1/15
Looks like no tags are added yet.
four questions to ask yourself before choosing a statistical test for data analysis:
what type of data do I have?
how many samples or groups are being compared?
what is the purpose of the analysis?
what assumptions can I make about the data? (distribution, sample size)
what does the correct test depend on?
type of data
study design
research question
assumptions made
when is a parametric test used?
when data is numerical and normally distributed
when is a nonparametric test used?
when data is categorical, true distribution is unknown
what are examples of parametric tests?
testing for a mean
determining the difference between independent / dependent means
correlation and regression
what are examples of nonparametric tests?
testing for a proportion
determining the difference between two proportions
chi-square test of independence
what does it mean to have two independent groups?
you’re comparing two separate groups of individuals
two proportions or two means
what does it mean to have one sample vs. a population or hypothesized value?
you have one sample and want to determine whether it differs from a known or expected value
what does it mean to have paired or related measurements?
you have two sets of measurements from the same individuals or matched units
ex. before or after, relationship between two variables
when do you use a test for proportion?
when one categorical outcome is compared with an expected proportion
when do you use a test of two proportions?
comparing proportions between two groups
when do you use a chi-square test of independence?
when determining whether two categorical variables are associated
when do you test for a mean?
when comparing one sample mean with a hypothesized mean
when do you use two independent means test?
comparing the means of two separate groups
when you use two paired means test?
comparing two related measurements
comparing blood pressure before and after treatment
when do you use a correlation and regression test?
when investigating a relationship between numerical variables