Word Stem Completion

[jswissmisspriss]

Hello,

I only found one post on word fragment completion. I am instead trying to use a word stem completion task (where, based on the stem of a word, you fill in the first word that comes to mind. Words here may vary in length, so you don't want a set number of blanks).

Does MediaLab allow you to do this? Also, will MediaLab code for you (for example, could it code, based on set parameters (obviously the words you are anticipating would have to be set), that the filled in word stem creates an appearance-related word versus a non-appearance related word?)

Thank you -- I'm not finding this answer anywhere.

[jarvis24]

Trying to picture the format you want to use for this. First thing I imagined was a fill-in-the-blank item where the question wording was the stem. Coding could be possible but it would depend. Usually MediaLab doesn't re-code automatically by any kind of open-ended content criteria. But it might be possible using the Advanced Features (see user's guide for detail). Want to give a concrete example of the recoding you are describing?

[jswissmisspriss]

Hello,

Thank you, first of all, for replying to my post.

I, too, thought fill-in-the-blank might be my best option, but at this point, I'm wondering if it's really a better option than paper-and-pen.

So, here is the task:

I am priming people with either media portrayals of hooking up or commitment.

In the stem completion task, I'm hoping to evaluate the prime. Therefore, each stem could potentially either be completed with a word that is unrelated to the prime or the prime (hooking up or commitment).

For example, the participant would see the word stem HOO ____

They could potentially fill it in with a word like "hood" or "hookup."

I would want to code for whether the filled in stem equated to a prime-related word (I could, potentially, anticipate what words could be filled in for parameters) or a non-prime-related word. I'm not sure if MediaLab allows for anything this advanced, though.

I'm thinking, though, that any kind of coding by a computer risks that unanticipated answers may slip through the cracks.

What do you think?

[jarvis24]

Well you could certainly write a custom item to do that but, if it were me, I'd simply do it as a recode in SPSS (or whatever). "if stem01 = "hood" then prime = 2" or that kind of thing. Can't imagine a faster solution and would certainly beat coding manually. Hope I'm understanding what you mean.

[jswissmisspriss]

Hello Again,

This sounds like a good plan, actually. I hope this is not too basic of a question, but how would I actually go about doing a recode? I know I would probably do a simple fill-in-the-blank for the task, but how does one go about recoding responses? Is there somewhere in the manual I should refer to?

Thank you again -- this is incredibly helpful!

[jarvis24]

Can't recode open-ended items in MediaLab. But you can in most data analysis programs like SPSS and SAS etc. Just look their user guides for the keyword "RECODE". It's usually a pretty simple syntax.