[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: RE: missing F statistic [was: ""]

From   "Maarten Buis" <>
To   <>
Subject   st: RE: missing F statistic [was: ""]
Date   Fri, 3 Aug 2007 15:50:23 +0200

--- Renuka Metcalfe wrote:
> I do not get missing F statistics if I do not just
> take greater tha or equal to 5 hired.

So apparently something goes wrong after you have 
performed this data manipulation. 

> I am bringing over a the 'new hires'
> variable from the employee dataset to the employer
> dataset. That is, I wanted all establishments that
> hired 5 or more people. Otherwise, I would have
> establishments that hired 0, 1s' etc.
> So I did
> .collapse sum newhires, by  (firmid)
> keep if > = 5
> I then merged it to the main file.

So you have two files: one on firms and one on 
their employees. You used the employee file to 
create a file with only the firms that hired
5 or more employees. Than you merged that file
with the firm file. 

Did you look at -tab _merge- to see if the 
merge was successful? Did you know that -merge- 
will not delete the firms that hired less than 
5 employees?

-- Maarten

Maarten L. Buis
Department of Social Research Methodology 
Vrije Universiteit Amsterdam 
Boelelaan 1081 
1081 HV Amsterdam 
The Netherlands

visiting address:
Buitenveldertselaan 3 (Metropolitan), room Z434 

+31 20 5986715

Thank you. 
But the results had missing F and Prob>F.

 No standard
errors are missing. No variables are dropped. Some of
the estimates are small; but they may be "meant" to be

Many thanks in advance

Maarten's quote:
It is not clear to me what you have done. What do you
mean by restrict:
are you imposing a constraint, or are you estimating
something on a
sub-sample, or are you doing something else? It helps
if you tell us
exactly what you typed. A lot of the problems are
caused by tiny
things, so if you don't tell us exactly what you
typed, we won't be
able to spot those. Do you get normal results if you
don't retrict
(whatever that may mean)? Are any (or all) of the
standard errors also
missing? Are any variables dropped? Are some of the
estimates unusually
large (or small)? If you are estimating a model on a
subsample, you
should have a good look at the variables in that
subsample: it may be
that a dumy variable consists almost entirely of zeros
(or ones) in a
subsample, which may cause problems, or there may be
much higher
multicolinearity in the subsample than in the entire

-- Maarten

--- Renuka Metcalfe <> wrote:

> Dear Statalist
> I tried to run a simple OLS, with a merged employee
> and employer dataset, but I restricted one variable
> from the employee dataset to establishments who
> 5 employees or more employees, whilst all employees
> are normally in the raw survey if they have less
> or equal to 25 employees or 25 employees are
> selected if they have > 25 employees.
> After running a simple OLS, the F (16, 1092) appear
> blank along with prob>F (see below).
> Regression with robust standard errors
>                 Number of obs =    1110
>                  F( 16,  1092) =       .
>                  Prob > F      =       .
>                  R-squared     =  0.6248
>                  Root MSE      =  .09573
> I would be grateful, if anyone would tell me why
> is the case.
> Thanks in advance

Yahoo! Mail is the world's favourite email. Don't settle for less, sign up for
your free account today*
*   For searches and help try:

*   For searches and help try:

© Copyright 1996–2017 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index