SAS Data Integration Studio, DataFlux Data Management Studio, SAS/ACCESS, SAS Data Loader for Hadoop and others

which SAS procedure to use?

Posts: 0

which SAS procedure to use?


I have a question about selecting appropriate procedure of data analysis for my data.

I am studying the biology of an insect which feeds on cereal crops over cereal growing season and want to study the effect of temperature variations over the season on the abundance of this insect in cropped fields. The data have two variables: insect count and average temperature for the particular sampling week.

Insects were collected weekly starting from April 1st 2010 to August 31 2010 in randomly selected cereal fields. On each weekly sampling date we performed a sweep survey using insect sweep nets using specific number of sweeps (usually 200 sweeps per field). The insects so collected were used as the variable "insect count" which was actually the average of number of insects collected/ no. of sweeps used per field. As mentioned before, I had the temperature data for every sampling week. for example- If my sampling period was 1 April -7 April 2010, I calculated the daily average temperature for this week and so on for each sampling week.

Now, I have to analyze the effect of temperature on the abundance of insect over a season. Normally, it is found that as the temperatures rise the insect abundance increases too. Thus, there exists linear relationship between temperature and insect abundance and simple linear regression can be used to analyze this.

In my case, however, the insect counts in most cases were 0 as the insect abundance for that particular year were very low in general. Due to this, I could not have my data normal and could not perform linear regression. I read about the SAS procedure on count data which is Proc Countreg and learned that with more 0 counts in data such distributions are Poisson or Negative Binomial Distributions and I can opt for zero-inflated negative binomial distribution models. Although I have insect count data, I have calculated the average of insects trapped per specific number of sweeps used making it continuous data. Can I use Proc Countreg or Genmod to study the effect of temperature on insect abundance as I want to in this case? Is this an appropriate procedure for my experiments?

Many thanks for your time and help with this. Message was edited by: swar
Ask a Question
Discussion stats
  • 0 replies
  • 1 in conversation