Non-Parametric Group Comparisons Exam 2

Question 1

Parametric statistics

Answer 1

• Assume data follow a normal distribution
• Make inferences about data parameters
• Used when you have continuous data: interval, ratio

Question 2

Nonparametric statistics

Answer 2

• Assume data do not follow a normal distribution
• Do not make inferences about data parameters
• Used when you have discrete variable data: nominal or ordinal (includes dichotomous)
• Can also be used for something that starts out as a continuous variable but do not meet parametric assumptions (Data that are not normally distributed, or have heterogeneous variance despite being interval or ratio scale)

Question 3

Advantages of Parametric statistics

Answer 3

• More accurate and precise estimations

- More statistical power (less Type II errors, less false negatives)

Question 4

Disadvantages of Parametric statistics

Answer 4

• Not very robust (doesn’t perform well if assumptions are violated)
• can use nonparametric methods if assumptions are violated

Question 5

Advantages of Nonparametric statistics

Answer 5

• Simpler

- More robust (resistant to outliers, heterogeneity of variance)

Question 6

Disadvantages of Nonarametric statistics

Answer 6

• Less statistical power (uses less information for calculations)
• Not well suited for numeric interpretation

Question 7

Which tests are Parametric?

Answer 7

• Paired t-test
• t-test
• ANOVA

Question 8

Which tests are Nonparametric?

Answer 8

• McNemar
• Chi-square
• Fisher exact
• Sign test
• Wilcoxon signed rank test
• Wilcoxon-Mann-Whitney test
• Kruskal-Wallis test

Question 9

When would you use Sign test?

Answer 9

• Used when you have two dependent groups, have ordinal dependent variable or outcome
• Test for direction of change
• Number with change is compared to number with opposite direction of, or no, change

Question 10

When would you use Wilcoxon signed rank test?

Answer 10

• Used when you have two dependent groups, have ordinal dependent variable or outcome
• Test for direction and magnitude of change
• Compare rankings among two dependent groups
• Can be directional or nondirectional
• More powerful than sign test
• More sensitive than paired t-test when n < 50

Question 11

When would you use Wilcoxon-Mann-Whitney test?

Answer 11

Used when you have two independent groups, have ordinal dependent variable or outcome

Question 12

When would you use Kruskal-Wallis test?

Answer 12

• Used when you have more than two independent groups, have ordinal dependent variable or outcome
• Uses mean ranks instead of sums of ranks

Question 13

Null hypothesis of Sign test

Answer 13

H0 -> the amount of people with negative data will be equal to those that have positive data or data that stays the same

Question 14

Null hypothesis of Wilcoxon signed rank test

Answer 14

H0 -> no difference in group medians

Question 15

Null hypothesis of Wilcoxon-Mann-Whitney test

Answer 15

H0 -> no difference in the distribution of the scores in each group (uses sum of ranks, can be used to test whether two independent groups have equal medians)

Question 16

Null hypothesis of Kruskal-Wallis test

Answer 16

H0 -> all the mean ranks equal one another

Question 17

Parametric equivalent of Sign test

Answer 17

Paired t test

Question 18

Parametric equivalent of Wilcoxon signed rank test

Answer 18

Paired t test

Question 19

Parametric equivalent of Wilcoxon-Mann-Whitney test

Answer 19

Independent groups t-test

Question 20

Parametric equivalent of Kruskal-Wallis test

Answer 20

ANOVA