Published on January 23, 2008
Examining Participation in Sporting and Cultural Activities : Examining Participation in Sporting and Cultural Activities Patrick Sturgis, University of Surrey Jon Jackson, London School of Economics Study Aims: Study Aims DCMS desire to enhance understanding of participants and non-participants Identify characteristics associated with participation in sport, and ‘culture’ Answer specific questions from Ministers and policy advisors Analytical Approaches: Analytical Approaches OLS and logistic regression models to identify demographic correlates of activity. CHAID modeling to identify complex interactions amongst predictors. Latent class and cluster analysis of sporting and cultural activities. Modeling predictors of class and cluster membership. Measuring Activities: Measuring Activities In addition to the diary, UKTUS contained traditional ‘reference period’ questions for sport and cultural activities: “In the last 4 weeks, have you been to … Where possible we used the diary data for our analyses. But, because of short reference period, we sometimes had to rely on retrospectives. Two Cultures?: Two Cultures? Are sports participators different from those with higher rates of cultural participation? Series of logistic regressions predicting likelihood of participating in named sports and cultural activities. Summary models (OLS) predicting # different activities in each domain. Factors Associated with # Cultural Activity: Factors Associated with # Cultural Activity Culture Factors Associated with # Sport Activity: Factors Associated with # Sport Activity Sport Culture Higher Qualifications = less sport?: Higher Qualifications = less sport? This is a somewhat counter-intuitive finding. But the model is predicting # ‘sporting’ activities in past 4 weeks. Predicting probability of doing any sport on average day from diary reverses the effect. Highlights importance of careful consideration of operationalisation of activity dependent vars. Latent Class Models: Latent Class Models Initial analyses treated ‘sport’ and ‘culture’ as homogenous classes of activity. However, water polo is clearly quite a different proposition than crazy golf! The LCA groups respondents into classes on the basis of their probability of doing different activities. We then use regression models and CHAID to examine the characteristics of these more refined activity variables. Sport: Sport Virtually no sport at all Swimming, cycling, racket sports, ball games, athletics Swimming, cycling, gymn, walking High probability all sports Golf, bowls, pub games, walking Culture: Culture Pub/café Theatre, concerts, opera, museums stately homes, museum, library Theme park, zoo, car boot, sport spectator Shopping, pub/café, cinema No Sport: No Sport N=10,648 67.6% 16-24 years N=1,172 52.8% 45-64 years N=2,840 80.5% Age Gender Female N=666 62.4% No N=454 94.2% Yes N=2,386 77.7% Male N=506 42.1% Household income <£33,800 N=1,112 60.9% >£33,800 N=325 39.5% Use of a car, motor bike or other motor vehicle Yes N=1,612 67.2% No N=326 85.4% Socio-economic classification Routine and other N=992 86.3% Managerial / professional N=844 67.5% Intermediate N=550 77.4% 8-15 years N=1,496 27.6% 25-44 years N=3,381 63.8% Gender Male N=1,437 56.0% Female N=1,944 70.5% Qualifications Below A levels N=535 78.0% Above A levels N=696 57.8% A levels and equivalent N=381 69.5% Qualifications Above GCSE level N=653 54.6% GCSE level and below N=459 69.9% Use of a car, motor bike or other motor vehicle Qualifications Degree N=471 60.8% Below degree N=373 75.8% Good N=650 82.5% Fair, poor N=342 93.4% Health 65+ years N=1,759 94.2% Yes N=1,005 90.9% No N=754 98.5% Use of a car, motor bike . . . <£20,860 N=906 27.6% >£20,860 N=590 15.6% Household income No Sport No Culture: No Culture Conclusions: Conclusions There is great variation in patterns and rates of sporting and cultural activity in the UK. The majority do very little of either. There are differences in the profiles of sports and cultural participators. Evidence of a general ‘participation’ effect. This is related to social class, education and income. Comments: Comments More attitudinal and contextual variables. Narrow time sampling poses limitation for covariance-based analyses. Representativeness – net diary response = 44%. Analysis must take account of hierarchical structure of the data. Use ‘svy’ commands in Stata!