2

We have $n$ observations of human performance before and after a training program. We have a random sample of $n$ individuals and assume population distributions are normal.

We thus have: $x_{1,t},\ x_{2,t},\ ...,\ x_{n,t}$ and $x_{1,t+1},\ x_{2,t+1},\ ...,\ x_{n,t+1}$. It is clear from the example that ${\rm Cov}(x_t,x_{t-1}) \neq 0$.

What is the convention and reasonable assumptions on the variance of the difference $d = x_{t} - x_{t+1}$?

$$\widehat{\rm var}(d) = \widehat{\rm var}(x_t)+\widehat{\rm var}( x_{t+1}) - 2\widehat{\rm cov}(x_t,x_{t+1}) ?$$

Or assume constant variance (homoscedasticity) over time?

$$\widehat{\rm var}( d) = 2\widehat{\rm var}(x) - 2\widehat{\rm cov}(x_t,x_{t+1}) ?$$

Is it correct to simply take the variance of $d$? How do you recommend start thinking (rigorously) about a problem like this?

gung - Reinstate Monica
  • 132,789
  • 81
  • 357
  • 650
snoram
  • 437
  • 3
  • 11
  • 1
    Thank you. It looks like you are asking people to offer opinions about changes in weight within some undescribed experiment. By providing no information about that experiment (is it a force feeding of mice? Observation of growth of human infants? Charting growth of normal human adults?) you can expect a whole range of answers, any of which might be correct. – whuber Oct 22 '14 at 15:26
  • The thing is I don't have much more information myself. It is observations of human performance before and after a training program. We have a random sample of n individuals and assume population distributions are normal. – snoram Oct 22 '14 at 15:33
  • 1
    Please, then, include that information in your question. Consider opening it up a little bit and asking readers about how one might go about deciding whether such data are homo- or hetero-scedastic. That might produce more useful and objective answers compared to asking vaguely about "conventions" and "assumptions." – whuber Oct 22 '14 at 15:39
  • I deleted it on purpose, I was stuck when I wrote it but after a bit of a rest I got past my problem and saw that my method 1 and 3 are essentially the same! so the only discussion left is about hetereoskedasticity, which I guess has a better place (and is unnecessary for me). – snoram Oct 23 '14 at 11:41

0 Answers0