To calculate the SE of the ln(RR) from the CI we can either use. Note that the 95 CI for the ln(RR) is usually calculated as: These figures are then exponentiated to give the CI of the relative risk. Therefore we need to use the 95 CI to work backwards and calculate the SE of the ln(RR).
The standard error of being an estimator of this fixed parameter divided by the sample size, converges to zero as the sample size grows. display bsigma sigma not found r(111) sigma is derived from lnsig. Assuming that the observed values of Z are independently and identically distributed, the standard error of is equal to: Note that the standard deviation of Z, Z, is a fixed parameter. display lnsigbcons-1.4256592 From the output above, you might also guess that the bsigma would work, but it does not. The normal regress functions dont allow me to give them as an input though. When you see /something, the coefficient is somethingbcons and the standard error is somethingsecons. Now I would like to find the t-statistics of coefficient a and b. An observation regarding robust standard errors in R and StataĬreated: Ap``` library(tidyverse) library(sandwich) library(lmtest) # Fit the model fit % janitor::clean_names() %>% mutate(log_price = log(sale_price)) fit <- glm(log_price ~ overall_qual + gr_liv_area, data = dat, family = gaussian(link = "identity")) coeftest(fit, vcov = robust(fit, stata = FALSE)) ``` The same model in Stata: !(/Users/rap168/Documents/GitHub//files/stata_se4. For studies 2 and 5 we are unable to do this as we do not have the information to construct a 2x2 table. Hi, I found the coefficients of a simple regression Y aX1+bX2 using a maximum likelihood optimization.