Measurement

1
Measurement
PSY 301
2
PSY 301
Science (psychology) starts with observation
The scientific method is based upon obtaining
data through planned, systematic, controlled and
repeatable observations.
Most scientific observations are translated into
numbers, i.e. they are quantitative -- even if
only to count the observations of a certain
quality
3
Measuring Behavior

• Psychologists observer behavior
• How do we associate numbers with behavior
• This is the issue of measurement!
• What do the numbers mean?
• Are the numbers consistant

4
PSY 301
The importance of measurement
We gain nothing if we cannot assess, gauge,
appraise, evaluate, classify, rank, grade, sort,
arrange, rate, weigh, etcetera, the outcomes.
It has been said that psychology cannot come up
with theories and laws of behavior until we can
describe and measure things with reasonable
accuracy.
5
PSY 301
Measurement (definitions)
The assignment of numerals to objects or events
according to rules (Stevens, 1951).
A systematic way of assigning values (numbers or
names) to objects and their attributes.
The assignment of numerals to objects or events
according to rules that permit important
properties of the objects or events to be
represented by the properties of the number
system.
6
PSY 301
Examples of psychological measurements

• - Your grade (in any class)
• - Ratings of perfume on a scale from 1-5
• (1liked least, 5liked most)
• Judgments of line length
• Number of males in Psych club
• Average length of time to graduate

7
PSY 301
Experimental Method (one process used in the
scientific method)
Manipulate independent variables
independent
Measure dependent variables
dependent
Control other relevant variables
Control
- eliminate confounding variables
confounding
- minimize extraneous variables
extraneous
8
PSY 301
Variables that need to be measured
independent variable - the variable/s that the
experimenter manipulates to understand how
it/they influence behavior
dependent variable - the variable/s (behavior,
outcome) that the experimenter is interested in
control variables - variables that the
experimenter wants to keep the same across all
conditions so that they do not affect behavior
extraneous variables - variables that may affect
the outcome behavior but that the experimenter is
not directly interested in but cannot control
confounding variables - variables that vary
systematically with the independent variable/s
(bad - avoid if at all possible)
9
PSY 301
Variables that cannot be measured directly
Often in psychology, we are interested in
variables that are not tangible (cant be touched
or observed directly) - things like honesty,
anxiety, fear, attention, etcetera.
These variables are called constructs (Goodwin,
pg 85).
In order to study these constructs, as well as
any other variable, we must operationalize them.
10
PSY 301
Operational definition
Defining exactly the operations that will be used
to create/measure the variable.
11
PSY 301
Why the need for operational definitions?
As stated, some variables (constructs) cannot be
observed directly so researchers must decide how
they will measure them.
Researchers need to be able to state exactly what
they did so others can replicate the study (i.e.
in the Method section of an APA style paper).
Researchers may have different ideas about how
best to measure a variable under certain
conditions.
Limitation Operational definitions restrict the
generalizability of the results.
12
PSY 301
Reliability
Definition how repeatable the measure is - if I
took the same measure more than once, will I get
the same answer.
If a measure is different each time you take it,
we say that there is variability in the measure.
The more variability there is, the less reliable
the measure is.
13
PSY 301
Types of Reliability (not in text)
Interrater reliability - the level of agreement
between different raters.
Test-retest reliability - the level of agreement
between scores on a test taken at 2 different
times.
Parallel-forms reliability - the level of
agreement between test scores of two tests that
are parallel forms of each other.
Split-half reliability - the level of agreement
between the scores of two tests taken
simultaneously.
14
PSY 301
Reliability - but what about accuracy?
Accuracy - refers to how closely a measure
corresponds (matches) a known standard.
Can a measure be reliable but not accurate?
Yes. Think about a thermometer in which the
glass tube has slipped down so it no longer is
lined up correctly with the temperature markings.
Can a measure be accurate but not reliable?
Depends how you think about it
15
PSY 301
Validity
Definition in regards to a measure, validity
refers to how well it measures what it was
intended to measure.
16
PSY 301
Types of Validity
Face validity - how well a measure appears (face)
to measure what it was designed to measure.
Construct validity -- Is there such a construct?
does the measure reflect the construct? Tied to
theory.
Content validity (not in text) - how adequately
the contents of a test samples the knowledge,
skills, or behaviors that the test was supposed
to measure.
Criterion (outcome) validity - whether the
measure 1) accurately predicts future behavior
and 2) is meaningfully related to another measure
of the same behavior.
17
PSY 301
Validity Reliability - how are they related?
Example - skull circumference as a measure of
intelligence.
Is this a reliable measure?
Yes. Sources of measurement error?
Is this a valid measure?
No. Why?
18
PSY 301
Validity Reliability - how are they related?
Example - Performance on pop quiz as measure of
knowledge of research methods
Is this a reliable measure?
Probably not. Test anxiety. Bad day. Studying
Is this a valid measure?
Probably yes. Main determinant is your knowledge
19
PSY 301
Scales of Measurement
Numbers can be used in a variety of ways
Identification (as labels) example addresses
Magnitude (to indicate more or less
than) example relative cost of candy bar and
bicycle
Equal intervals (to indicate exact
relationships) example 500-300 900-700
Absolute zero (0 has an absolute
meaning) example money - 0 means NO money
Each of these specifies a different scale of
measurement.
20
PSY 301
Types of Measurement Scales
Nominal scale - sorts objects into categories
Ordinal scale - ranks objects/events by magnitude
Interval scale - intervals between adjacent
values are equal
Ratio scale - all of above AND a meaningful zero
point
21
PSY 301
Types of Measurement Scales
22
PSY 301
Choosing the Measurement Scale (height)
Nominal category names A (1) (without regard to
actual height) B (2)
Ordinal meaningful labels tall medium
short
Interval differences 8 taller 4
taller same 4 shorter 8 shorter
Ratio actual height in inches (0 means no
height)
23
1 Chicago 2 Cincinnati 3 Houston 4 Milwaukee 5
Pittsburgh 6 St. Louis
24
1 2 3 4 5 6
25
PSY 301
Bottom line about measures
How you choose to measure a variable (and thus
the measurement scale) will determine what you
can do with the measures and what you can say or
conclude.
26
(No Transcript)