BCPS- Non-clinical Exam Questions With
Complete Answers
Types of hypothesis testing & statistical methods - ANSWER Difference > traditional
2-sided t-test; CI
Equivalence > Two 1-sided t-test (TOST); CI
Superiority > traditional 1-sided t-test; CI
Noninferiority > CI
Parametric tests - ANSWER student t-tests
ANOVA
ANCOVA
Student t-tests - ANSWER *student t-tests*
- one sample test: compares mean of study sample with the population mean
- two sample, independent samples, or unpaired test: compares means of two
independent samples
--> equal variance test
- paired test: compares mean difference of paired/matched samples (this is a related
samples test)
*common error = use of multiple t-tests with more than 2 groups; should use ANOVA*
,ANOVA - ANSWER Analysis of variance
t-test that can apply to >2 groups
one-way ANOVA: compares ONE factor between 3 groups
two-way ANOVA: compares an additional factor
(i.e. splits groups 1, 2, and 3 into "young groups" and "old groups")
repeated measures ANOVA: a related samples test; 1 group, multiple measurements
ANCOVA - ANSWER analysis of COvariance
helps explain influence of categorical (independent) variable on continuous variable
(dependent variable) while statistically controlling for confounding variables
Nonparametric tests - ANSWER *compare 2 independent samples* -similar to t-test
- Wilcoxon rank sum
- Mann-whitney U
- Wilcoxon Mann-whitney
- Kruskal-Wallis one-way ANOVA by ranks (3+ independent groups)
*related/paired samples* -similar to paired t-test
- Sign test
- Wilcoxon signed-rank test
- Friedman ANOVA by ranks (for 3+ paired/matched groups)
, Nominal data tests - ANSWER - chi square (test of independence and goodness of fit)
--> compares expected and observed proportions between 2 or more groups
- fisher exact test: chi square but for cells containing < 5 predicted observations
- McNemar: paired samples:
- Mantel-Haenszel: controls for influence of confounders
Chart- nominal - ANSWER - 2 groups (independent): chi square or fisher exact test
- 2 groups (related): mcNemar
- > 2 groups (independent): chi square
- > 2 groups (related): Cochran Q
Chart- ordinal - ANSWER - 2 groups (independent): Wilcoxon, Mann Whitney, etc
- 2 groups (related): wilcoxon signed rank, sign test
- > 2 groups (indep.): Kruskal-Wallis
- > 2 groups (related): Friedman. ANOVA
Chart- continuous, no factors - ANSWER - 2 groups (independent): equal variance t-test,
unequal variance test
- 2 groups (related): paired t-test
> 2 groups (independent): One way ANOVA
> 2 groups (related): repeated measures ANOVA
Chart, continuous, 1 factor - ANSWER 2 groups (independent): ANCOVA
Complete Answers
Types of hypothesis testing & statistical methods - ANSWER Difference > traditional
2-sided t-test; CI
Equivalence > Two 1-sided t-test (TOST); CI
Superiority > traditional 1-sided t-test; CI
Noninferiority > CI
Parametric tests - ANSWER student t-tests
ANOVA
ANCOVA
Student t-tests - ANSWER *student t-tests*
- one sample test: compares mean of study sample with the population mean
- two sample, independent samples, or unpaired test: compares means of two
independent samples
--> equal variance test
- paired test: compares mean difference of paired/matched samples (this is a related
samples test)
*common error = use of multiple t-tests with more than 2 groups; should use ANOVA*
,ANOVA - ANSWER Analysis of variance
t-test that can apply to >2 groups
one-way ANOVA: compares ONE factor between 3 groups
two-way ANOVA: compares an additional factor
(i.e. splits groups 1, 2, and 3 into "young groups" and "old groups")
repeated measures ANOVA: a related samples test; 1 group, multiple measurements
ANCOVA - ANSWER analysis of COvariance
helps explain influence of categorical (independent) variable on continuous variable
(dependent variable) while statistically controlling for confounding variables
Nonparametric tests - ANSWER *compare 2 independent samples* -similar to t-test
- Wilcoxon rank sum
- Mann-whitney U
- Wilcoxon Mann-whitney
- Kruskal-Wallis one-way ANOVA by ranks (3+ independent groups)
*related/paired samples* -similar to paired t-test
- Sign test
- Wilcoxon signed-rank test
- Friedman ANOVA by ranks (for 3+ paired/matched groups)
, Nominal data tests - ANSWER - chi square (test of independence and goodness of fit)
--> compares expected and observed proportions between 2 or more groups
- fisher exact test: chi square but for cells containing < 5 predicted observations
- McNemar: paired samples:
- Mantel-Haenszel: controls for influence of confounders
Chart- nominal - ANSWER - 2 groups (independent): chi square or fisher exact test
- 2 groups (related): mcNemar
- > 2 groups (independent): chi square
- > 2 groups (related): Cochran Q
Chart- ordinal - ANSWER - 2 groups (independent): Wilcoxon, Mann Whitney, etc
- 2 groups (related): wilcoxon signed rank, sign test
- > 2 groups (indep.): Kruskal-Wallis
- > 2 groups (related): Friedman. ANOVA
Chart- continuous, no factors - ANSWER - 2 groups (independent): equal variance t-test,
unequal variance test
- 2 groups (related): paired t-test
> 2 groups (independent): One way ANOVA
> 2 groups (related): repeated measures ANOVA
Chart, continuous, 1 factor - ANSWER 2 groups (independent): ANCOVA
