PN 3002 EXAM 6 2026 TEST SCRIPT
QUESTIONS AND SOLUTIONS GRADED A+
◉ Mann Whitney Test info.
Answer: nonparametric, outcome variable (ordinal, interval, ratio)
predictor variable (categorical), 2 groups, independent design
◉ Wilcoxon Signed Rank Test info.
Answer: nonparametric, outcome variable (ordinal, interval, ratio)
predictor variable (categorical), 2 groups, repeated design
◉ Kruskall Wallis Test info.
Answer: nonparametric, outcome variable (ordinal, interval, ratio)
predictor variable (categorical), > 2 groups, independent design
◉ Friedman's ANOVA info.
Answer: nonparametric, outcome variable (ordinal, interval, ratio)
predictor variable (categorical), > 2 groups, repeated design
◉ Pearson's chi-square test info.
Answer: nonparametric, outcome variable (none or categorical), 2 or
more groups, independent design
,◉ Independent T test info.
Answer: parametric, outcome variable (interval or ratio) predictor
variable (categorical), 2 groups, independent design
◉ Paired samples t test info.
Answer: parametric, outcome variable (interval or ratio) predictor
variable (categorical), 2 groups, repeated design
◉ One way ANOVA info.
Answer: parametric, outcome variable (interval or ratio) predictor
variable (categorical), 2 or more groups, independent design
◉ fully independent factorial ANOVA info.
Answer: parametric, outcome variable (interval or ratio) predictor
variable (over 1 categorical), 4 or more groups, independent design
◉ Nonparametric Statistics
For independent groups design:.
Answer: - Mann-Whitney U & Wilcoxon rank-sum test (2 groups)
- Kruskal-Wallis Test (more than 2 groups)
◉ Nonparametric Statistics
, For repeated measures designs:.
Answer: Wilcoxon signed-rank test (2 groups)
Friedman's ANOVA (more than 2 groups)
◉ Nonparametric tests do not assume.
Answer: linearity, normality or homogeneity of
variance/homoscedasticity (always assume independence)
◉ Disadvantages of Nonparametric Stats.
Answer: (1) Relatively few non-parametric tests are available
(2) Nonparametric tests are less powerful
- Reasons: start with conversion of scores (interval/ratio scale) to
ranks (ordinal scale) which greatly reduces the information content
of the data, so you need larger sample sizes/effect sizes for a
significant result
◉ Bootstrapping allows parametric testing without the assumption
of normality. However....
Answer: 1- Not all analyses can be bootstrapped
2- Bootstrapping does not solve problems associated with violating
the other analysis assumptions of parametric tests (other than
normality)
3- If you have outliers and a small sample size, bootstrapping may
not be the best approach
QUESTIONS AND SOLUTIONS GRADED A+
◉ Mann Whitney Test info.
Answer: nonparametric, outcome variable (ordinal, interval, ratio)
predictor variable (categorical), 2 groups, independent design
◉ Wilcoxon Signed Rank Test info.
Answer: nonparametric, outcome variable (ordinal, interval, ratio)
predictor variable (categorical), 2 groups, repeated design
◉ Kruskall Wallis Test info.
Answer: nonparametric, outcome variable (ordinal, interval, ratio)
predictor variable (categorical), > 2 groups, independent design
◉ Friedman's ANOVA info.
Answer: nonparametric, outcome variable (ordinal, interval, ratio)
predictor variable (categorical), > 2 groups, repeated design
◉ Pearson's chi-square test info.
Answer: nonparametric, outcome variable (none or categorical), 2 or
more groups, independent design
,◉ Independent T test info.
Answer: parametric, outcome variable (interval or ratio) predictor
variable (categorical), 2 groups, independent design
◉ Paired samples t test info.
Answer: parametric, outcome variable (interval or ratio) predictor
variable (categorical), 2 groups, repeated design
◉ One way ANOVA info.
Answer: parametric, outcome variable (interval or ratio) predictor
variable (categorical), 2 or more groups, independent design
◉ fully independent factorial ANOVA info.
Answer: parametric, outcome variable (interval or ratio) predictor
variable (over 1 categorical), 4 or more groups, independent design
◉ Nonparametric Statistics
For independent groups design:.
Answer: - Mann-Whitney U & Wilcoxon rank-sum test (2 groups)
- Kruskal-Wallis Test (more than 2 groups)
◉ Nonparametric Statistics
, For repeated measures designs:.
Answer: Wilcoxon signed-rank test (2 groups)
Friedman's ANOVA (more than 2 groups)
◉ Nonparametric tests do not assume.
Answer: linearity, normality or homogeneity of
variance/homoscedasticity (always assume independence)
◉ Disadvantages of Nonparametric Stats.
Answer: (1) Relatively few non-parametric tests are available
(2) Nonparametric tests are less powerful
- Reasons: start with conversion of scores (interval/ratio scale) to
ranks (ordinal scale) which greatly reduces the information content
of the data, so you need larger sample sizes/effect sizes for a
significant result
◉ Bootstrapping allows parametric testing without the assumption
of normality. However....
Answer: 1- Not all analyses can be bootstrapped
2- Bootstrapping does not solve problems associated with violating
the other analysis assumptions of parametric tests (other than
normality)
3- If you have outliers and a small sample size, bootstrapping may
not be the best approach
