Question

44-47. A researcher studying voting behavior ran a logistic regression model and got the following results. The dependent variable is Romneyvote, coded 1 if the respondent voted for Romney, 0 if the respondent voted for Obama. Pid3_01, which measures Party ID, is coded 0=Democrat, 1=Moderate, 2=Republican. Gender is coded 1=male and 2=female. Age is coded 1=17-39, 2=40-59, 3=60+.

Variables in the Equation
		B	S.E.	Wald	df	Sig.	Exp(B)
Step 1a	pid3_01	2.600	.079	1074.482	1	.000	13.467
	Gender	-.068	.095	.514	1	.473	.934
	Age group	.369	.061	36.148	1	.000	1.446
	Constant	-3.226	.214	228.073	1	.000	.040
a. Variable(s) entered on step 1: pid3_01, Gender, Age group.

True or False (based on the model above... just write a T of F in the blanks):

_________44. Each one unit increase in party corresponds to a 2.6% increase in the probability of voting for Romney.

_________45. The effect of gender is statistically significant.

_________46. The effect of age is statistically significant.

_________47. The Nagelkerke R-squared for this model is .605. This means that overall our model does a moderately good job of explaining the variation in presidential vote.

Answer 1

False 44. Each one unit increase in party corresponds to a 2.6% increase in the probability of voting for Romney.

(correct interpretation is: For one unit increase in party, we expect to see about exp(2.6)=13.46% increase in the odds of voting for Romney).

False 45. The effect of gender is statistically significant (since p-value=0.473>0.05)

True 46. The effect of age is statistically significant (since p-value=0.000<0.05)

True 47. The Nagelkerke R-squared for this model is .605. This means that overall our model does a moderately good job of explaining the variation in presidential vote