Millisecond Forums

Randomization and Subject ID

https://forums.millisecond.com/Topic30879.aspx

By clynzee - 2/20/2021

I currently have a study in which participants are directed from Qualtrics, to Inquisit, then back to Qualtrics. Everything is quite seamless, but I have a few issues that I need some help with before I send this out for participants to take part in:

1. Participants are intended to be randomly assigned to one of two conditions: 

<batch>
/ groups=(1 of 2)
/ file="stroop_deplete.iqx"
/ file="stroop manipulation instructions.iqx"
/ file="pasatc.iqx"
</batch>

<batch>
/ groups=(2 of 2)
/ file="stroop_nondeplete.iqx"
/ file="stroop manipulation instructions.iqx"
/ file="pasatc.iqx"
</batch>

It was recommended to select either Sequential or Random without Replacement for how groupid should be determined, and I have tried both ways (granted, not with a lot of participants) and I would like to know what might be best for what I am trying to do. I notice that most people who I have tried this out on have been assigned to group 1 (only 1/5 people were in group 2). I intend to collect data from ~215 participants, and I would like for the groups to be about even. It may be do to such a small sample, but can anyone tell me the difference between exactly what each option (Sequential or Random without Replacement) would do for my study?

2. Regarding the subject id--I have my survey broken into 2 parts (one Qualtrics survey for each), and I have my millisecond URL in the first Qualtrics survey set to send subject id. I have checked the data for all parts of my study, and while the first Qualtrics survey and the Inquisit data have matching subject/response id, the second Qualtrics survey has a different response (subject) id than the other two parts. Does anyone know how to make it so that the same ID carries through all 3 parts (Qualtrics-Inquisit-Qualtrics)?
 
3. Finally, I wonder if there may be a way to make the subject id numerical (and sequential) as opposed to a long alphanumeric string. This may result in a lot to sift through when all of my data is collected.

Any help is appreciated! Thanks!
By Dave - 2/21/2021

 
clynzee - 2/20/2021
I currently have a study in which participants are directed from Qualtrics, to Inquisit, then back to Qualtrics. Everything is quite seamless, but I have a few issues that I need some help with before I send this out for participants to take part in:

1. Participants are intended to be randomly assigned to one of two conditions: 

<batch>
/ groups=(1 of 2)
/ file="stroop_deplete.iqx"
/ file="stroop manipulation instructions.iqx"
/ file="pasatc.iqx"
</batch>

<batch>
/ groups=(2 of 2)
/ file="stroop_nondeplete.iqx"
/ file="stroop manipulation instructions.iqx"
/ file="pasatc.iqx"
</batch>

It was recommended to select either Sequential or Random without Replacement for how groupid should be determined, and I have tried both ways (granted, not with a lot of participants) and I would like to know what might be best for what I am trying to do. I notice that most people who I have tried this out on have been assigned to group 1 (only 1/5 people were in group 2). I intend to collect data from ~215 participants, and I would like for the groups to be about even. It may be do to such a small sample, but can anyone tell me the difference between exactly what each option (Sequential or Random without Replacement) would do for my study?

2. Regarding the subject id--I have my survey broken into 2 parts (one Qualtrics survey for each), and I have my millisecond URL in the first Qualtrics survey set to send subject id. I have checked the data for all parts of my study, and while the first Qualtrics survey and the Inquisit data have matching subject/response id, the second Qualtrics survey has a different response (subject) id than the other two parts. Does anyone know how to make it so that the same ID carries through all 3 parts (Qualtrics-Inquisit-Qualtrics)?
 
3. Finally, I wonder if there may be a way to make the subject id numerical (and sequential) as opposed to a long alphanumeric string. This may result in a lot to sift through when all of my data is collected.

Any help is appreciated! Thanks!

> It may be do to such a small sample, but can anyone tell me the difference between exactly what each option (Sequential or Random without Replacement) would do for my study?

Sequential selection is just that: The first participant visiting the start page gets group 1, the 2nd gets group 2, the 3rd again gets group 1 again, and so forth. Note that every visit to or refresh of the start page will trigger the assignment, regardless of whether that particpant actually ends up actually launching the study.

With random selection without replacement, for each set of two participants visiting the start page, one will randomly be assigned to group 1, and the other will be assigned to group 2. Here, too, every visit or refresh will trigger the assignment.

Participants visiting, but not actually completing the study will lead to imbalances or gaps.

If you prefer, you can also have Qualtrics perform the group assignment and have it send over the groupnumber in a URL parameter, just like the subject identifier.

> 2. Regarding the subject id--I have my survey broken into 2 parts (one Qualtrics survey for each), and I have my millisecond URL in the first Qualtrics survey set to send subject id. I have checked the data for all parts of my study, and while the first Qualtrics survey and the Inquisit data have matching subject/response id, the second Qualtrics survey has a different response (subject) id than the other two parts. Does anyone know how to make it so that the same ID carries through all 3 parts (Qualtrics-Inquisit-Qualtrics)?

See https://www.millisecond.com/forums/FindPost29950.aspx

> 3. Finally, I wonder if there may be a way to make the subject id numerical (and sequential) as opposed to a long alphanumeric string. This may result in a lot to sift through when all of my data is collected.

This is a Qualtrics question in this case, which you should direct to Qualtrics support.