Scribd
Upload
8 views7 pages

Testing for Normality & Specification Error

Notes

Uploaded by

Ivy Dasgupta
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
8 views7 pages

Testing for Normality & Specification Error

Notes

Uploaded by

Ivy Dasgupta
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 7

TESTING FOR NORMALITY

& SPECIFICATION ERROR


Dr. Ivy Das Gupta
Testing for Normality

Run the regression first


predict e_hat, resid
kdensity e_hat, normal
or
histogram e_hat, kdensity normal
Or
Standardize normal probability plot (pnorm) checks for non-normality
pnorm e_hat
Or
Testing for Normality

 Quintile-normal plots (qnorm) check for non-normality


 qnorm e_hat
 Shapiro-Wilk test for normality (for 4 ≤ n ≤ 2000 observations)
 swilk e_hat
 Null hypothesis is that the distribution of the residuals is normal.
 Shapiro-Francia test for normality (for 5 ≤ n ≤ 5000 observations)
 sfrancia e_hat
 Again, Null hypothesis is that the distribution of the residuals is normal.
Testing for Normality

 The values reported under W and W′ are the Shapiro–Wilk and Shapiro–Francia test
statistics.

 The tests also report V and V′, which are more appealing indexes for departure from
normality.

 The median values of V and V ′ are 1 for samples from normal populations.

 Large values indicate nonnormality.

 The 95% critical values of V (V ′), which depend on the sample size, are between 1.2 and
2.4 (2.0 and 2.8)
If residuals do not follow a ‘normal’
pattern

 Then you should check for omitted variables, model specification, linearity,
functional forms.

 Testing for omitted variable bias:

 In Stata we test for omitted-variable bias using the ovtest command

 regress var1 var2 var3 var4, robust

 ovtest

 H0: model has no omitted variables


If residuals do not follow a ‘normal’
pattern
 Specification error:
 To test model specification error use linktest.
 regress var1 var2 var3 var4, robust
 linktest
 It checks whether we need more variables in our model by running a new
regression with the observed Y against Yhat and Yhat-squared as
independent variables
 Null hypothesis is that there is no specification error.
 If the p-value of ‘_hatsq’ is not significant then we fail to reject the null
hypothesis and conclude that our model is correctly specified.
To check for outliers, we can use Box Plot
diagram.

graph box var1

use the avplots command (added-


Outliers variable plots) after regression

avplot var1 avplot var2 or

avplots

You might also like