Table 3.

Estimation results of bivariate probit model with selectivity correction and ordinary probit: male sample

Explanatory variables	Bivariate probit with selectivity correction				Ordinary probit
	Participate or not		Meeting the recommended level		Meeting the recommended level
	Coefa.	ME	Coefa.	ME	Coefa.	ME
Degree qualification	0.159***	0.064
Employed	−0.081*	−0.032
Age	−0.022***	−0.009	−0.017***	−0.004	−0.021***	−0.003
Access to vehicle			−0.184**	−0.047	−0.148**	−0.026
Ethnicityb
Mixed			0.130	0.033	0.016	0.003
Asian			−0.080	−0.018	−0.100	−0.016
Black			−0.042	−0.010	−0.104	−0.016
Chinese			−0.611**	−0.104	−0.519**	−0.060
Marital statusc
Other	0.028	0.012
Married (living)	−0.106**	−0.041
Income	0.000***	0.000	0.000***	0.000	0.000***	0.000
Full time work	−0.139**	−0.055
No. of children	0.050	0.019***
Drinkers	0.135**	0.052
Smokers	−0.215***	−0.085	−0.302***	−0.067	−0.304***	−0.045
Voluntary activity
Health statusd
Good health	0.717***	0.276			0.339**	0.050
Fair health	0.431***	0.163			−0.041	−0.022
Obese	−0.131***	−0.051	−0.301***	−0.065	−0.274***	−0.041
Seasonal effect e
Summer	0.309***	0.121	0.070	0.017	0.143**	0.025
Spring	0.131**	0.052	0.101	0.025	0.131**	0.023
Autumn	0.135***	0.053	0.177**	0.044	0.172**	0.030
Region of residencef
North east	−0.405***	−0.159
North west	−0.239***	−0.095
Yorkshire	−0.179**	−0.073
East Midlands	−0.128*	−0.053
West Midlands	−0.132*	−0.054
East	−0.074	−0.033
London	−0.268***	−0.107
South west	−0.102	−0.044
Constant	0.140		−0.263
Observations	6324		3425		6324
Rho	0.154		0.154
	P = 0.524		P = 0.524

Coef, Coefficients; ME, Marginal effects.

a: The estimated parameters and asterisks show significance level of 1% (***), 5% (**) and 10% (*).

b: Reference category: white.

c: Reference category: single.

d: Reference category: poor health.

e: Reference category: winter.

f: Reference category: south east.

Hausman specification test (null hypothesis: difference in coefficients in two models is not systematic): Chi² (7) = 10.13; P = 0.60.