Durbin-Watson Test

[mr_bennyc]

Hi,

I'm looking to analyse some vigilance data from monkeys. The data at the moment comes from 5minute samples, but these are separated depending upon which activity is occuring. There's also data on different heights, individuals nearby etc. So data might look like this:

Date Time Activity Height Time vigilance
06/02 1300 Feeding 4 120secs
06/02 1302 Resting 4 60secs
06/02 1303 Resting 8 120 secs
07/02 1500 Feeding 0 300secs
07/02 1600 Resting 2 120 secs
07/02 1602 Feeding 2 180 secs

So this means that I'm hoping to use each as a separate data points, even though they might come from the same sample. Therefore I assume there might be a problem with autocorrelation, as some data is obviously coming from the same time period. I understand that the Durbin-Watson test can test for autocorrelation, but I just wanted to check if it is ok to use with this kind of data, where say 3 data points might be within 5mins, but the next ones aren't until the next day?

Hope this all makes sense.

Thanks for the help

[GretaGarbo]

You started a tread yesterday about the same study.
http://www.talkstats.com/showthread....ng-assumptions

There you gave some information there and now you give some other information here. It is very confusing. “Have I not seen this somewhere”?

It is better if you put all information in one thread. Or, if it is a completely different question, at least link to that question. “I am all confused”!

[hlsmith]

Not having seen your other thread and ignoring these data, is you question whether or not the Durbin-Watson test can be used on a variable with unequal time intervals?

[mr_bennyc]

Yes, sorry, this is a different question to the one on another thread. Yep, my question basically is, will the Durbin-Watson test work on data with unequal time intervals?

Many thanks