IDEAS home Printed from https://ideas.repec.org/p/diw/diwsop/diw_sp129.html
   My bibliography  Save this paper

Analyse der Panelausfälle im Sozio-oekonomischen Panel SOEP

Author

Listed:
  • Tobias Gramlich

Abstract

Nonresponse is a severe problem in sample surveys, especially in panel surveys, where nonresponse results not only in reduced efficiency of estimates compared to the full initial sample but the sample also becomes more and more selective with each wave of the panel since nonresponse is cumulative over all waves. Nonresponsethreatens the possibility to make inference on the population the sample was drawn from if the mechanism that leads to nonresponse is nonrandom and (observed or unobserved) characteristics of nonrespondents differ from the characteristics of respondents. If nonrespondents differ systematically from respondents nonresponse can cause biased estimates. The analysis of nonresponse and the mechanisms that lead to nonresponse is therefore essential in order to examine possible differences between nonrespondents and respondents and to account for nonresponse in estimation to avoid nonresponse bias in estimates. However, the threat of possible nonresponse bias is complicated by the fact that nonrespondents can differ not only systematically from respondents, but also different types of nonrespondents can differ systematically, splitting the original sample not only in respondents and nonrespondents but in different types of nonresponse categories. Each of the different types of nonresponse can be a source of a possible nonresponse bias and has distinct effects. For instance, refusals may bias the mean downwards, whereas noncontacts bias the mean away from zero: If nonresponse is not analysed separately, possible effects of the different types can be mixed, makingit diffcult or impossible to account for possible nonresponse bias accurately. This thesis analyses nonresponse in the German Socio-Economic Panel Study (SOEP) from 1984 untill 2005. In the yearly gross data of individuals, the SOEP documents up to 29 different reasons for nonresponse; some of them lead to final drop out (deceased, moved abroad, explicit refusal), whereas other nonrespondents are contacted again the following year according to the tracking rules. Unfortunately, due to changes in fieldwork coding schemes or different coding behavior of the interviewers, the coding scheme for the 29 reasons for nonresponse is not consistent over time,yielding 10 distinctive and consistent categories of nonresponse over time after summarising the original 29 categories. The thesis shows the development of the different categories of nonresponse in the SOEP. By looking at the whole history of contact information for every single observation, drop-out sequences are constructed showing participation patterns and typical sequences one, two, and three years prior to leaving the panel. Typical drop-out sequences are, for example, definite refusals after successful previous interviews or two consecutive waves of refusing. The next most frequent drop-out sequences are leaving the panel due to death or moving abroad after being interviewed successfully. Refusals after missing the previous wave because of being incapable for an interview is another rather typical drop-out sequence. Closely related to the language of sequences, the thesis estimates in the following transition probabilites between the different states using a first-order stationary Markov-Model. Since drop-out categories are unordered states, this can be done using multinomial logistic regression. Both, the results from looking at typical sequences, as well as the estimated transition probabilities show, that there is a high probability being repeatedly interviewed successfully, but the overall probability decreases markedly if a person is missing the previous wave for any reason. Especially for refusals there is a high probability for refusing again the following wave and therefore leaving the panel de_nitely. The same holds for noncontacts (either not found at the old address or not reached during the fieldwork period) in the previous wave: there is a relatively high probability for dropping out again due to refusal if found and reached at the new address, or they get lost since they are not found repeatedly. Whereas the (inverse of the) predicted probabilites can be used as rough propensity weights, the results point to clear implications for the fieldwork procedures of the interviewers: by all means, temporary drop-outs have to be avoided. Nonresponse stellt ein ernstzunehmendes Problem für die Möglichkeit dar, von einer Stichprobe auf die Grundgesamtheit zu schließen. Durch Nonresponse verringert sich zunächst die Fallzahl der Stichprobe, sodass sich die Effizienz der Schätzer der Grundgesamtheitsparameter im Vergleich zu einer Stichprobe ohne Nonresponse verringert. Zudem besteht die Gefahr der Verzerrung der Schätzer, wenn sich Teilnehmer von Nichtteilnehmern systematisch unterscheiden (Kapitel 1.2) und der Mechanismus, der zu Nonresponse führt nicht zufällig ist (Kapitel 1.3). Diese Ausfallmechanismen können sich für verschiedene Arten von Nonresponse unterscheiden (Kapitel 1.4). Dabei besteht grundsätzlich kein Unterschied zwischen Nonresponse in einmaligen Querschnittserhebungen und Ausfällen bei Wiederholungsbefragungen; die Besonderheiten für Panelausfälle werden in Kapitel 1.5 beschrieben. In Kapitel 2 werden theoretische Erklärungsansätze für Nonresponse diskutiert und anschließend Verfahren zur Korrektur von Nonresponse vorgestellt (Kapitel 3), um trotz Nonresponse unverzerrte Schätzer der Grundgesamtheitsparameter zu erhalten. Alle diese Korrekturverfahren treffen (zumindest implizit) eine Annahme über den Ausfallmechanismus. Da sich Ausfallmechanismen für unterschiedliche Ausfallarten unterscheiden können ist für eine Korrektur von Nonresponse eine Unterscheidung der Ausfallarten von zentraler Bedeutung. Der empirische Teil der Arbeit analysiert Ausfälle aus dem SozioOekonomischen Panel (SOEP), dessen Grundgesamtheit und einzelne Substichproben in Kapitel 4 vorgestellt und beschrieben werden. Das SOEP ist eine jährliche Wiederholungsbefragung privater Haushalte und aller erwachsenen Personen in diesen Haushalten. In den Kontaktprotokollen der Interviewer werden bis zu 29 unterschiedliche Kontaktergebnisse festgehalten. Da diese über die Jahre nicht konsistent erfasst wurden, ist eine Aufbereitung der Kontaktergebnisse für jede Welle notwendig, die durch die komplexe Struktur des SOEP erschwert wird (Kapitel 5). Der deskriptive Teil der Analyse der Panelausfälle im SOEP (Kapitel 6 ab Seite 93) zeigt die Entwicklung der verschiedenen Ausfallursachen seit Start des SOEP im Jahr 1984. Da das SOEP eine Wiederholungsbefragung derselben Personen ist, können die Abfolgen der jährlichen Kontaktergebnisse untersucht werden. Die Teilnahmemuster von Teilnahme und Nichtteilnahme der Befragungspersonen des SOEP werden in Kapitel 7.1 beschrieben. Betrachtet man für Nonrespondenten zusätzlich die Abfolge der detaillierten Ausfallursachen, ergeben sich Teilnahmesequenzen. Bei der deskriptiven Analyse dieser Teilnahmesequenzen in Kapitel 7.2 zeigen sich typische häufige Ausfallsequenzen aus dem SOEP für die letzten zwei bzw. letzten drei Jahre, bevor eine Beobachtung aus dem SOEP ausscheidet. Diese Ausfallsequenzen zeigen klare Implikationen für die Feldarbeit und Erhebungspraxis des SOEP auf. Eng verbunden mit der Beschreibung der Sequenzen schließt sich im folgenden Kapitel eine Modellierung der Ausfälle aus dem SOEP als diskrete und stationäre Markov-Kette an (Kapitel 8 ab Seite 112). Ziel der Analyse ist die Schätzung der Übergangswahrscheinlichkeiten zwischen den Kontaktergebnissen zwischen den Wellen. Da die Ausfallursachen als ungeordnete Kategorien vorliegen, können diese Übergangswahrscheinlichkeiten mit einem multinomialen Logitmodell geschätzt werden. Die Ergebnisse des Markov-Modells bestätigen die Ergebnisse der deskriptiven Analyse. Ziel der Arbeit liegt aber weniger in den Übergangswahrscheinlichkeiten im Sinne einer Propensity-Gewichtung, sondern es soll vielmehr gezeigt werden, dass die unterschiedlichen Ausfallarten verschiedeneWahrscheinlichkeiten für eine erneute Teilnahme bzw. einen erneuten Ausfall besitzen. Dadurch soll gezeigt werden, dass eine Korrektur von Nonresponse aufgrund einer einfachen Unterscheidung Respondenten - Nonrespondenten zu kurz greift (vor allem, wenn bestimmte Gruppen eine höhere Ausfallwahrscheinlichkeit durch eine bestimmte Ausfallursache haben) und einen möglichen Nonresponsebias nicht oder nur unzulänglich korrigieren kann.

Suggested Citation

  • Tobias Gramlich, 2008. "Analyse der Panelausfälle im Sozio-oekonomischen Panel SOEP," SOEPpapers on Multidisciplinary Panel Data Research 129, DIW Berlin, The German Socio-Economic Panel (SOEP).
  • Handle: RePEc:diw:diwsop:diw_sp129
    as

    Download full text from publisher

    File URL: https://www.diw.de/documents/publikationen/73/diw_01.c.89659.de/diw_sp0129.pdf
    Download Restriction: no
    ---><---

    References listed on IDEAS

    as
    1. Small, Kenneth A & Hsiao, Cheng, 1985. "Multinomial Logit Specification Tests," International Economic Review, Department of Economics, University of Pennsylvania and Osaka University Institute of Social and Economic Research Association, vol. 26(3), pages 619-627, October.
    2. Alderman, Harold & Watkins, Susan Cotts & Kohler, Hans-Peter & Maluccio, John A. & Behrman, Jere R., 2000. "Attrition in longitudinal household survey data," FCND discussion papers 96, International Food Policy Research Institute (IFPRI).
    3. Duncan Thomas & Elizabeth Frankenberg & James P. Smith, 2001. "Lost but Not Forgotten: Attrition and Follow-up in the Indonesia Family Life Survey," Journal of Human Resources, University of Wisconsin Press, vol. 36(3), pages 556-592.
    4. Harold Alderman & Jere Behrman & Hans-Peter Kohler & John A. Maluccio & Susan Watkins, 2001. "Attrition in Longitudinal Household Survey Data," Demographic Research, Max Planck Institute for Demographic Research, Rostock, Germany, vol. 5(4), pages 79-124.
    5. John Fitzgerald & Peter Gottschalk & Robert Moffitt, 1998. "An Analysis of Sample Attrition in Panel Data: The Michigan Panel Study of Income Dynamics," Journal of Human Resources, University of Wisconsin Press, vol. 33(2), pages 251-299.
    6. Lee A. Lillard & Constantijn W. A. Panis, 1998. "Panel Attrition from the Panel Study of Income Dynamics: Household Income, Marital Status, and Mortality," Journal of Human Resources, University of Wisconsin Press, vol. 33(2), pages 437-457.
    7. Heckman, James, 2013. "Sample selection bias as a specification error," Applied Econometrics, Russian Presidential Academy of National Economy and Public Administration (RANEPA), vol. 31(3), pages 129-137.
    8. David T. Burkam & Valerie E. Lee, 1998. "Effects of Monotone and Nonmonotone Attrition on Parameter Estimates in Regression Models with Educational Data: Demographic Effects on Achievement, Aspirations, and Attitudes," Journal of Human Resources, University of Wisconsin Press, vol. 33(2), pages 555-574.
    9. Denise Hawkes & Ian Plewis, 2006. "Modelling non‐response in the National Child Development Study," Journal of the Royal Statistical Society Series A, Royal Statistical Society, vol. 169(3), pages 479-491, July.
    10. Gerard J. Van Den Berg & Maarten Lindeboom & Peter J. Dolton, 2006. "Survey non‐response and the duration of unemployment," Journal of the Royal Statistical Society Series A, Royal Statistical Society, vol. 169(3), pages 585-604, July.
    11. Gert G. Wagner & Jürgen Schupp & Ulrich Rendtel, 1991. "Das Sozio-ökonomische Panel - Methoden der Datenproduktion und Aufarbeitung im Längsschnitt," Discussion Papers of DIW Berlin 31, DIW Berlin, German Institute for Economic Research.
    12. Peter Lynn, 2006. "Editorial: Attrition and non‐response," Journal of the Royal Statistical Society Series A, Royal Statistical Society, vol. 169(3), pages 393-394, July.
    Full references (including those not matched with items on IDEAS)

    Citations

    Citations are extracted by the CitEc Project, subscribe to its RSS feed for this item.
    as


    Cited by:

    1. Klingwort, Jonas, 2017. "Nonresponse in aktuellen deutschen Viktimisierungssurveys," Duisburger Beiträge zur soziologischen Forschung 2017-01, University of Duisburg-Essen, Institute of Sociology.
    2. Anette E. Fasang, 2010. "Retirement: Institutional Pathways and Individual Trajectories in Britain and Germany," Sociological Research Online, , vol. 15(2), pages 1-16, May.
    3. Marcel Erlinghagen & Christoph Kern & Petra Stein, 2019. "Internal Migration, Social Stratification and Dynamic Effects on Subjective Well Being," SOEPpapers on Multidisciplinary Panel Data Research 1046, DIW Berlin, The German Socio-Economic Panel (SOEP).
    4. Pia S. Schober & Gundula Zoch, 2015. "Change in the Gender Division of Domestic Work after Mummy or Daddy Took Leave: An Examination of Alternative Explanations," SOEPpapers on Multidisciplinary Panel Data Research 803, DIW Berlin, The German Socio-Economic Panel (SOEP).

    Most related items

    These are the items that most often cite the same works as this one and are cited by the same works as this one.
    1. Farshid Vahid & Pushkar Maitra, 2005. "The Effect of Household Characteristics on Living Standards in South Africa 1993 - 98: A Quantile Regression Analysis with Sample Attrition," ANU Working Papers in Economics and Econometrics 2005-452, Australian National University, College of Business and Economics, School of Economics.
    2. Nic Baigrie & Katherine Eyal, 2014. "An Evaluation of the Determinants and Implications of Panel Attrition in the National Income Dynamics Survey (2008-2010)," South African Journal of Economics, Economic Society of South Africa, vol. 82(1), pages 39-65, March.
    3. Thomas, Duncan & Witoelar, Firman & Frankenberg, Elizabeth & Sikoki, Bondan & Strauss, John & Sumantri, Cecep & Suriastini, Wayan, 2012. "Cutting the costs of attrition: Results from the Indonesia Family Life Survey," Journal of Development Economics, Elsevier, vol. 98(1), pages 108-123.
    4. Nicole Watson & Mark Wooden, 2011. "Re-engaging with Survey Non-respondents: The BHPS, SOEP and HILDA Survey Experience," Melbourne Institute Working Paper Series wp2011n02, Melbourne Institute of Applied Economic and Social Research, The University of Melbourne.
    5. Nobuhiko Fuwa, 2010. "Should We Track Migrant Households When Collecting Household Panel Data? Household Relocation, Economic Mobility, and Attrition Biases in the Rural Philippines," American Journal of Agricultural Economics, Agricultural and Applied Economics Association, vol. 93(1), pages 56-82.
    6. Christopher J. Gerry & Georgios Papadopoulos, 2015. "Sample attrition in the RLMS, 2001–10," The Economics of Transition, The European Bank for Reconstruction and Development, vol. 23(2), pages 425-468, April.
    7. Arslan, Aslihan & Taylor, J. Edward, 2011. "Whole-household migration, inequality and poverty in rural Mexico," Kiel Working Papers 1742, Kiel Institute for the World Economy (IfW Kiel).
    8. Chapoto, Antony & Jayne, Thomas S., 2005. "Characteristics of Individuals Afflicted by AIDS-related Mortality in Zambia," Food Security Collaborative Working Papers 54472, Michigan State University, Department of Agricultural, Food, and Resource Economics.
    9. Denise Hawkes & Ian Plewis, 2006. "Modelling non‐response in the National Child Development Study," Journal of the Royal Statistical Society Series A, Royal Statistical Society, vol. 169(3), pages 479-491, July.
    10. Alan Sánchez & Javier Escobal, 2020. "Survey attrition after 15 years of tracking children in four developing countries: The Young Lives study," Review of Development Economics, Wiley Blackwell, vol. 24(4), pages 1196-1216, November.
    11. Islam, Asadul & Nguyen, Chau & Smyth, Russell, 2015. "Does microfinance change informal lending in village economies? Evidence from Bangladesh," Journal of Banking & Finance, Elsevier, vol. 50(C), pages 141-156.
    12. Yamano, Takashi & Jayne, T S, 2005. "Working-Age Adult Mortality and Primary School Attendance in Rural Kenya," Economic Development and Cultural Change, University of Chicago Press, vol. 53(3), pages 619-653, April.
    13. Michael Fertig & Stefanie Schurer, 2007. "Earnings Assimilation of Immigrants in Germany: The Importance of Heterogeneity and Attrition Bias," SOEPpapers on Multidisciplinary Panel Data Research 30, DIW Berlin, The German Socio-Economic Panel (SOEP).
    14. Shin, Jaeun & Moon, Sangho, 2006. "Fertility, relative wages, and labor market decisions: A case of female teachers," Economics of Education Review, Elsevier, vol. 25(6), pages 591-604, December.
    15. Banks, James & Muriel, Alastair & Smith, James P., 2010. "Attrition and Health in Ageing Studies: Evidence from ELSA and HRS," IZA Discussion Papers 5161, Institute of Labor Economics (IZA).
    16. Sarah Xue Dong, 2016. "Consistency between Sakernas and the IFLS for Analyses of Indonesia’s Labour Market: A Cross-Validation Exercise," Bulletin of Indonesian Economic Studies, Taylor & Francis Journals, vol. 52(3), pages 343-378, September.
    17. repec:dau:papers:123456789/5443 is not listed on IDEAS
    18. Maluccio, John A., 2005. "Coping with the “coffee crisis” in Central America: The Role of the Nicaraguan Red de Protección Social," FCND discussion papers 188, International Food Policy Research Institute (IFPRI).
    19. Behr Andreas, 2006. "Comparing Estimation Strategies for Income Equations in the Presence of Panel Attrition: Empirical Results Based on the ECHP," Journal of Economics and Statistics (Jahrbuecher fuer Nationaloekonomie und Statistik), De Gruyter, vol. 226(4), pages 361-384, August.
    20. Fertig, Michael & Schurer, Stefanie, 2007. "Labour Market Outcomes of Immigrants in Germany: The Importance of Heterogeneity and Attrition Bias," IZA Discussion Papers 2915, Institute of Labor Economics (IZA).
    21. Juergen Jung, 2022. "Estimating transition probabilities between health states using US longitudinal survey data," Empirical Economics, Springer, vol. 63(2), pages 901-943, August.

    More about this item

    Lists

    This item is featured on the following reading lists, Wikipedia, or ReplicationWiki pages:
    1. SOEP based publications

    Statistics

    Access and download statistics

    Corrections

    All material on this site has been provided by the respective publishers and authors. You can help correct errors and omissions. When requesting a correction, please mention this item's handle: RePEc:diw:diwsop:diw_sp129. See general information about how to correct material in RePEc.

    If you have authored this item and are not yet registered with RePEc, we encourage you to do it here. This allows to link your profile to this item. It also allows you to accept potential citations to this item that we are uncertain about.

    If CitEc recognized a bibliographic reference but did not link an item in RePEc to it, you can help with this form .

    If you know of missing items citing this one, you can help us creating those links by adding the relevant references in the same way as above, for each refering item. If you are a registered author of this item, you may also want to check the "citations" tab in your RePEc Author Service profile, as there may be some citations waiting for confirmation.

    For technical questions regarding this item, or to correct its authors, title, abstract, bibliographic or download information, contact: Bibliothek (email available below). General contact details of provider: https://edirc.repec.org/data/sodiwde.html .

    Please note that corrections may take a couple of weeks to filter through the various RePEc services.

    IDEAS is a RePEc service. RePEc uses bibliographic data supplied by the respective publishers.