Friday, December 24, 2004

Christmas Eve

So it is the day before Christmas and here I am, trying to figure out how to analyze my data, again. I wish that one of my committee could tell me if I am doing this right.

Issue 1: Estimate of means and variances -
Calculated with GLM of Trait = rep entry.
  1. Output residuals
  2. Square residuals
  3. Take abs value of residuals
  4. Aggregate file by genotype
  5. Divide square residuals by counts for each genotype
  6. Perform Modified Levene's Test

Issue 2: Do the models fit assumptions of GLM?

  1. Normal residuals - no
  2. Constant variance - residuals are linear with obs value

No, but what do I do?

  1. Transform - Each variable different?
  2. Weighted least squares - What should the calc be for determining weights?
  3. Ignore it all
  4. Run away to become a beach bum in Fiji

Issue 3: How to present in only 10 minutes for two populations, 19 traits

Issue 4: Missing data

  1. When a plant yields 0 should be missing or 0?
  2. What about must pop where there are mostly missing values due to maturity?
  3. Do I include mat score as a covariate?
  4. Ignore it all?

The day is cold and there is even a chance of snow. I am going to go home to get ready for christmas. I need to finish my wife's present; I am making her a necklace for christmas. I need to assemble my daughter's bike.

No comments: