Programming the statistical procedures from SAS

Designing an Experiment

Reply
Regular Contributor
Posts: 221

Designing an Experiment

I am quite unfamiliar with designing experiments.

I have an oven with 10 shelves.  Each shelf has a tray of 2000 viles.  It is only feasable to test about 30-40 viles per batch.  There will be 3 batches at one week apart.  This gives 90-120 possible viles.  The viles gather moisture inside, and I need to find if there is a difference in moisture accumulation throughout the oven (there shouldn't be).  In other words, does moisture gather more if it is near the top vs. the bottom, in the front vs. the back, on the left vs. the right, or in the center of a tray, etc.

My thought was to select 2 shelves (one at one-third down from the top and a second two-thrids down).  I would then select 4 sites on each of the two trays (8 sites total).  The sites would be like drawing a tic-tac-toe board inside a tray and selecting the 4 spots where the lines intersect.  At each site, I would select 5 viles that are all next to each other.  This would give me 40 viles per batch.

My first question is, does this design even make sense for my purposes?  Second, what is the name of this design so that I can run the correct ANOVA on it?  My first thought was to do a single-factor design with site as the factor with 8 levels, viles as replicates with 5 levels, and block on the 3 batches.  Then I wondered if I should do a 2-factor design where the two factors are shelf with 2 levels and site with 4 levels per shelf.  Am I on the right track?

VariableLevelsVariable Type
Batch3Blocking Factor
Site8Factor
Vile5Replicates
Moisture40Response
Ask a Question
Discussion stats
  • 0 replies
  • 104 views
  • 0 likes
  • 1 in conversation