quiz 2 Flashcards
psychology constructs
variable that represents something that is unobservable
- ex. mental health,
background, stress
Operational Definitions
concrete definition of an unobservable construct (important for replication)
Operational Measures
Important to define for replication
Self-Report Measures: measuring _ using a _ scale, how something is being measured
Behavioral Measures: fidgeting
Physiological Measures: pupil dilation, heart rate
levels of measurement
nominal (categories-eye color), ordinal (rank order), interval (SAT score, temperature), ratio (true zero, absence of trait, ex. number of children)
Reliability in Measurements
Alternate Forms: consistency between versions (good when measuring stuff repeatedly
-ex. SAT, make 2 tests and make sure results are consistent
btwn the two of them
correlation -1-+1,
measured with pearson’s r
(correlation, assesses
linear relationships)
Test-Retest: consistency over time
- ex. temperature, weight,
scale (should be consistent)
Internal Reliability: consistency across items to produce similar results
- split half (split
test/instrument in half
and compare), chronbach’s
alpha
- difficult to be consistent
validity in measures
Face validity: does measure appear to assess the construct? test/measurement tool appears to measure what it claims to measure (might not look valid, but could be important)
Content validity: does measure cover the content of interest?
- degree to which the items
on a measure the underlying
characteristic of interest
~ ex. 1st exam on 1st unit
goes over the content of
1st unit
criterion validity: does measure correlate with other measures of the same construct?
- does your measure of
intelligence correlate with
others test?
discriminant validity: is your measure/test unique?
Sources of Bias in Measurement (DEPS)
Participant Reactivity:
change behavior because they know they’re being observed (ex. recruiters while playing, or boss in workplace = hawthorne effect)
Socially Desirable Responding: ex. code switching, talking about drug use or sexual behavior
Demand Characteristics: guess at the purpose of the experiment and change their behavior based on cues (might have to use deception to get honest response)
Experimenter Bias: may have bias on race, gender, etc. (ex. Rosenthal: bloomers or standard students)
Ways to Minimize Bias in Measurement (ADS)
Automation: automate by using a computer
Double-Blind Procedures
Standardized of Industrializations, Scoring, etc. :every person who gets experimented on gets informed by a script
Types of Distributions
Frequency Distributions
- count of responses per
category
Histograms (frequency bars)
- freq. dist. in a graphical
format
- advantage: can see what
distribution looks like (i.e.
normal/skewed)
Measures of Central Tendency
MODE
- pros: good for nominal data,
always going to represent
something in your sample
- cons: could have mult
modes
MEDIAN
- pros: not affected by
extreme scores
- cons: could have number
that doesn’t exist in your
sample, not as stable as
mean
MEAN
- pros: has high sampling
stability, good for inferential
stats
- cons: cannot be used for
nominal data, sensitive to
extreme outliers, could be a
value that does not exist in
your sample
Measures of Dispersion (Variance)
Range
Variance: sum of squared deviations divided by # of scores (how much scores are spread around the mean)
Standard Deviation: square root of variance
Describing Distributions
Skew: Positive vs. Negative
Kurtosis (bell curve):
- Leptokurtic: skinny & tall
- Platykurtic (PLAT=FLAT): fat
& flat
Normal Distribution/Mesokurtic
Standard Deviation percentages for normal dist
-1 to +1= 68% of scores/sample
-2 to +1= 95%
-3 to +3= 99%
