Sample Size

what are the steps for estimating sample size

1. decide what type of variable your predictor and outcome are

2. select the appropriate statistical test

3. state null and alternative hypothesis

4. determine effect size

5. determine the variability of the outcome

6. set alpha and betat

what is the standard alpha and beta for sample size estimating

alpha: 0.05, beta: 0.2

what is alpha

the probability of making a type 1 error

• eg. incorrectly conclude there is an association, a false positive

• also known as the level of statistical significance

what is beta

the probability of making a type 2 error

• eg. incorrectly conclude there is no association, a false negative

how do we estimate sample size

esitmate based on previous research/theory what we expect our numerator and denominator to be, then we estimate how many people we would need to reach a certain level of precision, and consider the statistical test/data

what is the p-value?

probability of the test statistic assuming the null hypothesis is true

• don’t just aim to be wrong 5% of the time, this is simply just a cut off

what error are hypothesis tests conducted at

typically using a type 1 error rate of 5%, RA fisher suggested this value

what is the too large sample size problem

eg. increase sample from n = 100 to n = 1000, p value would decrease a lott

what was dr watties master thesis sample size?

135000

what paper did ploutz snyder, fiedler, and felvson write

justifying small n research in scientifically amazing settings

what are some factors to keep in mind

the actual size of the population, small groups, rare conditions/disease, and how to navigate those