I'm using the "linear regression t-test" guide at https://stattrek.com/regression/slope-test.aspx

The guide shows calculating t =b1/SE, where b1 and SE are provided by the regression function (here lm() - using R.) The guide shows the p-value gets

The output from R, using a binary predictor, gives a p-value for the coefficient (other than the intercept) that is equal to the reported F-statistic's p-value. This is the same F-statistic return from aov().

In the guide, the p-value associated with the coefficient is getting

Doesn't ANOVA measure the significance of the predictor in this case? So why is the "linear regression t-test" saying the p-value (associated with a slope of 0) is twice this?

Also, I can't get math tags to work. This, with square braces: {math}b_1{math} : gives \(b_1\)

The guide shows calculating t =b1/SE, where b1 and SE are provided by the regression function (here lm() - using R.) The guide shows the p-value gets

**doubled**as this is a two-sided test (hypothesis: the slope is zero).The output from R, using a binary predictor, gives a p-value for the coefficient (other than the intercept) that is equal to the reported F-statistic's p-value. This is the same F-statistic return from aov().

In the guide, the p-value associated with the coefficient is getting

**doubled**(at least with their continuous variable example).Doesn't ANOVA measure the significance of the predictor in this case? So why is the "linear regression t-test" saying the p-value (associated with a slope of 0) is twice this?

Also, I can't get math tags to work. This, with square braces: {math}b_1{math} : gives \(b_1\)

Last edited: