Using Response File to Skip Between Conditions (case 10536)

[bradley41980]

Hi,
After looking at the forums, I have been unable to find an appropriate solution to the problem I am having. I have an experiment in which I would like to assign participants to a condition based on their answer to a particular question (e.g., gender). That is, I would like participants to note their gender and have MediaLab use that information and assign them to a condition congruent with their gender. However, I would still like the conditions to be psuedo-randomized (i.e., still want equal n's). Currently, I am trying to use a response excel file to skip participants to one of the male or female conditions. The response file contains the gender variable and has a formula in the SkipTo column that randomizes them into one of gender congruent conditions based on their answer: (=IF(B2 = 1, RANDBETWEEN(2,4),RANDBETWEEN(5,7)). My intent is if participants answer 1 (or Male) to the gender question they will be randomly directed to one of the male conditions (i.e., 2-4). Conversely, if they answer 2 (or Female) they will be randomly assigned to one of the female congruent conditions (5-7). However, I am unsure if MediaLab will skip between conditions (rather than items or questionnaires). In sum, I would like to randomize participants to condition based on gender but have been unable to find a solution that will allow this to work. The Advanced example provided in the MediaLab Samples folder answers some questions but it only appears to be skipping within a questionnaire (rather than assigning condition). Any help you can offer is greatly appreciated. Thanks in advance!

[jason_reed]

I'm happy to help with your inquiry. It would be helpful if you could provide some more detail for me to make sure that I understand what you are trying to do.

Am I correct in saying that you have 3 conditions for each gender that you wish to ensure equal numbers of men in each of the 3 men conditions and equal numbers of women in each of the 3 women conditions? Also, could you describe the structure of each of these male and female conditions? That is, is each condition in a separate .que file, are they combined in .que files, etc.?

Thanks!

[bradley41980]

Thanks so much, Jason. I truly appreciate your help.

In the example below I was trying to simplify the design to see if I was able to get MediaLab to do what I needed. The full design is a 3 x 2 design. We would need equal numbers per cell. The catch is that the design requires individuals to be shown stimuli that are consistent with their gender (i.e., males should see male profiles and females should be shown female profiles). Thus, in the full design I have 6 conditions that need to be matched per gender (6 female and 6 male) with equal numbers of men and women in their appropriate conditions.

Right now, each condition is made up of several .que files (i.e., a separate que file per measure). That is a condition may be made up of 5+ .que files.

The goal was to have each participant randomly placed into one of the conditions that is consistent with their gender such that males would be randomly assigned to one of the 6 cells of the design that contain male profiles and females would be assigned to one of the 6 cells of the design that contain female profiles. And, that this condition assignment would take place using the gender information supplied by the subject. So, participants report their gender and are then randomly assigned to one of the gender congruent conditions based on that answer.

[jason_reed]

Do you need to rely on your participants self-reporting their gender in a MediaLab question to make the random assignment? I ask this because if the person running the experiment session can see the participant's gender at the start of the session, would it be more practical to create 12 different conditions in your MediaLab files (6 for the men and 6 for the women) and rely on the person running the session to choose a condition number from a premade table for each participant? The person in charge of the session would be responsible for making sure that equal numbers of men and women are assigned to the 6 conditions for that participant's particular gender.

[bradley41980]

Thanks for the suggestion, Jason. We were trying to avoid this to reduce error with research assistants pressing the wrong gender button and having to decipher whether names like "Jamie" were a male or female prior to the experimental session. However, if there is no easy way to accomplish this using MediaLab we can certainly use a pre-made randomization sheet specific to gender.

Thanks for the help.

[jason_reed]

Thanks for the clarification on the point of manually assigning participants. Let me think on any kind of automated assignment and I'll get back to you.

[jason_reed]

In thinking about your study, the major hangup I can't come up with a way to get past is to ensure an equal number of participants in each cell of your design as participants come into the study. It would be one thing if we knew the gender and final number of participants ahead of time, but I don't know of a way that you could set up the responses.xls formula to keep cell counts even based on the gender information provided as the study is run.