chapter 4 – research methods different methods to answer different questions a.does one factor...
TRANSCRIPT
Chapter 4 – Research Methods
Different methods to answer different questions
A. Does one factor cause another?The Scientific Method
- obtain reliable information under controlled conditions
• Example: Does psychotherapy cure snake phobia?
• Therapy vs. nothing => independent variable (IV)
• Level of snake fear => dependent variable (DV)
• DV depends upon IV
• How will we know if differences in snake fear are caused by therapy?
• Treatment of subjects (IV)- that which you are manipulating, systematically altering to see its effects
• CONFOUND = any other difference between the groups
• If no confounds, only thing different between the groups is the IV, then high internal validity
• Fairly sure that changes in the DV were due to IV
How to increase internal validity?
(make groups same except for IV)
1.Random assignment of subjects to groups
Groups: Experimental vs. Control
• Experimental = receive treatment being tested
• Control = comparison
• How to divide sample into groups?
• Random assignment*****
Random assignment makes the 2 groups equivalent
2.No other differences between groups
(“holding everything constant”)
• These 2 factors decrease confounds, & increase internal validity
Subjects/Who is in the experiment
• Population = all people of interest
• Sample = subset; those in the experiment
• Sample of convenience
• Random sampling
- everyone in population has an equal chance of being chosen
• Why random sampling?
• Sample is representative of the population of interest
• Can apply (“generalize”) results to population
• Increases external validity
• External validity = generalizability
• To other people, places, situations, etc.
• Key to Scientific Method = internal validity
• Controls to ensure that IV -> DV
• Rule out confounds
• Random sampling is not critical
• Increases external validity
Problems:
• Not always feasible or ethical
• Studies are analogues – simulations of real life (low external validity)
Advantage of Scientific Method
• Cause and effect
B. How strongly are two factors are related?
Correlational designs
• Longitudinal (how people change over time)
• Naturalist observation (watch people in natural settings)
• Not a true experiment
• No controls
• Is there a numerical relationship between 2+ factors?
Evaluating the outcome
• A correlation coefficient
indicates whether two variables are related
- 1.0 to +1.0
• Magnitude: absolute value of #= strength of relation
• Direction: sign
+ = as one increases, other increases
- = as one increases, other decreases
Relationships:
• Positive
• Negative
• Curvilinear
• None
Problem
• poor internal validity
-> don’t know WHY things happen
Reverse causality
Third-variable problem
Spurious relationships
Advantages
• easier, practical
• ethical, real-life
-> can have better external validity
C. What can we learn from one subject?
Three methods:
• Case study
• ABAB (Reversal) design
• Multiple-baseline design
Case study method
- documenting behavior of one person
Advantages:
• Real life (somewhat higher external validity)
• Suggests ideas
• Practical, easy (one person)
• Lots of information
Disadvantages:
• No controls/comparison
(poor internal validity)
• One subject not randomly selected
(poor external validity)
ABAB (Reversal)
• Get baseline (A)
• Introduce treatment (B)
• Return to baseline (A)
• Reintroduce treatment (B)
Advantages:
• More controlled than case study
• Still requires only 1 subject
Disadvantages:
• One person = limited external validity
• Sometimes unethical to withdraw treatment
• If return to baseline, then no cure
Multiple-baseline design
= change several behaviors sequentially
• Get baseline for all behaviors• Introduce treatment for first behavior• Then, treatment for second, etc.• Different treatments affect different behaviors
Advantages:
• More controlled than case study
• Also requires only 1 subject
• No withdrawal of treatment
Disadvantages:
• One person = limited external validity
• Sometimes hard to disentangle effects on individual behaviors