EPPP Research Design
Terms
undefined, object
copy deck
- Use: protocol analysis
-
type of content analysis
to study cognitions underlying problem-solving
subjects asked to "think aloud" - Use: interval recording
-
method of behavioural sampling
devide a period of time into discrete intervals and record whether the behaviour occurs
used for behaviours that are (1) complex, or (2) have no clear beginning or end
e.g. laughing, playing - Use: event sampling
-
method of behaviour sampling
records each occurrence of a behaviour during a preselected event
used for behaviours that are (1) rare, or (2)leave a permanent record
e.g. completed test - Shortcomings of case studies (2)
-
1. cannot draw conlcusions about exact relationship between variables
2. information may not be generalizable to other cases -
A shortcoming with surveys
Methods to avoid this (4) -
Nonresponse bias
1. follow-up
2. info about study
3. small reward
4. pre-contacting individuals by phone - Diff: true experimental research vs. quasi-experimental research
- random assignment
- Diff: random assignment vs. random selection
-
RA - observed effect on DV more likely to be caused by IV
RS - generalize findings from sample to population -
Def: cluster sampling
Def: multistage cluster sampling -
selecting uinits or groups of individuals from population
e.g. schools, hospitals
can randomly select individuals from each unit - Def: placebo control group
-
comparison group exposed to nonspecific aspects of the treatment
e.g. expectations of improvement - Def & Use: ANCOVA
-
analysis of covariance
statistical control of extraneous variable (the covariate) - Def: systematic vs. random error
-
SE - error due to extraneous variables
RE - error due to random fluctuations in subjects, experiment conditions, methods of measurement - Controlling variability due to extraneous variables
-
random assignment
hold extraneous variable constant
matching
blocking
ANCOVA - What does internal validity demonstrate?
- casual relationship between IV and DV
- Def: statistical regression
- extreme scores closer to mean when test readministered
- Def: analogue studies
- conducted in lab or other artificial settings
- What does external validity demonstrate?
- relationship between IV and DV can be generalized
- Threats to external validity
-
Interaction bet testing & treatment
Interaction bet selection & treatment
Reactivity - Use: Solomon four-group design
- to control pretest sensitization
- reactivity
- awareness of being observed in study
- demand characteristics
- cues in experiment informing subject how to behave
- Experimenter expectancy
- bias an experimenter introduces into sutyd as a result of expectations about sutyd's outcomes
- Single-blind vs. double-blind study
-
S-B = subs don't know what group they are in
D-B = subs AND experimenter don't know - Def: Hawthorne effect
- subjects perform better due to attention
- Def: Solomon four-group design
- controls effects of pretesting
- Types of group designs
-
between-groups/subjects
within-subjects
mixed - Factorial design
- when a study includes 2 or more IVs
- Benefits of factorial design
-
can analyze:
1) main effects of each IV
2) interaction bet IVs - Main effect
- effects of different levels of a single IV
- Interaction
-
impact of one IV differs at different levels of another IV
If significant interaction... main effects should be interpreted with caution - How do you control for carryover effects or multiple tx interference
- counterbalancing
-
Which is more powerful?
Between-subjects design
Within-subjects design -
within-subjects design.
fewer subjects needed
controls for individual diffs - Autocorrelation or serial dependancy
- performance on post-tests likely to correlate with performance on pretests
- Eg: Mixed design or "split-plot"
-
counterbalnced design
makes comparisons bet groups and within groups - Def: reversal (withdrawal) design
-
single-subject desgin with at least 2 baselines and 1 tx.
tx is withdrawn or reversed
e.g ABA or ABAB - 3 types single-subject designs
-
AB
reversal (ABA or ABAB)
multiple baseline - Def: factorial design
- any research design that includes 2 or more factors (IVs)
- Use: formative program evaluation
-
when program is in process of being developed
to see if program is being implemented as intended - Use: summative program evaluation
-
once program has been developed
assessing program's effectivenss to decide on continuation or expansion