How to use
- Paste the data with one row per case and one column per variable. Put the names in the first row.
- Pick the dependent variable (Y) and tick the predictors (X) to include.
- Read the equation, R² and the model F test, then each predictor's p-value in the coefficient table.
- Enter X values under Predict to get the predicted Y with a prediction interval.
Options explained
- B and Beta
- B is the coefficient in the original units; a one-unit increase in X changes Y by B on average. Beta is the standardized coefficient for comparing predictors measured in different units.
- R² and adjusted R²
- R² is the share of the variation in Y the model explains. It never drops when you add predictors, so compare models with adjusted R².
- VIF
- The variance inflation factor shows how much a predictor is explained by the other predictors. Above 10 (some texts say 5) signals strong multicollinearity, which makes individual coefficients unreliable.
- Durbin–Watson
- Checks whether neighbouring residuals are correlated. Values near 2 are good. Only meaningful when rows are in time order.
Worked example
The sample has 10 rows with Y as the dependent variable and X1, X2 as predictors.
Equation: Y = 0.4023 + 0.7876 × X1 + 1.0054 × X2
| Model summary | Value |
|---|---|
| R² | 0.9986 |
| Adjusted R² | 0.9982 |
| Std. error of the estimate | 0.1993 |
| F(2, 7) | 2462.4, p < 0.001 |
| Coefficient | B | SE | Beta | t | p | VIF |
|---|---|---|---|---|---|---|
| (Constant) | 0.4023 | 0.1625 | 2.476 | 0.042 | ||
| X1 | 0.7876 | 0.0825 | 0.511 | 9.550 | < 0.001 | 14.13 |
| X2 | 1.0054 | 0.1082 | 0.497 | 9.290 | < 0.001 | 14.13 |
Both predictors are significant, but their VIF of 14.1 is above 10 because X1 and X2 are strongly correlated with each other. The model still predicts well; it’s the separate effect of each predictor that is hard to pin down. Untick X2 for a simple regression: Y = 1.2267 + 1.5261 × X1, R² = 0.981. At X1 = 11 the predicted Y is 18.01, with a 95% prediction interval of 16.11 to 19.91.
APA style: R² = .999, F(2, 7) = 2462.39, p < .001.
How to read the output
- Model F test: p < 0.05 means the model as a whole explains something.
- R² or adjusted R²: how much of the variation in Y is explained.
- Coefficient p-values: which predictors still matter after controlling for the others.
- Residual plot: you want a shapeless band around 0. A curve means the relationship isn’t linear; a funnel means unequal variance.
- VIF: with several predictors, check for collinearity.
Frequently asked questions
Does it match SPSS and R?
Yes. Coefficients, standard errors, t, p, R² and F match SPSS Linear Regression and R's lm(); Beta is SPSS's "standardized coefficients".
What about blank cells?
A row with a blank in any selected column is left out (listwise deletion). The result says how many rows were dropped.
How do I include a categorical predictor such as gender or region?
Code it as 0/1 dummy columns first. For k categories make k − 1 columns and leave one category as the reference.
What does "perfectly collinear" mean?
One column can be computed from the others, for example a total score next to its parts, or two identical columns. Untick one of them.
Is my data uploaded?
No. The calculation runs in your browser.
Privacy
This tool runs entirely in your browser. Your files and text are processed on your device and are never uploaded to our servers.