Psychology 2020
Studying Behavior
Unit 2
1
Variables
► Variables
in research are those things which can
change (have more than one value).
 Some examples are task difficulty (high or low, amount
of reinforcement (1,2, or 3 candies), or number of
responses made in an hour.
► There




are four general categories of variables.
Situation variables
Response variables
Individual differences in participants variables
Mediating variables
2
Research Variables
► Situation
variables are often manipulated by the
experimenter.
 They become the “independent” variables in the
experiment or they are held constant (controlled)
► Response
variables are often measured as the
results of the experiment (what the participants
are doing differently after receiving the
experimental manipulation)
 These variables become the “dependent” variables of
the experiment.
3
Other Variables
► Individual
difference variables related to the
participants.
 These are often minimized or controlled such that the
experimental results are more clearly seen.
► Mediating
variables are inferred psychological
processes that are often used to explain research
outcomes.
 These variables are invented by the experimenter and
are related to the theory that produced the experiment.
4
Operational Definitions
► Measured
variables must be defined clearly
► When we define a variable in terms of the specific
method used to produce, manipulate and/or
measure that variable we have defined it
“operationally”.
► Examples:
 Intelligent is operationally defined as a score above 120 on an IQ
test.
 Courage is operationally defined as doing things that could produce
physical pain.
 A tantrum is operationally defined as screaming loudly for more
than 30 seconds when a toy is removed from play.
5
Relationships Between Variables
► Linear
relationships
 Positive linear relationship.
►Both
variables increase or decrease at the same rate
in the same direction.
►Represented as a straight line going up on a graph
 Negative linear relationship
►As
one variable increases the other variable
decreases at the same rate.
►Represented as a straight line going down on a
graph
6
Linear Relationship Graphs
Positive
Negative
7
Relationships Between Variables
► Curvilinear
relationship
 As one variable increases the other variable
both increases and decreases over time
(nonmonotonic).
 Represented as a curved line on a graph.
► No
Relationship between variable changes
 Represented by a flat, horizontal line on a
graph.
8
Curvilinear Relationship Graph
9
Experimental vs.
Nonexperimental Methods
► Nonexperimental
methods simply measure
the variables of interest in an attempt to
discover a relationship between them (if
there is a correlation between them).
► Experimental methods systematically
change some variables and measure the
effects of these changes on other variables
to assess the relationship between them.
10
Nonexperimental Methods
► These
methods are good for discovering if two or
more variables covary.
 Covariation or correlation is useful for prediction (if you
know the value of one variable you can predict the
value of correlated variable.
► Examples:
High SAT scores are positively
correlated with success in college, seat belt use is
negatively correlated with severe injury in auto
accidents.
11
Nonexperimental Methods
► Causality
MUST NOT be inferred from
correlational methods
 The problem of determining the direction of
cause and effect.
►Which
change came first?
 The third-variable problem
►Did
some other variable produce the changes in the
two measured variables?
12
Experimental Methods
► Experimental
methods introduce techniques of
control that allow inferences of causality.
 Extraneous variables are held constant and/or their
effects are randomized across experimental conditions.
► This
solves the third variable problem.
 One variable is systematically changed (the independent
variable) while changes in the other variable (dependent
variable) are measured.
► This
solves the “cause/effect direction” problem.
13
Experimental Methods &
Causality
► To
infer causality we must consider the
following
 Temporal order (cause comes before effect)
 There must be covariation between variables
(as one variable changes there must be a
corresponding change in the other variable)
 Other explanations for the change must be
eliminated
14
Determining which Research
Method to Use
► Requirement
for experimental control can
make experiments less real
 Limits the questions asked and reduces the
generalization of the findings
 Nonexperimental field studies are more realistic
but extraneous variables are less controlled
► Ethical
and practical concerns limit what can
be manipulated and controlled
15
Determining which Research
Method to Use
► When
description and prediction are all that
is required nonexperimental methods are
used.
► Sometimes both experimental and
nonexperimental methods are needed to
answer the research question.
16
Validity in Research
► Validity
refers to “truth” and accuracy.
 If you have accurately and truthfully defined your
variables (good operational definitions) your research
has good CONSTRUCT validity.
 If your experiment allows you to accurately determine
that the independent variable changes produced the
changes in the dependent variable your experiment has
good INTERNAL validity.
 If your experimental results generalize to other
populations and settings your experiment has good
EXTERNAL validity.
17
The Theoretical Construct
►A
Construct is an abstract or general idea
inferred or derived from specific instances.
► Theories are made up of constructs.
 Theories of mental illness are made up of
theoretical constructs like “schizophrenia”,
“paranoia”, “obsessive-compulsive”, etc.
 Theories of learning are made up of theoretical
constructs like “memory”, “reinforcement”,
“expectation”, etc.
18
Measurement Reliability
► Reliability
is analogous to consistency.
 If you measure something several times and get the
same result each time, the measurement is considered
reliable.
► Reliability
is different from validity.
 A valid measure truly measures what it purports to
measure.
 A reliable measure produces the same result each time
the measurement is taken.
► Both
validity and reliability are needed in research.
19
Measurement Reliability
► Measurements
consist of two components. Both
must be inferred because they can’t be directly
observed.
 True score
► The
real score of the variable that exists hypothetically and is
considered a constant.
 Measurement error
► The
portion of the measurement result that varies from
measurement to measurement
higher the correlation (r near 1) between two
measurements, the more reliable the
measurement.
► The
20
Measurement Analogy
21
Assessing the Reliability of
Psychological Measures
► Test-Retest
Reliability
 Take the measurement at two different times and
correlate the results (r =.8 to be considered reliable).
► Internal
Consistency Reliability
 Split test in half an correlate the scores on the first half
with the scores on the second half.
 Correlate every item score with every other item score
(Cronbach’s alpha)
 Correlate every item score with the total score.
► Interrater
Reliability
 Correlate rater scores with each other.
22
Measurement Validity
► If
a measurement tool is valid, it is really
measuring the construct it purports to
measure rather than some other
characteristic.
 A valid measurement tool is said to have
“Construct Validity”.
23
Methods to Assess Construct
Validity
► Face
Validity (does it appear valid)
 An IQ test that only asked questions about a
person’s income, address and clothing
preferences would have little face validity.
► Criterion-Oriented
Validity (does it allow
accurate predictions)
 A safe-driving test that passed people who
continued to have driving accidents would lack
criterion-oriented validity.
24
Forms of Criterion-Oriented
Validity
Validity- ability to predict something it
should theoretically be able to predict.
► Predictive
 A measure of math ability accurately predicts who will
be a successful engineer
► Concurrent
Validity- ability to distinguish between
groups that it should theoretically be able to
distinguish between
 A measure of anxiety distinguishes individuals with an
anxiety disorder from those with major depression.
25
Forms of Criterion-Oriented
Validity
► Convergent
Validity- results from the measure are
similar to other measures of the same construct
 The block assembly IQ test yields high IQ scores with
the same people who got high IQ scores on a spatial
problem solving IQ test.
► Discriminant
Validity- results from the measure are
NOT similar to measures of theoretically different
constructs.
 An abstract-reasoning test yields different results that a
rote-memory test
26
Measurement Reactivity
► Reactivity
in measurement is when the very
act of measuring a variable changes that
variable.
 Example: recording the amount of food eaten
each a day causes the person to eat less
► Using
unobtrusive measures reduces the
probability of measurement reactivity.
27
Measurement Scales
► Nominal
 No numerical properties, only named categories
 Examples: male or female, married or single
► Ordinal
 Has numerical properties but only for the
purpose of ordering the categories from first to
last
 Examples: Academic grades, places in a race,
ranking schools or sports teams
28
Measurement Scales
► Interval
 Has numerical properties with equal distance between
each number on the scale but no zero point
(representing a total absence of the thing being
measured)
 Example: Rating beauty on a scale of 1-10
► Ratio
 Same as the interval scale (equal intervals between
numbers) except there is a zero point.
 Examples: time measures, response rate measures,
other physical measurements.
29
Name the Measurement Scale
•
•
•
•
A food critic rates restaurants according to the quality of
food, service, and atmosphere. She assigns 4 forks for
excellent, 3 for good, 2 for fair, and 1 fork for poor.
Ordinal
Researchers have identified styles of leadership as
relationship-oriented or task-oriented.
Nominal
Organizing pictures of your past lovers from “best sex to
worst sex”.
Ordinal
A Psychological measure of an individual’s shyness.
Interval
30
Name the Measurement Scale
•
•
•
•
Getting a count of the number of automobiles on the
highway that were made in Europe, Japan and the USA
(totals for each group).
Ratio
The score you receive on the Beck Depression scale (118 is not depressed, above 18 is depressed).
Interval
Deciding if a house style is Victorian, Tudor, or Greek
Revival.
Nominal
Measuring how long it takes someone to drink a beer.
Ratio
31
Class Discussion Question
Page 97 question number 2 (identify the scale
of measurement in each of the examples)
32
33
General Observational
Approaches
► Quantitative
approaches
 Emphasize numerical properties of the thing
being observed.
 Observational results can be statistically
analyzed.
 Examples
►Hours
spent studying
►Ranking of task difficulty
►Percentage of on-task behavior
34
General Observational
Approaches
► Qualitative
approaches
 Emphasizes the nonnumerical aspects of the thing being
observed.
 Usually consists of descriptive paragraphs.
 Statistical analysis of data in this form is not possible.
 Examples
► Diaries
► Film
documentaries
► Written summaries of main themes and/or styles observed
35
Applying the Concepts
During a previous class period you watched
brief cartoons and were asked to decide
which one contained the most violence.
a) What did you do that is considered an
example of the “quantitative observation
method”?
b) What would be done differently if I asked you
to base your decision on a “qualitative
observation method”?
36
Naturalistic Observation
► The
researcher observes and records
behavior in its natural setting over a period
of time in a descriptive format.
 A qualitative research method if the data is not
converted to numbers, if numbers are recorded
it can become a quantitative method.
► Observations
are interpreted and these
interpretations are confirmed or
disconfirmed by multiple observations.
37
Issues in Naturalistic Observation
► Participation
 Does the observer participate in the setting that is
being observed and will participation disrupt the
setting making the observations less accurate?
► Concealment
 The purpose of the observation and/or the
observer is hidden making observations more
accurate but potentially unethical.
► Scope
of observation
 Not everything can be observed so the scope of
observation is limited by the experimenter and this
could bias the results.
38
Systematic Observation
► Careful
observation of specific behaviors in
specific settings using structured,
quantifiable methods.
► Observations are not described in paragraph
form but numerically coded on specially
designed data sheets.
► Systematic observation results can be
analyzed statistically.
39
Methodological Issues
Systematic observations can be aided with the use of
technical equipment (video tape, computers, clocks,
etc.)
► The use of systematic observational methods may
influence what is being recorded (reactivity problem).
► Reliability of observations can be measured by using
more than one observer or recording method.
► When all behavior cannot be measured, sampling
methods are used to get accurate estimates of total
behavior.
►
40
Case Studies
► Case
studies are usually extensive descriptions of
one individual or organization over a period of
time.
► Case studies are one form of naturalistic
observation research
► Case studies are conducted when an individual is
unusual, has a rare disorder or noteworthy
condition.
► Case studies provide valuable information that is
not available in other forms of research.
41
Archival Research
► Archival
research uses data that was
previously collected and compiled (usually
for other reasons) to answer new questions.
► Two problems exist for doing archival
research
 The necessary records or documents may be
difficult to obtain
 The accuracy of information collected by
someone else is always suspect
42
Examples of archival research
data
► Statistical
Records: using US census records to
study immigration trends.
► Survey Archives: using college survey results to
predict employment trends
► Written and Mass Communication Records: using
soldiers war letters to study the impact of combat
 A Content Analysis is the systematic analysis of
existing documents to answer a research question
43