Wednesday, May 26, 2021 WEEK 3 Lecture 4 Special designs Single case - ABA Reversal design - ABAB Reversal design - Multiple baseline design Quasi-experiments When we want to do an experiment but cannot do it in real life - One-group post-test only - One-group pretest-post-test - Non-equivalent control group - Non-equivalent control group pretest post-test Multiple repeated measures - Interrupted time series - Control series Developmental - Cross-sectional - Longitudinal - Sequential Reversal Designs Measure of his talking speed 5 days normal, then with the drink mixed into his food for 5 days, then stop for 5 days and measure his talking speed. Special drink seems to have an impact on the participant's speech. - A single reversal (ABA) design is susceptible to one important problem o Baseline- Treatment- Baseline - When you go back to A it might take longer to go back to normal, making it difficult to see the contrast - How can we solve this problem? - Multiple reversals (e.g., ABA, ABAB) minimize the chance of this problem affecting your results. - Single case designs, in general, suffer from lack of generalisability. Case studies can be an example.
Quasi-experimental Designs - No randomization (people choose to participate or not) - Uses pre-existing groups (or allows participants to sort themselves into groups) - Randomization is not possible or realistic One-group post-test only design (only one group and you make one manipulation) Smoking cessation program offered -> Participant group: Students who sign up -> Treatment: Smoking cessation program -> post-test: Smoking frequency ffi pack/week The program might have had an effect One group pretest-post-test design Smoking cessation program offered -> Participant group: Students who sign up -> pretest: Smoking frequency 4 packs/week -> Treatment: Smoking cessation program -> post-test: Smoking frequency ffi pack/week This program might have had an effect! Non-equivalent group designs - Non-equivalent control group design Participant Group Students who sign up Treatment Smoking cessation program Smoking cessation program offered Students who DON'T sign up No program - Non-equivalent control group pretest-post-test design Participant Group Pretest Treatment Students who sign up Smoking frequency 5 packs/week Smoking cessation program Smoking cessation program offered Students who DON'T sign up Smoking frequency 5 packs/week No cessation program Posttest Smoking frequency 1 pack/week Smoking frequency 3 packs/week Posttest Smoking frequency 1 pack/week Smoking frequency 4 packs/week
Quasi-experiment and Correlations - How are they similar, how are they different? Quasi-Experiments Correlation studies No randomization Outcome variables are measured Regarding group-based data: Can deal with multiple discrete groups without Inherent order without inherent order Can deal with only two discrete groups Hair length 1 2 Both Correlation & Quasi-experiment If we add gender identities for instance, we cannot calculate correlation. Internal validity is ... Our ability to infer that changes in the DV are attributed to the changes in the IV Threats to Internal Validity - History: Any external event happening that affected participants (especially between first and second measurements) - Maturation: People change over time (fatigue effect, get hungrier, more