Key statistical concepts Flashcards
3 domains of healthcare
Health protection
Health improvement
Healthcare public health
How do we make inferences about a population
Take a sample
define population
all possible observations of an experimental/study population
primarily interested in
Sample define
Selection of observations taken from a population
sample not of interest - want to generalise population
How to reduce chance - random error?
Increase sample size
True vs observed
True probability = underlying
Observed = may be different to true
What is an observed value?
Best estimate of the true/underlying value
What is a hypotheses?
Statement that an underlying truth takes a particular quantitative -
eg
- prevalence of TB in a population is 2 per 10,000
- the coin is fair
- new drug is neither better or worse
What is hypothesis testing?
Calculating the probability of getting an observation as extreme as the one observed (assuming the hypotheses is true)
What does it mean if p value is small?
Probability small so can conclude that the observation and hypotheses is incompatible
What does it mean if p value is <0.05?
If p value less than 0.05, data is inconsistent with hypothesis so there is strong evidence to reject it
- observations statistically significant
What does the p value >0.05 mean?
Does NOT mean that the hypothesis has been proven
Consistent with idea that it may be true
Just failed to reject hypothesis
Problem with hypothesis (p value) testing?
0.049 and 0.051 are excluded but WHY?
Statistical significance depends on sample size
Statistically significant is different to CLINCIAL significance
Problems with observed quantities
Subject to variation by chance (statistical variation)
How do we make rational inferences about true value given variation?
Use confidence intervals - 95% confidence limits (95% sure that true value lies between these two points)
Point estimate/best guess will be in the middle of these limits
