Skip to main content
. Author manuscript; available in PMC: 2009 Jul 14.
Published in final edited form as: Drug Alcohol Depend. 2008 Apr 21;96(1-2):155–164. doi: 10.1016/j.drugalcdep.2008.02.011

Table 4.

Selected Estimation Results for Gambling Outcomes Past Year

Explanatory Variables Any Gambling Related Problemsa Number of Gambling-Related Problems (conditional)b Number of Gambling-Related Problems (unconditional)c
Baseline Means/Proportions 0.030 2.275 0.069
Weekly or more frequent drinking
Single-equation estimation 0.013***
[0.009,0.018]		 0.041
[-0.214,0.296] 		0.036***
[0.021,0.051]
Simultaneous-equation estimation 0.065**
[0.001,0.128] 		0.133
[-1.147,1.413]		 0.044***
[0.020,0.067]
Wald test for independent equations -0.299**
[-0.535,-0.020]		 -0.029
[-0.386,0.336]		 -0.009
[-0.043,0.016]
Wald test for exclusion restrictionsd 3.32 3.04 6.19
Any drinking to intoxication
Single-equation estimation 0.021***
[0.015,0.028]		 0.442**
[0.093,0.791] 		0.070***
[0.050,0.089]
Simultaneous-equation estimation 0.066**
[0.004,0.128] 		0.345
[-0.188,0.879]		 0.079***
[0.060,0.098]
Wald test for independent equations -0.240*
[-0.461,0.009]		 0.030
[-0.093,0.152]		 -0.011*
[-0.023,0.002]
Wald test for exclusion restrictionsd 3.25 3.20 6.05
Alcohol abuse and/or dependence
Single-equation estimation 0.044***
[0.032,0.056] 		0.570***
[0.222,0.917] 		0.162***
[0.124,0.200]
Simultaneous-equation estimation 0.029
[-0.039,0.097]		 0.495*
[-0.38,1.028]		 0.166***
[0.129,0.204]
Wald test for independent equations 0.070
[-0.269,0.393]		 0.023
[-0.077,0.122]		 -0.004
[-0.012,0.003]
Wald test for exclusion restrictionsd 3.82 3.48 8.00**
Observations 41,270 1,203 41,270

Notes: All coefficient estimates are marginal effects and were estimated at mean values for other regressors. All specifications include the same list of covariates as in Table 2. Bolded coefficient estimates correspond to the preferred model (single-equation or simultaneous-equation) based on the specification tests. 95% confidence intervals in brackets.

a

Estimated with probit in single-equation estimation and recursive bivariate probit in simultaneous-equation estimation.

b

Estimated with OLS in single-equation estimation and a treatment effects regression model in simultaneous-equation estimation. Sample includes respondents who experienced one or more gambling-related problems within the past year.

c

Estimated with OLS in single-equation estimation and treatment effects regression in simultaneous-equation estimation.

d

Just identified models were estimated following Bollen, Guilkey, and Mroz (1995) and Rashad and Kaestner (2004). The tests were preformed multiple times with each instrument selected as the excluded instrument. The most conservative test results (i.e., smallest p-values) are reported.

*

Statistically significant, p ≤ 0.10;

**

Statistically significant, p ≤ 0.05;

***

Statistically significant, p ≤ 0.01.

HHS Vulnerability Disclosure