R robust se - Revision history

Rb: /* Heteroskedasticity Robust F-tests */

2015-08-26T14:35:21Z

‎Heteroskedasticity Robust F-tests

2015-08-05T22:59:18Z

‎Heteroskedasticity Robust F-tests

2015-08-05T22:58:43Z

‎Heteroskedasticity Robust F-tests

2015-08-05T22:57:25Z

2015-07-10T09:58:29Z

‎Which package to use

2015-07-10T09:58:06Z

‎Which package to use

2015-07-10T09:51:04Z

‎Which package to use

2015-07-10T09:50:06Z

‎Which package to use

2015-07-06T16:14:40Z

‎Heteroskedasticity robust standard errors

2015-04-13T15:23:40Z

@@ Line 90: / Line 90: @@
 As you can see it produces slightly different results, although there is no change in the substantial conclusion that you should not omit these two variables as the null hypothesis that both are irrelevant is soundly rejected.
 == Footnotes ==
   <references />

@@ Line 87: / Line 87: @@
       Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
-The input <code><span style="color:blue">vcov=vcovHC</span></code>
+The input <code><span style="color:blue">vcov=vcovHC</span></code> instructs R to use a robust version of the variance covariance matrix.
 As you can see it produces slightly different results, although there is no change in the substantial conclusion that you should not omit these two variables as the null hypothesis that both are irrelevant is soundly rejected.

@@ Line 76: / Line 76: @@
 But this procedure assumed that the error terms were homoskedastic. If you want to allow for for heteroskedastic error terms you merely have to add <span style="color:#blue">another input</span> to the <code>waldtest</code> function call.
-      > waldtest(reg_ex2, .~. - age - educ, <span style="color:#blue">vcov=vcovHC</span>)
+      > waldtest(reg_ex2, .~. - age - educ, <span style="color:blue">vcov=vcovHC</span>)
       Wald test
@@ Line 87: / Line 87: @@
       Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
-The input <code><span style="color:#blue">vcov=vcovHC</span></code>
+The input <code><span style="color:blue">vcov=vcovHC</span></code>
 As you can see it produces slightly different results, although there is no change in the substantial conclusion that you should not omit these two variables as the null hypothesis that both are irrelevant is soundly rejected.

@@ Line 49: / Line 49: @@
      coeftest(reg_ex1, vcov = vcovHAC(reg_ex1))
 == Footnotes ==
   <references />

@@ Line 13: / Line 13: @@
      library(sandwich)
-All code snippets below assume that you have done so. <span style="color:red">In fact, you may instead want to use another package called "AER" which contains the sandwich package</span> and other relevant packaes (such as the one used for instrumental variables estimation [[IV_in_R]]. See [http://cran.r-project.org/web/packages/sandwich/index.html the relevant CRAN webpage]
+All code snippets below assume that you have done so. <span style="color:red">In fact, you may instead want to use another package called "AER" which contains the sandwich package</span> and other relevant packaes (such as the one used for instrumental variables estimation [[IV_in_R]]). See [http://cran.r-project.org/web/packages/sandwich/index.html the relevant CRAN webpage]
 == Heteroskedasticity robust standard errors ==

@@ Line 29: / Line 29: @@
 This saves the heteroscedastic robust standard error in <source enclose=none>vcv</source><ref>Predictably the <source enclose=none>type</source> option in this function indicates that there are several options (actually "HC0" to "HC4"). Using "HC1" will replicate the robust standard errors you would obtain using STATA.</ref>. Now you can calculate robust t-tests by using the estimated coefficients and the new standard errors (square roots of the diagonal elements on <source enclose=none>vcv</source>). But note that inference using these standard errors is only valid for sufficiently large sample sizes (asymptotically normally distributed t-tests).
-You may actually want a neat way to see the standard errors, rather than having to calculate the square roots of the diagonal of this matrix. This is done with the following function:
+You may actually want a neat way to see the standard errors, rather than having to calculate the square roots of the diagonal of this matrix. This is done with the following function (this is part of the lmtest package which will be automatically installed if you installed the AER package as recommended above):
      coeftest(reg_ex1, vcv)

@@ Line 46: / Line 46: @@
      vcv <- vcovHAC(reg_ex1)
-Everything is as for heteroskedastic error terms.
+Everything is as for heteroskedastic error terms. I.e. you would print these standard errors along with the coefficient estimates, t-statistics and p-values from:
 == Footnotes ==
   <references />