On 10/6/2012 7:21 AM, Anoop Balachandran wrote:
> I had 2 simple stats questions:
>
> *1.* In our lab, we did grip tests and we had both males and
> females. After the test, we see the average (3 measurements) of the
> males and females to differ. If so, can we say that there is a
> difference between male and female grip strength *in our sample *(
> not trying to generalize to apopulation)? If we say that there is a
> difference, what if it was just due to chance?
I'm not sure I follow. Do you have three males and three females each
with one grip strength measurement, or do you have a different number of
males and females, each of who repeats the grip test three times?
There is a difference in the sample if there is a difference in the
sample. It requires just looking at the data. If you want to generalize,
then you need a t-test, assuming one measurement per patient. With three
measurements per patient, it gets a bit trickier, but you can always
average the three measurements per patien
>
> *2.* To use correlations, do we need a normal data? I am trying to
> compare 2 variables collected from our lab. And I am not trying to
> infer anything about the population. I just want to see if there a
> correlation for my variables in my subjects. I have read that
> correlation is calculated from Z scores so I am guessing it has to be
> a normal curve.
Computation of a correlation does not require any assumptions about the
data, but any p-value or confidence interval calculation would require this.
But I'm a little bit concerned. Twice now, you've mentioned that you
don't care to extrapolate beyond the patients in your sample. There are
settings where extrapolation to a larger population is not interesting
or important, but they are fairly rare. Unless your sample was very
special, like the King and Queen of England, I can't imagine why you
wouldn't be interested in extrapolation.
> Thanks you and sorry for these silly questions
Not a silly question. A silly question would be "Why don't they invent a
cordless extension cord?"
Steve Simon, [log in to unmask], Standard Disclaimer.
Sign up for the Monthly Mean, the newsletter that
dares to call itself average at www.pmean.com/news
|