meta analyses Flashcards
what is the post hoc fallacy
after this therefore because of this. make connections between events even if there isn’t any (correlation does not equal causation asf)
what is confirmation bias
seeking out information that confirms existing views.
try to disprove your own hypothesis
what is regression to the mean
conditions fluctuate around a mean (chronic conditions). people likely to seek new treatments when illness at its worst, therefore are likely to improve anyway. (an extreme measurement less likely to be extreme next time)
what is spontaneous remission
self limiting illnesses already likely to get better
littlewoods law: miracles happeb every 35 days
some cases coincide with innfecetive treatments
what is observer bias
researchers expectations influence results/evaluation of experiment. bias of the observer
what is feedback bias
people are more likely to report positive effects.
what is publication bias
publishers reluctant to publish negative findings. funnel plots can be used on publications to see if they are reliable.
what is the placebo effect
mock treatments producing effects on the body if the patient expects to get better
what is the nocebo effect
mock treatments producing a negative effect if the patient expects to get worse.
what is factors should be adjusted for a well designed clinical trial
- eliminated sources of bias
- large sample size
- was it randomised
- was subjective or objective factors measured
- how was data analysed (blinded)
double blind placebo controlled trials are very reliable. eliminates observer and participant biases by double blinding. placebo trials to eliminate conformation bias in data analysis.
what are systematic reviews
comprehensive summaries of existing evidence from studies to answer specific research questions
what are meta analyses
statistical technique used to combine and analyze the quantitative data from multiple studies included in a systematic review. (combining and contrasting clinical trials to find trends)
what factors could increase perception of efficacy for CAM
- post hoc fallacy
- regression to mean
- spontaneous remission
- confirmation bias
- observer bias (poorly designed trials)
- publication bias
- feedback bias
- placebo effect
how could bias and chance be minimized for CAM studies
placebo effect and observer bias effects studies.
divided into control and treatment groups randomly then these conditions should be reversed to decrease participant variability. placebo to allow for control condition to see if significant difference between groups.
observer bias if study cant be double blinded.
discuss ethics and participant selection and what will be measured based on what cam is being assessed
repetitions decrease chance.
how can observer bias be minimised
1.Assign random subjects to each group (can choose healthier looking patients to control group)
- Use double blind experiments (neither researcher or subjects know which treatments are being used or which group they belong to)
- Having multiple researchers involved
