1

I am conducting an ordinal logistic regression and my alternative hypothesis is one-directional (I predict that my independent variable positively affects my dependent variable) and my null hypothesis is that my independent variable does not positively affect my dependent variable.

As I am slightly confused about differentiating between one-and-two tailed tests and P-values, would the code written below give me the correct P-Value I could use to reject my null hypothesis (assuming that it meets my 0.05 predetermined threshold of course).

ctable <- coef(summary(Model_1))

p <- pnorm(abs(ctable[, "t value"]), lower.tail = FALSE) * 2

ctable <- cbind(ctable, "p value" = p)

ctable
kjetil b halvorsen
  • 63,378
  • 26
  • 142
  • 467
Joey
  • 11
  • 1
  • I am not sure what you are asking. The p-values here are associated with the regression's coefficients and intercept, they are not the result of a test of equality. There are no p-values directly produced by the ordinal regression but you can produce some using likelihood ratios, e.g. car::Anova(Model_1) – GuillaumeL Mar 18 '21 at 22:07

0 Answers0