censcor is an R package for estimating correlations between censored variables. censcor works by fitting a multivariate normal distribution to the data using Stan.
devtools::install_github("silentspringinstitute/censcor", args = "--preclean")Documentation and vignettes for the package are available on the censcor website.
# Generate 100 x and y values with a 0.5 correlation coefficient and no censoring
df <- generate_censored_data(N = 100, rho = 0.5, L_prob = 0)
fit <- censcor(x ~ y, df)
# Fit dataset with left censoring in both x and y
df <- generate_censored_data(N = 100, rho = 0.5, L_prob = 0.5, direction = -1)
fit <- censcor(x | cens(x_cens) ~ y | cens(y_cens), df)
You can find more examples in the Introduction to censcor vignette.