Chapter 11 - Special Designs Flashcards
single case designs
- participants’ behaviour is first measured during a baseline control time period, then the manipulation is introduced during a treatment period.
Ex. ABA reversal design, ABAB reversal design, Multiple baseline design
ABA (single-reversal) design
- Go from baseline to treatment back to baseline
- Ex. Give food without energy drink for 5 days, give food with energy drink for 5 days, give food without energy drink for 5 days
- Big problem: because you’re only measuring something 1 time with 1 person, it’s hard to tell whether or not the treatment was effective or if something else caused that change
- How to solve this problem: multiple reversals (ex. Abab, ababab)
- Generally suffer from lack of generalizability (because the 1 person you’re running the tests on may not be representative of everyone else) -> must be replicated if you want to infer something larger
quasi-experiments
- No randomization
- Uses pre-existing groups (or allows participants to sort themselves into groups)
- Randomization is not possible or unrealistic
types of quasi-experimental designs
- One-group post-test only
- One group pretest-posttest
- Nonequivalent control group
- Nonequivalent control group pretest-posttest
- Multiple repeated measures
one-group post-test only
- type of quasi-experimental design
- no pre-test, no control group -> no internal validity
- ex. Quit smoking program offered -> students sign up (forming your participant group) -> implement the program (treatment) -> smoking frequency is now 1 pack/week (posttest)
one-group pre-test post-test
- type of quasi-experimental design
- no control group, but there is a pre-test which gives us some kind of comparison -> a bit of internal validity
- ex. quit smoking program offered -> students sign up (forming your participant group) -> smoking frequency is measured as 4 packs/week (pretest) -> implement the program (treatment) -> smoking frequency is now 1 pack/week (posttest)
nonequivalent control group
- type of quasi-experimental design
- Experimental group is tested alongside a group not receiving treatment
- Problem: you don’t know whether your groups were equivalent because no random assignment is used (you group based on existing natural groups)
nonequivalent control group pre-test post-test
- type of quasi-experimental design (and one of the most internally valid)
- Experimental group is tested alongside a group not receiving treatment
- Although random assignment isn’t used, you can give a pretest to see whether your groups were equivalent
Multiple repeated measures
- Interrupted time series
- Control series
similarities between quasi-experiments and correlational designs
- No randomization
- Outcome variables are measured
- Challenges to causal claims
- Can deal with discrete groups or continuous variables (Quasi-experiments can deal with multiple discrete groups; correlational designs can deal with only two discrete groups)
can experiments be both quasi-experimental and correlational?
- Yes!
- Ex. Men’s hair length vs. Women’s hair length
- Measuring variables, not manipulating (correlational)
- Looking at two groups that can’t be randomly assigned because you can’t assign someone to be male or female (quasi-experimental)
types of developmental designs
- Cross-sectional
- Longitudinal
- Sequential
threats to internal validity
- History
- Maturation
- Testing
- Instrument decay
- Regression towards the mean
- Mortality
- Selection effects
- Cohort effects
history
- threat to internal validity
- any event happening between first and second measurements
maturation
- threat to internal validity
- people change over time (fatigue effect, get hungrier, more mature, etc.) independent of manipulation
testing
- threat to internal validity
- taking a pretest is enough to change a participant’s posttest (eg. Practice effect)
instrument decay
- threat to internal validity
- characteristics of a measure, or people’s use of a measure, changes over time (ex. TA’s marking tests)
regression towards the mean
- threat to internal validity
- participants, who are selected because of their extreme scores, tend to subsequently score closer to the mean (ex. If you get 1% on a midterm, odds are very likely that you’ll do better than 1% on the second one)
Strengths and Weaknesses of ABA reversal design
weak on all – history, maturation, testing, instrument decay, regression towards the mean
Strengths and Weaknesses of one group post-test only design
weak on history, maturation, and regression towards the mean
Strengths and Weaknesses of one group pre-test post-test design
weak on all – history, maturation, testing, instrument decay, regression towards the mean
Strengths and Weaknesses of nonequivalent control group
Good on history and maturation, weak on regression towards the mean
Strengths and Weaknesses of Non-equivalent control group pretest-posttest
Good on history, maturation, testing, and instrument decay, weak on regression towards the mean
cross-sectional design
- type of developmental design
- taking people of different age groups and testing them all at the same time (ex. Taking 40-year-olds, 60-year-olds, and 80-year-olds and testing their willingness to believe things they read on the internet)
- Problem: cohort effects
