Hello. I would really appreciate some feedback on this:
-Between- subject experiment with two conditions A & B: Each condition has separate 10 teams(Player 1 & Player 2 make a team)Pairs in both conditions(control & experiment) play a game. There are 20 trails per condition. That is I will have 20 data points from the same player and from the same team.
Hypothesis: The score of group A is higher than Group B. Can I use mixed modeling with random intercept as Player and team as intercepts?
What are the other options than one-way Anova or a non-parametric hypothesis test to analyze this data?
Hi, since nobody else answered, I will give it a try.
If I understand your question correctly, this might be a reasonable first model to try. I however cannot completely parse your question - is the A&B condition the control&experiment you talk about later? Or are those different dimensions? Do you observe a player only in a single team or in more teams? How many players are there?
@martinmodrak Thank you answering( I didn’t expect any at this period of time)
So A is my control condition and B is my treatment condition.
Let’s say totally there are 20 teams ,10 in each condition.
Each team consists of 2 players.They play over 20 trails.The no of trails (here 20 ) is constant in both the control and the treatment.
I don’t get the dimension part of your question here. As I wrote before its a between subject design which means that I observe each team(player1 & player 2) either in A or B.They can’t be in both.
Should I also include trails as random slope here along with looking at Team and player as intercept?
Then your proposed model might work well. However, good practice is to not rely on intuitions and perform posterior predictive checks to test whether your model is adequate for your data (see pp_check and browse these forums for more details)
I would do that only if the model without those fails to capture some aspect of the data (as determined by a posterior predictive check).