PADP 7110 Lecture Notes - Lecture 7: Interrupted Time Series, Time Series, Random Assignment
Document Summary
Lecture 7 : research design quasi-experimental: conditions, violates some conditions of experimental design, treatment & comparison groups are not formed by random assignment, multiple observations before & after treatment/interventions. Examples: non-equivalent control group design (comparison group pretest/posttest, interrupted time series design, interrupted time series design with comparison group. Causal claims: plausibility > logical theoretical connection, statistical association > x & y must covary. Use statistical analyses to compare values in dependent variable & inde- pendent variable across many cases at one point in time: temporal sequence > x must proceed y. May not introduce/manipulate the independent variable by can compare the value of the dependent variable before and after being exposed to the inde- pendent variable: eliminate alternative explanations (rival independent variables) Use multiple pre/post tests or comparable groups (e. g. matching) to help. Rule out the likelihood of other/alternative variables can explain changes in the dependent variable.