Re: st: how to check the reliability and validity of the data

[Daniel Schneider <daniel.schneider@stanford.edu>]

You cannot assess representativeness in any such way. And it has nothing 
to do with reliability or validity (well, not a lot). What you COULD do 
is to evaluate how likely it is that your two branches are different 
from all branches by mere chance - but for that you would need to define 
a hypothesis about all branches. For example, if you said 'I believe the 
average time is 10 minutes' you can a) investigate what the mean time is 
in your two branches and how likely it is that b) that mean is different 
from 10 minutes (that is whether the difference is statistically 
significant). Now, with two out of 129 branches you will not have a lot 
of power to do this and you should consider estimators that incorporate 
the fact that your population is finite (because 129 is pretty small 
number). You could say that you have more than 2 observations because 
you have many observations within those two branches, so you are 
actually looking at a multi-level model or at least something that 
adjusts for the clustering.

All of this might sound like gibberish. I am sorry if it does, but what 
you are asking is much more a question about statistics in general than 
a question about Stata. you need to clearly define what you hypothesis 
is and then some of us might be able to give you an indication for what 
you command you should use. But you also first need to clearly 
understand the concepts of reliability, validity and representativeness.

George Huang wrote:

> Dear Dave,
>
> Thanks for your opinion. Actually, 72,878 observations are measured by 
> stopwatches from 2  branches (out of 129 branches). They represent 2,800 
> activity time of a bank. I would like to know whether the data close to 
> mean (that is, whether these data are representative). However, I don't 
> know how to use stata to get the proper results.
>
>
>
> Thank you,
>
>
>
> George
>
>
>
>
>
>
>
> ----- Original Message ----- From: "David Airey" 
> <david.airey@vanderbilt.edu>
> To: <statalist@hsphsun2.harvard.edu>
> Sent: Sunday, March 23, 2008 11:29 AM
> Subject: Re: st: how to check the reliability and validity of the data
>
>
>
> > .
> >
> > I will bet that others will also say you have not provide enough 
> > information to give a good answer. What do you mean be reliability and 
> > validity, exactly?
> >
> > -Dave
> >
> > On Mar 22, 2008, at 10:04 PM, George Huang wrote:
> >
> > > Dear all:
> > >
> > > I have a database with 72,878 observations which can be grouped into 
> > > 2,800 average time of different activities. Therefore, there are  
> > > 2,800 means and 2,800 S.D. for the activity time. How can I know the 
> > > reliability and validity of each activity time? Should I use 
> > > nonparametric statistics or chisquare test?
> > >
> > > Thanks a lot for your help!
> > >
> > > George Huang
> > > *
> > > *   For searches and help try:
> > > *   http://www.stata.com/support/faqs/res/findit.html
> > > *   http://www.stata.com/support/statalist/faq
> > > *   http://www.ats.ucla.edu/stat/stata/
> > *
> > *   For searches and help try:
> > *   http://www.stata.com/support/faqs/res/findit.html
> > *   http://www.stata.com/support/statalist/faq
> > *   http://www.ats.ucla.edu/stat/stata/ 
> *
> *   For searches and help try:
> *   http://www.stata.com/support/faqs/res/findit.html
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/
*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/