Dear Xianghong
I am very interested in your work, but when reviewing your code section I encountered a question regarding how your core function handles c1*se.exp^2. The code shows:
s = matrix(0, m, 4)
s[1:m, 1] = se.exp^2 * drop(C[1,1])
s[1:m, 4] = se.out^2 * drop(C[2,2])
s[1:m, 2] = s[1:m, 3] = drop(C[1,2]) * se.exp * se.out
s11 = s[1:m, 1]
#s11 = se.exp^2
When c1 is close to 1, the results may be similar, but is this approach still valid when c1 deviates significantly from 1? In your paper description, the numerator of C_1 and C_2 should be "-Crse.exp^2", yet in the code implementation (if I understand correctly), it appears to use "-Crc1*se.exp^2", where c1 represents the exposure intercept term. I would like to clarify this apparent discrepancy between the theoretical framework and the implementation.
best,
Chongyang Yu
Dear Xianghong
I am very interested in your work, but when reviewing your code section I encountered a question regarding how your core function handles c1*se.exp^2. The code shows:
s = matrix(0, m, 4)
s[1:m, 1] = se.exp^2 * drop(C[1,1])
s[1:m, 4] = se.out^2 * drop(C[2,2])
s[1:m, 2] = s[1:m, 3] = drop(C[1,2]) * se.exp * se.out
s11 = s[1:m, 1]
#s11 = se.exp^2
When c1 is close to 1, the results may be similar, but is this approach still valid when c1 deviates significantly from 1? In your paper description, the numerator of C_1 and C_2 should be "-Crse.exp^2", yet in the code implementation (if I understand correctly), it appears to use "-Crc1*se.exp^2", where c1 represents the exposure intercept term. I would like to clarify this apparent discrepancy between the theoretical framework and the implementation.
best,
Chongyang Yu