Term
| for a factor to be associated with disease, it must have what relation between exposed and non-exposed individuals |
|
Definition
| it must be different among the exposed and non-exposed individuals |
|
|
Term
mean is the measure of what for which kind of data?
- it doesn't tell us what?
- problem |
|
Definition
central tendency for continuous data
- doesn't tell us about how values are distributed
- highly influenced by extreme values |
|
|
Term
Central limit theorem
-results in
-basis for which test? |
|
Definition
take a large sample of observations and calculate the mean of the values, and repeat over and over again.
wind up with:
- normal distribution***
- variance (s^2) which describes the variability of the sampled means in all directions
- a mean (ยต)
- standard deviations
*basis of z-test |
|
|
Term
| limitations of Central limit theorem |
|
Definition
large sample size required
need to know standard deviation of population |
|
|
Term
| standard error or mean formula |
|
Definition
|
|
Term
compare t-distriubtion to a normal distribution
- when do we have to use t-dist? |
|
Definition
t distribution is shorter and wider than normal.
can't have normal distribution when use samples. |
|
|
Term
| what two things could we be testing for when we use a t-test? |
|
Definition
test if single mean significantly different from zero
or
test if two means are significantly different from one another |
|
|
Term
| compare proportions and ratios |
|
Definition
proportions is a fraction measure where numerator is included in denominator (0.5, 1/2, 50%)
ratios is a fraction measure where the numerator is NOT included in the denominator (A to B, A:B) |
|
|
Term
| when is X^2 (chi-square) test used? |
|
Definition
| when there are proportions or ratios since they can't be negative (so can't use t-distirubtion) |
|
|
Term
|
Definition
the probability that the observed difference occurred due to chance or sampling variation. the probability that the test statistic would be larger or as large than calculated if the null hypothesis was true . probability of seeing test statistic calculated if no true difference between groups.
high p-value - likely taht difference was due to chance, should accept null.
low p-value < 0.05, only 5% chance association due to chance, so we reject null and state there is a difference/ association |
|
|
Term
|
Definition
| no difference between groups or association between exposure/disease |
|
|
Term
| equation for confidence interval |
|
Definition
|
|
