Question

Dr. Jimenez is interested in the relationship between depression in adulthood and childhood emotional trauma. He hypothesizes that adults who are depressed may have experienced emotional trauma, such as emotional abuse, earlier in life which predisposed them to developing depression as adults. In preparation to conduct this research, he decides to create his own questionnaire that will measure early childhood trauma (the Childhood Emotional Trauma Questionnaire; CETQ) where he has adults respond to 10 questions (1-10 point scale) about potentially emotionally traumatic events in their childhoods. When he gives the CETQ to a sample of 30 adults (Time 1), and then gives the measure to the same 30 adults again 3 weeks later (Time 2), he finds a correlation of .75 between the two scores. The split-half correlation coefficient for the CETQ is .73 before the Spearman-Brown Formula is applied, while the Cronbach’s coefficient alpha is .70 at Time 1. The standard deviation of Time 1 scores is 7.5.

What types of reliability evidence are addressed. Select all (no partial score will be given).

		Test-retest
		Parallel forms
		Split-half
		Cronback's coefficient alpha
		KR20

Answer 1

Answers: Test-retest, Cronback's coefficient alpha

In test-retest reliability, the same test is given to the same set of test respondents on two separate occasions or two different times and the scores obtained on the two occasions are correlated. This is what is done in the example given above.

In parallel forms reliability, test A is given to a group of people on one day and another test B is given to the same group of people on another day. The scores from test A and test B are correlated. This is not done in our example above.

In split-half reliability, a group of students is split into two groups and both groups take the test at the same time. This is not done in our example above.

Cronbach's alpha is usually used for scale data while KR20 is usually used for reliability test of binary data (yes or no).