Converting 95% CI to SE

[itchyshin]

There is a function to do this

convert <- function(lower, upper, n){
    width <- upper - lower 
    se <- width/(2*qt(0.975,n - 2)) 
    se
}

lower = lower bound, upper = upper bound and n is n used for getting that estimate for se

[itchyshin]

@elmacartney - when we convert estimates (and SE) into r (correlation) - we will need to know whether these estimates are a comparison of two groups or it is a slope (relationship between trait and % success). If there are two groups, I will need n for each group (I guess it is usually balanced?). Let me know

[elmacartney]

Thanks @itchyshin I will take note of if it was between two groups or not.

For the studies that provide no error for the estimate, is there anything we can do or should we contact authors? estimates come with p value

[itchyshin]

@elmacartney - if you have p value we do not need SE - that will be fine.

Yes, thank you for noting if these are from two groups or not