a setitem problem or a selection problem

jens — Thu, 26 Nov 2015 01:59:47 GMT

Hello,

In a dot-probe-experiment, I assign the dots (.. or :) on the basis of a list. If a particular value is selected from the list, either a vertical or a horizontal dot-pair is presented. At least, that is the idea. However, vertical dots are presented more often than horizontal ones (about three times as often as horizontal ones) according to a tally made during the experiment. I tried to check the presentation frequency by storing the picture-filename in a variable. According to the analysis of this variable (values.file.name) horizontal and vertical dots appears equally often. So, where is my error?

RE: a setitem problem or a selection problem

jens — Thu, 26 Nov 2015 01:59:47 GMT

Cool. Jens

RE: a setitem problem or a selection problem

Dave — Wed, 25 Nov 2015 08:33:26 GMT

Jens,

<picture dot.pic>
/items = ("DotHor.png", "DotVer.png")
</picture>

Defines a <picture> element with two items. Since no /select attribute is specified, the default is used -- which is 'noreplace'.

In your <trial> elements, you constantly change what the *first* item in that <picture> element is:

<trial target.neg.left>
/ontrialbegin = [if (list.pos.tar.neg.left.nextvalue == 1)       {picture.dot.pic.hposition = values.left; values.pos = values.left; picture.dot.pic.setitem("DotHor.png", 1); values.file.name = "DotHor.png";}
...
                  else if (list.pos.tar.neg.left.nextvalue == 4) {picture.dot.pic.hposition = values.right; values.pos = values.right; picture.dot.pic.setitem("DotVer.png", 1); values.file.name = "DotVer.png";}]
...
</trial>

That, however, *does not change the fact that there are two items in the element which are being sampled randomly without replacement*. There is no reason whatsoever why only the 1st item -- which is what you set -- should be sampled and thus displayed in any given trial.

Suppose that in the very first trial you run, you set the 1st item to "DotVer.png". Your <picture> element now effectively looks like this:

<picture dot.pic>
/items = ("DotVer.png", "DotVer.png")
</picture>

i.e., you now sample without replacement from *two identical items*. Further suppose that in that first trial the 1st item is randomly drawn. The trial will thus display "DotVer.png".

Since you are sampling *without replacement*, in the *next* trial the 2nd item will *inevitably* be sampled -- which is also "DotVer.png". And this is true *regardless* of whatever you set the 1st item to at the beginning of that 2nd trial. That's where you're apparent "mismatches" come from.

You either need to make sure that the <picture> constantly selects the 1st item as in

<picture dot.pic>
/items = ("DotHor.png", "DotVer.png")
/ select = 1
</picture>

or abandon the setitem() approach, which is what I would *highly* recommend due to performance reasons alone. Instead use a simple variable to select the proper item as in

<values>
...
/ dot.pic.item = 1
...
</values>

<picture dot.pic>
/items = ("DotHor.png", "DotVer.png")
/ select = values.dot.pic.item
</picture>

with

<trial target.neg.left>
/ontrialbegin = [if (list.pos.tar.neg.left.nextvalue == 1)       {picture.dot.pic.hposition = values.left; values.pos = values.left; values.dot.pic.item = 1; values.file.name = "DotHor.png";}
...
                  else if (list.pos.tar.neg.left.nextvalue == 4) {picture.dot.pic.hposition = values.right; values.pos = values.right; values.dot.pic.item = 2; values.file.name = "DotVer.png";}]
...
</trial>

and so forth.

Millisecond Forums » Millisecond Forums » Inquisit 4 » a setitem problem or a selection problem

a setitem problem or a selection problem

RE: a setitem problem or a selection problem

RE: a setitem problem or a selection problem