CODEWITHSUNDEEP
rregressionlmstandard-error
Fabian Stolz
Fabian Stolz

Reputation: 2095

Extract standard errors from lm object

We got a lm object from and want to extract the standard error

lm_aaa <- lm(aaa ~ x + y + z)

I know the function summary, names and coefficients.

However, summary seems to be the only way to manually access the standard error.

Have you any idea how I can just output se?

Upvotes: 36

Views: 93785

Answers (5)

Whitebeard13
Whitebeard13

Reputation: 455

I think that the following lines can also provide you with a quick answer:

lm_aaa<- lm(aaa~x+y+z)
se <- sqrt(diag(vcov(lm_aaa)))

Upvotes: 5

Annerose N
Annerose N

Reputation: 487

If you don't want to get the standard error/deviation of the model, but instead the standard error/deviation of the individual coefficients, use 

# some data (taken from Roland's example)
x = c(1, 2, 3, 4)
y = c(2.1, 3.9, 6.3, 7.8)

# fitting a linear model
fit = lm(y ~ x)

# get vector of all standard errors of the coefficients
coef(summary(fit))[, "Std. Error"]

For more information on the standard error/deviation of the model, see here. For more information on the standard error/deviation of the coefficients, see here.

Upvotes: 8

smillig
smillig

Reputation: 5351

To get a list of the standard errors for all the parameters, you can use 

summary(lm_aaa)$coefficients[, 2]

As others have pointed out, str(lm_aaa) will tell you pretty much all the information that can be extracted from your model.

Upvotes: 14

csgillespie
csgillespie

Reputation: 60522

The output of from the summary function is just an R list. So you can use all the standard list operations. For example:

#some data (taken from Roland's example)
x = c(1,2,3,4)
y = c(2.1,3.9,6.3,7.8)

#fitting a linear model
fit = lm(y~x)
m = summary(fit)

The m object or list has a number of attributes. You can access them using the bracket or named approach:

m$sigma
m[[6]]

A handy function to know about is, str. This function provides a summary of the objects attributes, i.e.

str(m)

Upvotes: 29

Roland
Roland

Reputation: 132999

#some data
x<-c(1,2,3,4)
y<-c(2.1,3.9,6.3,7.8)

#fitting a linear model
fit<-lm(y~x)

#look at the statistics summary
summary(fit)

#get the standard error of the slope
se_slope<-summary(fit)$coef[[4]] 
#the index depends on the model and which se you want to extract

#get the residual standard error
rse<-summary(fit)$sigma

Upvotes: 11

Related Questions