Millisecond Forums

I've got a question regarding the randomized presentation of 8 versions of an Ebbinghaus task.
The participants are presented these tasks in a randomized order and they are supposed to adjust the red element to the correct size. Up to this point, the adjustment always takes place on the right side. To rule out differences between the hemispheres of the field of vision I would like to randomly let the adjustment take place on both the right and the left side. It is important though, that of the 8 trials exactly 4 are adjusted on the left side and 4 on the right side, while kepping the order randomized. Also one of the eight versions of the task should be presented either with adjustment on the right or on the left (not on both sides, I want to remain with only 8 trials i.e. every version once), so that for example one participant sees the version with the adjustment in the big flower with adjustment on the left and the next participant sees it with adjustment on the right. The sides where the adjustment takes place is the one where the picture has an arrow added to it and where the red element with the dynamic value in the /size attribute is situated.
To achieve this I implemented a list element with the /items (1,1,1,1,2,2,2,2), wrote two values, namely /adjustside and /compareside and then wrote /ontrialbegin attributes in each trial which give these values one of two different values, depending on which item is drawn from the list. The values are used in the /hposition attribute of the elements I want to position dynamically and represent either the right or the left side of the screen (indicated by (display.width/2) + expressions.vectorposition respectively (display.width/2) - expressions.vectorposition).

You can see how it should look where I didn't try the randomisation yet and you can sadly see how it shouldn't whith the trials where I tried the randomisation. Somehow the elements are all in the left corner/outside of the screen.
I would greatly appreciate if you could help me out here. I don't know if this approach is usable at all, but it was the only one I could come up with.

The file with the pictures is attached.

Best regards,
Pablo