Someone hypothetically comes to you with some regression results that are confusing him:
$\begin{array}{lcc}
\text { Independent Variable: } & \text { Model 1 } & \text { Model } 2 \
\text { Education (in years) } & 3.47^{* * *} & 3.47 \
\text { Uses a computer at work }(Y=1, N=0) & -- & 7.80 \
\text { Constant } & -20.45 & -24.20 \
\text { R-Squared } & .14 & .14 \
n & 1000 & 100
\end{array}$ He is confused by what happens to the effect of education between Model 1 and Model 2: the value of the slope is the same, but it is no longer statistically significant.
What is the most likely explanation for this?
A) It is not that education has an effect, but that people with more education are more likely to use a computer at work.
B) The value for R-squared stayed the same, so having two independent variables is no better than having one independent variable.
C) The constant in Model 2 decreased by a value greater than the slope for education.
D) The sample size has decreased.

Question

Quizplus · Accepted Answer

The most likely explanation for the change in statistical significance of the education variable between Model 1 and Model 2, while the slope remains the same, is the decrease in sample size from 1000 to 100. A smaller sample size reduces the statistical power of the test, making it harder to detect a statistically significant effect.

Someone Hypothetically Comes to You with Some Regression Results That