random chance - internal validation Flashcards
What is internal and external validity
Internal validity: How accurate are the results of our study to the population
- effected by: confounding, bias, random chance
External validity: How applicable is the results to other population from our population
- effected by: What is studied
What is sampling error
Random variation within our samples due to variation of parameter within the population
Cannot be removed but can be limited by increasing sample size
What is clinical significane
Clinical significane is a self defined level which an association must be above for the exposure to be deemed a clinically siginificant exposure
CI must be above the clincial siginficane value to be deemed clinically siginifcant
What is a confidence interval
A confidence interval is a defined level of confidence in which the true populaiton perameter will fall within
- If contains the null, cannot make a claim about the importance
What does a p-value measure
A P-value is the probability that your observed value would be observed under the assumption that there is no association between two groups.
Lower p value, more unlikely that H0 is true
What are the possible outcomes from a hypothesis test
We get it right,
- H0 is true and we fail to reject H0
- H0 is wrong and we acept H0 (power)
We get it wrong,
- T1 error: H0 is true and we reject (defined by alpha)
- T2 error: H0 is false and we fail to reject (defined by 1-beta- power)
(reduced T2 by incraese sample size)
How do we link confidence intervals to the P-value
If the pvalue > alpha value -> not a statistically significant difference
if the pvalue < alpha -> there is a statisitically significant difference
What are 3 problems with the P-value
Thershold of 0.05 is abirary
Only gives information about H0
