1/66
Looks like no tags are added yet.
Publicizing norms
Using communication to make these unspoken rules visible
EXAMPLE: Does putting posters with inclusivity make the climate seem more inclusive? Yes
Control Variable
Things held constant on purpose (same for both groups)
Independant vs Dependant variable
Independant is what you use to see how dependant changes
Conditions
The levels of the indepenant variable (how much of it)
How can you tell if a study is expiremental?
It has a manipluated variable
Requirements for internal validity
Covariance: needs a comparision group, also needs a difference between the groups
Confounds: The independant variable is the actual cause and not a third variable
Temporial precednence
Can you say which variable comes first? You should be able to
Covariance
Needs a comparision group, also needs a difference between the groups
Confounds
The independant variable is the actual cause and not a third variable
Design Confound
The expirementer makes a mistake when implementing IV which causes variation within the IV (not always problematic) (bad if systematic)
Selection effects
When the participants in one level are systematically different than participants in the other level — Use RANDOM assignment to AVOID
How to avoid Selection effects
use matched groups
Control Variable(s)
Factor(s) you keep the same on purpose: like the room both groups take a test in
Whats the best design?
Depends on the context
Pretest/postest design
Tested on something then exposed to IV then they are tested again
Postest only design
Randomly assigned to have independant variable 1 or 2 and then measure the two in comparison
Independant groups vs. In group designs
Independant is different groups get different levels of IV, in group gets like all levels of the IV
Measuring/Intrument effects to internal validity
When the measuring changes over time
Testing threat prevention
Have it be post-test only
Preventing Measuring/Intrument effects
Make sure pre and post test measures are the same, or posttest only design
Testing threats
Scores change if people take a test more than once (they learn the answers)
Attrition threats prevention
Remove the people who dropped out’s scores for pretest as well
Selection-history threats
Outside event impacts only one level of the IV
EXAMPLE: Collecting data on if people like ICE per state, people in minnesota extra against it bc of events which went down
Selection-attrition
Only one level of the IV drops out of the study
EXAMPLE: Collecting data on if people like ICE, Illegal immigrent group drops out bc deported
Attrition threats
When certain types of participants drop out
EXAMPLE: All the introverts droup out
Second fix to regression threats
Anaylizing the pattern of results to see what the other studies got
Regression to the mean threat
When the pretest score is extreme
History threats
An external factor which effects most members for a treatment
EXAMPLE: Experiment about happiness then 9/11 happens
Preventing threats to the study
Having a comparison group
pretest and postest
Maturation Threats
Traits changing over time
Observer bias
Expectations for results influences interpretation of results
Selection effect
Different IV groups have different types of participants
Design Confound
Poorly designed expirement where variable varies with the IV
Why is something a factorial design?
2 or more variables
How to tell if there is an interaction on a graph
Top of chart minus bottom
How to tell if there is a main effect on a graph
Top plus (+) bottom divided by top
2×2 factoral design
2 variables with 2 levels
Whats more important: interactions or main effects?
Interaction is more exciting bc it’s effect is dependant
Interactions
Is there a difference in differences?
Top of chart minus (-) bottom
How to fix observer bias
Do a double-blind study
Marginal means
Arthimatic mean for each level of the IV, averaging over the levels of the over iV
Demand characteristics
Participant is able to figure out what the study is about and because of this changed their behavior
How to fix demand characteristics
Do a double-blind or masked study (hide the purpose initally)
Main effect
Overall effect of one of the IVs on the DV and average over the other IV
How large is the overall difference?
Overall effect of one of the IVs on the DV and average over the other IV
How large is the overall difference?
Retrieval vs Pure Studies
Pure studies are better for Short term
Retrieval is better for Long term
Can you use factorial designs to test hypothesis?
Yes
Particpant variables (IVs)
Things that already exist and can’t be changed (race, sex, etc)
Can same situations effect different people differently??
Factoral designs
Studies with two or more IVs
Most common thing: Cross the two, study all combos
Spreading interaction
“especially” — Spreads out
Bacon, no bacon, avacado, no avacado. All of them togheter makes most delicious
Crossover interaction
“it depends” — X shape
Hot/cold and pancakes/ice cream, no one likes hot ice cream and no one likes cold pancakes and vise versa
Placebo Effects
Improvement is due to believing what they are doing works
Interaction effect:
Does the effect of one IV depend on level of other IV
Simple expirement
Expirements with one IV
How to prevent placebo effects
Control group
Effect size
Large D = IV causes large change in DV
Small D = Groups are more similar
Construct validity
DV: how well was is measured
IV: how well was it manipulated
Theory testing: How well do the measures capture the conceptual variable
How to avoid selection effects
Matching groups and repeated measure design
Selection effects
Participants or data points chosen for a study do not properly represent the broader target population
Disadvantages of in group designs
Order effects (can be fixed with counterbalancing) and sometimes it just doesn’t work
How to combat Order effects
Counterbalancing: give people different orders, some do step A first some do step B first
Order effects
Order effects
being exposed to one condition might change how you respond to a later condition
Pratice effects
Repeated testing might make them better
Fatigue effects
Gets tired of doing stuff
Reapeated-Measures (Within-group-designs) designs
Participants are exposured to each different measure and then measured on each level’'
dont need as many participants
less variations in participants
Participants are exposured to each different measure and then measured on each level’'
dont need as many participants
less variations in participants
aaa