David Airey
>
> I was speaking to a scientist who records electrical
> potentials from
> living neurons (brain cells) in awake behaving animals. The
> sampling
> rate is very high, such that 100s to 1000s of data points
> are available
> per neuron, and 100 neurons are sampled per animal. There are never
> more than 10 animals in an experiment, which could be a
> problem. The
> levels of the treatment are each exposed to each animal or
> neuron. So
> it's like a nested repeated measures ANOVA, but the sampling rate
> suggests a time series analysis is more appropriate. What
> kind of time
> series analysis _in Stata_ allows grouping of correlated
> measures in
> addition to handling the autocorrelation problem? By
> grouping I mean
> all the neurons in one animal are more similar than neurons between
> animals, and cannot be treated as independent. I've read a
> little about
> time series, but what I've read seems to describe following
> a sample of
> experimental units through time, and doesn't talk about following
> groups of correlated samples through time.
>
This sounds like a job for -xt- i.e. you have panel data
in which each animal is a panel and each neuron is
a separate variable. You might end up working with
reduced variables, e.g. principal components. It
seems to me that the bigger question is the reverse,
i.e. what methods are going to be helpful scientifically, and
can you apply them in Stata.
Nick
[email protected]
*
* 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/
