Commonly Used Statistical Tests
1-sample t-test
determines whether the average of a set of observations is equal to the standard value of the reference population.
t-statistic
difference between group means divided by variability within groups
paired t-test
Compares 2 responses or outcome measures in the same person. Each subject is his or her own control.
r value
Correlation coefficient which determines if 2 variables are associated. Affected by sample size, measurement error, and types of variables used
regression
Examines whether 1 variable causes another to occur. Types include linear, multiple, weighted, and logistic
Pearson's correlation
Examines whether 2 variables are associated.
one-way ANOVA
Parametric test that compares 3 or more groups. Null hypothesis claims that all groups are equal while alternative hypothesis claims that at least 2 means differ significantly. Common in biomedical literature
t-tests
Parametric tests which compare the average response to a drug between 2 distinct groups. Most common types are 1-sample, 2 sample, and paired.
chi square test
Tests baseline characteristics of groups to ensure there are no underlying compounding factors between experimental groups. Can be used with nominal or ordinal data Most frequently reported test in medical literature.
2-sample t-test (independent-samples, independent-groups, or unpaired)
Used to compare the responses or outcome measures in 2 independent groups. Examples: response of active drug compared with control, placebo, or standard therapy; response of 2 different groups (men vs. women, young vs. old). Can be 1 tailed or 2 tailed. Very common in biomedical literature.
linear regression
Y=mX+b (dependent variable = slope of the relationship x independent variable). Shows whether the independent variable can predict the dependent variable.
variance homoscedasticity
type of 2-sample t-test which assumes the variances of the 2 groups are equal
variance heteroscedasticity
type of 2-sample t-test which assumes the variances of the 2 groups are not equal