Sniper-spotter
Problem description
Factors
- Feedback: Private (18 teams) v Public (19 teams)
- Teams have 1 sniper and 2 spotters
- 4 trials for each team:
- first three trials have the same sniper
- last trial has a different sniper
Responses (unclear what type of responses there are)
- some team responses
- some individuals respones
Questions
- Do the teams/individuals learn?
Consulting response
Two possible approaches are 1) simplify and analyze with simpler models or 2) model the whole structure and build contrasts. This experiment is a repeated measures, partial cross-over, split-plot design and thus will be complicated to construct.
Simplify and analyze with simpler models
This approach would involve calculating the appropriate differences to address the scientific questions of interest. For example, to understand the impact of feedback on learning through differences in trial 1 and trial 3 at the team level, you can calculate differences for each team between trial 1 and trial 3 and then you just have a t-test model to compare public vs private.
Model the whole structure and build contrasts
Team responses
A model for team responses will be relatively straight-forward but will need a random effect for team that may call for a more sophisticated error structure than exchangeability.
Individual responses
The individual responses have a split-plot structure with cross-over due to the switching between trial 3 and trial 4.
blog comments powered by Disqus