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

Re: st: which statistical analysis to use

 From Nick Cox To statalist@hsphsun2.harvard.edu Subject Re: st: which statistical analysis to use Date Thu, 19 Apr 2012 12:39:27 +0100

```I can't even say whether the question makes sense. For example, you
have not told us anything about the sampling process which leads to
these companies being the data. You have not told us anything about
what kinds of data generation process would allow significance
calculations. My instinct here is that this is descriptive statistics
and quite tough in that even thinking about what graphs and
tabulations will be really helpful is quite a challenge.

As hinted by others in this thread, very general questions like this
are too difficult to answer and "How should I analyse my data?" is not
a good question to ask in any case. But if you keep to specifics you
are more likely to get specific answers.

On Thu, Apr 19, 2012 at 12:27 PM, Deborah Beckers
<deborahbeckers@hotmail.com> wrote:
> I've looked at the data for each skill separately by using tab and sum, and looked like you say at median and mean scores, and I did see some clear differences, but without statistical analysis, there's no way to tell if these differences are significant, right?
>
> ----------------------------------------
>> Date: Thu, 19 Apr 2012 12:03:54 +0100
>> Subject: Re: st: which statistical analysis to use
>> From: njcoxstata@gmail.com
>> To: statalist@hsphsun2.harvard.edu
>>
>> A more general point is that you are not wedded to the scores as given
>> as long as there is a logic to how you treat or re-present them. For
>> example, if any skills are graded by 0 by everybody then I am not sure
>> you can do much with those except list them. As far as the other
>> skills are concerned, you could look at median and quartiles for
>> scores as well as mean scores.
>>
>> Some years ago in an internal discussion about workload weights for
>> different kinds of administrative responsibilities we first rejected
>> the idea of keeping diaries and quantifying time spent because that
>> would be a pain and reward the inefficient and penalise the efficient.
>> something like this. Consider the first few Fibonacci numbers 1, 2, 3,
>> 5, 8, 13, 21. Let's have a system in which being Chair of Dept gets
>> 21, being in charge of a major area gets 13, and so on down to being
>> just a committee member gets 1. This was just plucked out of the air
>> as a piece of pure mathematics, but what was interesting was the quick
>> consensus was that would produce as good a quantification as any other
>> scheme,
>>
>> On Thu, Apr 19, 2012 at 11:30 AM, Deborah Beckers
>> <deborahbeckers@hotmail.com> wrote:
>> > Ok, thank you Nick, I will give that a try!
>> >
>> > Deborah
>> >
>> > ----------------------------------------
>> >> Date: Thu, 19 Apr 2012 11:25:10 +0100
>> >> Subject: Re: st: which statistical analysis to use
>> >> From: njcoxstata@gmail.com
>> >> To: statalist@hsphsun2.harvard.edu
>> >>
>> >> I don't understand this point about it being too bad that 1 is most
>> >> important. You can just reverse the scale so that 7 is most important.
>> >>
>> >> gen scale2 = cond(scale >=1, 8 - scale, 0)
>> >>
>> >> That gives you 7 -> 1, 6 -> 2, ..., 1 -> 7, 0 -> 0.
>> >>
>> >> Other ways of doing this include -recode-.
>> >>
>> >> Whether one-way anova is a good fit is another question. I suspect
>> >> that few off-the-shelf techniques are quite right here.
>> >>
>> >> Nick
>> >>
>> >> On Thu, Apr 19, 2012 at 11:14 AM, Deborah Beckers
>> >> <deborahbeckers@hotmail.com> wrote:
>> >> > Dear David,
>> >> >
>> >> > Thank you for your help!
>> >> > I am definitely going to look up that paper and books, I'm certain it will be useful, thankyou!
>> >> > Just like you said I was also thinking about replacing all the chosen skills by '1', this will certainly make things easier I think..
>> >> > Too bad that '1' is the most important, otherwise I could maybe have use a oneway anova like in this example:
>> >> > http://nd.edu/~rwilliam/stats1/Oneway-Stata.pdf
>> >> >
>> >> > Anyway, I'm going to have a look at those references you gave me, and if that doesn't work out I'll probably make the data dichotomous.. Again, thank you very much for your help!
>> >> > Deborah
>> >> >
>> >> >
>> >> >> Date: Wed, 18 Apr 2012 15:33:22 -0400
>> >> >> Subject: Re: st: which statistical analysis to use
>> >> >> From: dchoaglin@gmail.com
>> >> >> To: statalist@hsphsun2.harvard.edu
>> >> >>
>> >> >> Deborah,
>> >> >>
>> >> >> By an interesting coincidence, the issue of Computational Statistics &
>> >> >> Data Analysis that arrived today contains a paper on ranking data:
>> >> >>
>> >> >> Lee PH, Yu PLH. Mixtures of weighted distance-based models for ranking
>> >> >> data with applications in political studies.  Computational Statistics
>> >> >> & Data Analysis 2012; 56:2486-2500.
>> >> >>
>> >> >> That paper is probably not directly relevant to the analysis that you
>> >> >> are trying to do, but its list of references may be helpful in making
>> >> >> contact with that literature.  In particular, I noticed two books:
>> >> >>
>> >> >> Marden JI. Analyzing and Modeling Rank Data. Chapman and Hall, 1995.
>> >> >>
>> >> >> Fligner MA, Verducci JS (eds.). Probability Models and Statistical
>> >> >> Analyses for Ranking Data. Springer-Verlag, 1993.
>> >> >>
>> >> >> I hope this information is useful.
>> >> >>
>> >> >> David Hoaglin
>> >> >>
>> >> >> On Tue, Apr 17, 2012 at 7:59 AM, Deborah Beckers
>> >> >> <deborahbeckers@hotmail.com> wrote:
>> >> >> > Hello everybody,
>> >> >> >
>> >> >> >
>> >> >> > I'm having a problem with statistical analysis for my thesis. I am using stata 11 for windows.
>> >> >> > My data consists of a survey filled in by 360 companies, and the question I want to use is a question where they get a list of 27 employee skills, and they have to choose the 7 most important skills, by giving them a score from 1 to 7. The other skills (which they find less important) are not given any score (they are zero in my data). The data for that question thus looks somewhat as follows (example for 3 companies, one row per company:
>> >> >> >
>> >> >> >
>> >> >> > My question is: what kind of statistical analysis should I do, and how, to find out whether certain skills are ranked as more (or less) important than others by the companies, and if this difference is significant?
>> >> >>
>> >>
>> >> *
>> >> * For searches and help try:
>> >> * http://www.stata.com/help.cgi?search
>> >> * http://www.stata.com/support/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/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/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/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/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
```