Conference Paper BHPS-2009 Conference: the 2009 British Household Panel Survey Research Conference, 9-11 July 2009, Colchester, UK
Panel attrition in the BHPS - a question of personality?
Panel attrition and nonresponse are seen as major threats for the quality of survey data.
Panel attrition means that participants of one wave of a panel reject to take part in a next
wave. If the attrition shows a systematic character, there might be a strong bias in the
panel data. Nonresponse occurs in cross sectional surveys, if sample persons refuse to
participate. If those substantially differ from participating subjects, the survey data
might be biased.
The literature on nonresponse mentions several hypotheses about the mechanisms that
lead to nonresponse. One of these hypotheses is the social isolation hypothesis.
Following this hypothesis, persons tend to refuse a survey request more likely, if they
feel socially isolated. In this sense, the phenomenon of nonresponse is seen as related to
specific, marginalised subgroups of a society. The social isolation hypothesis was
empirically tested in several studies. Social isolation was usually operationalised
through socio-demographic and socio-economic variables. But research on personality
psychology has shown that social isolation is primarily a question of personality. One of
the most comprehensive concepts of personality is the “Five Factor Model” - a purely
descriptive model of personality, which includes five different personality dimensions
(Big 5): extraversion, agreeableness, conscientiousness, neuroticism, and openness.
In my studies on nonresponse in the German General Social Survey in 2004 and 2006 I
followed the idea that social isolation, caused by specific personality traits, causes
nonresponse. I came to the result, that neuroticism and conscientiousness are negatively
correlated with the willingness to participate in the GGSS, while extraversion,
agreeableness, and openness are positively correlated.
These results of the GGSS shall serve as basis for further analyses on the BHPS. In my
paper I analyse the data of the BHPS in order to answer the following question:
Compared to my results on the impact of personality on nonresponse, do personality
traits affect panel attrition in the same way?