Estat ovtest

[Mary Anne]

Hello everyone,

I regress a fixed effect model with heteroscedastic consistent and autocorrelation adjusted standard errors. After the regression, I performed the command -ovtest- and the following message appeared:

powers of fitted values collinear with explanatory variables
(typically because all explanatory variables are indicator variables)

Is this something bad?

Thank you.

[Carlo Lazzaro]

Mary:
nothing bad indeed; this message simply states that all your preditors are categorical.
It's easy to see from the following toy-example why -estat ovtest- cannot work in this instance:

. use "C:\Program Files\Stata16\ado\base\a\auto.dta"
(1978 Automobile Data)

. regress price i.foreign

      Source |       SS           df       MS      Number of obs   =        74
-------------+----------------------------------   F(1, 72)        =      0.17
       Model |  1507382.66         1  1507382.66   Prob > F        =    0.6802
    Residual |   633558013        72  8799416.85   R-squared       =    0.0024
-------------+----------------------------------   Adj R-squared   =   -0.0115
       Total |   635065396        73  8699525.97   Root MSE        =    2966.4

------------------------------------------------------------------------------
       price |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
     foreign |
    Foreign  |   312.2587   754.4488     0.41   0.680    -1191.708    1816.225
       _cons |   6072.423    411.363    14.76   0.000     5252.386     6892.46
------------------------------------------------------------------------------

. predict fitted, xb

. estat ovtest
powers of fitted values collinear with explanatory variables
(typically because all explanatory variables are indicator variables)
test not possible
r(499);

. pwcorr foreign fitted

             |  foreign   fitted
-------------+------------------
     foreign |   1.0000
      fitted |   1.0000   1.0000

.

Deafult or non-default standard errors have no bearing on that.

As an aside, I assume that you ran a fixed effect regression via -regress- plus -i.panelid- plugged in the right-hand side of your regression equation, as -estat ovtest- cannot be invoked after -xtreg,fe- (that is way more advisable if you have a panel data set to deal with).

[Mary Anne]

Carlo,

My mistake. I did a multiple linear regression, not a fixed effect (sorry for the confusion, it was really late when I wrote that post).
My problem is that my predictors are not categorical, but continuous. I already put "c." before continuous variables on the regression but the message still appears.

Thank you for your help.

[Carlo Lazzaro]

Mary Anne:
thanks for clarifying.
Putting a -c.- before the predictor may be not enough:

. use "C:\Program Files\Stata16\ado\base\a\auto.dta"
(1978 Automobile Data)

. regress price c.foreign

      Source |       SS           df       MS      Number of obs   =        74
-------------+----------------------------------   F(1, 72)        =      0.17
       Model |  1507382.66         1  1507382.66   Prob > F        =    0.6802
    Residual |   633558013        72  8799416.85   R-squared       =    0.0024
-------------+----------------------------------   Adj R-squared   =   -0.0115
       Total |   635065396        73  8699525.97   Root MSE        =    2966.4

------------------------------------------------------------------------------
       price |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
     foreign |   312.2587   754.4488     0.41   0.680    -1191.708    1816.225
       _cons |   6072.423    411.363    14.76   0.000     5252.386     6892.46
------------------------------------------------------------------------------

. estat ovtest
powers of fitted values collinear with explanatory variables
(typically because all explanatory variables are indicator variables)
test not possible
r(499);

That said, sharing an example/excerpt of your data via -dataex- can help enormously more positive replies. Thanks.

[Su hendra]

Mary Anne:
thanks for clarifying.
Putting a -c.- before the predictor may be not enough:

. use "C:\Program Files\Stata16\ado\base\a\auto.dta"
(1978 Automobile Data)

. regress price c.foreign

Source | SS df MS Number of obs = 74
-------------+---------------------------------- F(1, 72) = 0.17
Model | 1507382.66 1 1507382.66 Prob > F = 0.6802
Residual | 633558013 72 8799416.85 R-squared = 0.0024
-------------+---------------------------------- Adj R-squared = -0.0115
Total | 635065396 73 8699525.97 Root MSE = 2966.4

------------------------------------------------------------------------------
price | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
foreign | 312.2587 754.4488 0.41 0.680 -1191.708 1816.225
_cons | 6072.423 411.363 14.76 0.000 5252.386 6892.46
------------------------------------------------------------------------------

. estat ovtest
powers of fitted values collinear with explanatory variables
(typically because all explanatory variables are indicator variables)
test not possible
r(499);

That said, sharing an example/excerpt of your data via -dataex- can help enormously more positive replies. Thanks.

So what is the solution for this problem Carlo? I have the same problem with Mary

[Carlo Lazzaro]

Su.
welcome to this forum.
You may want to take a look at -linktest-.