Autocorrelation refers to the correlation of a variable with its past values, and when it appears in the residuals of a regression model, it suggests that the model is missing some important variable or structure.
- Validity of the model is questioned because autocorrelation indicates that the model does not capture all the patterns in the data.
- The reliability of the model might also be affected, but autocorrelation primarily impacts the validity
.- Statistical significance can be misleading when autocorrelation is present, but it’s the validity that is directly challenged