PSCH 443 Multiple Regression 3 Evaluating Assumptions Part 2

Normality

The assumption that predictor and outcome variables should be normally distributed in statistical analyses.

Linearity

The assumption that the relationship between predictor and outcome variables is linear in regression analyses.

Homoscedasticity

The assumption that the variance of residual errors is consistent across all levels of the predicted values.

Multivariate normality

The assumption that combinations of multiple variables should have a joint normal distribution.

Residuals

The differences between predicted values and observed values in a regression model.

Statistical significance

A determination that a result is unlikely to have occurred due to random chance, often assessed using p-values.

Suppressor effect

A phenomenon where the inclusion of a predictor variable changes the relationship between another predictor and the outcome variable, often enhancing the model's fit.

Durbin-Watson statistic

A statistic that tests for autocorrelation in the residuals of a regression analysis, with a value around 2 indicating no autocorrelation.

Autocorrelation

A situation where residuals or errors are correlated with each other over time or across observations.

Heteroskedasticity

A condition in regression analysis where the variance of errors is not constant across all levels of the independent variable.

Correlation matrix

A table showing the correlation coefficients between a set of variables, helping to identify relationships before regression analysis.

Curvilinear relationship

A type of relationship between variables that is not linear, where the association between variables changes at different levels.

Errors in prediction

The discrepancies between predicted and actual values in a regression model.

Independent errors

An assumption that the errors of prediction are not correlated with each other.

Bivariate correlation

The correlation between two variables, providing insight into their relationship prior to more complex analyses.

Outliers

Data points that differ significantly from other observations, potentially impacting model assumptions and results.

Phantom data

Fictitious or manipulated data that may not accurately represent real-world observations, potentially leading to incorrect conclusions.