SOAN 2120 Lecture Notes - Lecture 8: Hurricane Katrina, Internal Validity, Dependent And Independent Variables
Document Summary
Combines elements of both the posttest and pretest-posttest control group designs. Random assignment or event matched pairs assignments is not practical. Ex: new teaching technique will influence a student"s performance. Asside from the absence of random assignment, these designs look much like a regular experiment: A group is exposed to an event, and then a dependent variable observation is recorded. X is the hurricane, o is the observation. Problem: you don"t have a comparison group. = limited in saying that x is what causes your scores on o. Internal validity threats which confound pretest and posttest comparisons. Selection is not an issue in this kind of design. You need to have 2 groups in order for selection bias to be an issue. Group 1 gets the treatment, group 2 doesn"t get the treatment and then it"s given the posttest. If the two scores are different then it is attributable to x.