R reg diag - Revision history

Rb: /* Heteroskedasticity in cross sections */

2016-07-22T13:27:35Z

‎Heteroskedasticity in cross sections

Rb: /* Heteroskedasticity in cross sections */

2016-03-02T00:32:19Z

‎Heteroskedasticity in cross sections

Rb at 00:31, 2 March 2016

2016-03-02T00:31:55Z

Rb at 00:31, 2 March 2016

2016-03-02T00:31:06Z

Show changes

Rb at 00:30, 2 March 2016

2016-03-02T00:30:02Z

Show changes

Rb: /* Heteroskedasticity */

2015-07-15T21:58:26Z

‎Heteroskedasticity

Rb: /* Heteroskedasticity */

2015-04-14T09:07:10Z

‎Heteroskedasticity

Rb: /* Heteroskedasticity */

2015-04-14T09:06:40Z

‎Heteroskedasticity

Rb at 09:05, 14 April 2015

2015-04-14T09:05:00Z

Rb at 08:58, 14 April 2015

2015-04-14T08:58:49Z

@@ Line 48: / Line 48: @@
 The p-value comes from the asymptotic distribution (Chi-Square distribution with 2 degrees of freedom) of the test statistic assuming that the null hypothesis of homoskedasticity is valid. In this particular case this p-value is fairly small (smaller than 1%) which woul lead us to reject the null hypothesis of homoskedasticity at a 1% significance level.
-This was pretty easy, but I want to demonstrate what is really happening behind this test so that you know how to change it according to your problem at hand. As discussed above, we need to decide on which explanatory variables to use in order to execute the test. The default choice of the bptest is to use the same explanatory variables as those used in the original regression model, here exper and <code>log(huswage)</code>. You could get the same result by running this regression yourself and calculating the test statistic <math>n???R2</math>
+This was pretty easy, but I want to demonstrate what is really happening behind this test so that you know how to change it according to your problem at hand. As discussed above, we need to decide on which explanatory variables to use in order to execute the test. The default choice of the bptest is to use the same explanatory variables as those used in the original regression model, here exper and <code>log(huswage)</code>. You could get the same result by running this regression yourself and calculating the test statistic <math>n*R2</math>
 <pre class="r">mydata$res1_sq &lt;- (reg_ex1$residuals)^2 # add squared residuals to dataframe

@@ Line 17: / Line 17: @@
 <pre class="r">library(AER)  # allow access to AER package</pre>
-## Loading required package: splines</pre>
 <pre class="r"># This is my first R regression!
 setwd(&quot;C:/Users/Ralph Becker/Dropbox/ECLR/R/RegressionDiagnostics&quot;)              # This sets the working directory

@@ Line 16: / Line 16: @@
 <pre class="r">library(AER)  # allow access to AER package</pre>
-<pre>## Warning: package 'AER' was built under R version 3.1.3</pre>
 ## Loading required package: splines</pre>
 <pre class="r"># This is my first R regression!
@@ Line 39: / Line 24: @@
 # Now convert variables with &quot;.&quot; to num with NA
 mydata$wage &lt;- as.numeric(as.character(mydata$wage))</pre>
-<pre>## Warning: NAs introduced by coercion</pre>
 <pre class="r">mydata$lwage &lt;- as.numeric(as.character(mydata$lwage))</pre>
 <pre>## Warning: NAs introduced by coercion</pre>

← Older revision		Revision as of 21:58, 15 July 2015
Line 52:		Line 52:

	which returns exactly the same test statistic and p-value. Note that the function <span style="color:#0000ff"><source enclose=none>pchisq(teststat,2)</source></span> returns the size of the left hand tail of the Chi-Square distribution with 2 degrees of freedom.		which returns exactly the same test statistic and p-value. Note that the function <span style="color:#0000ff"><source enclose=none>pchisq(teststat,2)</source></span> returns the size of the left hand tail of the Chi-Square distribution with 2 degrees of freedom.
		+	In this procedure you could easily change the variables that should be included in the auxiliary regression by changing the explanatory variables in <code>lm(res1_sq~exper+log(huswage), data = mydata)</code>. But note that you will have to adjust the degrees of freedom in <code>pchisq(teststat,2)</code> to reflect the number of variables included.

	= Autocorrelation =		= Autocorrelation =

@@ Line 51: / Line 51: @@
      print(1-<span style="color:#0000ff">pchisq(teststat,2)</span>)   # print p value
-which returns exactly the same test statistic and p-value. Note that the function <source enclose=none><span style="color:#0000ff">pchisq(teststat,2)</span></source> returns the size of the left hand tail of the Chi-Square distribution with 2 degrees of freedom.
+which returns exactly the same test statistic and p-value. Note that the function <span style="color:#0000ff"><source enclose=none>pchisq(teststat,2)</source></span> returns the size of the left hand tail of the Chi-Square distribution with 2 degrees of freedom.
 = Autocorrelation =

@@ Line 49: / Line 49: @@
      teststat <- hs_test_sm<source enclose=none>$</source>r.squared*nrow(mydata)
      print(teststat)  # test statistic n*R^2
-     print(1-pchisq(teststat,2))   # print p value
+     print(1-<span style="color:#0000ff">pchisq(teststat,2)</span>)   # print p value
-which returns exactly the same test statistic.
+which returns exactly the same test statistic and p-value. Note that the function <source enclose=none><span style="color:#0000ff">pchisq(teststat,2)</span></source> returns the size of the left hand tail of the Chi-Square distribution with 2 degrees of freedom.
 = Autocorrelation =

@@ Line 47: / Line 47: @@
      hs_test <- lm(res1_sq~exper+log(huswage), data = mydata)
      hs_test_sm <- summary(hs_test)
-     print(hs_test_sm$r.squared*nrow(mydata))  # calculates R^2 * n
+     teststat <- hs_test_sm<source enclose=none>$</source>r.squared*nrow(mydata)
 which returns exactly the same test statistic.

@@ Line 44: / Line 44: @@
 This was pretty easy, but I want to demonstrate what is really happening behind this test so that you know how to change it according to your problem at hand. As discussed above, we need to decide on which explanatory variables to use in order to execute the test. The default choice of the <source enclose=none>bptest</source> is to use the same explanatory variables as those used in the original regression model, here <source enclose=none>exper</source> and <source enclose=none>log(huswage)</source>. You could get the same result by running this regression yourself and calculating the test statistic <math>n*R^2</math>
-     mydata$res1_sq <- (reg_ex1$residuals)^2 # add squared residuals to dataframe
+     mydata<source enclose=none>$</source>res1_sq <- (reg_ex1<source enclose=none>$</source>residuals)^2 # add squared residuals to dataframe
      hs_test <- lm(res1_sq~exper+log(huswage), data = mydata)
      hs_test_sm <- summary(hs_test)