Assume that a researcher is interested in developing a measure that can be used with adolescents to estimate their levels of stress due to exposure to traumatic events. Her interest is in using the measure to determine if therapy helps to alleviate stress associated with the experience of traumatic events.
Also assume she creates a measure with 10 items and administers it to a group of adolescents from a single inner city middle school in a high crime area. The measure is given to the same group of students at two time points spaced 12 months apart resulting in the following total scores at each time point:
Student ID Time 1 score Time 2 score
_________________________________________
1 13 18
2 14 20
3 17 14
4 20 22
5 24 25
6 25 13
7 27 34
8 31 36
9 32 25
10 35 27
_________________________________________
Assume that none of the students received therapy or any other type of intervention in between the two test administrations.
Using only the pretest data on the 10 items, the researcher estimates internal consistency reliability using Cronbach’s alpha and finds it to be .77.
Discussion points (REMEMBER to provide a rationale for all of your answers):
A. What type of measurement error does Cronbach’s alpha index? Is the value obtained by the researcher acceptable?
B. In class we talked about two types of reliability requiring two test administrations. For this example, calculate the correct type of reliability (either alternate forms or test-retest) using the time 1 and 2 scores above.
C. Tell what type of measurement error is indexed by the reliability coefficient you calculated for B and tell if the coefficient is at an acceptable level.
D. Based on your estimate of reliability, do you think it would be a good idea for the researcher to use this test to determine if a form of therapy designed to reduce stress is effective?
E. Was the time interval between the researcher’s test administrations reasonable or should it have been longer or shorter? Why?
F. Describe a procedure that the researcher could use to evaluate the construct validity of the test.
G. Describe a method that could be used to demonstrate concurrent validity.
