SOAN 2120 Lecture Notes - Lecture 14: Random Assignment, Data Analysis, Selection Bias
Document Summary
1: combines elements of both the post-test and only the pretest-posttest control group designs. Non-experimental research designs: random assignment or even matched pairs assignment is not practical. Non-equivalent control group design (field experiments use experimental design: new teaching technique (unlikely that classes split up for purposes of experimentation). Research designs: aside from the absence of random assignment, theses designs look" much like a regular experiment: 01 x 02. 03 04: not true experiments (quasi-experiments, internal validity issues. One shot case study: x 0, a group is exposed to an event (condition x) and then a dependent variable observation is recorded (0) X would be the hurricane and 0 would be the findings that researcher"s found. The issue, however, it that there is no comparison group. One group pre-test/post-test design: 01 x 02. Professor david walters: internal validity threats which confound pretest-posttest comparisions, selection will be an issue if there are no r"s!!