Psyc.242 Lecture 7.1 Repeated Measures Design PDF

Psyc.242 Lecture 7.1 Repeated measures design Repeated Measures Designs  Each individual participates in all conditions of the experiment  Completes the DV measure in all conditions  Hence “repeated measures”  Also called “within-subject” design  Entire experiment is...

Psyc.242 Lecture 7.1 Repeated measures design Repeated Measures Designs  Each individual participates in all conditions of the experiment  Completes the DV measure in all conditions  Hence “repeated measures”  Also called “within-subject” design  Entire experiment is conducted “within” each subject  In the previous chapter, the independent groups design is also called “between-subject” design Within subject designs  In within-subject design, there’s only one group of participants – a very powerful design. “Noises” due to individual difference can be eliminated.  It’s very commonly seen on cognitive experiments such as reaction time studies.  NOT ALL research questions can be designed as a within-subject study. In fact, most can’t. For example: a target search study - In this task, the search target is the color blue. Only the blue X is the real target. - All other letters (red X and red O) are color distractors. - There are 30 distractors on the screen. How fast can a research participant pick out the blue X and click on it on the screen? Simple target search task with color distraction  The target search study can be a bit more complicated - In this task, the search target is the color blue + the shape X. Only the (blue X) is the real target. - All other letters are distractors (red X)(blue O)(red O). - There are 37 distractors on the screen. How fast can a research participant pick out the blue X and click on it on the screen? Complex target search with color & shape distraction “on the computer screen, click on the correct targets –blue X”  Within subject design with three conditions:  A: 10 distracters  B: 15 distracters  C: 20 distracter  DV: reaction time (total amount of time it takes the participant to click on all the correct targets)  Participants: it’s very common for within-subject, reaction time study to have very few participants Design procedure A B C 8 Results. (Notice how different participants go through the conditions in different order?) Conditions A B C Participant Order (10) (15) (20) 1 ABC 18.33 22.39 24.97 2 ACB 15.96 20.72 21.79 3 BAC 19.02 22.78 25.46 4 BCA 25.36 27.48 27.91 5 CAB 19.52 24.64 26.75 6 CBA 23.27 24.96 25.49 Mean Scores 20.24 23.83 25.40 What if we don’t vary the order of conditions for different participants? What if everyone just goes through ABC, or CBA? Important issue in within-subject design  Since all participants go through all conditions, we need to watch out for sequence effect: the experience with one condition affecting performance in subsequent conditions Sequence effect  1. Positive practice effect (training effect)  Solution: (1)counterbalance (2) random selection presentation  2. Negative practice effect (fatigue, boredom)  Solution: rest between trials  3. Carryover effect (contamination). The effect from one condition will be carried to the other condition. For example, an experimental drug in the treatment condition may have cured the patient. Another example: a mood manipulation from the experimental condition is very strong that the participant still feels that way when they start the control condition phase.  Solution: usually the carryover effect will make a within-subject design undesirable – this is the reason why a within subject design is rare in certain area of research. (1)Counterbalancing: systematic way to control sequence effect (we’ll discuss counterbalancing more in the next lecture)  For three conditions, how many variations are there for the presentation order?  ABC - participant#1 will go through this order  ACB - participant #2  BAC - participant #3  BCA - participant #4  CAB - participant #5  CBA - participant #6  Six! The number of participants should be multiples of six (2)Random selection presentation (random order of presentation) – broadly applicable, especially when the number of condition is too large  Randomly select the next condition that’s presented to the participants  For example, in a very complex cognitive experiment, there are 5 conditions. The total number of variations for the presentation order would be 5! (5 factorial), which is 120.  In this case, it’s unrealistic to completely exhaust all the possible order combinations  A computer program will just randomly select the next condition for each participant (abced, cbaed, ecadb, etc..) Strength of within-subject design  More sensitive to small group difference  Variability caused by individual difference is eliminated (reduced error variance)  Fewer participants needed (because of eliminated individual difference - increased power)  Less time spent on instructions Weakness of within-subject design  Participants could figure out the hypothesis (subject effect)  Sequence effect  When sequence effect cannot be taken care of by your design (counterbalancing etc.), you should NOT do a within-subject design What is the critical factor that determines whether we should go with a within-subjects design? Or independent groups design?  How great is the sequence effect and whether it can be taken care of by the design!  When sequence effect is controllable, go with within-subjects design because of its sensitivity  When sequence effect cannot be controlled by the study design, go with between-subject design. Could a within-subject design be used for this research question?  Researchers want to see if exposure to different information may influence people’s opinion. The stimulus materials the researchers want to use are arguments that either support or oppose capital punishment. The dependent variable is people opinion on capital punishment. What research design is more appropriate? Could a within-subject design be used for this research question?  A researcher wants to test whether bright light makes participants’ memory work better compared to dim light. Assignment  Connect: Chapter 7