Biostats Final (Statistical Tests)

One sample t-test (Assumptions)

Variable is normally distributed

Sample is random

One sample t-test (Definition)

Compare mean to a constant

Parametric and non-exact

Two-tailed, right-tailed, or left-tailed

Paired t-test (Assumptions)

Paired differences are normally distributed

Sample is random

Paired t-test (Definition)

Compare means of paired groups

Parametric and non-exact

Two-tailed, right-tailed, or left-tailed

Two-sample t-test (Assumptions)

Both groups are normally distributed

Both groups are homoscedastic

Both samples are random

Two-sample t-test (Definition)

Compare means of 2 unpaired groups

Parametric and non-exact

Two-tailed, right-tailed, or left-tailed

Welch’s Approximate t-test (Assumptions)

Both groups are normally distributed

Both samples are random

Welch’s Approximate t-test (Definition)

Compares means of 2 unpaired groups

Parametric and non-exact

Two-tailed, right-tailed, or left-tailed

F-test (Assumptions)

Both samples are normally distributed

Both samples are random

F-test (Definition)

Compares variances of 2 groups

Parametric and non-exact

Two-tailed, right-tailed, or left-tailed

Levene Test (Assumptions)

All samples are approximately normally distributed

All samples are random

Levene Test (Definition)

Compares variances of 2 or more groups

Parametric and non-exact

Omnibus test

Always right-tailed

Shapiro-Wilk Test (Definition)

Compares sample distribution to normal distribution

Sign Test (Assumptions)

Dichotomous outcomes

Trials are independent and have the same probability of success

Samples (or the paired differences) are random

Sign Test (Definition)

Compares median to a constant

Non-parametric and exact

Non-parametric version of one-sample or paired t-test

Two-tailed, right-tailed, or left-tailed

Mann-Whitney U Test (Assumptions)

Both populations have the same distribution shape

Both samples are random

Mann-Whitney U Test (Definition)

Compares medians of 2 unpaired groups

Non-parametric and non-exact

Non-parametric version of two-sample t-test

Two-tailed, right-tailed, or left-tailed

One-way ANOVA (Assumptions)

Residuals are independent, normally distributed, and homoscedastic

One-way ANOVA (Definition)

Compares means of a numerical response variable across 2 or more levels of a factor

Parametric and non-exact

Omnibus test

Always right-tailed

Kruskal-Wallis Test (Assumptions)

Data in each group have the same distribution shape

Samples are random

Kruskal-Wallis Test (Definition)

Compare medians of a numerical response variable across 2 or more levels of a categorical explanatory variable

Non-parametric and non-exact

Non-parametric version of ANOVA

Omnibus test

Always right-tailed

Multi-Way ANOVA (Assumptions)

Residuals are independent, normally distributed, and homoscedastic

Multi

-Way ANOVA (Definition)

Compare means of a numerical response variable across 2 or more level of 2 or more factor

Parametric and non-exact

Omnibus test

Always right-tailed

Tukey-Kramer HSD (Assumptions)

Both groups are normally distributed

Both groups are homoscedastic

Both samples are random

Tukey-Kramer HSD (Definition)

Pairwise comparison of group means

Parametric and non-exact

Always right-tailed

Pearson Product-Moment Correlation Coefficient (Assumptions)

X and Y are linearly related

Bivariate normality

Both samples are random

Pearson Product-Moment Correlation Coefficient (Definition)

Linear relationship between variables

Parametric and non-exact

Two-tailed, right-tailed, or left-tailed

Spearman Rank Correlation (Assumptions)

X and Y are monotonically related

Ranks X are normally distributed with equal variance for ranks Y

Ranks Y are normally distributed with equal variance for ranks X

Both samples are random

Spearman Rank Correlation (Definition)

Monotonic relationship between variables

Non-parametric and non-exact

Non-parametric version of Pearson Product-Moment Correlation Coefficient

Two-tailed, right-tailed, or left-tailed

Simple Linear Regression (Assumptions)

X and Y are linearly related

Residuals are normally distributed, homoscedastic, and independent

Simple Linear Regression (Definition)

Finds the best linear relationship between a numerical response variable and a numerical explanatory variable

Parametric and non-exact

Two-tailed by default

Multiple Linear Regression (Assuptions)

X_k and Y are linearly related

Residuals are normally distributed, homoscedastic, and independent

The explanatory variables are not colinear

Multiple Linear Regression (Definition)

Finds the best linear relationship between a numerical response variable and 2 or more numerical explanatory variables

Parametric and non-exact

Two-tailed by default

ANCOVA (Assumptions)

X_i and Y are linearly related

Residuals are normally distributed, homoscedastic, and independent

Explanatory variables are not colinear

ANCOVA (Definition)

Relates a single response variable to a set of explanatory variables, at least one factor and one covariate

Parametric and non-exact

Two-tailed or right tailed

General Linear Model (Assumptions)

X_i and Y are linearly related

Residuals are normally distributed, homoscedastic, and independent

Explanatory variables are not colinear

General Linear Model (Definition)

A statistical framework used to relate a single response variable to one or more categorical and numerical explanatory variables

Parametric and non-exact

Always right-tailed⁣