Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: Making a variable overall follow a normal distribution


From   Nick Cox <[email protected]>
To   "[email protected]" <[email protected]>
Subject   Re: st: Making a variable overall follow a normal distribution
Date   Mon, 5 Aug 2013 12:29:21 +0100

What you say about missings isn't clear. But the Shapiro-Wilk test
just ignores the missings in your data; it can't do anything else.

What is missing (different sense) is quite what "theory" makes you
expect a normal distribution. But supposing that you are right and if
the missings would have been certain non-missing values had they been
measured, and the whole set would have been approximately normal, then
this is a problem for multiple imputation, but there are limits on
when that makes sense, which I won't try to summarise here. -mi- is
the magic word in Stata (or a spell that can be misused).

As already implied, I don't get a good sense of your situation but
I'll add that there is as much folklore as theory about normal
distributions. It's not even true that variables being normally
distributed is required for correlation and regression, despite
numerous statements to that effect.
Nick
[email protected]


On 5 August 2013 12:09, Ruth Young <[email protected]> wrote:
> Hello,
>
> My question involves missing data:
>
> I currently have generated a combined variable (let's call it a+b+c) and according to tests for normality (SW test) it does not follow a normal distribution but I know in theory that it should. Is there some way in Stata where I can generate a new variable (let's call it d) based on the mean of a+b+c that follows a normal distribution.
>
> The reason why I think it does not follow a normal distribution is that I think there was such a large proportion of missing data that influenced the resultant curve and I have 887 observation out of 2785.
>
> Thanks!
>
> Ruth
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/statalist-faq/
> *   http://www.ats.ucla.edu/stat/stata/

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index