82 | Chapter 1 3 membership is a subset of the risk set for stopping a sport, consisted of 13,955 person-years in which 1530 respondents were at risk of ending sport club membership, because they were practising one or more sports in a club context. On average, respondents were at risk to stop practising a sport for 11 years (24,497/2272=10.8), and at risk of ending a sport club membership for 9 years (13,955/1530=9.1). Measurements Sport participation was measured in accordance with prevailing Dutch guidelines for sport participation research (Bottenburg, 1999; Bottenburg & Smit, 2000; The Netherlands Institute for Social Research (SCP), n.d.). A show card depicting 44 different sports was used to prompt respondents. Respondents were asked whether they had participated or were currently participating in any of these sports at least once every 14 days. Life-course information was asked on each sport performed, including starting and stopping ages, participation context, frequency and motivation. If a respondent started (and stopped) a certain sport multiple times, sport history details were included for all of these different periods. Our first dependent variable, stop practising a sport, was scored 1 in years in which a respondent stopped practising a sport (2854 events), and 0 in years they did not stop practising the sport. The dependent variable ending a sport club membership was scored 1 for years in which a respondent ended membership in a sport club (1447 events), and 0 in all other years. The risks or “hazard rates” of stopping a sport and ending a sport club membership in young adulthood, which is what event history refers to, are presented in Figure 3.1. The figure shows that at age 18 22.4% of all respondents stopped practising a sport. This risk declined to around 10% after age 25. The risk of ending a sport club membership in young adulthood followed a similar path but with lower percentages, indicating that stopping a sport often coincides with ending a club membership. Information on the four life events, beginning to work, living on one’s own, cohabiting or getting married, and birth of one’s first child, was also gathered retrospectively. For every life event we created a time-varying variable which we assigned a value of 1 in the year a respondent experienced that specific event. In all other years, respondents were scored 0. We constructed four