Multiple Linear Regression do not shown its statistics

[Zaki Rasyadi]

How do you deal with multiple regression using stata that show no output for its statistics of regression? Here I attached my command and result for this case.

It shows that there is no result for Std. Error, t stat, P>|t|, and 95% conf. interval since it should be filled when I do multiple linear regression in Excel.

Does my calculation of regression lack treatment to make this data available to have its regression statistics? Please help me regarding this issue, Appreciate it so much. Thank you

[Nick Cox]

At a guess, it's a matter of definition that FLARE = PROD - SALES - OWN so your regression has rediscovered an identity. The relationship shows R-square of 1 as it is should but in principle there is no uncertainty. Now in practice the constant or intercept is estimated to be very small and not exactly zero but that's a side-effect of the fact that Stata can do all calculations infinite precision. Or perhaps there is also some measurement error.

Here is a demonstration of this phenomenon.

. sysuse auto, clear
(1978 automobile data)

. set seed 2803

. gen frog = runiformint(1, 10)

. gen toad = runiformint(1, 10)

. gen diff = frog - toad

. regress diff frog toad

      Source |       SS           df       MS      Number of obs   =        74
-------------+----------------------------------   F(2, 71)        =         .
       Model |  1034.86486         2  517.432432   Prob > F        =         .
    Residual |           0        71           0   R-squared       =    1.0000
-------------+----------------------------------   Adj R-squared   =    1.0000
       Total |  1034.86486        73   14.176231   Root MSE        =         0

------------------------------------------------------------------------------
        diff | Coefficient  Std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
        frog |          1          .        .       .            .           .
        toad |         -1          .        .       .            .           .
       _cons |  -2.28e-15          .        .       .            .           .
------------------------------------------------------------------------------

.

This is not a problem: there is no bug or misfeature, but nothing to write up either. This can bite on various levels. I dimly remember a book proclaiming some economic finding and an excoriating review maintaining that the author didn't realise his findings were just a matter of rediscovering how business accounts are produced through double-entry book-keeping.

If you're saying that MS Excel fills in the table, that's alarming but no surprise for anyone who has learned not to use Excel for more than what it does well.

[Nick Cox]

… can only do calculations with finite precision.