Exam 2 - Handout 6b - Bivariate Analysis & Comparing Groups - Practice Questions Flashcards
Which statistical test is used to compare means when there are more than two groups?
ANOVA
What is the null hypothesis for a paired t-test comparing A1C at baseline and post-treatment
H0 = MeanB - MeanPT = 0
Which nonparametric test is used for ordinal or heavily skewed continuous data when comparing two independent groups?
Wilcoxon-Mann-Whitney test
What does the Pearson correlation coefficient (r) measure?
The strength and direction of the liner relationship between two continuous variables
Which test is appropriate for comparing the proportion of patients at A1C goal between two groups
Chi-square test of homogeneity
What is the two-step analysis process used in ANOVA when comparing means for more than two groups?
- Calculate F-statistic and p-value to determine if any differences exist among group means
- If p < 0.05, conduct post-hoc tests (eg. tukey’s/ HSD, Scheffe’s test) to determine which groups differ
How does the sign test differ from other statistical tests in terms of data requirements and what is measures?
The sign test does NOT require continuous data, only the direction of change
It tests the null hypothesis that the number of patients w/ decreased values is equal to the number w/ increased values
In the statistical test section flowchart, what are the 3 key factors to consider when choosing an appropriate test?
- Independence of dependence of the comparison groups
- Level of measurement of the dependent variable
- Assumptions of parametric tests vs nonparametric tests
What is the Spearman Rank Correlation Coefficient (rs) used for? What does it measure?
Used for ordinal or skewed data and measures the degree of correspondence between the ranks of two variable
What is the main difference between parametric tests in terms of their assumptions about the data?
Parametric tests (eg. t-test, ANOVA) make assumptions about the population distribution
NONparametric tests (eg. Wilcoxon, kruskal-wallis) have fever assumptions about the underlying distribution of the data