It performs a t-test on recursive residuals. If the relationship is non-linear, the mean of these residuals will differ significantly from zero. Hypothesis: Null Hypothesis ( H0cap H sub 0 ): The true relationship is linear. Alternative Hypothesis ( H1cap H sub 1 ): The relationship is non-linear (convex or concave).
If "HARVTEST V1.4" refers to a different utility, it may be confused with: Harvest: Time Tracking Software With Invoicing HARVTEST V1.4
When utilizing this test within a regression diagnostic workflow, the following parameters and outputs are typical: formula : A symbolic description of the model to be tested. It performs a t-test on recursive residuals
The test is used to determine if the relationship between variables in a linear regression model is truly linear or if it is instead convex or concave. Alternative Hypothesis ( H1cap H sub 1 ):
In R, the function harvtest(formula, order.by = NULL, data = list()) is used to execute the test. Key Components and Usage
It is frequently used alongside other diagnostic tools like the Breusch-Pagan test (for heteroskedasticity) and the Durbin-Watson test (for autocorrelation). Related "Harvest" Software