1 a This is the sqrt(n) factor in the denominator of the standard error.
2 d This is the Empirical Rule in the regression context. 3 b See Class 13 notes , slide 7.1. 4 a A 99.7% CI would have width +/- 3 std.err, whereas a 95% CI would have width +/-2 std.err. 5 e Definitional. 6 d 10 * 0.2 + 20 * 0.5 + 30 * 0.3 21 7 c Note the tracking in the residual plot. 8 a Definitional. 9 e Collinearity is correlation between the x-variables. It says nothing about the distribution of the x-variable. 10 b Log transforms reduce right skewness of positively valued data. 11 a Simply parse the output. 12 a The t-stat is (3.946 - 3.5)/0.1459 = 3.05. The p-value must be less than 0.003. 3.056888 13 a Both residual plots look good. 14 b Find the 95% CI for the slope, then multiply both ends by 5. 111.916 227.674 15 e Find the 95% CI for the slope and see if 35 is in there. (It is). 22.3832 45.5348 16 c Multiply the slope by 5 and interpret. 169.7975 17 c The log-log model has percent change in x to percent change in y, interpretations for the slope. 18 b Create a prediction interval on the log scale, then exponentiate both ends. 15480.32 99454.2 19 e You can't spot leverage from a multiple regression residual plot. 20 c West. They are the reference group. Both other regions have negative coefficients, so smaller predicted values. 21 d The partial-F test in the Effects Tests table answers this question. 22 a Definitional. Answer c would require a three-way interaction. 23 c The interaction term isn't significant. And we would never remove Region and keep the interaction. 24 c Use the marginal probability to back out W, then the fact that probabilities add to 1 for Z. 25 b Identify the two-simple events that make up this more complex event: 0.2 + 0.1. 26 d This is a conditional probability: 0.1/0.2.