loughborough london school of sport & exercise sciences the economics of sports participation:...
TRANSCRIPT
Loughborough
London
School of Sport & Exercise Sciences
The Economics of The Economics of Sports Participation: Sports Participation: Some Longitudinal Some Longitudinal Analysis.Analysis.
Paper presented to the GHS user group meeting 13Paper presented to the GHS user group meeting 13 thth March 2009 March 2009
Paul DownwardPaul Downward& & Joe RiordanJoe Riordan
Background:Background:1. Promotion of physical activity is now central to public
policy concerns2. Relatively Little economic analysis and large-scale data
testing3. Builds upon Downward (2007); Downward and Riordan
(2007) to analyse (and seek advice?) Participation decisions Social interactions Over time
THAT IS……………………………….
Why do we do this?Why do we do this?
What has happened over time? Should we seek to promote this?
OverviewOverview
• Policy Context
• Literature; Theoretical Predictions
• Data Variables
• Empirical Strategy
• Results
• Discussion
Policy ContextPolicy Context
Seven Drivers of Change Five Settings for Change Ageing population The Home Time Pressures The Community
Well-Being and Obesity The Workplace Levels of Investment Higher and Further Education Utilising Education Primary and Secondary School
Variations in Access Volunteers and Professionals
Sport England Analysis of Determinants of
Participation (2004)
In the UK a ‘Twin track’ approach
Increase mass participation Enhance international success
Increase quantity and quality of participation
Creating a talent identificationand development pathway
A fit, active population A first class successful sporting nation
Sport England Strategy 2008-11Build on school provision → work with NGBs → develop
community sport → Excel, Sustain, Grow
DCMS Game Plan 2002
Literature/TheoryLiterature/Theory
• Heterodox:– Gratton and Tice (1991) explore the psychological
foundations of consumer choice in sport and, in particular, learning by doing (Scitovsky, 1976; Earl, 1986, 1983).
– Post Keynesian consumer analysis draws upon this concept and also insights from the studies of Leisure by Veblen (1925) and Galbraith (1958) and, by implication, Bourdieu (1984, 1988, 1991) that individual preferences are shaped by social values.
– Prior experience in sports activities is likely to raise participation in any specific activity, and that social interactions, or lifestyles, will also affect participation.
• Uncertainty, preferences evolve, social constraints
Literature/TheoryLiterature/Theory
• Neoclassical– Income-Leisure Trade off of Labour Supply (Gratton
and Taylor, 2000)
– Becker (1965, 1974). • The latter paper is directly concerned with the
accumulation of personal-consumption capital and social interactions in consumption.
Literature/TheoryLiterature/TheoryFunction Interpretation
Uit = Uit (Pit) Utility depends on participation Pit = Pit(xit, t, Cit, Sit) Participation is produced
Mit = pxxit + pesESit + pecECit Money income constraint Cit = C0
it-1 + ECit Consumption skill enhancement Sit = S0
it-1 + ESit Social characteristics enhancement Mit +pes S
0it-1 + pec C
0it-1 = pxxit + pesSit + pecCit Wealth constraint
Marginal utility mediated through marginal productivity as
dUit = (δUit /δPit)(δPit /δxit)dxit + (δUit /δit)(δPit /δit)dit + (δUit /δPit)(δPit /δCit)dCit +
(δUit /δPit)(δPit /δSit)dSit
Literature/TheoryLiterature/TheoryFunctions Marshallian Demands
Uit = CitαSit
β Sit = (β/β+α) (Mit + pes S0it-1 + pec C
0it-1)/ pes
Mit +pes S
0it-1 + pec C
0it-1 = pxxit + pesSit + pecCit
Cit = (α/β+α) (Mit + pes S
0it-1 + pec C
0it-1)/ pec
The ‘Marshallian’ demands for consumption skills and social
characteristics are augmented by their previous endowment. Rise with money income, fall with their prices, but rise as a result of
their previous acquisition. Increases in participation through the production function will
mediate this resource transfer.
Can integrate time cost explicitly in a simple way e.g. add wst and wct to r.h.s. so that economic shadow price includes time (If ws ≠ wc then an element of ‘depreciation’ so that per period allocations are different).
Literature/TheoryLiterature/TheoryFunction Interpretation
Uit = C1itαC2it
β Utility comes from participation in 2 sports
C1i2 = C1i0 + C1i1
Participation in C1 evolves over time C2i2
= C1i0 + C2i1
Mit = p1C1it+ p2 C2it And augments the ability to consume C2
Money budget constraint C2i2 = (β/β+α) (Mit + p2 C1i0)/ p2
Consumption skill enhancement of C2 by initial skills from C1
Allow for time
Mit = p1C1it+ p2 C2it + C1it w1t+C2it w2t Budget constraint C2i2 = (β/β+α) (Mit + p2 C1i0)/ (p2 + w2t)
Consumption skills now also
diminish/barriers decrease with time
In general this suggests that sports participation will be likely to vary directly with the acquisition of specific personal consumption and social capital, and also with the decline in any initial obstacles to participation through time (no reinvestment!).
Empirical WorkEmpirical WorkCountry Author Theory Indicative Findings
US Cicchetti et al (1969)
Neoclassical Demand
Age (-); Non-white (-); Male (+); Income (+); Education (+); Facility supply (+)
US Adams et al (1966)
Neoclassical Demand
Age (-); Income (+); Male (+); Education (+); White (+);
US Stempel (2005) Bourdieu Income (+); Education (+)
US Humphreys andRuseski
(2006)
Becker Age (-); Married (-); Children (-); Income (+); Employed (-); Retired (+); Education (+); Female (-); White (+); Urban (+); Health (+);
Empirical WorkEmpirical WorkCountry Author Theory Indicative Findings
UK Gratton and Tice (undated)
Heterodox Male (+); Age (-); Socio-economic status (+); Income (+); Illness (-); Number of Activities (+)
UK Farrell and Shields (2002)
(Implicit Becker Household preferences)
Male (+); Age (-); Married (-) Children for males (+); Infant (-); Ethnic Minority (-); Education (+); Drinling (+); Smoking (-); Health (+); Income (+); Unemployment (+); Household membership (+)
UK Sturgis and Jackson (2003)
Unstated Age (-); Number of Adults in the household (-); Income (+); Male (+); London/SE (+); Own House (+); Education (+)
UK Downward (2007) Neoclassical and Heterodox
Working (+); Skills/Professional (+); Education (+); Married (+); Regions not SE (+); Male (+); White (+); Health (+); Smoking (-); Drinking (+); Access to vehicle (+); Age (-); Children (-); Number of Adults in the household (-); Income (+); Work hours (-); Unpaid work hours (-) Volunteering (+); number of leisure activities (+)
UK Downward and Riordan (2007)**
Becker and Heterodox Age (-); Skills/Professional (+); Drinking (+); Regions not SE (-); Access to a vehicle (+); Sports, other club membership (+); Volunteer (-); Number of sports (+); Sport lifestyle (-)
Empirical WorkEmpirical WorkCountry Author Theory Indicative Findings
Flanders Scheerder et al (2005a)*
Bourdieu Age (-); Female (-); Class (+); Family size (+); Urban (+)
Flanders Scheerder et al (2005b)*
Bourdieu Humanities school (-); Parents participating in sport (+)
Flanders Taks and Scheerer (2007)*
Bourdieu/Market segmentation
Female (-); Age (+); Socio-economic (+);Parents participating in sport (+) In sport (+) 5. Age (-); Socio-economic (+); Parents participating in sport (+)
Australia Stratton et al (2005) No explicit theory Age (-); male (+); State (+); Suburb (+); Professional (+); Income (+); Socio-economic (+)’ Couple no children (+); Single (+); Education (+); English speaker (+); Health (+); Easy transport (-); Not safe environment (-); Weekly contact family friends (-)
Norway Skille (2005) Bourdieu Female (-); Academic school (+); Active family members (+); Volunteer family members (+); Peer and Media information (+)
Germany Breuer (2006) Becker Income (+); Working time (-); Education (+); Age (-); Immigrant (-);
Germany Lechner (2008) No explicit theory
Males: German (+); Education (+);
Year (-); Technical occupation (-); autonomy at work (+); never smoked (+); High life satisfaction (+); Unemployment (+)
Females: Year (-); German (+); Children < 3 years (-); Children > 10 years (+); Family income (+); Office work (-); low autonomy at work (-); autonomy at work (+); illness (+); unemployment (+); inhabitants per km2 (+); City centre (-)
Empirical WorkEmpirical Work
*Studies used either Factor analysis of Cluster analysis to group activities.
**Study used cluster analysis to identify lifestyles
Empirical WorkEmpirical Work
Increase Participation Decrease Participation
1. Male
2. Socio-economic status
3. Income
4. Health
5. Education
6. Transport
7. Drinking
1. Smoking
2. Children
3. Marital status
4. Work hours
5. Ethnicity
•Time (Investment; preferences shifting)? •Social Interactions (groups of characteristics)?
Data\VariablesData\VariablesThe General Household Survey (GHS) was the data source for the research.• A continuous survey, which began in 1971, and is conducted by the
Office for National Statistics. It collects data on a range of topics, by face-to-face interview, from private households in Great Britain. As well as core topics such as household and family characteristics, education, health, income and demographics, it also investigates other topics, such as Sport and Leisure, periodically.
• Data from the 1980, 1986, 1990, 1996/7, 2002 are available.Some analysis done in Downward and Riordan (forthcoming, 2009)
• But only 22 activities participated in or not over the last four weeks• Conformity?
– Income a problem. Household and individual data, gross and net. A series of net income per week per individual identified by proportionate adjustment. This was deflated by the Retail Price Index for the year.
– Some socio-economic; regional characteristics identifiable at more aggregate levels– Odd patterns of participation
1980-20021980-2002
Sample Year Sample Size % 1980 12157 25.5 1986 7055 14.8 1990 6893 14.4 1996 7691 16.1 2002 13943 29.2 Total 47739 100.0
1980-20021980-2002
Participation in any sport Sample Year
(anysport) 1980 1986 1990 1996 2002 Overall No Participation 71.8% 65.2% 54.7% 54.0% 67.1% 64.1% Participation 28.2% 34.8% 45.3% 46.0% 32.9% 35.9%
Sample Size 12157 7055 6893 7691 13943 47739
Sample sizes?
1996/7 -2002
1. 40 Activities
2. Easier to match variables
Pooled data, not a panel, from 5 different years of GHS Survey
Empirical StrategyEmpirical Strategy
• Cluster Analysis (Two-Step)– Personal Consumption capital; Social Characteristics
• Regression analysis (Controlled for household selection; robust errors)– Individual factors, plus cluster membership variable.
– Participation Decision• Ln(Pit/1-Pit) = β0 + ∑βjXjit + vit
– Number of Sports• Numsportit = α0 + ∑αjXjt + uit
Adequate strategy?
Empirical Strategy: Selection?Empirical Strategy: Selection?
Tobit Model: But: Lacks robust SE corrections, cluster sampling and weighting
options• Numsport*it = α0 + ∑ α jXjit + uit
if Numsportit* ≤ 0 Numsportit = 0
if Numsportit* > 0 Numsportit = Numsportit*
Heckman Model to distinguish decisions/correct for sample bias:
• (H1) Numsport*it = α0 + ∑ α jXjit + uit Numsporti > 0 only if Pit =1.
• (H2) Pit = β0 + ∑βjXjit + vit Pit = 1 and 0 otherwise
Where u is N(0, σ)v is N(0, 1)
Corr (u, v) = ρ
It could be the case that the choice set
comprises voluntary decisions to
participate on any number of independent
occasions which could include not at allCoefficient Z P > z
Mills lambda 0.44 1.49 0.14rho 0.34sigma 1.29lambda 0.44
Descriptives (1)numsport1 n %
0 16970 55.8
1 6767 22.22 3188 10.53 1630 5.44 855 2.85 457 1.56 244 0.87 153 0.58 61 0.29 47 0.2
10 19 0.1
11 11 012 8 0
13 5 014 1 015 1 016 1 017 2 0
Total 30420 100
anysport1 n %
No 16970 55.8Yes 13450 44.2Total 30420 100
Descriptives (2) Decline
1997 2002numsport1 n % n %
0 8522 54.5 8448 57.11 3539 22.6 3228 21.82 1694 10.8 1494 10.13 869 5.6 761 5.14 454 2.9 401 2.75 250 1.6 207 1.46 127 0.8 117 0.87 85 0.5 68 0.58 38 0.2 23 0.29 27 0.2 20 0.1
10 13 0.1 6 0.011 6 0.0 5 0.012 5 0.0 3 0.013 3 0.0 2 0.014 1 0.0 0 0.015 1 0.0 0 0.016 1 0.0 0 0.017 1 0.0 1 0.0
Total 15636 100.0 14784 100.0
anysport1 1997 % 2002 % Total
No 8522 54.5 8448 57.1 16970Yes 7114 45.5 6336 42.9 13450
Total 15636 100 14784 100 30420
Sport 1997 2002
Swim indoors 12.80% 12.10%Cycling 11.00% 8.80%Keep fit 12.30% 12.00%
Weight training 5.60% 5.50%Football outdoors 3.80% 3.60%
Rugby 0.60% 0.40%Cricket 0.90% 0.60%Hockey 0.30% 0.30%Netball 0.50% 0.30%
Track events 0.20% 0.20%Jogging 4.50% 4.80%
Cluster analysis
Cluster Cluster 1 Cluster Cluster 1Variable 1 2 Cluster 2 Variable 1 2 Cluster 2
numsport1 Mean 0.78 1.51 Female n 4550 3246 1.40S Dev 1.04 1.97 % 58.40% 41.60%
realinc Mean 385.43 359.81 Male n 5052 3104 1.63S Dev 268.95 582.12 % 61.90% 38.10% > male
ndepchldd Mean 0.79 0.47 Don't play Golf n 9035 5981 1.51S Dev 1.05 0.85 % 60.20% 39.80%
tothrsd2 Mean 29.60 27.91 Golf n 567 369 1.54S Dev 19.43 19.33 % 60.60% 39.40% > golf
nadmalesd Mean 1.19 0.96 Don't Swim n 8371 5319 1.57S Dev 0.45 0.83 % 61.10% 38.90%
nadfemsd Mean 1.14 1.03 Swim n 1231 1031 1.19S Dev 0.40 0.75 % 54.40% 45.60% < swim
Age Mean 45.53 36.50 Don't bowl n 9520 6313 1.51S Dev 11.80 14.57 % 60.10% 39.90%
genhlthd1 Mean 2.05 2.19 Indoor bowls n 82 37 2.22S Dev 0.84 0.82 % 68.90% 31.10% > bowls
Non white n 410 340 1.21 No football outdoors n 9408 5833 1.61% 54.70% 45.30% % 61.70% 38.30%
White n 9192 6010 1.53 Football outdoors n 194 517 0.38% 60.50% 39.50% > White % 27.30% 72.70% < football
non-married n 0 6030 0.00 Don't play Netball n 9584 6305 1.52% 0.00% 100.00% % 60.30% 39.70%
Married n 9602 320 30.01 Netball n 18 45 0.40% 96.80% 3.20% > Married % 28.60% 71.40% < netball
non-professional n 9082 6099 1.49 Don't drink n 907 666 1.36% 59.80% 40.20% % 57.70% 42.30%
Professional n 520 251 2.07 Drink n 8695 5684 1.53% 67.40% 32.60% > professional % 60.50% 39.50% > drink
no Higher ed n 6811 4534 1.50 Don't smoke n 6827 3736 1.83% 60.00% 40.00% % 64.60% 35.40%
Higher ed n 2791 1816 1.54 Smoke n 2775 2614 1.06% 60.60% 39.40% > higher ed % 51.50% 48.50% < smoke
Logistic Regression
anysport1 Coefficient z P>z Probability z P>z anysport1 Coefficient z P>z Probability z P>z
cohort -0.08 -8.25 0.00 -0.02 -8.25 0.00 cohort -0.09 -8.32 0.00 -0.02 -8.32 0.00cignnowd1 -0.37 -9.28 0.00 -0.09 -9.34 0.00 cignnowd1 -0.36 -9.01 0.00 -0.09 -9.05 0.00drinknowd1 0.61 9.23 0.00 0.15 9.61 0.00 drinknowd1 0.61 9.30 0.00 0.15 9.62 0.00genhlthd1 0.12 4.37 0.00 0.03 4.37 0.00 genhlthd1 0.12 4.21 0.00 0.03 4.21 0.00tothrsd2 0.00 0.37 0.71 0.00 0.37 0.71 tothrsd2 0.00 0.47 0.64 0.00 0.47 0.64realinc 0.00 4.25 0.00 0.00 4.25 0.00 realinc 0.00 1.80 0.07 0.00 1.80 0.07married -0.23 -1.84 0.07 -0.06 -1.84 0.07 married -0.22 -1.76 0.08 -0.05 -1.77 0.08single -0.09 -0.69 0.49 -0.02 -0.69 0.49 single -3.10 -10.30 0.00 -0.60 -16.70 0.00
sepdivorce -0.10 -0.77 0.44 -0.03 -0.77 0.44 sepdivorce -0.16 -1.24 0.22 -0.04 -1.24 0.22professional 0.58 4.87 0.00 0.14 5.13 0.00 professional 0.62 5.04 0.00 0.15 5.37 0.00managetech 0.58 6.40 0.00 0.14 6.55 0.00 managetech 0.59 6.51 0.00 0.15 6.68 0.00
skillnonm 0.52 5.95 0.00 0.13 6.09 0.00 skillnonm 0.52 5.80 0.00 0.13 5.95 0.00skillman 0.40 4.57 0.00 0.10 4.64 0.00 skillman 0.42 4.75 0.00 0.10 4.85 0.00partskill 0.26 2.98 0.00 0.07 3.00 0.00 partskill 0.27 3.03 0.00 0.07 3.06 0.00highered 0.54 9.79 0.00 0.13 9.98 0.00 highered 0.53 9.25 0.00 0.13 9.43 0.00
alevel 0.50 8.36 0.00 0.12 8.60 0.00 alevel 0.49 8.01 0.00 0.12 8.26 0.00olevel 0.33 6.36 0.00 0.08 6.43 0.00 olevel 0.31 5.86 0.00 0.08 5.93 0.00asian -0.63 -3.11 0.00 -0.15 -3.28 0.00 asian -0.59 -2.91 0.00 -0.15 -3.03 0.00black -0.51 -2.46 0.01 -0.12 -2.55 0.01 black -0.48 -2.30 0.02 -0.12 -2.36 0.02white 0.03 0.17 0.86 0.01 0.17 0.86 white 0.01 0.07 0.95 0.00 0.07 0.95
ndepchldd 0.02 0.75 0.45 0.00 0.75 0.45 ndepchldd 0.02 0.74 0.46 0.00 0.74 0.46nadmalesd -0.09 -2.42 0.02 -0.02 -2.42 0.02 nadmalesd -0.04 -1.07 0.29 -0.01 -1.07 0.29nadfemsd -0.12 -3.42 0.00 -0.03 -3.42 0.00 nadfemsd -0.10 -2.70 0.01 -0.02 -2.70 0.01
age -0.12 -12.08 0.00 -0.03 -12.08 0.00 age -0.12 -11.82 0.00 -0.03 -11.82 0sex 0.61 15.58 0.00 0.15 15.82 0.00 sex 0.53 13.19 0.00 0.13 13.34 0
Cluster 3.14 11.34 0.00 0.78 11.42 0_cons 169.26 8.27 0.00 _cons 168.25 8.19 0.00
n = 15824 n = 15824χ2 = 1836.11 χ2 = 1920.75P > χ2 = 0.000 P > χ2 = 0.000Pseudo R2 = 0.1124 Pseudo R2 = 0.13229829 clusters used to correct Ses 9829 clusters used to correct Ses
Regressionsnumsport1 Coefficient t P>t numsport1 Coefficient t P>t
cohort -0.04 -7.55 0.00 cohort -0.05 -8.79 0.00cignnowd1 -0.27 -10.94 0.00 cignnowd1 -0.24 -10.73 0.00drinknowd1 0.20 6.02 0.00 drinknowd1 0.20 6.66 0.00genhlthd1 0.04 2.59 0.01 genhlthd1 0.04 2.83 0.01tothrsd2 0.00 -0.32 0.75 tothrsd2 0.00 -0.50 0.62realinc 0.00 5.06 0.00 realinc 0.00 -1.46 0.15married -0.36 -6.49 0.00 married -0.33 -6.61 0.00single -0.07 -1.17 0.24 single -3.13 -35.31 0.00
sepdivorce -0.17 -2.89 0.00 sepdivorce -0.26 -5.06 0.00professional 0.30 3.66 0.00 professional 0.32 4.48 0.00managetech 0.23 4.60 0.00 managetech 0.25 5.42 0.00
skillnonm 0.22 4.48 0.00 skillnonm 0.20 4.32 0.00skillman 0.05 0.91 0.36 skillman 0.10 2.15 0.03partskill 0.03 0.63 0.53 partskill 0.05 1.18 0.24highered 0.32 9.09 0.00 highered 0.26 8.21 0.00
alevel 0.30 7.71 0.00 alevel 0.25 7.05 0.00olevel 0.17 5.39 0.00 olevel 0.12 4.30 0.00asian -0.44 -3.68 0.00 asian -0.34 -3.01 0.00black -0.33 -2.66 0.01 black -0.25 -2.12 0.03white 0.05 0.54 0.59 white 0.03 0.36 0.72
ndepchldd 0.02 1.45 0.15 ndepchldd 0.02 1.76 0.08nadmalesd 0.00 -0.13 0.89 nadmalesd 0.09 4.24 0.00nadfemsd -0.08 -3.72 0.00 nadfemsd -0.04 -2.05 0.04
age -0.07 -12.50 0.00 age -0.07 -12.75 0.00sex 0.50 20.62 0.00 sex 0.31 14.38 0.00
Cluster 3.39 47.66 0.00_cons 89.68 7.70 0.00 _cons 90.23 8.59 0.00
n = 15824 n= 15824
F( 25, 9828) = 103.73 F( 26, 9828) = 211.20
Prob > F = 0.0000 Prob > F = 0.0000
R-squared = 0.1711 R-squared = 0.3541
9829 clusters used to correct Ses 9829 clusters used to correct Ses
Interactions?numsport1 Coefficient t P>t numsport1 Coefficient t P>t
cohort -0.13 -3.21 0.00 cohort -0.16 -3.52 0.00cignnowd1 -0.24 -10.87 0.00 cignnowd1 -0.28 -11.07 0.00drinknowd1 0.20 6.68 0.00 drinknowd1 0.20 6.11 0.00genhlthd1 0.04 3.31 0.00 genhlthd1 0.05 3.21 0.00tothrsd2 0.00 0.20 0.84 tothrsd2 0.00 0.57 0.57realinc 0.00 -1.45 0.15 realinc 0.00 5.07 0.00married -0.31 -6.17 0.00 married -0.33 -5.97 0.00single -3.10 -34.60 0.00 single -0.03 -0.50 0.62
sepdivorce -0.23 -4.55 0.00 sepdivorce -0.14 -2.34 0.02professional 0.32 4.40 0.00 professional 0.29 3.58 0.00managetech 0.25 5.39 0.00 managetech 0.23 4.56 0.00
skillnonm 0.19 4.27 0.00 skillnonm 0.22 4.41 0.00skillman 0.09 2.06 0.04 skillman 0.04 0.80 0.42partskill 0.05 1.14 0.26 partskill 0.03 0.54 0.59highered 0.26 8.16 0.00 highered 0.31 9.00 0.00
alevel 0.24 6.85 0.00 alevel 0.29 7.43 0.00olevel 0.11 4.03 0.00 olevel 0.16 5.06 0.00asian -0.32 -1.55 0.12 asian -0.54 -2.21 0.03
asiantime -0.01 -0.05 0.96 asiantime 0.13 0.46 0.65black -0.21 -1.00 0.32 black -0.36 -1.56 0.12
blacktime -0.05 -0.19 0.85 blacktime 0.05 0.19 0.85white -0.04 -0.26 0.80 white -0.08 -0.42 0.68
whitetime 0.10 0.48 0.63 whitetime 0.17 0.75 0.46ndepchldd 0.03 2.05 0.04 ndepchldd 0.03 1.77 0.08nadmalesd 0.09 4.26 0.00 nadmalesd 0.00 -0.12 0.91nadfemsd -0.04 -1.93 0.05 nadfemsd -0.08 -3.57 0.00
age -0.16 -3.79 0.00 age -0.20 -4.25 0.00agetime 0.01 4.43 0.00 agetime 0.01 5.62 0.00
sex 0.32 9.36 0.00 sex 0.54 14.15 0.00sextime -0.02 -0.49 0.62 sextime -0.06 -1.45 0.15Cluster 3.39 47.48 0.00_cons 264.42 83.16 3.18 _cons 330.72 3.54 0.00
n = 15824 n = 15824 F( 30, 9828) = undefined F( 30, 9828) = 87.18 Prob > F = undefined Prob > F = 0.000 R-squared = 0.3553 R-squared = 0.17329829 clusters used to correct Ses 9829 clusters used to correct Ses
DiscussionDiscussion• Some standard drivers of participation receive large-scale
empirical support– Income– Human capital – education; employment– Health– Minor impact of family (aggregate measure?)– Evidence that social and consumption characteristics matter– Age (-); Sex (M>F)
• Time?– Cohort variable suggests declining general interest in sport– Interaction effects suggest reducing impact of traditional
constraints/choices except BAME when allow for lifestyles• Access to a given sport seems to be easing (policy?)
– Combined effects are decline.