Research 1 Midterm Flashcards
Pretest/Posttest control group design
INDEPENDENT GROUPS
- controls for history, maturation, selection bias, testing, instrumentation effects
- threat to internal validity is attrition
- threat to external validity is interaction of treatment and testing
Cause & effect; RCTs
Change scores
Two group pretest/posttest design
INDEPENDENT GROUPS
- controls for history, maturation, selection bias, testing, instrumentation effects
- threat to internal validity is attrition
- threat to external validity is interaction of treatment and testing
Difference between two treatments
Unpaired t-test: interval/ratio data
Mann-Whitney U: ordinal data
Multigroup pretest/posttest
INDEPENDENT. GROUPS
- controls for history, maturation, selection bias, testing, instrumentation effects
- threat to internal validity is attrition
- threat to external validity is interaction of treatment and testing
Control vs. treatment 1 vs. treatment 2
ANOVA - interval/ratio data
Kruskal-Wallis ANOVA - ordinal
Posttest only control group design
INDEPENDENT GROUPS
True experimental design
*strong internal/external validity
Unpaired t-test: interval/ratio data
Mann-Whitney U: ordinal data
ANOVA - interval/ratio data
Kruskal-Wallis ANOVA - ordinal
Factorial
Multifactorial design
Independent groups
2+ IVs
2 way ANOVA
Randomized block
Multifactor design
INDEPENDENT GROUPS
Concern over an extraneous factor such as age, sex, health→ factor becomes attribute variable
- 2-way ANOVA
- Multiple Regression Analysis
Nested
MULTIFACTOR DESIGNS FOR INDEPENDENT GROUPS
Some attribute variables can’t cross; experience becomes 3rd independent variable
One way repeated measures
REPEATED MEASURES
All subjects exposed to all levels of one treatment variable
- Internal threat is carry over
- Latin Square used for sequencing treatment
2 way ANOVA
Cross over
Repeated measures
½ subjects receive treatment A followed by B/ ½ subjects receive treatment B followed by A
Paired t-test
2 way ANOVA (interval/ratio)
Wilcoxon signed ranks test (ordinal)
Two way design with two repeated measures
MULTIFACTOR DESIGNS FOR REPEATED MEASURES
Repeated measures with more than one independent variable
2 way ANOVA with 2 repeated measures to measure main and interaction effect
Mixed Design
MULTIFACTOR DESIGNS FOR REPEATED MEASURES
2 independent variables, 1 repeated across all subjects, and the other randomized to independent groups
2 way ANOVA with 1 repeated measure for main effect
One group pretest/post test
Quasi
Time becomes IV
- threat to internal validity: no comparison group
- threat to external validity: potential interactions with selection b/c of no comparison group
Paired t-test (ratio/interval)
Wilcoxon signed rank test (ordinal)
One way repeated measures over time
Quasi
Time becomes IV
Time course of a disease
Repeated measures ANOVA
Time Series
Quasi
Effect of treatment on physiological or psychological variables over time
Graphic visual analysis Multivariate methods (ARIMA)
Nonequivalent pretest/post test
Quasi
Subjects in fixed groups or self-selected
Threat to internal validity is interaction of selection with history and maturation
Unpaired t-test: interval/ratio data
Mann Whitney U: Ordinal data
Nonequivalent posttest only control group
Quasi
Pilot study or exploratory to generate hypothesis for further study
Regression
ANCOVA
Continuous Variable
can take on any value along a continuum within a defined range
Discrete Variable
described only in whole numbers (HR,BP)
Construct
an abstract concept that is invented or created to represent immeasurable behaviors or ideas
Sources of Measurement Error
- Rater error
- Inaccuracies in the measuring instruments
- Variability of the characteristics being measured
Development of testing instruments
Involve a specific protocol that maximizes the reliability of the instrument
Errors are identified and then controlled or eliminated ---Controlled or eliminated by: •Careful Planning •Clear Operational Definitions •Inspection of Equipment
Correlation
- Correlation-degree of association
- Not cause and effect research
- State two variables are related (X,Y)
- No true variable manipulation
- Positive and negative correlations
r
Pearson product
rs
Spearman rank
ICC
Intraclass
●Model 1-Raters randomly chosen for different subjects- difference among subjects
●Model 2-Each subject assigned a random set of raters -interrater
●Model 3-Each subject assigned the same set of raters -intrarater
k
Percent agreement
Test-Retest Reliability effected by:
- Testing Effects - practice or carry over effects
- Rater Bias - same rater can be influenced by the memory of the first score
- Controlled by blinding tester
- Test-Retest Interval Time