User Guide: Regression Doctor AI

Welcome to the Regression Doctor AI. Multiple Linear Regression is one of the most powerful tools in your statistical arsenal, allowing you to see how multiple Independent Variables (IVs) predict a single Dependent Variable (DV). This tool ensures you haven't violated strict statistical assumptions before drawing your conclusions.

1. The Regression Assumptions (L.I.N.E.M)

Before you can trust a regression model, your data must pass a "health check":

Linearity: A linear relationship exists between the IVs and the DV.
Independence of Errors (Autocorrelation): Residuals are independent. We check this with the Durbin-Watson statistic. It must be between 1.5 and 2.5.
Normality: Errors are normally distributed.
Equal Variance (Homoscedasticity): The variance of errors is constant.
Multicollinearity: Your IVs shouldn't be too similar to each other. We check this using VIF. If VIF is > 10, the assumption fails.

2. R² vs. Adjusted R²

R Square (R²): Tells you the percentage of variance in your DV that is explained by your IVs. For example, R² = .45 means your model explains 45% of the outcome.

Adjusted R²: Every time you add a new variable to a model, R² artificially goes up, even if the variable is useless. Adjusted R² mathematically penalizes you for adding useless variables. Always report Adjusted R² when you have more than one IV.

3. Reading the Coefficients Table

The Coefficients table tells you which specific variables actually matter in your model:

Sig. (p-value): If p < .05, the variable is a significant predictor. If p > .05, it is not.
Unstandardized B: Tells you exactly how much the DV will increase/decrease for every 1-unit increase in the IV. Useful for creating a literal prediction equation.
Standardized Beta (β): Converts all variables to the same scale. This allows you to directly compare which variable is the "strongest" predictor. A variable with β = .60 has twice the impact of a variable with β = .30.

4. Extracting Data from SPSS

Run: Analyze > Regression > Linear...
Under "Statistics", ensure you check Collinearity diagnostics (for VIF) and Durbin-Watson.
Copy the Model Summary, ANOVA, and Coefficients tables directly into the input fields in the app.