PALPSYU3349 Week 7

Which if the following statements about type I error is incorrect?

• The more hypothesis tests you run, the greater the probability of making a type I error.

• You will be making a type I error if you fail to reject the null hypothesis when it is actually false.

• Bonferroni adjustments can be used to minimise the probability of making a type I error.

• You will be making a type I error if you reject the null hypothesis when it is actually true.

The Hierarchical regression approach to model selection can be performed using which type of analysis?

• Simple linear regression

• Pearsons correlations

• Sequential regression (nestreg)

• Bootstrapping

If you have ordinal IVs and want to perform a multiple regression analysis, what must you do?

• Upgrade the ordinal IVs to numeric or downgrade the ordinal IVs to categorical.

• You have to centre the ordinal IV first.

• You will have to change your analysis (you cannot run a multiple regression analysis).

Which of the following statements about Hierarchical regression is incorrect?

• It can be used to determine whether a group of IVs predict variance in the dv over and above the other IVs.

• It can be used to determine the % of unique variance explained by individual predictors.

• It allows you to group similar predictors together (during analysis).

• It uses backwards elimination to find the most parsimonious model.

Which of the following is not one of the three main approaches to model selection?

• Backwards stepwise

• Part correlations

• Best subsets

• Forwards stepwise

To find % unique variance explained (AKA % variance fitted last) which two analyses can you perform?

• Sequential regression or backwards elimination (via bootstrapping)

• Pearsons correlations or one-way ANOVA

• Part/semipartial correlations or sequential regression

• Spearman's correlations or part/semipartial correlations

When doing backwards elimination via bootstrapping, what is the general rule when deciding which predictors to keep in your model?

• Only keep predictors that are retained at least 50% of the time.

• Only keep predictors that are retained 80-100% of the time.

• Only keep predictors that are retained more than 25% of the time.

• There is no general rule.