Quasi-Experiment Methods Flashcards
internal validity in quasi-experiments
- quasi experiments lack full experimenter control over the IV
- they can have better internal validity and can use methods to rule out threats to external validity
how to prevent selection effects as a threat to internal validity?
pretest-posttest design
- are subjects at one level of IV systematically different from subjects at another level of IV?
how to prevent design confound as a threat to internal validity?
measuring second variable can help rule out design confounds
how to prevent maturation/history as a threat to internal validity?
pattern of results in comparison group can help rule out these threats
characteristics of quasi-experiments
- cannot establish causality
- researchers take precautions to reduce threats to internal validity
- eliminates as many alternative explanations as possible
- lack random assignment to condition
non-equivalent groups
participants do not start from the level
they vary systematically from the beginning
experiment:
studying students from public/private schools and their attitudes regarding authority
non-equivalent groups need to be in place because students were not randomly assigned to a school
nonequivalent group design
compares 2 non-equivalently groups that are intentionally as similar as possible
1. treatment group
2. control group
group a measure DV -> scores
treatment compare
group b measure DV -> scores
control
pretest-posttest nonequivalent groups
comparing before and after treatment that participants were not assigned to
compares 2 non-equivalently groups that are intentionally as similar as possible
1. treatment group
2. control group
* difference between scores before and after treatment is compared across groups
surgery pretest treatment posttest
no surgery pretest treatment posttest
* compare results before and after treatment across groups
interrupted time series
treatment in the middle of multiple measurements from DV
take multiple pretest and posttest measurements to asses effects of both treatment and time passing
purpose: eliminate noise before and after data
controlled interrupted time series design
observe a non-equivalent control group over the same time period
comparing two sets of data and taking multiple measurements
treatment vs no treatment
*measures multiple DVs before and after treatment
structure of interrupted time series design
measure DV - measure DV - measure DV -treatment or intervention - measure DV - measure DV - measure DV
compare
compare average before and after treatment
longitudinal developmental design
observe a group of individuals as they age
- time consuming and can be confounded by attrition
looking at maturation itself
measure DV over the course of the participants lifespan
and compare the scores over time
cross sectional developmental design
compares different age groups to see how the DV changes with age
internal validity threated by cohort effects (differences due to cohort characteristics rather than development)
group 1 measures dv -> scores
(65+ years old)
group 2 meadures dv -> scores
(18-25 yrs old)
compare scores afterwards