Office Of Accreditation

UMMC / Education / Academic Affairs / Institutional Research

Test Scoring and Analysis

Whom should I contact about test scoring?
Contact Ruby Smith at 41202 or by email.
Where do I pick up the test answer sheets?
From Ruby Smith in Room U170 (across the wide hall from the entrance to the library). Before coming, please Contact Ruby Smith at 41202 or by email.
The class has taken the test. How do I communicate the correct answers to Institutional Research?
To identify the correct answers, bubble in the following on a blank answer sheet:
a) the correct response to each item
b) ZZZKEY in the first six columns of the NAME box.
c) 000000001 in the first 9 columns of the IDENTIFICATION NUMBER box.
The course director wants to have results reported for the different parts (instructors, objectives, etc.) How do I communicate the parts to Institutional Research?
To identify different parts, bubble in the following on a blank answer sheet:
a) A for all the items associated with the first part (instructor, objective), B for the items associated with the second part (instructor, objective), etc.
b) ZZZPARTS in the first eight columns of the NAME box.
c) 000000002 in the first 9 columns of the IDENTIFICATION box.
The course director wants to vary the credit students receive for given items, e.g., have some questions given 1 point, some 2 points and some 3 points. How do I communicate the question weights to Institutional Research?
To vary the credit students receive for given items, bubble in the following on a blank answer sheet:
a) A for all items that you wish to assign a weight of "1" and B for all items that you wish to assign a weight of "2", etc.
b) ZZZWEIGHTS in the first ten columns of the NAME box.
c) 000000003 in the first 9 columns of the IDENTIFICATION NUMBER box.
The course director wants to give credit to all students for a question. How do I communicate this to Institutional Research?
1) On the answer key (ZZZKEY) as described above, leave the bubble blank beside the question number for which you want to give credit to all students.
The course director doesn't wants to give credit to any students for a question. How do I communicate this to Institutional Research?
Prepare a weight sheet (ZZZWEIGHTS) as described above but leave the bubbles blank beside the item for which you do not want to give credit.
Where do I drop off the completed answer sheets?
Leave the answer sheets in the In Box above the table in the lobby of U170 (across the wide hallway from the entrance to Rowland Library).
How does Institutional Research know who the answer sheets are from?

Always attach a Test Scoring and Analysis Request Form. They are available in U170 and may be duplicated.

Can I receive my test report by email?
Yes. You may request that the test report be sent by hard copy or by email according to the policies of your school.
On the Test Statistics Page, what is the Reliability KR-20?

The Kudor-Richardson formula 20 is a measure of reliability which indicates inter-item consistency. The value of KR-20 ranges between 0 and +1.00. If a test consists only of items that measure the same type of material and which require the same kind of thinking behavior from the students the KR-20 should be close to 1.00.

If a test consists of items that measure the different types of material and which require different kinds of thinking, the KR-20 will be closer to 0.00.

The KR-20 can be affected by the number of items on the test. Longer tests tend to have scores closer to 1.00. Occasionally part scores will have higher KR-20's than the total test because the parts have more homogenous items that the test considered as a whole. The higher the KR-20 value the more confident the instructor can be in making judgments about the test results.

On the Test Statistics Page, what is the Standard Error?

The Standard Error of Measurement is a second measure of reliability. A student's score is made up of a "true" score and a number of chance factors. These chance factors may raise or lower the student's score from his/her true raw score in an unknown fashion.

The instructor can assume with 68 percent confidence (2 times out of 3) that the student's true score is in a range of the achieved score plus or minus the standard error. For example, if the student' s score on the score sheet is 82 and the standard error is 4.5, then about 2/3 of the time, the instructor can assume that his score without any chance factors would be between 77.5 and 86.5.

To be 99% certain, the instructor would have to increase the range to three standard errors above and below the achieved score.

A small standard error is preferable to a large standard error.

On the Item Analysis Page, what is the Discr Index and the R Point-Biserial?

Both these measures are designed to indicate to the instructor if students who scored well on the test as a whole correctly answered a given item and if students who scored poorly on the test as a whole answered a given question incorrectly. The values range from +1.00 to -1.00.

A +1.00 indicates that the students who scored well on the test as a whole correctly answered the item and the students who scored poorly on the test as a whole answered the item incorrectly.

A 0.00 indicates that the students who scored well on the test as a whole responded to the item correctly at the same rate as did students who scored poorly on the test as a whole.

A -1.00 indicates that the students who scored well on the test as a whole failed to answer the item correctly and the students who scored poorly on the test as a whole answered the item correctly.

If the numbers of students taking the exam and/or the number of items is less than 50, the R Point Biserial should be used.

If the numbers of students taking the exam and/or the number of items is less than 50, the discrimination index should be used.

Who should I contact to find out more about test scoring?
Contact Ruby Smith at 41202 or by email.
What should I do if Ruby Smith, who normally grades the tests, is out?
Drop off the test in the usual manner. Someone else in the department will grade the test.
What if I need help with test construction or inteterpreting test results?
Contact Dr. Olson (GroupWise or ext. 41202) or Dr. Ibrahim (GroupWise or ext. 41197) re test construction, administration, and interpretation of test results.