Then tell me at what percentage an increased risk reaches statistical significance in epidemiology.
I’m not looking to argue with you - I truly am interested in knowing
That depends on how large your sample is.
A 200% factor (twice as many incidences) will already show significance when there are less than 10 expected occurences. For example, if the incidence increases from 7% to 14%, a sample of 100 will show statistical significance (you get 14, you expect 7, probability of getting 14 by chance alone is under 1% so it’s a significant effect).
If you want to detect a relative risk factor of 150% (half again as many incidences) you need a larger sample size. For example, if the incidence increases from 7% to 10.5%, and you have a sample size of 400, you get 42 when you expect 28 and that also has a less than 1% probability of occurring by chance alone so you call it statistically significant.
There is a formula giving the relationship betweem the following variables
expected frequency
observed frequency
sample size
significance level.
The percentage you are working with is the ratio of expected frequency to observed frequency. In my examples, I used a 1% significance level (meaning the result is declared significant if it has less than a 1% probability of occurring by chance alone if there were really no systematic difference).
That depends on how large your sample is.
A 200% factor (twice as many incidences) will already show significance when there are less than 10 expected occurences. For example, if the incidence increases from 7% to 14%, a sample of 100 will show statistical significance (you get 14, you expect 7, probability of getting 14 by chance alone is under 1% so it’s a significant effect).
If you want to detect a relative risk factor of 150% (half again as many incidences) you need a larger sample size. For example, if the incidence increases from 7% to 10.5%, and you have a sample size of 400, you get 42 when you expect 28 and that also has a less than 1% probability of occurring by chance alone so you call it statistically significant.
There is a formula giving the relationship betweem the following variables
expected frequency
observed frequency
sample size
significance level.
The percentage you are working with is the ratio of expected frequency to observed frequency. In my examples, I used a 1% significance level (meaning the result is declared significant if it has less than a 1% probability of occurring by chance alone if there were really no systematic difference).