Documente Academic
Documente Profesional
Documente Cultură
INTRODUCTION
Psychology science of behavior
Scientific approach to understand behavior
Knowledge on psych processes based on
scientific evidence (gotten through research)
Scientia knowledge
Knowledge
o Content what we know, facts
o Process systematic ways of gathering,
noting
relationships,
offering
explanation
Nonscientific Inference
o First data we collect about others:
traits/first impressions
o ^Bias to overlook situational data in
favor of data that substantiate our trait
claims
o Generally more accurate when we
know someone well
o Traits useful predicting behavior long
run
o Situations predictor: momentary
behavior
o Overconfidence bias predictions seem
more correct than actually are, we
have more data (accurate or not), more
confidence
Characteristics of Modern Science
data
observable/experienced
o ^ can be verified or disproved
(investigation)
CH. 1:
Good Thinking
o Central feature of sci. method
o Systematic, objective, rational approach to
collection and interpretation of data
o being open to new ideas even when they
contradict our beliefs
o Principle
of
parsimony
simplest
explanation
is
preferred;
simplicity,
precision, clarity of thought
Self-Correction
o Content of science changes with new sci.
info.
o Changes important for sci. progress
o More evidence supporting theory, more
confidence theory is correct
Publicizing results
o Continuous exchange of info vital to
sci. process
Replication
o Repeat our procedures, get same
results (others should as well)
Objectives of Psych. Science
Description
o Initial step
o Systematic & unbiased account of
observed characteristics of behaviors
o Good
descriptions
greater
knowledge of behaviors
o Info about what behavior will be like
Prediction
o Capacity for knowing in advance
certain bh. expected to occur
o Correlational, quasi-experimental used
to predict behavior
Explanation
o Explained
behavior,
understanding
what causes it to occur
o Includes knowledge of conditions that
reliable reproduce occurrence of bh.
o To explain, experimental
research
design in which we systematically
manipulate aspects of the setting
o Control other factors that might
influence bh
Control
o Application of what has been learned
about the bh
Observation
o Systematic noting and recording of
events
o Only observable events can be studied
scientifically
o Thinking, feeling as long as we
develop observable signs
o Ex. moods cannot be observed
reliably; typically questionnaires
o System for observing should be
applied the same way every time
Measurement
o Assignment of numerical values to
objects or events etc. according to
conventional rules
o Standardized units (conventions)
o Standards determined by context of
study
o Measurement must be consistent
across each set of conditions same
instruments and procedures
o Use of stat to evaluate findings
Experimentation
o Process undertaken to test hypothesis
(behavior will occur reliably under
certain, specifiable conditions)
o Predictions must be testable
o Minimum requirements:
cond./treatment
(within-subjects
design)
o Usually new problems with ^
Random assignment
Presenting
treatment
identically
Wilhelm
Wundt
first
experimental
psychologist
and
using
advancing
understanding of feelings, thoughts that will
benefit humanity
*Institutional Review Board (IRB) evaluate
proposed studies before being conducted
o Journals require researches to be
approved by IRB first
o Primary duty ensure safety of
research
participants,
adequately
protected
o At risk subject more likely to be
harmed in some way
o IRB must determine if risks are
outweighed by potential benefits,
knowledge
Risk/benefit analysis
o Should have members that are skilled
in research mehods
Poorly designed research can lead to
unwarranted and inaccurate conclusions
Informed consent nature of the study,
procedure etc.
o Part. should give consent freely
o Free to drop out of experiment
o Full explanation of procedures, offer to
answer questions
o How long it will take
o Make clear potential risks and benefits
possibility of injury should be
explained too
o Provide assurance data will remain
private and confidential
o *No need to reveal true hypothesis
(exp.
expectation
exposedmight
influence behavior)
o Should be obtained in writing, with
participants copy
o Minor consent from guardian/parent
(though they could still refuse to
participate)
o ^Also applies for the cognitively
impaired
Evolution of Ethics in research
o Came to forefront after discoveries of
brutal experiments on Jews (conc.
camp)
o National Commission for Protection of
Human Subjects of Biomedical and
Behavioral Research
o Belmont report three basic ethical
principles
Researchers
publishing
false
data
Psychologists do not fabricate data
Research articles reviewed by editor of journal
+ several experts peer review
o Assess merit of a submission
o Something strange about findings, will
be discovered
o Make recommendations to editor
o First line of defense
Replication
o Second line of defense
o Especially if results surprising, novel or
important
Competitive nature of academic psych
o Tenure and promotion
o Publish or perish
o Works against fraud but also could be
the major cause
Sir Cyril Burt
o Major fraud!
o Inheritability of IQ
Data falsification can have important real-world
implications
National Science Foundation
o Misconduct: fabrication, falsification,
plagiarism
o Plagiarism
CH.
3:
ALTERNATIVES
NONEXP.DES.
TO
EXPERIMENTATION:
Recall:
o Primary purpose of exp: establish
causal relationship between treatments
(antecendent, specific) and observed
behavior
Individuals
immediate
experience
Personal experience as a
source of data
Antecedents
not
manipulated
Comparison of behaviors
under different treatment
cond. not required
Conclusions
are
instantaneous
Difficult
to
replicate
experiences,
apply
scientific criteria
Cannot
be
used
to
understand
causes
of
behavior
Frequent
component of
qualitative research
No
manipulation
of
antecedent conditions
Descriptive record of a
single
individuals
experiences (kept by an
outside observer)
^Patient
or
other
knowledgeable
source
provides information on
^Difference
have
impt. implications
for etiology, origin,
psychopathology
We cant be sure if could be
applied
to
general
population
If not observed all the time,
not sure if aware of all
relevant
aspects
of
persons life
Neglect to mention all impt
info
Retrospective
data
recollections
of
past
events,
accuracy
not
guaranteed
(altered,
reconstructed)
^only
a
relationship
between two occurences
Field Studies
Degree
of
constraint
varying
Naturalistic
Observation
Studies
o Observing
as
events
occur
spontaneously
o Descriptive
o No manip
o Low-low
o Animal behavior
o Try
to
remain
inconspicuous
o Behaviors
should
not be altered by
presence
of
observer
o Vast
array
of
responses
(which
one to observe?)
o Essential,
preferable
to
experiments
in
some cases
o Designed to answer
specific questions
o INTERSECT
=
Special
coding
system
(allow
observations to be
quantified, permit
stat analyses)
o ^Systematic
observation limit
type of responses
recorded (low-high)
o Lab
findings
vs
natural settings
o ^Some behaviors
may be distorted in
lab setting, best
studied in natural
o Behaviors different
when
watching
known
o Reactivity
tendency to alter
behavior
when
aware of observers
presence
o ^Confirm
or
sabotage
Unobtrusive
measures
behavioral
indicators
Participant observer study
o Researcher
becomes part of
group studied
o ^when
mems
might
not
cooperate
o Sometimes
they
are not told
o Usually does not
include systematic
obs.
o Qualitative data
o Difficult to remain
objective
and
unbiased
o ^when friendships
form
o Privacy?
o Public place where
privacy
not
expected ethical
o Disclose
status
when in a private
place (IRB)
o Great potential for
reactivity
o No contaminating
laboratory setting
effects
Archival Studies
o Descriptive
research
o Already
existing
records reexamined
for new purpose
o Data from govt,
agencies
o Causal inferences
cannot
be
supported
o ^Valuable info tho
Qualitative research
Self-reports,
personal
narratives, expression of
ideas
No
understanding
of
qualitative methods not
really appreciate breadth of
r. methodology used by
contemporary
psy.
Scientists
Paradigm
attitudes,
values, beliefs, methods
generally accepted in a
discipline
^hypothesis
testing,
experimentation,
quanti.
data
To
study
contextual
phenomena
o
Interpretation might be
influenced by researchers
own viewpoint (potential
bias)
Accuracy of self-reports,
retrospective data
Qualitative
data
=
reproducibility
of
conclusions?
Goodness of qualitative
data = replicability
Nominal scale
Classifies response in
categories
Lowest
level
of
measurement
No info on magnitude
^nor
differences
in
magnitude
Ordinal scale
Rank ordering of a
response
Magnitude measured in
the form of ranks
Interval scale
Magnitude
of
quantitative size with
equal
intervals
between values
No zero point
Ex.
Fahrenheit
and
centigrade (no true 0,
the true absence of any
measurable temp)
Ratio scale
Equal
intervals
between values
Nature of variable
Response styles
o Subj. personal characteristics affect
answer
o Willingness, preferences, yea-saying
and nay-saying
o Willingness
to
answer:
questions
require specific knowledge
o ^no right or wrong
o Position preferences: doubt, answer c
etc.
o The way of answering could tell a lot
about u
o Manifest content - plain meaning of
words that appear
o Yea-sayers agree with questions
regardless of ^
o Nay-sayers disagree no matter what
o Pretest questions before data collection
o Get volunteers, go over each question
o Context effects
Use
buffer
questions
to
separate related questions
Collecting Survey Data
Self-administered questionnaires
o Consider possibility of reactivity
o Collect data anonnymously
Mail surveys
o Questionnaire and return procedures
protect anonymity
o Keep track of people who do not return;
report
o Keep as short as possible
o Those who returned not same as those
who did not
Computer, internet
o As long as theres privacy
o Less concern with social desirability
o No experimenter present, reduce
coercion
o More diverse sample
o Cannot see subjects
o Multiple times answering
Telephone surveys
o Listed no. different from unlisted
o Interviewers to establish rapport,
maintain interview style
o Keep simple
Social desirability
Unusual: open ended questions
Males:
more
authoritative
and
persistent
Interviews
o For high quality survey data
o Most expensive
o Different in questions (open-ended)
o Establish rapport, avoid judgmental
statement, know how to keep interview
flowing
o Confidence, trust
o Subtle changes can influence subject
o Structured vs. unstructured
o Interviews appearance
o Interview should match respondent
Focus Groups
o Small groups of people with similar
charac.
o Facilitator, guides group in discussion
o ^should be skilled for useful data
o
o
o
Reliability
o Extent to which survey is consistent
and repeatable
o Responses similar questions consistent
o Different
survey
givers:
similar
responses
o Same person multiple tests same
response
Validity
o Extent to which survey actually
measures intended topic
Portion
of
whole
pop
selected
unbiasedly
o *Random selection does not guarantee
sample being truly representative of
entire pop
o List down all members then random
(1ina100)
Subsets
are
sampled
separately so impt minorities
are represented in total
Cluster Sampling
o Population too large to randomly select
one by one
o Entire
clusters/naturally
occurring
groups
o Whole groups selected rather than
individuals
o Ex. zip code areas, school districts,
cities, countries etc.
o Main advantage: can sample data
efficiently from relatively few locations
o Potential disadvantage: subjects within
clusters may resemble one another
o Desirable to sample many clusters,
large sample as possibly
Quota Sampling
o Predetermined quotas intended to
reflect makeup of population
o No constraints on selection as long as
quota is filled
o Resulting sample might not be good
representation
o Quota sampling lacks rigor
o Low external validity
Convenience Sampling
o Any groups who happen to be available
o Weak form of sampling
o
CH. 5:
Correlation
o Degree of relationship
o If correlated, changes in one, changes
in another
o Selected traits or behaviors measured
first
o Low-high
o Better prediction of persons score on
another measure highly related to it
o Higher correlation, more accurate
prediction
o Simple correlation between pairs of
scores
o Scatterplots visual representations of
scores of each subject; first step in
analyzing
o Regression lines lines of best fit
(direction of line = direction of
relationship)
o Positive correlation direct relationship
o Negative
correlation
inverse
relationship
o Direction of relationship one of the
most difficult concepts to grasp
o Sign tells direct/inverse, absolute value
how strong it is
o Bigger absolute value, more accurate
o Nonlinear trend could be curvilinear
o Range truncation artificial restriction
of the range values of X and Y
(restriction can make trend weak)
o Outliers extreme scores
multiple
regression
analysis
o Regression equation determines weight
of each predictor
o Path
analysis
possible causal
sequences of related behaviors
Causal Modeling
Path Analysis
o When subjects are measured on
several related behaviors
o Creates models of possible causal
sequences
o Another
descriptive
method
but
generates impt info for prediction,
generate experimental hypotheses
o Model can only be constructed using
the
behaviors
that
have
been
measured
o Goodness of it
o Ex. empathy manipulation affects level
of empathy, affect attitude toward drug
addicts, produced helping behavior
o Internal validity low (based on corr.
data)
o Cause and effect cannot be established
with certainty
o Third variables can never be ruled out
completely
o Useful for generating hypotheses for
future
research,
potential
causal
sequences
in
instances
where
experimentation not feasible
Recap:
o Correlation degree of relationship
o Multiple Correlation - ^between three
or more variables
No manipulation of antecedent
No random assignment
CH. 6:
Longitudinal design
o Same subjects at different points in
time, look to see how things have
changed
o Influence of time on behaviors
o Same sample must be tracked for a
long time
Cross-sectional Studies
o Subjects already at different stages
compared at a single point in time
o Stat tests to analyze effects less
powerful than within subjects group
o Runs risk of people in groups might
differ in other characteristics
o Neither ^ can be used to infer cause
and effect
o Antecedents not manipulated
o Imposition of units high
Pretest/Posttest design
o Assess effects of naturally occurring
events (approval ratings before and
after presidential speech as example)
o If studies beyond single brief research
session, too many other things that
could influence
o Used to test effects of foreseeable realworld events
o Can be extremely low in internal
validity
o This much flexibility, ought to use
experimental instead
Quasi lack artificiality of experiments; used as
sources
Recap:
o Quasi differences in preexisting
groups of subjects, group diff in some
variables or different treatments to
preexisting groups
o Ex
post
facto
characteristics
(naturally occurring) of preexisting
groups of subjects
o Nonequivalent compares effects of
diff.
treatment
conditions
on
preexisting grps
o Longitudinal changes across time by
measuring behavior of same subjects
(diff points in time)
o Cross-sectional changes across time;
subjects at already different stages
o Pretest/posttest
effects
of
event/treatment
by
comparing
behavior before and after
Formulating the Hypothesis
Most psychological research is designed to test
hypothesis
Approaches to arriving: induction, deduction,
building on prior research, serendipity and
intuition
The end of the long process of thinking of a
topic
Is the thesis/main idea of the experiment
Statement on predicted relationship between
at least two variables
Speculation, guess, hunch
Nonexperimental hypothesis how events,
traits, behaviors might be related (not cause
Synthetic Statements
o Can either be true or false (some
chance it is true, some chance false)
o Can be supported/contradicted
o Nonsynthetic
Testable Statements
o The means for manipulating the
antecedent
and
measuring
the
resulting behavior must exist
o Elaborate space travel, dreaming in
dogs etc. untestable (for now)
o Untestable
hypo
not
necessarily
useless
Falsifiable Statements
o Disprovable by research findings
o Need to be worded so failures to find
predicted effect can be evidence that
hypo is false
Parsimonious statements
o Simplest explanation is preferred
o Simple over one that requires many
supporting assumptions
Fruitful statements
o Could generate new research, leads to
new studies
o Fruitful when we could think of new
studies if hypo is supported
The Inductive Model
Usual approach
Take
note
of
all
potentially
relevant
observations
Basic
components:
IV,
DV,
operational
definitions, control
Independent Variable
o Dimension that the experimenter
intentionally manipulates
o Antecedent chosen to vary
o Independent values are created by
experimenter, not affected by anything
else in the experiment
Subject
variables
characteristics
of
subjects
themselves
quasiexperimental
variables
as
independent variables
Choice
between
diff.
levels
of
measurement: choose highest level
possible provides more info,
statistical tests can be used (for
interval and ratio, more powerful)
o Recap:
Reliability
o Consistency and dependability
o Apply in more than one experiment,
work in similar ways each time
o Ex. Hungry condition, always show
signs of hunger
o Haphazard definitions better needed
o Measured operational definitions
several sets of measurement same
results each time
o Select
measuring
instruments:
standardized tests
o The more accurate, the more likely it is
reliable
o Interrater reliability
Different
observers
take
measurements
of
same
responses
Agreement
between
measurements
Eval
using
statistical
techniques
o Test-retest reliability
Internal consistency of
items test for single
construct
high
degree
of
internal
consistency if they are
reliably
measuring
same variable
Split-half reliability
split test into two
halves
at
random,
compute coefficient of
o
reliability
between
scores
^two halves should
correlate strongly
Cronbachs
a
considers correlation of
each item with every
other item
Validity
o Actually studying variables we intend
to study
o Could be reliable but not valid not
accomplish goals of experiment
o Compare consequences of various
procedures if all produce anxiety,
they should all lead to same observable
signs
o Necessary to provide evidence for
validity manipulation check
o Face validity
Measured
operational
def:
whether we measured what we
intended
to
measure
All
aspects
of
content
represented appropriately?
High
content
validity
measuring
instrument
not
evaluating other qualities we
do not intend to meas.
They
should,
if
we
are
measuring what was intended
to be measured
o Concurrent Validity
Comparative
rather
than
predictive
Comparing
scores
from
instrument with other known
standard
Transition
from
theory
to
research application
o
o
o
o
o
o
Boredom, fatigue
People
perform
differently
second time they are tested
Could
be
mistaken
for
treatment effects
Selection
Not
assigning
subjects
randomly
Nonrandom
assignment
Attrition rate
in particular
treatment high red flag
Confounding variable
Selection Interactions
A family of threats
If
subjects
not
randomly
assigned to groups (or if
random assignment failed to
balance out differences among
subjects) + another threat
Planning the Method Section
For replication
Participants
o Describe subjects
o Track genders, ages
o How many participated
o Did not complete study (report) +
reasons
o Not filling out questionnaire, not
included in data, report
o How they were recruited
o What they received in return
Materials
o All items presented to subjects
o Any special or unusual equipment
o Careful notes about everything you did
in your experiment
o Verbal instructions
Procedures
o Describe all procedures used in
experimental sessions
o List in chronological order
CH. 8: Controlling Extraneous Variables
Physical Variables
Elimination
o EV should be taken out amap
o If cannot be eliminatedmake them
constant
Constancy
o Keep
all
aspects
of
treatment
conditions as nearly similar as possible
o Not eliminate extraneous variable,
make them stay the same in all
treatment conditions
o Keep mechanics of procedure the same
o Write out instructions to subjects
o They should get exactly the same
conditions
o Same amount of time allowed
Balancing
o Confounding: something systematically
changes along with IV
We want participants to be as
nave as possible
Controlling
demand
characteristics: Single blind
experiment subjects do not
know the treatment they are
getting
*Placebo effect
Cover
stories
false
explanation
for
procedures
used in study; subjects would
not
look
for
another
explanation
deception
(should be used sparingly)
o Experimenter Bias
Might
also
treat
subjects
differently depending on what
o
Social
they
expect
from
them
(Rosenthal effect)
Errors might also be committed
in recording data (misread
scale/score item to support
hypo)
To
control:
Double-Blind
Experiments: both participant
and experimenter not aware
which
treatment
is
given
(cannot bias responses in a
systematic way)
Not possible for single/double
try not to assign subj to
conditions until after we have
finished interacting with them
Give subj written instructions,
listen to tape etc.
Personality Variables
Experimenters
o Warm and friendly vs. cold and aloof
o High social desirability likeable
experimenters
o Improves subjects performance might
affect experiments external validity
o Extremely
important
to
maintain
consistency
in
interactions
with
subjects
Volunteer subjects