Psyc212/667 Practice Problems
Psyc212/667 Practice Problems

13. Correlation and regression

Find the correlation and the regression line for the following x,y pairs:

3,3
3,3
4,6
5,7
6,9
7,8

What is the predicted y for an x of 6?

14. Procedures using Fisher's r to Z'

A researcher finds that the correlation between age and number of strangers approached is -.40 in the 20 cats he examines (i.e., as cats' age increased their approach of strangers decreased). In the 40 dogs he examines, he finds a correlation of +.50. Is the age-"social anxiety" relation different across species?

What if he wanted to created a 99% confidence interval around the population correlation (rho) for the dogs? How would that be computed?

15. Significance of regression coefficients
Modelunstandardized coeffstd errorstandardized coefficient
intercept2.018
DispPrej.337.259.199
Training.381.089.651
An applied researcher was interested in predicting discriminatory hiring recommendations from dispositional prejudice (DispPrej) and diversity training (Training) among 10~0 middle-managers. Show how you would test the contribution of each predictor.

16. Significance of R2 and change in R2 The researcher is interested in examining the effects of a Kaplan-type computer aid on SAT-Q performance, and recognizes that math ability could add noise to the prediction. There are 24 high school seniors in his sample for this pilot study. He enters standardized math scores on the Iowa test first to predict SAT-Q; R=.166. In the second step, he enters the number of minutes of computer time with the aid used by each student. R=.695.
Test the significance of R at each step, then test whether the change in R2 is sufficient to conclude that instruction predicts SAT-Q, above and beyond general math ability.

17. Chi-square goodness of fit

A psychologist studying reaction times found that 9 of his 100 participants produced reaction times that were 4 standard deviations above the mean, and wished to justify excluding them from his analyses. Of the 9, 2 were from the sad mood condition, 1 from the happy mood condition, and the remainder from the angry mood condition. Assuming that producing an extreme score should be equally probable across conditions, is there any cause for concern about the distribution of the current extreme scores?