EXAM 2 Flashcards
Research designs that use an intervention…
Expereimental, RCTs, Quazi- expereimental studies
What type of comparisons will be made?
- You can compare the same group of individuals at different time or after an intervention
- you can compare two groups with one group getting an intervention
How are cofounding variables controlled?
WHich cofounding variables will be controlled?
Variables that can influence an outcome that weaken a link between independent and dependent variable. Randomization, crossover, homgenity, matching, and statistical control are methods to limit the influence of cofounding variables
What is blinding
. Blinding participants, blinding the people performing the interventions, blinding other staff, and blinding the data collectors are all options for blinding. The more groups blinded in a study, the less bias present in a study.
How often will data be collected? When, relative to other events, will data be collected?
Cross- sectional or longitudinal
When will information on independent and dependent variables be collected, looking backward or forward?
Relative timing. Will it be a retrospective study (looking backwards), or will it be a prospective cohort study (looking forward). Note a retrospective study is a secondary type of study since the data was already collected before the study by someone other than the researcher.
Where will study take place?
Location. Setting selection, using one site versus multiple sites.
Three traits for establishing causal relationships are temporal, relationship, and confounders
Temporal requires a cause to precede an effect.
Relationship requires an association to be shown between the cause and effect.
Confounders requires it to be proven that the effect cannot be caused by a third variable
Experimental Designs
Some RCTs will use a pretest-posttest design which measures the outcome indicator before (baseline data) and after the intervention (posttest data).
Randomization is achieved with computer programs, but more mundane examples include pulling a name out of a hat or flipping a coin.
A crossover design is used when you give more than one treatment but separate the treatments and use randomization to determine the ordering.
Limitation of RCTs
Limitations of RCTs include the inability to test all interventions in this format, the inability to randomize certain variables, and difficulties getting administrative approval to carry out a study.
In the PICO question for C what are the options
Usual care is when there is a therapy in place for something and you want to compare it with the new intervention.
An alternative treatment is testing out two treatments for one outcome to see which one gets better results. Ethical since both treatments should benefit the patient, but if outcomes are similar the study loses power to determine if an intervention is better.
A placebo is a treatment expected to have no effect to compare with the intervention.
An attention control condition is when the control group gets attention but not the intervention. An example of this is both independent and control groups receiving education on weight loss, but the intervention group is given a supplement to take in addition to the education.
Delayed treatment gives both groups the intervention, but the control group waits until later to get it. This option is the most ethical of the C options.
Quasi-experimental stuff
- examines an intervention but does not include randomization.
- sometimes no control group
- Intervention is offered to everyone
-In order to compare researchers would need to find a another area with same population - collect data on outcome before intervention started
and after a set time interval - second population does not receive intervention
Quasi- experimental
- some may only collect post data causing them to be less relaible then those tests that collect both pre and post test data
- if they have no control group, it can be difficult to know how intervention affected outcome
Quaisi time series design
- collects data over an extended period
- introduce intervention, then collect data for same time
Advantage of Quasi
- more practica land eaasier to conduct
- particpants can feel like they get intervention
Disadvantage of Quasi
- leads to bias, results are not generalized, and cofounding variables can influence results
Non-expereimental studies
intervention is not introduced and known as observational
Correlational research
- determines relationship but not cause or effect
Prospective design
- looks for presumed effect
CAse- control
- starting with effect and looks at patinet datat fro potentia cause of variables
Descriptive research
- observes for the prevalence of something without dtermining a relationship
descriptive correltaional study- measures instance without linking a cause
Cross-sectional
data collected at one point of time
longitudinal
collect data at multiple times over a period
Methods of control
script- researcher read exact same thing
Hogeneity- retricting indiviudals with cofoundiing varaible
Matching- developing a comparision gorup with same cofounding varaibles aas control gorup
Statistical control- accounts fro cofounding variables
Statistical valididty
- statitical power- sample size
- inteventions that are defined (needs to be diff than control)
internal validity
- independent variable causes outocme
- minimizing threats
external validity
concerns ability of applying stduy to gen poppulation
Construct valididty
- inferences from participants to high order constuct the yare to represent
weakness: lack of blinding
