(no subject)
Locked
(no subject)
 
|
Request you to help me on the below mentioned scenario: An educational institution conducts various online exams. As these have been conducted over the years, there are number of questions that have been piled up. Now the requirement is we need to understand how many times a question has been exposed during the exams, as we need to retire those based on the exposure and replace them with the new ones. For this we need to understand the exposure of each question. Request you to please guide me through this by using SPSS.Best Regards. Get your own FREE website and domain with business email solutions, click here
|
Re: (no subject)
|
How is the data stored?
Is each test in an SPSS file? When the question is re-used does it have the same name? Art Kendall Social Research ConsultantsOn 7/29/2013 1:11 AM, Ramya Mridul [via SPSSX Discussion] wrote:
Art Kendall
Social Research Consultants |
Frequency of a Test Item Across Semesters (was Re:
|
|
In reply to this post by Ramya Mridul
The first question you have to answer is each test item/question
identified
with a unique variable name (e.g., CAR8EC04M023 which can be
decoded as: CAR is the first three characters of the authors whose
textbook
was used such as CARROLL, 8E identifies the that item is associated
with
the 8th edition of the text, C04 identifies that the item is for chapter
4,
and M023 identifies the items as Multiple choice question #23). The
above
applies if one is using test items provided by the book
author/publisher.
The problem then becomes how often was a particular item used in
different tests over a period of time. If the variable names are like
I specify
above, then one can use ADD CASES for courses that were supposed to
cover the same material with the same text book. The items common to
the
different files will be identified and then you will be asked if you want
variables
found in one file but not another also included -- respond yes.
After all of the
relevant courses have been put through the ADD CASES, then use
aggregate
to get the N of values for each item which should range from 1 to the
number
of exams used (i.e., the number if an item was on every test).
If the items are simply named Q001 to Q100, then you'll have go through
each of tests and assign a unique name for the item. Then repeat the ADD
CASES
and aggregate procedure.
If the test items did not come from a test bank but were written by the
instructor,
then you'll also have to go through each test and determine a unique
identifier for
each item. Again use ADD CASES and then aggregate for the number of
values
each item has. Each item should have a minimum value of "1"
(one)
indicating that it was used only once and some number greater than
one
if it was repeated in different administrations of the exam.
If my assumptions for the above are correct, somebody is going to
spend
a lot of time trying to uniquely identify each item. But I could be
wrong and
maybe one of the other wizards around here can see a simpler way of
doing
things.
-Mike Palij
New York University
|
| Free forum by Nabble | Edit this page |