[Rasch] Using Winsteps to compute scores
Rense Lange
rense.lange at gmail.com
Tue Feb 5 03:05:32 EST 2013
You will not find a straight line, but a fuzzy line at best. Using common calibration (all items) you could use the method shown in BTD to find if you get above chance # of people deviating from y = x. Assuming that the items and students fit of course.
From the iPhone of
Rense Lange
On Feb 4, 2013, at 4:05 AM, Juanito Talili <talilij at yahoo.com> wrote:
> We developed a 36-item self-administered survey instrument that will be used to measure the performance of "recollection facilitators". The instrument consists of five subscales with items that used a five-point ordered response options. For the scoring, it is tedious in the part of the user if the scores shall be computed by subscale. Thus, we plan to rasch analyze the entire scale (the 36 items together) and then produce the scores by respondents across items using Winsteps. Then, from that scores, the user can easily compute the scores by subscale. However, we afraid because Rasch analysis of the entire scale, instead of by subscale, might be wrong. To check if we are not wrong, we do the two analyses as follows:
> Analysis 1: We Rasch analyze by subscale and then produce the scores by subscale.
> Analysis 2: We Rasch analyze the entire scale and then produce the scores by subscale.
>
> Then, for each subscale we will draw scatterplot of the scores generated from the two analysis. Analysis 1 and analysis 2 are the same if the points in the scatterplot will form a straight line. Are we correct?
>
> Any comments are welcome.
>
> Thank you.
> J
> _______________________________________________
> Rasch mailing list
> Rasch at acer.edu.au
> Unsubscribe: https://mailinglist.acer.edu.au/mailman/options/rasch/rense.lange%40gmail.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://mailinglist.acer.edu.au/pipermail/rasch/attachments/20130204/38220d57/attachment.html
More information about the Rasch
mailing list