“Unit Zero” – Research
Design
Basic Vocabulary:
•Hypothesis – tentative explanation – must
be FALSIFIABLE – able to be supported
or rejected
•Operational Definition: clear, precise,
quantifiable definition of your variables –
allows replication and collection of
reliable data
•Qualitative data: descriptive data (eye color)
•Quantitative data: numerical data – IDEAL
and necessary for statistics
•Population – everyone the research could
apply to
•Sample – the people (or person) specifically
chosen for your study
Research Designs
•CORRELATION: identify relationship
between two variables
Adv: useful when experiments are unethical
Disadv: CORRELATION DOES NOT
EQUAL CAUSATION
Directionality problem – which
direction does the correlation go?
(depression cause low self-esteem, low
self-esteem causes depression, or a 3rd
variable?)
3rd variable problem diff. variable is
responsible for relationship (ice cream
and murder)
oPositive Correlation – variables increase
& decrease together
oNegative Correlation – as one variable
increases the other decreases
oThe stronger the # the stronger the
relationship REGARDLESS of the
pos/neg sign. Cannot be < or > than 1.
oStronger relationships = tighter clusters
on graph
•EXPERIMENTS: purposefully
manipulate variables to determine cause
/effect
Adv: only type that establishes cause and
effect Disadv: can be unethical, too artificial
oIndependent Variable: purposefully
altered by researcher to look for effect
Experimental Group: received the
treatment (part of the IV); can have
multiple exp, groups
Control Group: placebo, baseline
(part of the IV); can only have 1
oDependent Variable: measured variable
(is DEPENDENT on the independent
variable)
Vocab unique to experiments:
oPlacebo Effect: any observed effect on a
behv. That is “caused” by the placebo
(shows effectiveness of exp. Treatment).
Usually fixed w/ blinded studies
oDouble-Blind: Exp. where neither the
participant or the experimenter are aware
of which condition people are assigned
to (drug studies)
oSingle-Blind: only participant blind –
used if experimenter can’t be blind
(gender, age, etc)
oConfound: error/ flaw in study that is
accidentally introduced (can be called a
confounding variable)
•Random Assignment: assigns
participants to either control or
experimental group at random –increase
chance of equal representation among
groups (spreads the lefties across both
groups) – allows you to say Cause / Effect
OTHER STUDY TYPES
•NATURALISTIC OBSERVATION:
observe ppl in their natural settings
Adv: real world validity
Disadv: No cause and effect
•CASE STUDY: Studies ONE person
(usually) in great detail
Adv: – collect lots of info
Disadv: No cause/effect
•META-ANALYSIS: combines multiple
studies to increase sample size and
examine effect sizes
STATISTICS
•Descriptive stats: show shape of the data
oMeasures of Central Tendency:
Mean: Average (use in
normal distribution)
Median: Middle # (use in
skewed distribution)
Mode: occurs most often
•Bimodal – has two
modes – usually
indicates good bad
scores
•Skews – created by
outlers
oNeg skew = mean is
to the left (neg side),
mode is to the right
oPos skew = mean is to
the right
•Measure of variation
•Range – distance bw
smallest and biggest #
•Standard deviation – avg. amount the
scores are spread from the mean
(bigger # = more spread)
•INFERENTIAL STATISTICS:
establishes significance (meaningfulness)
oSTATISTICAL SIGNIFAN CE =
results not due to chance,
exp.manipulation caused the
difference in means
p<.05 = stat. sig, smaller = better
oEFFECT SIZE = data has practical
significance – bigger = better
•ETHICAL GUIDELINES (IRB
APPROVAL NEEDED FOR PPL)
oConfidentiality: names kept secret
oInformed Consent: must agree to be
part of study
oInformed assent – minors AND their
parents must agree
oDebriefing: must be told the true
purpose of the study (done after for
deception)
oDeception must be warranted
oNo harm– mental/physical
Additional Vocabulary:
•Surveys: usually turned into correlation.
Subject to self report bias - errors when
collect surve data due to:
oSocial desirability – ppl lie to look
good
oWording effects – how you frame
the question can impact your
answers
•Random Sample (selection): method for
choosing participants for your study –
everyone has a chance to take part,
increases generalizability
•Representative Sample: Sample mimics
the general pop. (ethnic, gender, age)
•Convenience Sample: select participants
on availability – less representative and
less generalizability this way
•Sampling bias – sample isn’t
representative, due to conv. sampling
•Cultural norms – behvs of a particular
group can influence research results
•Experimenter bias / Participant bias:
experimenter/participant expectations
influences the outcome
•Cognitive bias– bias in thinking/judgment
oConfirmation bias – find info that
supports our preexisting beliefs
oHindsight bias – “I knew it all along”
oOverconfidence – overestimate our
knowledge / abilities
oHawthorne effect – ppl change
behavior when watched
•Research needs peer review and adequate
sample sizes
DO NOT MIX Random Sample and Random Assignment.
Sample = Generalize. Assignment = Cause/Effect
