Psyc 321_09 within groups
Download as PPTX, PDF2 likes648 views
This document discusses within-groups or repeated measures designs. In within-groups designs, the same group of participants receives all levels of the independent variable. The key advantages are that all groups are equal at the beginning and the design is very sensitive to changes. Disadvantages include potential carryover effects and sensitivity to time variables like practice and fatigue. Counterbalancing techniques like Latin squares can help control for order effects. Mixed designs combine within- and between-groups approaches to rule out carryover effects while maintaining control. Matched subjects designs aim to reduce error variance by matching and randomly assigning participants.
Psyc 321_09 within groups
- 1. Developmental Research Methods Topic 9: Within Groups Designs 10/1/2014 CEDP321 Ryan Sain, Ph.D. 1
- 2. Within groups designs Also called repeated measures We are comparing groups of people to themselves Our own best controls Each person in every group receives all levels of the IV
- 3. Advantages All groups are equal at the beginning of the study Fewer participants Very sensitive to changes (statistically) ◦ Within subject variance is going to be low (error variance)
- 4. Disadvantages Carryover effects ◦ Seeing a picture vs. not seeing one ◦ Permanent change – cannot study in a within subjects design ◦ Clinical treatments Drug trials Psychotherapy Sensitive to time variables (practice, fatigue, order of IV) ◦ Order effects and practice effects
- 5. Counterbalancing Complete counterbalancing ◦ Each condition must occur with the same frequency ◦ Each condition must precede and follow all conditions an equal number of times Controls for time and order effects 5!
- 6. Counterbalancing Group 1 Treat 1 Treat 2 Treat 3 Group 2 Treat 1 Treat 3 Treat 2 Group 3 Treat 2 Treat 1 Treat 3 Group 4 Treat 3 Treat 1 Treat 2 Group 5 Treat 2 Treat 3 Treat 1 Group 6 Treat 3 Treat 2 Treat 1
- 7. More counterbalancing Incomplete counterbalancing ◦ Just balance the order. Group 1 Treat 1 Treat 2 Treat 3 Group 2 Treat 2 Treat 1 Treat 3 Group 3 Treat 3 Treat 2 Treat 1 Latin square – doing this randomly with the number of levels randomized for each participant Differential order effects ◦ certain orders may produce certain outcomes – ◦ a sort of interaction = treatment x order
- 8. Mixed designs Using between groups and within groups logic Measures change over time within a group and compares a factor between groups as well ◦ 2(between factor) x 3 (within factor - often a time variable)
- 9. Mixed Designs cont. Rule out carryover effects ◦ (using the between groups part) High levels of control of a within design. Main effects ◦ Time ◦ Other IV Interactions ◦ Time x IV
- 10. Matched subjects designs Reduce error variance Combine the benefits of within subjects and between groups Twin studies – not always possible Match participants ◦ age, gender, SES, ethnicity, etc. ◦ Then randomly assign one of each match to a condition.
- 11. Matching - types Use a matched group as a control group ◦ Random assignment and equal groups. Use a matched group as an experiment in unto itself ◦ Two groups matched ◦ Each group has both males and females – treat as normal. Compare males treatment to males control, etc
- 12. Credits www.sxc.hu ◦ Mixing table 1 - hovard ◦ Mixing table 2 - riverstorm ◦ People – biewoef ◦ Yin Yang - hisks ◦ Orchids – flajs01 ◦ Drugs - alaasafei Scrum – www.blackfire.net