| Reference | Downloads | Github

A preference Judgement test

Mac OS Sierra 10.12.4
PsychoPy version 1.84.2
Standard Standalone.

Hello everyone!

I’m trying to build a test for Preference Judgement.

It consists two parts:
In the first part, the participant rates different items between 0-1 (I used the visual analog scale, and a picture to represent the item).
On the second part the participant is making choices between two pictures - one of a highly rated item (0.7 and up) and one of a poorly rated item (0.3 and below), using the left and right arrows. I also inserted a “correct answer”, which supposed to be the highly rated item.
The items are divided into different categories, and the choice has to be made between items on the same category (for an example - fruits, cellphones, car brands, etc.)

I built the first part, and it generates a data file that consists the different items and their picture name, their category and the rating that they got.
I also built a draft of the second part, but I don’t know how to “connect" the two parts - Ideally, the program will choose automatically highly rated and poorly rated items from the same category and compare them on the second part. Right now I have no idea how to do that on the builder, and I don’t know how to code, so I’m doing it manually…

I would very much appreciate the forum help with the “connecting” issue.

Thank you very much!!


Hi Tiran,

This sounds like an interesting experimental design. The good news is that you have implemented the first part in Builder and got it working. Other good news is that doing the second part will be very similar. The bad news is, as you suspect, that joining the two stages is going to be tricky.

This will inevitably require some Python coding to do what you want. It could just be a matter of going through the data from part 1, and constructing a new conditions file that could be used for part 2 for each subject, so that the second part would still use a standard Builder approach. This linking code won’t be straightforward, particularly because your data is based on subject responses. e.g. you can’t guarantee in advance that there will be the same number of items rated above 0.7 as there are above 0.3, and you might even need to cater for cases where there are no judgments in one (or both) of those categories, or in the different picture types. i.e. this code will need to be very flexible and fault-tolerant, unlike a lot of the relatively mechanical code we can write when we know that all of the experimental conditions are nicely balanced.

So to implement this will either require you learning how to program in Python, making friends with someone who can, or looking to see what technical support you can access (e.g. does your department have any programming technical support available)? Unlike a lot of the problems that get solved on this forum (with relatively simple solutions), I suspect that this code would need a continuous cycle of writing and then testing to make sure it does what is needed, and that is hard to do remotely.

Whichever way you go, the process will require a very careful description of the problem and what needs to be achieved. You could try doing that here and see if anyone can at least help you along the way. There might even be a PhD student here who is desperately avoiding their own thesis work and might be prepared to do some remote support :wink:

I’ll try to find someone who know how to code!.. Maybe I’ll gain a new abillity on the way haha

Thank you for your answer :slight_smile: