Difference in p-values between summary() vs glance()?

Question

I'm noticing a difference in the p-values when calling summary() on a linear model object versus calling broom::glance(). I think the floating point precision when calling summary() is limited to 2.2e-16 while glance can reach beyond 1e-100. Is my suspicion correct, or are these values inherently different?

x <- c(1, 2, 3, 4, 5, 6, 7, 8, 9)
y <- c(10, 20, 30, 40, 50, 60, 70, 80, 90)

mod <- lm(y~x)

summary(mod) # p < 2.2e-16

broom::glance(mod) # p = 4.66e-112

score 2 · Answer 1 · answered Jan 23 '19 at 01:11

2

They are the same. Please see the P value from summary after we print the coefficient.

summary(mod) 

s$coefficients
#                 Estimate   Std. Error      t value      Pr(>|t|)
# (Intercept) 9.473903e-15 3.161050e-15 2.997075e+00  2.002483e-02
# x           1.000000e+01 5.617334e-16 1.780204e+16 4.661081e-112

answered Jan 23 '19 at 01:11

www

38,575
12
48
84

So the 4.66e-112 is the p-value for the `x` coefficient, but not for the entire model? These are two different statistics then? – philiporlando Jan 23 '19 at 01:13
For coefficient. – www Jan 23 '19 at 01:15

score 2 · Accepted Answer · answered Jan 23 '19 at 01:12

2

It is just a matter of visualization,

summary(mod) says that p-value is less than 2.2e-16 (<2.2e-16), not equal.

For example, if you run

summary(mod)$coefficient you get 4.66e-112

Best

answered Jan 23 '19 at 01:12

LocoGris

4,432
3
15
30

Do you how to extract the model p-value (<2.2e-6) instead of just the coefficient p-value? – philiporlando Jan 23 '19 at 01:17
1

https://stackoverflow.com/questions/5587676/pull-out-p-values-and-r-squared-from-a-linear-regression – LocoGris Jan 23 '19 at 01:24
Just found this too. This ended up being the issue. Thanks! – philiporlando Jan 23 '19 at 01:25

Difference in p-values between summary() vs glance()?

2 Answers2