Statistics Flashcards
Odds ratio important in what study?
Larger OR means?
OR precision and accuracy determined by?
Case-control study - comparing exposed and unexposed group’s outcome
Higher odds the exposure leads to outcome
Confidence interval - smaller CI = precise OR
Difference between arithmetric and geometric mean?
Arithmetric mean = MEAN affected by outliers, skewed data
Geometric mean = MEAN NOT affected
when to use non-parametric test?
Sample size tooo small
Median better represent study
Ordinal/interval data
Outliers that can’t be removed
Intention to treat analysis often do what to treatment effect?
underestimate
what is use to measure publication bias?
Funnel plot = symmetric inverted funnel shape shows no publication bias
chi squared (X^2) test assumes what null hypothesis?
That all studies are homogeneity ( similar studies) or producing same results
I^2 statistic test function?
Variation across studies due to heterogeneity rather than chance
75% high heterogeneity
0% no heterogeneity
Prevalence definition?
Prevalence Point ( remember this) -proportion of cases that have the disease at a specific point of time
Incidence RISK definition?
Incidence RATE definition
Incidence risk = proportion of cases that develop the disease within specified time ( new cases)
Formula : New Case / People not having the disease
Incidence rate = new cases including those suspected to be positive but still investigating
Formula : New case ( including suspected ones) / total people at risk
Attack RATE definition?
similar to Incidence RATE
-measure of speed of spread/morbidity in an at-risk population
Information/classification bias - has 2 types
Differential and non-differential = tell me the difference?
Random/systemic inaccuracy in measurement
differential = one variable affects another variable Non-differential = errors in variable is independent of another variable
Population Attributable Risk (PAR) definition?
incidence of disease in population that would be removed if exposure is eliminated
PAR =Prevalence x AR
sequential clinical study means?
Results are assessed as they are produced and total number of participants going to next round depends on the results
Crossover clinical study 2 main advantage?
Confounding variables reduced
Need less sample size ( participants will eventually try all treatments)
Allocation concealment versus blinding?
Allocation concealment - prevents RESEARCHERS from allocating participants into which group = prevents selection bias
Blinding
- single blinding = participants
- double blinding = participants and observers
- triple blinding = participants, observers, researchers
