Multiple Regression Issues - Stata not showing any information

[shadi sab]

Hi! I am trying to do a multiple regression but when I try to do it, I don't get any of the p-values, st errors or t values. Please let me know what I am doing wrong! Thank you so much!

All of my variables are binary and I am using the following code:

regress V162034a V161345 V161320 V161157



Source | SS df MS Number of obs = 6
-------------+---------------------------------- F(3, 2) = .
Model | .833333333 3 .277777778 Prob > F = .
Residual | 0 2 0 R-squared = 1.0000
-------------+---------------------------------- Adj R-squared = 1.0000
Total | .833333333 5 .166666667 Root MSE = 0

------------------------------------------------------------------------------
V162034a | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
V161345 | -5.07e-17 . . . . .
V161320 | 2.03e-16 . . . . .
V161157 | -1 . . . . .
_cons | 3 . . . . .
------------------------------------------------------------------------------

[William Lisowski]

Recall that Stata omits from the model any observation for which any independent or dependent variable has a missing value. The results you posted show that only 6 observations were included in the regression once this was done.

For the six observations that remained, the model V162034a = 3 - V161345 essentially first the data perfectly - hence the R² value of 1. And with a perfect fit, there are no standard errors, and thus no t statistics or p values for those statistics.

[shadi sab]

Thank you for your response! When I run bivariate regressions of each independent variable I have much larger observed values. Do you know why it shows as 6 when I run a multiple regression? Is there any way to fix this? I have already recoded missing values.

[Carlo Lazzaro]

Shadi:
there's simply no scope in performing any regression with a sample size=6.
That said, it seems that your results show no variation in the predictors, the number of which is also dangerously near to sample size.
The following toy-example seems to mimick your results:

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

. regress price mpg if _n<=2

      Source |       SS           df       MS      Number of obs   =         2
-------------+----------------------------------   F(1, 0)         =         .
       Model |      211250         1      211250   Prob > F        =         .
    Residual |           0         0           .   R-squared       =    1.0000
-------------+----------------------------------   Adj R-squared   =         .
       Total |      211250         1      211250   Root MSE        =         0

------------------------------------------------------------------------------
       price |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
         mpg |       -130          .        .       .            .           .
       _cons |       6959          .        .       .            .           .
------------------------------------------------------------------------------

[William Lisowski]

To respond to the question in post #3, when you run

regress V162034a V161345

those observations with missing values for either V162034a or V161345 are eliminated. When you add two more independent variables

regress V162034a V161345 V161320 V161157

now additionally observations with missing values for either V161320 or V161157 are also eliminated.

Perhaps the commands

misstable summarize V162034a V161345 V161320 V161157
misstable patterns V162034a V161345 V161320 V161157

will help you understand the effect of missing values on your sample.