Biostats previous exam info

Chi-squared test statistic (interp)

Difference between observed data and expected data

Larger = bigger discrepancy = potential association

Chi square critical value (what, how)

Threshold for statistical significance. If calculated value >, suggests sig diff

df and alpha/significance level

df = rows-1*columns-1, sig commonly .05

SD levels and %

0.5 = 38%

1 = 68%

1.5 = 87%

2 = 95%

2.5 = 98.7%

3 = 99.7%

Assumptions of statistical procedures (4)

Normal hom0 in line

Normal distribution

Homogeneity of variance

Independence

Linearity

Standard error of the mean (SEM) (def, equ, note)

How close your sample mean is likely to be to the true population mean

SEM = sample SD/square root of sample size

Smaller SEM = greater confidence

SEM decreases with sample size increase

Mann-Whitney U test (def)

Non-parametric for t test

When data is non-normally distributed or small samples

Coefficient of variation (def)

Shows relative variability

Ratio of SD to mean

Lower = greater consistency, less variability

Coefficient of determination (name, def)

R2

How much of the variation in dependent variable is explained by the independent variable in a regression

Range from 0 to 1

x% of the data’s variation is explained by the model

CI calculation (95%)

mean +- 1.96*SEM

SEM = SD/(sqrt of n)

Commenting on CI

“We are 95% confident that the true mean systolic BP of the population lies between x and y. this interval is narrow, reflecting the large sample size”

Log rank test (def, alt name, use)

Non-parametric test to compare survival distributions of two+ groups

Mantel-cox test

Determine if there is a significant difference in the time until an event