Question

You are a program manager for a welfare agency. You want to consider variables that migh affect the length of unemployment. The table below lists categories of different factors that are hypothesized to be the most important in affecting the length of unemployment. Which are statistically significant? Is the overall model a strong one? Interpret and write up the results.
Multiple Regression Output
Model
R	R-square	Adjusted R2	SEE
0.66	0.435	0.423	0.092
Dependent variable: Unemployment duration
Coefficients
Model	b	Std. Err.	t	P>|t|
Constant	0.231	0.03	7.740	0.000
Receives job training	-0.010	0.04	-2.579	0.010
Marital status	-0.072	0.017	-4.125	0.000
Medical condition	0.013	0.005	2.540	0.012
Number of dependents	0.000	0.001	0.252	0.802
Education	-0.003	0.003	-0.834	0.405
Note: SEE = standard error of the estimate; Std. Err. = standard error; P>|t|= significance
Marital status: 1 = married; 0 = not married

Answer 1

H0: βi = 0

H1: βi ≠ 0

Since p-values for Receives job training, Marital status and Medical condition is less than 0.05, we reject the null hypothesis and conclude that β1 ≠ 0, β2 ≠ 0 and β3 ≠ 0.

Since p-values for Number of dependents and Education is more than 0.05, we do not reject the null hypothesis and conclude that β4 = 0 and β5 = 0.

So, Receives job training, Marital status and Medical condition are statistically significant. Number of dependents and Education are not statistically significant.

Adjusted R2 = 0.423

This means that the 5 factors explain only 42.3% of variation in length of unemployment. So, the model is not a strong one.