Tuesday 14th July      Wednesday 15th July      Thursday 16th July      Friday 17th July     

Thursday 16th July, 09:00 - 10:30 Room: O-202

Longitudinal surveys - challenges in running panel studies 1
Convenor	Dr Jutta Von Maurice (Leibniz Institute for Educational Trajectoires )
Coordinator 1	Ms Joanne Corey (Australian Bureau of Statistics)

Session Details

Longitudinal surveys - challenges in running panel studies.

This session will focus on the organisation of panel studies, including panel maintenance, panel engagement, sample review processes, choice of data items and methodologies, and interviewer training.

The focus is on the particular challenges faced by those running panel studies such as:

. maintaining up-to-date contact information and tracking of respondents, including privacy concerns;

. engaging repsondents over the life of the survey, particularly for different age groups, for example how to keep young people interested as they move from children to young adults and they become the primary consenter;

. how successful are different modes for making contact, e.g. mail, phone, text;

. do targeted approach stategies work, e.g. different approach letters depending on past wave response;

. decision making guidelines about when a respondent should be removed from the sample;

. the debate between longitudinal consistency and using a better/updated measure;

. how to conduct training for a mix of experienced and new interviewers, balanced with the amount of new content and methodologies; and

.testing techniques for longitudinal surveys.

Paper Details

1. Changing panel survey items: the dilemma of stability vs. innovation
Dr Frederic Malter (Max-Planck-Society)

The essence of panel studies is asking identical questions to the same people over time for the assessment of change over time. Although keeping items identical is the only solution to allow for panel analyses, there is impetus for changing items over time: wanting new items (innovation) and improvement of existing items. Four aspects will be addressed: I will outline concepts of “identity”, then demonstrate item changes from SHARE, and provide descriptive statistics of item (in-) stability over time. Finally, I will discuss possible solutions to reconcile innovation and stability.

2. Assessing targeted approach letters: effects in different modes on response rates, response speed and sample composition
Professor Peter Lynn (University of Essex)

This presentation reports the results of a randomised experiment in which members of a general population panel (n = 2,733) were sent either targeted or standard letters in advance of a wave of data collection. The experiment was run both in a single-mode CAPI context and in a mixed-mode web/CAPI context. Effects on response rate and response speed are assessed overall and for subgroups and conditional on previous wave response behaviour. Overall sample composition is also assessed. Implications for targeted procedures are discussed.

3. The impact of providing a preferred mode on response rates in a longitudinal design
Mr Yamil Nares (ISER)

Numerous elements have been developed in order to counter non-response in longitudinal studies. A potential option to overcome the drop in response is the use of mixed-mode designs, providing respondents with their preferred mode. This paper examines responses to survey requests within a leverage-salience theory framework, hypothesising that preferred mode constitutes effective leverage. Furthermore, using the preferred mode to make the leverage more salient may further increase response rates. The basis of this research will be the fouth and fifth wave of the UK Household Longitudinal Study Innovation Panel.

4. Changing from CAPI to CAWI in an ongoing household panel – experiences from the German Socio-Economic Panel (SOEP)
Professor Jürgen Schupp (DIW Berlin/SOEP)
Dr Denise Saßenroth (DIW Berlin/SOEP)

The longitudinal German Socio-Economic Panel survey (SOEP) is characterised by a continuously refreshment of the sample. In 2014, four subsamples of the study „Families in Germany (FiD)“ were included in the SOEP. All four subsamples were formerly surveyed by CAPI mode for 4 waves. A replacement of CAPI by the CAWI mode was realised in four of the three subsamples in order to reduce survey costs. For the purpose of panel maintenance, we used a CAWI-CAPI sequential mixed-mode design. The experiment allows drawing preliminary conclusions regarding the conflicting goals of low survey costs and high response rates

5. Testing Adults’ Competences in a Panel Study: Experimental Evidence on Immediate and Long-Term Consequences of Test-Induced Burden on Respondent Engagement
Mr Bernhard Christoph (Institute for Employment Research (IAB))
Professor Corinna Kleinert (Leibnitz Institute for Educational Trajectories (LIfBi))

One of the major challenges in panel surveys is to keep respondents continuously engaged. This paper looks at how respondent burden induced by in-survey competence assessments influences respondents’ survey engagement.
Using data from the German National Educational Panel Study, we test the consequences of such assessments by applying an experimental split design, varying test duration (one versus two 30-minute tests).
Persons in the longer version were more prone to switch to a “last resort” CATI-option and to leave the panel subsequently. Moreover, respondents with low test-results had a higher probability for dropping out, thus increasing selectivity.