Statistics Flashcards
Parametric Data =
Tests used to assess difference between groups
= normally distributed data
Unpaired or paired T-tests
Non-Parametric Data =
Test used to assess difference between groups
= non-normally distributed data
Wilcoxon-signed rank test
Absolute risk reduction
= control event rate - experimental event rate
(remember to divide by number in each group)
Standard error of the mean =
SD of sample/ sq. root of sample size
Type 1 error
= rejecting the null hypothesis which is true
Type 2 error
= rejecting the alternative hypothesis which is true
Unpaired vs paired T test
Paired = two small paired observations
e.g. comparison of effects of two drugs at different points in time
Unpaired = comparison of two independent samples e.g. two different groups of patients
Non Parametric tests
Chi Squared test
Wilcox Rank test
Mann U Whitney test
Chi squared test
= compare percentages or proportions
Continuous data
Non normal distribution
Wilcox Rank Sum
= unpaired data (before and after)
Mann U Whitney
= similar to Wilcox but for ordinal/interval data
Incidence =
number of new cases / total population at risk
Prevalence =
total cases / total population at risk
Incidence vs absolute risk
SAME THING
Absolute risk = new cases/total population at risk
Relative risk =
experimental event rate
control event rate
Sensitivity =
= ability of test to correctly identify those who truly have the disease
Sensitivity formula =
true positive
true positive + false negative
Specificity =
= test’s ability to correctly identify those who do not have the disease
Specificity formula =
true negative
true negative + false positive
Positive predictive value
All of the positives
True positive
——————-
True positive + false positive
Negative predictive value
All of the negatives
True negative
——————–
True negative + false negative
Positive result likelihood
- definition
- formula
= how much odds of disease increase when test is positive
sensitivity
1-specificity
Power =
1 - chance of making a type II error
Number needed to treat =
1
absolute risk reduction
