Panel attrition 2 |
|
| Chair | Dr Christian Bruch (University of Mannheim ) |
The measuring of skills in the form of competence tests in personal interviews is a particular challenge for panel surveys. Effects of non-response and panel mortality are to be expected in this special interview situation. Kleinert et al. (2015) have shown that this problem does not apply to adult respondents in general. However, they identified a co-existing negative link between the test result and the willingness to participate in the follow-up, especially for individuals with low competence levels. This paper focuses on youths and young adults who have just left the general education system by analyzing the impact of competence measurements on their participation. Since this target group is in the transition to apprenticeship or the labor market, its constituents are highly mobile and sometimes difficult to reach.
Indications in the literature convey that the risk of a withdrawal from participation increases with the burden of a respondent. According to Bradburn (1978), the burden consists of four main factors: 1) interview length (duration of a survey), 2) effort required for a target person, 3) stress level for a target person (e.g. competence tests), and 4) number of interviews. All these factors are particularly relevant to the target group considered here.
Our analyses are based on wave 5, 6, and 7 of the transitional cohort in apprenticeship and the labor market (starting cohort 4) of the National Educational Panel Study (NEPS). All participants have already participated in several surveys conducted within the framework of the NEPS study. Some of the interviewees were already interviewed in a non-school context, while others were questioned in a school context only. An experimental two-stage split-half design was implemented in wave 5. The entire sample of the school-leaver survey was divided into two equal samples before the survey started. One half was assigned to the CATI-field (without the competence tests), while the other half was requested to perform tests during a CAPI interview. In addition, the CAPI sample was split again into two parts: one half received just one test and the other half was asked to perform two tests. With this design, it is possible to analyze the willingness to participate in the test wave as well as the participation behavior in the following waves, depending on the tests assigned.
Panel attrition is one of the main problems in longitudinal surveys. To keep respondents constantly engaged is one of the great challenges in panel surveys. Respondents experience in the previous interview is one of the main determinants for survey participation in subsequent waves (Laurie et al., 1999).
Empirical research has shown that interview lengths affects survey participation, particular in longitudinal surveys. In general, methodological literature suggest to reduce interview lengths, since time-consuming interviews increase the respondents’ burden and decrease the respondents’ cooperation (Dillman et al., 2009; Groves et al., 2009; Schnell, 2012). However, Lynn (2014) finds no effect of interview lengths on subsequent survey participation. In a similar vein, Kleinert, Christoph & Ruland (2015) report that varying test duration has no direct impact on subsequent survey participation.
Therefore, the aim of the paper is first, to examine the patterns of interview lengths, and second, to analyse the effect of interview lengths on panel attrition. The analysis is based on data of the National Educational Panel Study (NEPS) adult starting cohort. The sample members were drawn from residents’ register office and represent individuals living in private households in Germany born between 1944 and 1986. The NEPS adult study is conducted annually since 2009 in a mixed-mode design with computer assisted telephone interviews and computer assisted face-to-face interviews. Every second wave, respondents are asked to additionally take part in competence assessment with paper and pencil, or computer-based.
By using this data set, we are able to measure the effect of interview lengths on panel attrition over the panel duration of seven waves. First results show, that particular respondents with a large number of events between the waves have higher interview lengths. However, there is no evidence for a direct effect of individual interview lengths on panel attrition.
The popularity of web surveys is on the rise, and simultaneously with the increasing availability of different devices to access the internet, a growing number of international surveys consider the use of web mode in data collection. The presentation gives an overview about the results of the recruitment process of an international web panel (CRONOS) in Estonia. We compare our results with the recruitment outcomes of a previous single-country web-survey. The particularity of international, centrally conducted comparative surveys is compared with single-country web surveys.
The recruitment experiment is part of the European Social Survey (ESS) web panel CRONOS, a pilot project to study the feasibility of a cross-national probability-based web panel. ESS 2016 face-to-face respondents in Estonia got an invitation to participate in the following panel study in web mode. Respondents without internet access were offered an opportunity to have internet access and a tablet for free.
Data from the ESS Round 8 face-to-face interviews enable us to analyse the selectivity of respondents in the web panel based on their socio-demographic characteristics as well as living conditions and personality traits. The particularities of the recruitment process in Estonia also allow for an investigation into the reasons for refusal to participate in the international panel, especially the concerns about the international character of the survey and transfer of personal data into international databases.
Previous research has demonstrated the importance of motivation, social involvement, topic salience, trust and importance of data protection issues on response behaviour (Abraham et al, 2009; Groves, Singer, and Corning, 2000; Haunberger, 2011).
In our presentation, we concentrate on the analysis of how attitudes toward international data transfer affect (1) respondent recruitment and (2) later survey behaviour in the web panel. We address the following questions:
• To which extent are refusals related to the international character of a survey?
• Can we convince the respondents who initially refuse to participate due to the international transfer of their data to become active members of the panel?
• Does the subsequent survey behaviour of the respondents who were initially reluctant towards data transfer differ from other panel respondents?
Abraham, K. G., Helms, S., & Presser, S. (2009). How Social Processes Distort Measurement: The Impact of Survey Nonresponse on Estimates of Volunteer Work in the United States. American Journal of Sociology, 114(4), 1129-1165.
Groves, R. M., Singer, E., & Corning, A. (2000). Leverage-saliency theory of survey participation: description and an illustration. The Public Opinion Quarterly, 64(3), 299-308.
Haunberger, S. (2011). To participate or not to participate: decision processes related to survey non-response. Bulletin of Sociological Methodology, 109(1), 39-55.
There is growing evidence that biological and social factors from early life onwards affect how well we age in later life. Much of the evidence comes from birth cohort and historical cohort studies. High participation rates are needed in these studies so they remain as representative of the original population and so that the risk of selective attrition biasing associations between lifetime risk factors and ageing outcomes is reduced. In the oldest British birth cohort study, the MRC National Survey of Health and Development (NSHD), an overall participation of 94% at age 68-69 years was achieved with 84% of study members completing a postal questionnaire and 80% having a home visit by a research nurse (Kuh et al, in press). To inform strategies to maintain participation in long-term cohort studies, we investigated how participation varied by lifetime and recent contact, health status, previous clinical feedback and study engagement, and to what extent these associations were explained by prior socioeconomic and cognitive characteristics already known to affect attrition. Overall and home visit participation was higher in those with higher levels of prior contact; tracing lost study members was worthwhile given their subsequent participation rates. Study members with the poorest health status had lower home visit participation rates. Previous feedback on clinical assessments generally did not affect subsequent participation, but certain types of feedback on blood results were associated with lower subsequent home visit participation. Celebrating key birthdays and involving study members in public engagement activities were associated with increased home visit participation.