Millisecond Forums

Response variable saves correct not valid response

https://forums.millisecond.com/Topic14031.aspx

By rogorman - 8/20/2014

It seems that the script I am using prompts the data element to write the correct response by participants, not the first valid response, as far as I can tell. There are two responses that are valid for the bulk of the script, "E" and "I", scan codes 18 and 23. When someone shows as having the incorrect answer, but where they have to then correct their response to advance, the recorded response is the correct one. I got this script from Inquisit's samples, and weirdly the script author notes that the script will record the correct response. However, I cannot see how they can force this, yet Inquisit's support material says that response could capture the first valid response. Any help would be appreciated. I am attaching the script.
By Dave - 8/20/2014

The script's behavior is enforced by the /responsemode=correct setting at the <block> level. Participants are required to press the correct response key before a given trial terminates.
By rogorman - 8/20/2014

Hi Dave,
Perhaps I was unclear. In the datafile, under 'response', the script is delivering the correct (final) response, not the first valid one. This is clear from the responses that have been recorded as incorrect, yet have a response that is the correct one. They must have first pressed the wrong valid one, yet this is not being captured.
The Inquisit support documentation says this:
"A correct response may be given at any time. If a valid but incorrect response is given, the response is treated as an error. The first valid response, correct or incorrect, is recorded in the data file. The recorded latency is the that of the correct response."
This is what I find even for 'response' under 'block' in the support docs.
But this isn't what seems to be happening. Are you telling me that the support docs are wrong on this?
By Dave - 8/20/2014

Thanks for clarifying. You are correct -- and in fact I have had a bug open for this for a while. /response=correct currently isn't working as advertised in the documentation. This isn't a major problem in cases where you only have two response keys (such as yours) because you can fully infer which key was pressed first via the 'correct' column. But it would be a problem for cases where you have 3 or more response keys and the specific key that was pressed first is of interest (because it may indicate the *kind* of error that was committed).

Hope this clarifies.
By rogorman - 8/20/2014

Yes it does! Many thanks Dave. I had been over the script a bit and was baffled as to what I was missing. In my case, as you note, it is okay once I am sure what the problem is.