Lecture 9 Pscy 2900U Quasi-Experimental Research PDF

11/4/24 2 Lecture Overview Ch. 8 – Quasi-Experimental...

11/4/24 2 Lecture Overview Ch. 8 – Quasi-Experimental u One-group designs Research u Non-equivalent groups designs PSYC 2900U – RESEARCH METHODS u Assignment #3 KARLA EMENO 1 1 2 Quasi-Experimental Research 3 Types of Designs 4 u Quasi-Experimental Research – Research that resembles Non-Equivalent Groups Designs: experimental research, but it is not true experimental One-Group Designs: (no random assignment) research. (no control group) u Posttest Only u An IV is manipulated, but it is not possible to include one u Posttest Only u Pretest-Posttest of the following: u Interrupted Times Series u Pretest-Posttest 1. A control group OR u Pretest-Posttest with Switching u Interrupted Time Replication 2. Random assignment to groups Series Switching Replication with u Eliminates the directionality problem but NOT u confounding variables. Treatment Removal Design 3 4 5 6 One-Group Posttest Only Design One-Group Posttest Only Design u A treatment (the IV) is given to just one group and an outcome (the DV) is measured just ONCE. u Weakest type of quasi-experimental design. Group1 A treatment is A posttest is (the only given given u Main limitation – No control or comparison group. group) u Results from this type of study are frequently reported in the media. 5 6 1  11/4/24 7 8 One-Group Pretest-Posttest Design One-Group Pretest-Posttest Design u A treatment (the IV) is given to just one group and an outcome (the DV) is measured TWICE (i.e., before and after treatment). Group 1 u Similar to a within-subjects experiment, but the order A pretest is A treatment is A posttest is (the only of the conditions is NOT counterbalanced. given given given group) u No counterbalancing means there could be alternative explanations for differences in pretest to posttest scores. 7 8 Why are control/comparison groups 9 10 Alternative Explanations so important? u History u History – Something else happened during the course of the study (e.g., TV show, 9/11) u Maturation u Maturation – Participants matured or developed over u Testing the course of that time (they were learning or growing) u Instrumentation u Testing – Measuring the DV at pretest affected u Regression to the mean responses at posttest (e.g., learned the test, primed u Spontaneous remission thinking about he issue) 9 10 Alternative Explanations 11 Interrupted Time Series Design 12 u Instrumentation – The measurement of your device has changed (e.g., human observer less sensitive to u Measurements are taken in intervals over a period of behaviour) time u Regression to the Mean – Extreme scores tend to u Similar to the pretest-posttest design in that it includes normalize to the mean over time measurements of the DV before and after treatment u Spontaneous Recovery – Things would have improved u Unlike a pretest-postest design in that it includes over time anyway multiple pretest and posttest measurements (and not u Address these alternative explanations by adding a just one before and one after) control group – but then it’s no longer a one-group u Accounts for natural variability in responses design 11 12 2  11/4/24 13 14 Interrupted Time Series Design Treatment did NOT work Multiple Multiple Group 1 pretests A treatment is posttests (the only are given given are given group) over time over time Treatment did work 13 14 15 16 Posttest Only Nonequivalent Groups Posttest Only Nonequivalent Groups Designs Designs u Participants in one group are exposed to a Group 1 A treatment is A posttest is treatment, a nonequivalent group is not exposed to (treatment) given given the treatment, and then the two groups are compared. u Researcher should take steps to ensure their groups Group 2 No treatment A posttest is are as similar as possible (but confounding variables (control) is given given could still be an issue). 15 16 Pretest-Posttest Nonequivalent Groups 17 18 Pretest-Posttest Nonequivalent Groups Designs Designs u A treatment group is given a pretest, receives a treatment, and then is given a posttest. At the same time, a nonequivalent control group is given a pretest, does not receive the Group 1 A pretest is A treatment is A posttest is treatment, and then is given a posttest. (treatment) given given given u Want participants who receive the treatment to improve more than those who do no receive the treatment. u Helps account for history and maturation (but not ALL A pretest is Group 2 No treatment A posttest is confounding variables). given (control) is given given u Becomes a between-groups experiment IF random assignment to conditions is used. 17 18 3  11/4/24 19 20 Interrupted Times-Series Design with Interrupted Times-Series Design with Nonequivalent Groups Nonequivalent Groups u Involves taking a set of measurements at intervals over a period of time both before and after an intervention of interest in two or more nonequivalent groups. u Is an improvement over the interrupted time-series design because you now have a control group. 19 20 Pretest-Posttest Design with Switching 21 22 Pretest-Posttest Design with Switching Replication Replication u Nonequivalent groups are administered a pretest of the DV. One group then receives a treatment while a nonequivalent control group does not receive a treatment, and the DV is A A Another Group 1 A pretest Treatment treatment posttest posttest assessed again. The treatment is added to the control group, (treatment) is given continues is given is given is given and finally, the DV is assessed one last time. u Strengths of the design are: (1) it includes a built-in replication and (2) provides more control over history, maturation, and No A The Another Group 2 A pretest instrumentation. (control) treatment posttest treatment posttest is given is given is given is given is given u Demand characteristics, placebo effects, and experimenter expectancy effects could still be an issue. 21 22 23 24 Switching Replication with Treatment Switching Replication with Treatment Removal Design Removal Design u The treatment is removed from the first group when it A A Group 1 A pretest Another Remove is added to the second group. (treatment) is given treatment posttest treatment posttest is given is given is given u Can provide strong evidence for treatment efficacy. u Not always possible depending on the treatment. A Another No The Group 2 A pretest treatment posttest treatment posttest u The quasi-experimental designs highest in internal (control) is given is given is given is given is given validity are those with a switching replication. 23 24 4  11/4/24 Example of Switching Replication with 25 26 One-Group Design Recap Treatment Removal Design One-Group Design Description (i.e., no control group) A treatment is implemented (or an IV is One-Group Posttest Only manipulated) and then a DV is measured once after the treatment is implemented. The DV is measured once before the One-Group Pretest-Posttest treatment is implemented and once after it is implemented. A set of measurements taken at intervals Interrupted Time Series over a period of time that is "interrupted" by a treatment. 25 26 27 Non-Equivalent Groups Design Recap (cont’d) 28 Non-Equivalent Groups Design Recap Non-Equivalent Groups Design Description (i.e., no random assignment) Non-Equivalent Groups Design Description Involves taking a set of measurements at intervals (i.e., no random assignment) Interrupted Time Series with over a period of time both before and after an Participants in one group are exposed to a Nonequivalent Group intervention of interest in two or more Posttest Only Nonequivalent treatment, a nonequivalent group is not exposed nonequivalent groups. Adds a control group. Groups to the treatment, and then the two groups are Nonequivalent groups are administered a pretest compared. of the DV, then one group receives a treatment There is a treatment group that is given a pretest, Pretest-Posttest with Switching while a nonequivalent control group does not receives a treatment, and then is given a Replication receive a treatment, the DV is assessed again, and Pretest-Posttest Nonequivalent posttest. Then, at the same time there is a then the treatment is added to the control group, Groups nonequivalent control group that is given a and finally the DV is assessed one last time. pretest, does not receive the treatment, and then Switching Replication with The treatment is removed from the first group when is given a posttest. Treatment Removal it is added to the second group. 27 28 Assignment #3: Method (20%) 29 Upcoming Dates 30 u Due: Submit in Canvas by 11:59 pm on Friday, November 22nd u Tuesday, Nov. 12th – Ch. 9 Factorial Designs – In-person lecture u You will begin Assignment 3 with a Current Study section (revised from and tutorial Assignment 2 based on TA feedback). u Tuesday, Nov. 19th – Ch. 10 Single-Subject Research – In-person u You will then write a Method section, which will explain – in detail – how lecture and tutorial you would carry out your proposed study. This will include a detailed description of the participants, materials, and procedure. u Friday, Nov. 22nd – Assignment 3 due in Canvas u This assignment will end with a Study Limitations and Future Research u Tuesday, Nov. 26th Directions section where you discuss some limitations of your proposed NO IN-PERSON LECTURE (slides only will be uploaded to research study and mention some future research that could follow from your proposed study. Canvas on Ch. 11 Presenting Your Research) u Detailed instructions for Assignment 3 (including the grading rubric), an NO IN-PERSON TUTORIAL (virtual exam Q&A session instead) example template, and an example paper, have been added to Canvas. u Monday, Dec. 9th , 12-2 pm (Regent Theatre) – Final Exam 29 30 5  11/4/24 31 Questions about Chapter 8? 31 6