KLA210 Lecture Notes - Lecture 2: Western Electric, Confounding, Quasi
Social Psychology week 2: methods in social psychology
- Research methods change over time
Scientific method:
- What makes science a science:
oApplying principles and methods
oPsychology applies this method to things that are usually only
described
oThe way we do research makes psychology a science
- Do not assume theory is true because it is logical:
oTest theories based on hypotheses: testable
oMake predictions based on theory and evaluate these against data
oResearch methods: create a study to test these predictions
oStats are an argument for or against a theory
They don’t tell you this is true or not true
They offer suggestions to accept or deny the theory
- What is a theory:
oSomething that helps us break down complex phenomena
oBreaks into main constituents
oExample: students learn better if the lecturer tells jokes
Concepts:
Learn better: outcome
Tells jokes: cause
How could we test this: operationalize
Learn better: exam results before and after
Tells jokes: number of jokes
Criticism:
Individual differences (if using control and
experimental groups)
oIntelligence
oHumor
oHowever this factor is eliminated due to the
experimental process
oRandom assignment
- Science defines the truth as:
oSomething that cannot be falsified at the moment
oCircling in on the truth:
Observation
Generate theory
Make prediction
Test the prediction
Refine theory
New prediction
New test
“Truth”: work with the best we have at the moment
look at the strength of evidence
oImplications:
find more resources at oneclass.com
find more resources at oneclass.com
there is no “truth” as long as someone could provide to the
contrary of the theory
Sir Karl Popper generated this idea
- Statistics:
oAllows us to examine whether the results could have been generated
by chance alone
oAlpha 0.05: null hypothesis
oType one and type two errors: false positives etc.
oPrevents against dogmatism: believing something is true because you
are told it is true
oReplication is essential:
Someone else somewhere in the world should be able to run the
same study and find the same results
1/20 is false
- Causality:
oExperiments:
Independent variable: manipulated
Dependent variable: varies as a result of change in independent
variable
Random assignment: reduce risk of confounding variables
Requires a large enough sample
Evens out the influence of confounding variables
Allows causal interpretation
Representativeness is low:
The extent to which this is a problem depends on how
much the researcher wants to generalize the data
Only holds true for the condition that we held it
Manipulation differs between groups
oRandomization: no rules for allocation
Achieved by generating a string of random numbers
oExamples: TV and aggression:
Watching violent TV vs. not watching violent TV
UNEVEN conditions: the other group didn’t watch TV at all
No way we can infer this
Even if another TV show
Need to do pretesting
Differences in agitation
oRealism/external validity/mundane realism is low:
Often the conditions are simplified
So we can measure actual behaviour and to avoid confounding
with everyday objects
oControl is high: of independent variables and confounding
oDemand characteristics:
Experimenter may give clues as to hypothesis
Social desirability
oConfounding variables
oFloor effects/ceiling effects:
find more resources at oneclass.com
find more resources at oneclass.com
