Lect 2 - Biostats Flashcards
(12 cards)
define standard error aka SD of population
variability we might expect in means of repeated samples taken from population
(SD) of sample/SQT (sample)
how variable sample might be compared to true population
7 steps of hypothesis testing
- Research question
- Sample and conduct study
- Null & alternative hypothesis
- Identify level of significance / probability
- Calculate test statistic
- Obtain p-value / confidence interval
- Interpret and make conclusions
what does the question need to do
Identify population of interest
Define outcome / DV of interest - and parameters
Define factors / IVs of interest - and parameters
what do we need to think about before defining sample parameters
while it is ideal to use random sampling to get the best representation of the population, however, there is a balancing act between cost and precision
we prefer samples to be larger, this increases the probability of mean and SD being closer to true values. This also reduces error and increases power
what does the null hypothesis mean
that our results are due to chance
differentiate between type 1 and 2 error types
type 1: incorrectly rejecting null hypothesis
type 2: incorrectly failing to reject the null hypothesis
what affects type 1 and 2 errors
type 2: higher the power, less likely you’ll make type 2
type 1: has to do with alpha.. how much error am I willing to accept
define power and what affects it
Probability of finding an association (result) in our
sample if there is a true association in the
population
It depends on
-Probability of a type I error-α
-Alternative hypothesis is true (size of effect)
-Sample size
-Statistical test used
what does p-value mean
p value= .006: Probability that the two samples
come from the same population is .006 or 0.6%
what is the central limit theorem
Central limit theorem – the sampling distribution of the mean approaches normal as the number of samples (n) increases – regardless of the underlying distribution in the population
how is CI different to p-values
CI describes precision and uncertainty whereas p value only compares it to chance
clinical significance?
effect sizes not just statistical differences
