Back to quiz

6. What standardised residual is worth checking as an outlier?

  • They're all worth checking, you can never be too sure
  • >2
  • >3
  • >1

7. What's a type 2 error

  • Clara
  • False positive
  • False negative
  • Elisa

8. What should you be careful of with hierarchical multiple regression

  • It is not as good as backwards multiple regression
  • Order of entry of predictors may affect the model
  • It only works if a large number of predictors is entered
  • I can't think of a 4th answer

9. What is an OK range for kurtosis and skew

  • Less than plus or minus 1
  • Less than plus or minus 2
  • As close to 2 as possible
  • Less than 1

10. Simpson's paradox, When a trend appears in several groups of data but disappears when they are combined, can be caused by

  • The wrong analysis being performed
  • Not having a control
  • confounding variables
  • Small sample size

11. PLus or minus 3.3 standard errors is equal to what percentage CI?

  • 99.9
  • 99.3
  • 99
  • 95

12. If ...% of residuals are >2 the model might be a poor fit

  • 20%
  • 5%
  • 1%
  • 10%

13. How many standard errors cover 99% CI?

  • plus or minus 2.6
  • plus or minus 3.3
  • plus or minus 3
  • plus or minus 1

14. What can't you predict wth multiple linear regression

  • All of these
  • Whether Maddie McCann will ever be found
  • Whether I will pass my stats exam
  • Categorical things such as alive vs dead

15. What is the problem with forwards and backwards variable selection?

  • They both lead to false negatives, though forwards is a bit better
  • They both lead to false positives, though forwards is a bit better
  • They both lead to false positives, though backwards is a bit better
  • They both lead to false negatives, though backwards is a bit better

16. If SEMs for two groups DO NOT overlap what can you conclude about p?

  • It could be anything
  • p is a man-made construct and is as futile as our existence
  • It has to be smaller than 0.05
  • It has to be bigger than 0.05

17. Which of these statements regarding colinearity is false?

  • Tolerance factor of more than 0.01 is problematic
  • When there is colinearity, very little variance is explained by one factor alone
  • As colinearity increases so does the standard error of b coefficients
  • VIF above 10 or average VIF above 1 helps you spot colinearity

18. What is the likelihood that the population mean lies within plus or minus 1 standard error

  • 175%
  • 68%
  • Needs more information to be answered
  • 95%

19. Model 1 has an R2 of 0.32 and model 2 has an R2 of 0.54, how much more of the variance can model 2 explain

  • 22%
  • Can't tell
  • 64%
  • 200%