Representative, big sample

Hello there,

we need a representative sample of 4000 persons. Does anybody have an idea to work around the 1500 persons barrier? Or did anbody something similar in the past? Could need some advice and help, thanks :slight_smile:

1 Like

@Community_Leaders any thoughts? :slight_smile:

1 Like

hi @Elena_Groben , off the cuff, one opinion would be to run three studies and test whether your models differ in any substantive way by “wave”. But that won’t help if the pool gets too low - so that’s something to consider.

Hey @P.P . Thank you for your answer. I have also found one of your FAQs pages that I could help myself by copying the survey, but the redirected link then said that duplicating representative studies would not be possible:

“Unfortunately it is not currently possible to duplicate representative sample.”(Representative samples – Prolific) studies or studies that were published using our ‘balance sample across sex’ feature.

Hi @Elena_Groben , you’re right. The trick on the setup side is that you would need to make a separate study (rather than copy the existing one), and use a prescreener based on participation in the first study (then for the third, you would need to screen for both one and two). So you would run the studies back-to-back, after creating each one separately but making them the same. The unknown here is that when you go over 1500 participants, there may not be an actual representative sample left in the participant pool who’s willing to do the study. There are a lot of users, so it could be more than 1500, but that’s the “Prolific maximum”. Does that help at all?

Dear Elena
Welcome to the forums m(._.)m
You could, in theory at least, create your own representative sample. Prolific explains how they do it here

To get around the restrictions on duplicating studies with representative samples you can create drafts which you duplicate and add the balancing in the final stage (as suggested by Pau Alexander).

If you make a draft with all fields filled in except the age and race demographic that you want to focus upon, duplicate it (9 times 9 age brackets, and 5 races that Prolific uses, using the same data that Prolific uses) and then change the age ranges, then duplicate each of these 9 age range drafts again and add the five races, and finally add balance for sex, or enter the number of males and females you require again based on the demographic data from the US Census Bureau or the UK Office of National Statistics.

Bearing in mind that males and females are not equally prevelant in all the age ranges in the actual population I am afraid to create the same sort of representative sample as Prolific provides you would need to create 2 (sexes) x 9 (age groups) x 5 (races) which is 90 duplicates, but in theory at least it could be done for all 4000 persons. Whether there are enough participants on Prolific to fill up those slots however, I am not sure.

Tim

Dear @timtak
thank you very much for your answer, your explanation is very easy to understand, I’ll figure out if it could work that way, it’s worth a try :slight_smile:

But one question: As far as I’m informed, Prolific uses 5 age brackets for their representative sampling or am I wrong?

Hi Paul,

so you say I could run three representative samples when I duplicate the draft 3 times and then I prescreen for the other two “studies” in order to prevent people from participating again?

Elena

Glad to have been some small help.

Yes, you are right “five 9-year brackets” (not 9 five year brackets as I misread it) :woozy_face:

Tim

1 Like

Dear Elena

Jumping in, if I may.

You can only prescreen for studies after the other studies have finished. There is no way of doing this in real time, like dealing cards (participants) to poker players (studies).

You can prescreen for studies by selecting Participation on Prolific and Previous studies and the study but the study is only selectable after it is finished.

You can on the other hand prescreen for participants in studies as soon as you have their Prolific ID. But if you released more than one study simultaneously, participants could enrol in more than one before you add their ID to a custom blacklist.

So ordinarily studies are carried out in series, or the “dealing” is carried out by a survey platform.

Tim

1 Like

hi @Elena_Groben (& @timtak ),
yep, you should be able to run the same representative study three times, as Tim mentioned, sequentially. You would wait until the first “wave” is complete, then submit a request to Prolific to start the “second wave study” (which is the same as the first) but with a screener based on participation in the previous study. Then the “third wave” is the same except based on the previous two studies.

(you can’t include the prescreener yourself in representative samples which is why you have to request Prolific to initiate the second and third studies)

You also could run a representative sample as mentioned above in which you create many studies with their own custom criteria. However, neither approach can get around the fact that there may or may not be a representative sample of that size in your target population on Prolific. When you create the studies with the screeners as Tim mentioned, you will see how many Prolific users there are who meet the criteria, so that is one advantage to trying that approach first. If you find a group that does not have enough people, you’ll know there is a limitation.