Quiz 4 Flashcards
6 test Validity
Face, Content, Predictive, Concurrent, Convergent, Discriminant
Internal/External Validity
Internal: the extent to which an experiment accurately measures what it intended to measure (depends on how well the experiment is designed and controlled for relevant extraneous factors
External: an experiment can be generalized to a larger content (good study design)
Threats to Internal Validity
History, Maturation, Reactive Pretest, Instrumentation, Statistical Regression, Differential subject selection, Attrition, Interaction of factors
Generalizability
Threats to Generalization are threats to external validity! Do patterns from the sample apply to the population?
- Subject selection (does it accurately represent the population)
- Interactive pretest (when a reactive pretest interacts with the dependent variable)
- Reactive Arrangements (effect seen is specific to experimental context; Hawthorne effect: subjects’ behavior changes because they know they are being studied/observation)
- multiple treatment interference
Face
Does the test look and sound like it accurately measures the construct (ex: title)
Content
Does the test cover the relevant domains of the construct
Predictive
does the test predict things it should predict (ex: do lower scores correlate with lower reading proficiency?)
Concurrent
Does the test discriminate between relevant groups (ex: are scores significantly lower for kids with reading delays vs those with typical reading development?)
Convergent
does the test get similar results to other tests measuring the same or similar constructs (ex: do test scores correlate well with other test scores covering the same topic?
Discriminant
does the test get different results than other tests measuring different constructs (ex: do test scores fail to correlate with with scores of different constructs?)
Single-Subject Designs
focus is on individual behavior; can include multiple subjects, but they are not analyzed primarily as a group
Between-Subjects Group Design
different groups of subjects are exposed to different levels of the independent variables; need to account for group equivalence
- subject randomization
- subject matching
Within-Subjects Group Design
same subjects are exposed to different levels of the independent variables; requires fewer subjects; need to account for sequencing effects
- sequence randomization
- counterbalancing
Mixed Group Design
Use both between-subject and within-subject variables