moreover, another source of changes in income variables in a given wave from release to release, which is that imputed income variables can change from release to release. This is because the imputation methods draw on the longitudinal information to improve the quality of the imputations. For both of these reasons, it is important when comparing across waves to use a single data release, and preferably the most recent release, since this will apply a consistent method for constructing income variables across all waves, with the most recent release providing the most accurate estimates.
The plan of this paper is as follows. Section 2 provides an overview of the income model, describing the income components included in the model (and what cash flows are excluded) and how they are combined to produce income aggregates. Section 3 describes the process by which regular and irregular income components are distinguished, while Section 4 explains the methods for estimating government benefit income components that are not collected by the HIDA Survey. Section 5 explains how income tax is estimated in order to produce disposable income estimates. Concluding comments are presented in Section 6.
18 March 2014
FLoSse Research Repository
Just a reminder that you must, under your Deed of Licence, provide DSS with copies of your research. You can do this by directly entering the bibliographic details of your research into the FLoSse Research repository. DSS’s Longitudinal Surveys Electronic (FLoSse) Research repository contains details of research based on DSS’s three longitudinal surveys. The research is not made directly available in FloSse, but provides links/directions on how interested researchers may obtain copies of the research that you submit.
On entering the FloSse repository you can follow the links from ‘deposit research details’ to enter links or directions on how to obtain copies of your completed research. If working collaboratively, you may like to check first to see if your research is already listed before proceeding.
FLoSse is publicly available but is primarily for use by Commonwealth agencies to source relevant evidence based research for input into policy development.
Researchers must also acknowledge data ownership. Wording for this can be found in Fact Sheet 8.
Many thanks DSS Longitudinal Surveys Team
6 March 2014
A hands-on introductory HILDA Survey training course will be held on 14-16 April 2014 at the Australian National University in Canberra.
Please read on if this is of interest to you…
Getting started: Analysing HILDA with Stata
Course information
This 3-day training course is designed for people who are interested in using the HILDA Survey data but have not yet done so. It will take a descriptive research problem such as “What happens to single mothers who receive Parenting Payment Single? Do they stay on the program from one year to the next? Do they move to another income support program? Do they leave income support?” and go through all the steps necessary to answer this question. We begin with the HILDA DVD and assume participants have relatively little experience in working with large data sets. The training will then cover the following topics:
There will be scope within the course to deal with individual questions which participants might have about data extraction and structure in panel data contexts.
Course instructors
The course will be presented by Professor Robert Breunig and Ms Nicole Watson.
Professor Robert Breunig is a member of the Crawford School of Public Policy at the Australian National University. His primary applied research interests are in analysing individual and firm behaviour, evaluating government tax and transfer programs and analysing large data sets.
Nicole Watson is part of the HILDA Survey team at the Melbourne Institute of Applied Economic and Social Research at the University of Melbourne. She has worked on the HILDA project since 2000 and has been involved in the sample design, weighting, imputation and overall survey management.
Pre-requisites
You do not need to have used the HILDA data or Stata before. You should have some experience with a statistical software package of some sort (eg. SAS, SPSS, Stata). While the course will be taught in Stata, SAS equivalent code will be provided.
Course details
| Location: | Crawford School, Australian National University |
| Date: | Monday 14 April to Wednesday 16 April 2014 |
| Time: | 9.15am – 4.00pm |
| Catering: | Morning tea, lunch and afternoon tea provided |
| Fees: | $730 ($550 PhD and Masters students) |
| Registration: | Places are limited. Please register here. |
You may register your interest in the course with Victoria Lane prior to 28 March though payment must be made within 5 working days of registering your interest to secure your place.
Cancellations within 5 working days prior to the start of the training will not be refunded.
If you are unable to attend this training course, the next time we plan to offer it will be in September. If you would like to receive notifications of this and other upcoming courses, please subscribe to the HILDA mailing list.
27 February 2014
New HILDA Discussion Paper is now available online
DP 1/14 – Wooden - The Measurement of Sexual Identity in Wave 12 of the HILDA Survey (and associations with mental health and earnings)
Please find the abstract and download link below.
HILDA DP No. 1/14
Mark Wooden
Abstract:
An important characteristic of individuals, and one that is the subject of anti-discrimination laws in many countries, including Australia, but which was not measured during the first 11 waves of the HILDA Survey, is sexual orientation. In wave 12, however, this gap was filled, in part, through the inclusion in the self-completion instrument, of a single question about sexual identity.
Sexual orientation encompasses sexual behaviour, attraction and identity (e.g., Laumann et al. 1994), and hence would require a suite of survey questions that is both potentially quite long and detailed, and might be seen as very invasive by some sample members, especially given the sample for the HILDA Survey covers, and is representative of, the entire Australian population. It was, however, decided that the narrower concept of sexual identity could be identified within the HILDA Survey. This is a concept that is relatively easy to measure, requiring just the one question, and has recently been included in a number of large population-wide surveys conducted by official statistical agencies, both in Australia and elsewhere. Included here are the 2007 Survey of Mental Health and Well-being (SMHWB) conducted by the Australian Bureau of Statistics (ABS) and, since 2009, the UK Integrated Household Survey conducted by the UK Office for National Statistics (ONS). A question on sexual identity was also included in wave 3 of the UK Household Longitudinal Study (UKHLS), conducted in 2011-2012, a study that has very similar design features to the HILDA Survey.
The broad aim of this paper is to introduce users and potential users of the HILDA Survey data to the question on sexual identity that was included in wave 12 with a view to highlighting the strengths and weaknesses of the data collected. More specifically, the paper seeks to achieve four main tasks:
12 February 2014
Training opportunity
We are considering running an introductory HILDA User Training on 14-16 April and would like to gauge the level of interest you might have for this.
This course will be presented by Prof Bob Breunig and Nicole Watson (MIAESR). It is targeted to people who have little or no experience working with the HILDA data and documentation and would like to get some practical hands-on experience. It is similar to the introductory courses we have run over the last few years (for a description of the course content see http://melbourneinstitute.com/downloads/hilda/HILDA_User_Training/2013Training/Apr13_Getting_started_Analysing_HILDA.pdf).
If you are interested in such a training course can you please advise if you would prefer it to be held in
Also, would you prefer the package be:
We will choose the combination that suits most people. The course cost is likely to be around $730 ($550 for students). If there is not sufficient interest in this April training course, the next one will most likely be in September.