Dear statalisters!
I am going to run an economic experiment, but I have troubles calculating the sample size I'll be needing. Some general information about the experiment:
* There are two types of participants, let's call them A and B.
* Participants will allocate money. One of the possible allocations is some sort of investment which benefits both A and B (sort of public good).
* Three treatments:
1. Participant A decides allocation
2. Participant B decides allocation
3. Participant A and B needs to agree upon the decision.
What I want to compare is how much of the money they use invest in the different treatments.
I already did a pilot with the
Treatment 1: 10 pairs
Treatment 2: 10 pairs
Treatment 3: 11 pairs
In each treatment they allocated 6000. Of that, the following was used for investment in each treatment (on average):
Treatment 1: 2000 (33% of total amount)
Treatment 2: 1650 (27.5% of total amount)
Treatment 3: 1230 (20.5% of total amount)
How can I calculate the sample size I need? Especially, I need to keep in mind that for each treatment and need two participants per observation (and each participant only participate in one treatment). This implies that some participants will come to the experiment and not decide anything at this game (however, if I just ignore this for the moment and calculate how many pairs I need in each treatment, I can of course, just multiply by 2 to get total sample size).
I do not need to compare all the three investments at the same time, only whether T1=T2, T2=T3 and T1=T3. I want a power of 0.80 and an alpha equal to 0.05.
I am going to run an economic experiment, but I have troubles calculating the sample size I'll be needing. Some general information about the experiment:
* There are two types of participants, let's call them A and B.
* Participants will allocate money. One of the possible allocations is some sort of investment which benefits both A and B (sort of public good).
* Three treatments:
1. Participant A decides allocation
2. Participant B decides allocation
3. Participant A and B needs to agree upon the decision.
What I want to compare is how much of the money they use invest in the different treatments.
I already did a pilot with the
Treatment 1: 10 pairs
Treatment 2: 10 pairs
Treatment 3: 11 pairs
In each treatment they allocated 6000. Of that, the following was used for investment in each treatment (on average):
Treatment 1: 2000 (33% of total amount)
Treatment 2: 1650 (27.5% of total amount)
Treatment 3: 1230 (20.5% of total amount)
How can I calculate the sample size I need? Especially, I need to keep in mind that for each treatment and need two participants per observation (and each participant only participate in one treatment). This implies that some participants will come to the experiment and not decide anything at this game (however, if I just ignore this for the moment and calculate how many pairs I need in each treatment, I can of course, just multiply by 2 to get total sample size).
I do not need to compare all the three investments at the same time, only whether T1=T2, T2=T3 and T1=T3. I want a power of 0.80 and an alpha equal to 0.05.
Comment