mod 4 Flashcards
design validity
truthfulness or accuracy of results
believeability
types of design validity
external
construct
internal
statistical conclusion
statistical conclusion threats
low statistical power
violated assumptions
reliability & variance
failure to use ITT analysis
ITT intention to treat
“once randomized always analyzed”
-includes every pt that is randomized
-ignores no compliance, protocol deviations, withdrawal
-maintains baseline/prognostic balance generalized from og random assignment
internal validity
extent to which IV caused observed change in DV
internal validity threats
history
maturation
attrition
testing
instrumentation
regression to the mean
selection (assignment)
internal validity social threats
diffusion or imitation
compensatory equalization
compensatory rivalry
demoralization
history
did unanticipated events occur during the study that could affect the DV?
nature of threat
- events unrelated to treatment that influence outcome
- confounding effect of specific events after introducing IV or between pre & post test
*participants involved in other activities or intervention
history possible solution
-control group or random assignment to sep effect of history
-time or olán the study to avoid predicable event
-apply statistical adjustment
maturation nature of threat
were changes of DV due to normal development or simple passage of time?
- changes in human behavior over time
ex. growing up, recovery post stroke
maturation solution
-random assignment
-random testing or tx order
-statistical adjustment
attrition nature of threat
differential loss of subjects across groups?
-loss of subs = dec in samp size
-imbalance of baseline characteristics
attrition solution
-replacing lost subjects
-ITT analysis
-re-examine data to see existing status
testing nature of threat
did the pretest or repeated test affect scores on the posttest?
-repeated testing on subjects may lead in inc familiarity or inappropriate cues provided by tester
testing solution
-eliminate multiple tests
-randomly selected experimental/control groups
instrumentation nature of threat
did the DV change bc of how it was measured?
-problems with tool used to measure variables of interest
-inapp selection of tool
instrumentation possible solution
-select app technique/tool
- calibration
-user training or practice in taking measurement
regression to the mean nature of threat
is regression from pre or post test scores?
-when subjects enter a test w an outlier-extreme value
regression solution
-trim outliers
-take repeated baseline measures
-control group
assignment threats
bias in assignment?
-unequal baseline characteristics of groups might influence outcome
assignment solution
-random assignment
-adequate inclusion & exclusion criteria
-statistical adjustment
construct validity
concerns regarding IV/DVs:
-how they are conceptualized
-well establish/correctly labeled
-measure truly represents variable
construct validity threats
-operational definitions
-timeframe
-multiple treatment interactions
-experimental bias
hawthorne effect
effect of subjects knowledge that they are part of study on the performance
external validity
results may be generalized to outside experimental situation
external validity threats
-biased sample selection
-setting difference/ (diff from clinic)
-time (changed circumstances)
