RESEARCH · METHODS · EVIDENCE
Interpret a regression model
Browse statistics lessons
- Statistics for research
- Center and spread
- Distributions and standard errors
- Statistical inference
- P-values and interpretation
- Critical values and test direction
- Compare means with t-tests
- Read t-test output
- Compare several means with ANOVA
- Read an ANOVA table
- Analyze categorical counts
- Read chi-square output
- Describe an association
- Read correlation results
- Interpret a regression model
- Choose a statistical method
- Statistics resources
Regression relates an outcome to specified predictors. A coefficient represents the modeled change in outcome per unit of a predictor, conditional on the other included terms; it is not automatically a causal effect.
Check functional form, residuals, independence, influential observations, and the uncertainty relevant to the model. Standard linear-regression inference commonly assumes constant residual variance and an appropriate error model. Avoid predictions far beyond the data used to fit the model.
References: NIST model assumptions