Statistics Flashcards
null hypothesis
assumes there is not significance between control and treatment groups
studies try to disprove the null hypothesis
alternate hypothesis
states there is a difference between control and treatment groups
research tries to prove this
alpha
maximum permissable margin of error
normally 0.05
if P-value < alpha, reject null, accept alternate - significant result
if P-value > alpha, accept null, reject alternate - NOT significant result
confidence interval
if it crosses 0, not significant
type 1 error
false positive
when null hypothesis is rejected in error
type 2 error
false negative
when null hypothesis is accepted in error
relative risk (RR)
risk in treatment group / risk in control group
relative risk reduction (RRR)
1 - relative risk (RR)
number needed to treat (NNT)
1 / (risk in control - risk in treatment) or 1 / ARR
odds ratio (OR)
odds outcome occurs w/ treatment vs. occurs w/o treatment
hazard ratio (HR)
HR = hazard rate in treatment / hazard rate in control
OR or HR = 1
event rate same in treatment and control
no advantage to treatment
OR or HR > 1
event rate higher in treatment than control
don’t use treatment
OR or HR < 1
event rate lower in treatment than control
treatment is beneficial
nominal (categorical) data
categories, names
ex. male, female
ordinal data
logical order or scale
pain scale
continuous data
measurement, unlimited options
weight, temperature
T-test
assess significance w/ continuous data
used for 2 sample groups
ANOVA
used for 3+ sample groups
Chi-Square
for discrete date (nominal or ordinal)
gender, ethnicity, pain scales
independent variable
manipulated by researcher
dependent variable
affected by change to independent variable
composite endpoint
combines multiple endpoints
all must be similar in magnitude, importance
sum of individuals will NOT equal composite endpoint
intent to treat
analysis of all patients, even if dropped from study
per protocol
analysis of only those who completed trial per protocol
sensitivity
how effectively it identifies those w/ condition
specificity
how effectively it identifies those w/o condition
case control study
compares pts w/ disease to those w/o, retrospective
cause/effect not determined
cohort study
compares outcomes of exposed group to non-exposed group, prospective
cross-sectional survey
estimates prevalence at one time in defined group
do not determine causality
