Statalist


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

st: RE: RE: RE: -impute- variable creates only type 'float'


From   "Howrey, Bret T." <[email protected]>
To   "[email protected]" <[email protected]>
Subject   st: RE: RE: RE: -impute- variable creates only type 'float'
Date   Tue, 15 Jul 2008 20:20:11 -0500

;
;
;
Hewan,
hot deck is designed for multiple imputation, more than one pass.  thus, one must designate the names of the multiple datasets generated and then recombine them.  look at help -hot deck-  I have found it useful for large datasets with minimal missing data (<1%), though not tested it on small ones (e.g., <100k obs).

That point aside, I see your point - noone has responded with a reason why this occurs in reference to the float issue.  However, practically speaking, is there a reason you HAVE to use impute?  Or are you just trouble shooting ?  Sorry not more helpful in the particulars of how impute "works".  Or in your case, doesn't.
  -B
________________________________________
From: [email protected] [[email protected]] On Behalf Of Hewan Belay [[email protected]]
Sent: Tuesday, July 15, 2008 6:16 PM
To: Stata List
Subject: st: RE: RE: -impute- variable creates only type 'float'

Dear Bret and Maarten,

Indeed, it looks like both -hotdeck- and -ice- don't have the format-type problem that I highlighted with regard to -impute-. At the same time, substance-related merits and demerits of the three commands aside, it would have been nice to know if indeed I was right that -impute- doesn't allow creating the imputed variables in 'double' instead of 'float'.

So far apparently no "defender" of -impute- stood up to comment on this--I would still appreciate any remarks on this issue if anyone can confirm or otherwise correct my finding with regard to format type of -impute-, which apparently to-date is the only of the 3 that is an official stata command for imputing missing values.

Just a brief aside: Looks like hotdeck is doing a few strange things given the way the programme was coded. E.g. it generates the new stata dataset as "filename.dta1" instead of "filename.dta". To see this concretely, do this:

.. webuse abdata

.. hotdeck nL1 w k using my_imputed.dta, store
DELETING all matrices....

Table of the Missing data patterns
 * signifies missing and - is not missing

Varlist order: nL1 w k

    pattern |      Freq.     Percent        Cum.
------------+-----------------------------------
        *-- |        140       13.58       13.58
        --- |        891       86.42      100.00
------------+-----------------------------------
      Total |      1,031      100.00
 1031
WARNING: When the <command> option is not selected
then no analysis is performed on the imputed datasets

.. use my_imputed.dta, clear
file my_imputed.dta not found
r(601);

.. use my_imputed.dta1, clear

.. d

Contains data from my_imputed.dta1
  obs:         1,031
 vars:             3                          15 Jul 2008 19:10
 size:        16,496 (99.8% of memory free)
---------------------------------------------------------------------------              storage  display     value
variable name   type   format      label      variable label
---------------------------------------------------------------------------w               float  %9.0g
k               float  %9.0g
nL1             float  %9.0g
---------------------------------------------------------------------------Sorted by:
..

Regards,
Hewan


--------------------------------------------------------------------------



© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index