Final Exam Review Flashcards
Independent group designs
AKA?
What are they?
How are groups arranged?
- Aka between subjects
- Compare differences between groups while controlling for differences within groups
- separate groups for each level of IV
Experimental methods
Establish causation if random assignment is used
An experiment is used to infer causality by using?
Manipulation of the IV
Control requires balanced samples. What does this mean and why is random assignment a good way to achieve this?
Controls for individual differences within groups providing for greater internal validity
Why is one group pre test post test design not an experiment?
No control
What is required to make a casual inference? Why does an experiment make this possible?
Covariation, time order relationship between IV and DV, elimination of alternative explanations
What does co variation mean?
The IV and DV are correlated; both events occurred around the same time
Time order relationship
IV changes, then DV changes
Internal validity
What is it?
Requirements?
Extent to which you can make a casual inference
Only occurs when all 3 aspects are met
External validity
Does intervention work in the real world?
Threats to internal validity;
Confounds
Extraneous variables
Systematic differences between groups
What are intact groups and why are they a problem?
Using pre existing groups
Results may have alternative explanation
Attrition
Loss of participants in a study
Mechanical loss
Factors external to participants
Not a big deal if infrequent
Selective subject loss
What is it?
How to control for it? (2 ways)
Internal factors
Can destroy comparable groups
Pretesting, drop one as another drops out
Participant and experimenter biases
Unethical manipulation of any part of an experiment that will aid in obtaining desired results
Double blind design
What is it?
Can control what?
Neither experimenter or participants know if they are in the control group or the treatment group
Can control experimenter effects
What is an independent groups design?
Between groups design
Separate group of participants for each level of the IV
What are between groups and within groups differences?
Q
Why can’t random assignment always be used?
Pre existing variables such as quasi experiments
Why can’t matched and natural groups designs not use random assignment?
No control because manipulation is not technically possible with pre disposed variables
3 types of independent groups?
Random
Natural
Matched
Random independent groups
Balance out individual differences
Randomly assign participants to level of IV
Natural independent groups
What is it?
True experiment?
Why or why not?
Naturally occurring IVs
Not true experiment (correlational because the already existing IV cannot technically be manipulated)
Matched independent groups
What is it?
When to use?
Instead of random assignment, groups formed on relative dimension
Works best w Smaller sample sizes
Repeated measures design
What is it?
How does it differ from an IGD?
Each participant completes all conditions of the experiment; participants serve as their own controls
Compares within group effects rather than between group effects
What are practice effects?
Change in participant’s responses over time due to learning a new task
Boredom, fatigue
What are anticipation effects?
Q
Purpose of counterbalancing?
To correct for
Advantages of the RMD
Participants are own controls, so less subjects are needed
Disadvantages of the RMD
Practice effects are possible if not properly counter balanced
Types of counterbalancing: complete
Block randomization
ABBA
Types of counterbalancing: incomplete
All possible orders
Latin square
Random starting order with rotation
Block randomization
What is it?
Size of blocks =?
Requires what?
Counterbalancing in which each block contains all conditions in random order
Size of block = number of cond
Requires many presentations to balance
ABBA counterbalancing What is it? Conditions presented how? Practice effects? Can only be used when?
Conditions in one random sequence, then reversed
Presents conditions only a few times to each participant
Each condition has same amount of practice effects
Can only be used when practice effects are linear
All possible orders
Best choice for which design?
What is it?
# of possible orders=?
Best choice for incomplete design Each participant randomly assigned to all possible orders # of possible orders = N! With N conditions
Latin square
Uses what?
What is it?
Uses selected orders
Each condition appears in each ordinal position once
Each condition proceeds and follows each other condition only once
Random starting order with rotation
Start with random order, then for each row, rotate one to the left
What is differential transfer? Why is it problematic for RMD?
Performance on one condition is dependent on the condition that precedes it
What is a complex design?
Why are they useful?
1 DV, 2 or more IVs
Interactions are the main advantage
What is a mixed design?
A complex design which used IVs of both ind groups and repeated measures
What types of effects are concerned within a complex design?
How do we know if each is present?
Main effect
Interactions
Simple main effect
Main effects
An effect of a single IV alone
Interactions
What are they?
Why are they important?
When the effect of an IV is different at different levels of another IV
Simple main effects
The effect of an IV at a single level of another IV
What are ceiling effects?
What are floor effects?
Ceiling: when performance reaches a maximum in any condition of an experiment
Floor: performance reaches a minimum
Single case design
What is it?
Is it an experiment?
Why or why not?
Studies one subject at a time
IS an experiment
Uses manipulation and control
Small n research
Small number of subjects
Difference between a case study and a single subject experiment?
Case study is not an experiment
Advantages/Disadvantages of single case design
A: high internal validity
Useful in dismantling other studies
D: limited to interventions w immediate/specific effect
Limited when examining behaviors with high variability (no stable baseline)
Advantages/disadvantages of case studies
A: provides rich description of individual
New or rare phenomena
Provides counter evidence
D: 1 person not enough empirical evidence
Can’t make causal inferences
Bias
Threats to external validity
When would a researcher choose a case study?
When studying new or rare phenomena
When searching for counter evidence
Why can’t casual inferences be made from a case study?
Not a true experiment; does not use manipulation and control
Ideographic approach
Intensive study of an individual
Nomethetic approach
Approach that serves to establish broad generalizations or laws that apply to large groups of individuals
Anecdotes
What are they?
Why are they not adequate support for a hypothesis?
A
Testimonials
What are they?
Why are they not adequate support for a hypothesis?
Q
Requirements for single subject experiment
Behavioral DV with stable baseline
Potent IV that results in immediate change
Controlled conditions
Types of designs?
ABAB
Multiple baseline across individuals
Behaviors
Situations
Why is a stable baseline needed?
It allows us to detect a change
Concept of reversal
Why is it needed to infer causality in an ABAB design?
Behavior changes with implementation of the IV, then reverses back to original state with removal of IV
Quasi experimental designs
Resembles a true experiment of treatment/intervention performed in natural setting
Lack of full control (no randomization)
Difference between a quasi design and a true experiment?
Lack of full control
When would a researcher choose to use a quasi experiment?
When randomization is not feasible, some hypothesis can’t be tested in the real world
Threats to internal validity: 8
History Maturation Testing Instrumentation Regression to mean Attrition Selection Additive effects with selection
Threats to internal validity: history
Some non treatment produces change in participants behavior
Solution: control groups
Threats to internal validity: maturation
People change naturally over time
Solution: control group to compare performance
Threats to internal validity: testing
People get better when tested again
Solution: use equivalent but different measures
Threats to internal validity: instrumentation
Measure of your DV changes
Solution: ensure reliability and validity
Threats to internal validity: regression to the mean
People at extreme ends of a measure tend to move toward the middle over time
Solution: don’t choose samples of extreme individuals
Threats to internal validity: attrition
Participants lost over time
Solution: careful follow up procedures, statistically compare those who dropped out to those who teamin
Threats to internal validity: selection
One group systematically differs from the other in ways unrelated to the intervention
Solution: randomization
Threats to internal validity: additive affects with selection
When any of the first 6 threats exists for one group but not the other
Solution: randomization
5 threats to independent group designs (even true experiments)?
Contamination
Experimenter expectancy
Novelty
Hawthorne Effect
Threats to IGD: contamination
What is it?
Examples?
When groups communicate with each other
Resentment, rivalry, control group seeking the treatment
Threats to IGD: experimenter expectancy effect
What is it?
Control through?
Unintentionally influencing results through observation, errors, etc.
Control through double blindness
Threats to IGD: novelty effects
Newness of the tx has an effect rather than the tx itself
Threats to IGD: Hawthorne effect
Behavior changes simply because someone is interested in the participants
(They care; they’re judging me)
Control through having same effect in control group
3 types of quasi experimental designs?
- Nonequivalent control group
- Interrupted time series
- Time series with non equivalent control group
Nonequivalent control group
Type of?
What is it?
Vulnerable to what?
Quasi experiment
Groups that are truly comparable controls for many threats to internal validity
Vulnerable to additive effects with selection
Interrupted time series What is it? What is often used? Compares what? 2 requirements?
No control group, multiple observations before intervention (baseline)
Archival data often used
Compares baseline before and after intervention
Requirements: must be abrupt, evidence of effect
Time series with Nonequivalent control group
What is it?
What kind of groups does it have?
Rules out what effects?
Multiple pre and post tests
Intervention and control group
Rules out history and instrumentation effects
Descriptive methods
Do NOT manipulate IV
may establish correlation but not causation
1 problem in quasi experiments? Why?
Additive effects with selection. Because of lack of randomization
Selection with maturation
Selection with history
Selection with instrumentation
- participants mature at different rates in the different groups
- participants experience different events that affect their responses
- instrument is more or less sensitive to change in one group vs another