Reliability
ability to get the same results if you repeat your test/ measurement under similar conditions
more variability = less reliable
Reliability of physical measures
e.g. height or weight
repeatedly measure a fixed quantity of the variable
taking the variation in the variable to get a precise measurement result
Reliability of Population estimates
measurement of opinion, attitude
we can just estimate the average value of the variable when we take a sample of the whole population
Margin of error = there can be a variation from sample to sample
Reliability of psychological tests of measurement
tendency to change over time
test-restest reliability
that you administer the test twice with a longer amount of time in between
Use: assessing stable characteristics of individuals e.g. intelligence
variable is unlikely to change over time
-you can extend the amount of time that participants won’t remember the first test
Parallel forms reliability
Goal: not remembering the questions from the previous test
- 2nd administration: using a similar form (items are equivalent to the one in the first form)
- assessing the same knowledge, skills
Problem:
- if the forms are not that equivalent they may change the results/performance
Split- half reliability
two parallel forms of the test in one test
rated and scored separately
Validity
does a test/measurement really measure what it should/ is intended to measure
Forms of validity
Face validity
how well a measurement instrument (e.g. test of intelligence) appears to measure (just by its appearance) what it is designed to measure
e.g. mathematical ability - Mathematical problem
Content validity
how good does the test represents the learned knowledge, skills, behavior that it is actually designed to measure
e.g. final exam
Criterion-related validity
use of test score of an individual to make predictions on another scale.
e.g. using high school grades to predict the success at the college
Construct validity
not observable variable
developed to explain a behavior or idea with a theory (e.g. free will)
Having a concept in mind about
theoretical construct
(e.g. low self-esteem related to higher fast food consumption
Differences in measurement
Differences in validity
Difference in sensitivity
some dependent variable are more sensitive to manipulations than others
Range effects
2. ceiling effect
floor effect
variable reaches the lowest possible value
ceiling effect
variable reaches the highest possible value
Behavioral measurements
Reactivity of human
Demand characteristics
= cues from the experimenter/context that communicate the purpose of the study
asking what is the experiment about
Role attitude cues
signal for the participant to change the attitude; conforming to a new role of participant
Cooperative attitude
strong desire to please the experimenter
-helping science
-desire to perform as well as possible
= good participant effect