0

if I run the lm in R, after getting the results, can I find out the normality of the residual? or I need to run another test such as qqline? here is the example: In other words, what does the last section tells me? Thanks,

Call:
lm(formula = ydata ~ xdata)

Residuals:
    Min      1Q  Median      3Q     Max 
-4.5401 -1.3643  0.1956  1.8130  3.1046 

Coefficients:
            Estimate Std. Error t value Pr(>|t|)    
(Intercept) 3.188580   1.015338   3.140  0.00632 ** 
xdata       0.048447   0.008045   6.022 1.78e-05 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 2.297 on 16 degrees of freedom
  (17343 observations deleted due to missingness)
Multiple R-squared:  0.6939,    Adjusted R-squared:  0.6748 
F-statistic: 36.27 on 1 and 16 DF,  p-value: 1.776e-05
  • See https://stats.stackexchange.com/search?q=regression+residual+normal+answers%3A1+score%3A1 for many posts on this popular topic. The "last section" of the output tells you nothing about Normality of the residuals. – whuber Jul 17 '18 at 11:45

1 Answers1

1

The residual standard error only tells you exactly that, the standard error of your residuals. The Multiple R-squared, Adjusted R-squared, and F-stat won't tell you anything about whether your assumptions about your residuals are true. You need to perform additional analysis to determine whether or not your errors are uncorrelated white noise. You should look at a residual vs. fitted values plot, a qq-plot, etc. There are also hypothesis tests which you can perform such as the Ljung-Box test. I recommend looking over a linear regression text book for more details.