So, this is the question. I have recorded the behavior of individuals every 30 seconds and I want to compute the bout duration averages dedicated to each type of activity. I have already included this data in an excel file.
As I said, I prefer to do it in R, but recommendations on any other program will be welcome.
Thanks
What if your n? Once upon a time we did this by hand:
On SPSS you go to
1.analyze, the select descriptive statistics, then descriptives
2. click on your continuous variables and move them to the variables box
3. Click on options button. Make sure Mean, standard deviation, minimum, maximum are ticked and then click on skewness and kurtiosis
4. Click on continue and then on Ok.
We need more information. Do you have a 1/0 every 30 sec for each type of behavior? Is each row coding a 1/0 for n behaviors by individual y? Or, is each row coding individual y on behavior i?
If it's 1/0, then you can either sum the number for each behavior and do a Poisson regression as a function of behavior classification (glm(count~behavior, family=Poisson) in R) or a multilevel logistic regression.
The issue you'll face is that count/duration data will be heavily skewed so a standard arithmetic mean will be inappropriate. If you only want descriptive stats and not a test of difference, use geometric means of the counts/durations. This can be done in R on a column of counts by creating a new column (log.count
- Badges
- Science topic