Methods after first midterm: 8 Specialized designs Flashcards
what makes designs quasi-experimental or ex post facto designs
when something is missing from the design….: typically control groups, and/or random assignments of subjects
something is violated
what are teh major categories of designs
pre-experimental
experimental
quasi-experimental
ex post facto
quasi-experimental design: ____
do not fill all requirements for a true experimental (analytics) experiment
issue with :
- external validity
- internal validity
explain why:
random selection not usually done
random selection not usually possible- usually looking at a subsection of the whole population
ex: jet lag study has to find participants who travel a bit who have the same time zone and travek at the same time and get the same amount of sleep
what would make random assignment not assessible
if you’re unable to give someone an independent variable ex: cancer
ex post facto designs : ______
cannot manipulate independent variable; after the fact designs
reseacrher arrives after the fact; nature has implemented the treatment
ex: cancer, hurricanes
when is quasi-experimetnal deisgns useful
when there is low statistical power (low subject availability - ccept subjects with variationatino betweent them); or low subject availability; few measurements possible
what is the pooling falicy?
when people think they can compensate for low statistical power by having low # of partiicpants by getting more data on each of the people they have
counterbalancing : _____
having subjects do a reapeated measure experiment in different orders so we can account for order effects
whats another way to measure order effects other than counter balancing
measuring order effects
- make the orders into a factor, see if the order has an effect on the results
explain prospective ex post facto vs retrospective ex post facto designs
prosepctive - takes time and money - watch ppl over a long time - “10 year later”
retrospective - go through ppls history, studying what already happened
how are participants chosen for ex post facto designs and the rationale behind it
after the fact because of ethics concerns adn subject variables ( these are not true independent variables because you cannot technically manipulate them )
rationale - you can not randomly assign people to different age groups, genders ect
prospective ex post facto design interal validity issues
- no random sampling from population
- no random assignment to conditions
- confounding variables
- convienence sampling - sampling criteria may themselves be confounds
- accurate identification of members of the respective groups based on criteria or criteria changes over time (identifying some groups easier than others)
( air traffic controllers exposed to radiations from equipement…. different jobs, socioeconomic status, educational background))
why are confounding varaibles often found in ex post facto designs (after the fact designs)
- high stress participants smoke and drink more
- things that happen throughout a persons life influence what you are measuring
what is detection bias in prospective eEPF designs
detection of certain issues happens more often in one group than another, if both groups are put together in a study one group may result in one thing more often than another or they say one group more likely to do this than the other group even tho its due to something else like health checks (ex: air traffic controllers, rich prople, examined by doctors more often than janitor or poor people)
