Biostats Flashcards
Precision
Data points cluster around one POINT
Accuracy
Aka Validity - combo of specificity and sensitivity and it = “gold standard”
(Think of the targets)
Standard error of mean (as sample goes up or down what happens)
SEM becomes smaller as more samples are added to data set (power increase) so the data becomes more PRECISE
Z score
Shows how far above or below the mean for a given value (in standard deviations basically)
Median helps do what with outliers
Corrects for outliers
Confidence interval (CI)
Shows precision of a data set, it’s 2 times the SEM so if CI crosses 1 the results are not significant/precise enough to be useful
T-test good for what type of data?
Used to assess data from 2 groups (T for 2)
*same individual followed over time
ANOVA used in what situation?
To assess data from 3 or more groups
Chi square COMPARES what?
COMPARES multiple groups and indicates whether or not they’re statistically significant used when data comes in discrete CATEGORIES
What does randomization account for/do?
- Accounts for selection bias
- “intention to treat” preserves the benefit of randomization
- minimizes confounding bias
Cohort study uses what equation/parameter?
Relative Risk = (exposure with dz / all exposed) / (no exposure with dz / all not exposed)
Case control uses what equation/parameter?
Odds Ratio = (exposed with dz X no exposure with no dz) / (no exposure with dz X exposure with no dz)
*Case-control is subject to recall bias
P-value means what?
If < 0.05 it’s significant, means that if study were repeated there’s 95% chance of same results
* also means there’s 5% probability there’s no association to the null hypothesis (meaning it’s rejected)
Type I error (alpha)
False Positive (you Accept alternate hypothesis when it’s false) AKA P-value
Type II error (beta)
false Negative (Beta error, you’re Blinded to the fact the alt hypothesis is true) POWER = (1-Beta)
