Chapter 9 - Within Groups design Flashcards
What is a within subjects design?
- Single group of participants
* Tests/observes each individual in all the different treatments being compared
What is the other name of within subjects design?
Repeated Measures experimental design
What are the 2 possible structures for a within subjects design?
Sequential: Treatments administered one after the other
Concurrent: Treatments administered all at once
What are the 2 major sources of confounding in within subjects designs?
Enviromental variables and time-related variables
What are environmental variables?
Characteristics of the environment that may change from one treatment to another
What are time-related variables?
Factors influencing the participants between 1st and last measurements
What are the confounding variables related to history?
environmental events other than the treatment that change over time and may affect the scores in one treatment differently than in another treatment
EX: life at home, school, work between the treatments
What are the confounding variables related to maturation?
systematic changes in participants physiology/psychology that occur during a research study
What are the confounding variables related to instrumentation?
changes in a measuring instrument that occur over time (mostly with behavioural observation measures)
How is instrumentation also called?
instrumental bias or instrumental decay
What is observer drift?
when observers become inconsistent in their observations (ex: becoming more strict or lenient in their categorization of behaviour)
What are the confounding variables related to regression towards the mean?
tendency for extreme scores on any measurement to move toward the mean (regress) when the measurement procedure is repeated
How is regression towards the mean also called?
Statistical regression
What are order effects?
participation in one treatment has influence in participation in the following treatments
Any change in performance caused by participation in other treatments = order effects
What is fatigue?
decline in performance as participant moves through treatments
What is practice?
improvement in performance as participant gains experience through treatments
What is carry-over?
effects on one treatment linger in the following treatment and affects its performance
What are contrast effects?
type of carry over effect (EX: coming from a bright VS a dark room and entering a dimly lit room - perception will not be the same)
What are progressive error effects?
Carry over effects are caused by experiencing other treatments, other effects (fatigue, practice) are called progressive error effects and are due to simply being in the study
Name all time-related factors
History, maturation, instrumentation, regression
Name all order effects
All effects directly relates to experience in other treatments
What are the 3 methods to control for environmental factors?
- Randomization
- Holding constant
- Matching across treatments
What are the 3 main ways to control for time related or order effects?
Controlling time
Switch to a betwee-subjects design
Counterbalancing
What is controlling time?
Controlling the time it takes to complete the study (increases the potential of order effects)
Why would we switch to a between subejcts design?
• If order, practice or fatigue effects are expected to be strong, it might be a good idea to switch to a between subjects design so that participants do not experience all treatments.
What is counterbalancing?
• Changing the order of treatments between participants so that they don’t all develop experience/fatigue in the same order
What is the effect of counterbalancing on order effects?
• Counterbalancing helps to minimize the difference in scores between 2 treatments due to order effects, but in real life studies, we can never know how important/impactful the order effects are compared to the effects of the treatments
Does not eliminate order effects; it distributes them
What is the effect on counterbalancing on variance?
• Can increase variance within treatments (if half the scores receive a 5 points increase due to the counterbalancing in treatment 1, it obscures the true effect of treatment one)
What is the absolute level of performance?
The true mean
What are asymetrical order effects?
• We assume that order effects are the same for all treatments, but it might not be the case (one treatment might give more practice opportunity than another)
When this is the case, counterbalancing does not work well
How can we calculate the numer of possible sequences depending on the number of treatments (n)?
n! (n factorial)
EX: if n=4, then n!=1x2x3x4=24 groups needed to go over each sequence
What is complete counterbalancing?
When the researcher uses the same number of groups as the number of possible sequences of treatments in their experiment
What is partial counterbalancing?
§ Each treatment occurs once 1st, once 2nd, once 3rd, etc…
For 4 treatments, we only need 4 sequences
How can we determine which sequences to do or not do in partial counterbalancing?
Latin square:
1st row = ABCD
2nd row = D moves at the beginning (DABC)
Etc (CDAB,…)
What are the advantages of within subjects designs
- Requires few participants in comparison to between-subjects
- Eliminates the problem of individual differences (main concern of between subjects design)
What are the 3 aspects to consider when choosing a within or between subjects experiment?
1. Individual Differences Within Subjects eliminates this 2. Time-related factors and order effects Eliminated in Between Subjects 3. Fewer Participants In within Subjects design
What is a matched-subjects design?
- Uses a separate group for each treatment condition, but each individual in one group is matched to an individual in every other group (matching based on a variable crucial to the study)
- Goal: have all advantages of within and between subjects without the disadvantages
What is a 2 treatment design?
Within subjects design with only 2 treatments
How do we statistically analyze a two-treatment design?
- If data allows it, we can compute the 2 means for each group (then t test or ANOVA)
- Ordinal/nominal variables - Wilcoxon Signed-Ranks test can be used to determine significance
- Some within subjects only show the direction of the difference between the 2 treatments (sign test is used to determine if it’s significant)
What is a multiple treatment design?
Within subjects with more treatments
