log bi u kontekstu kliničkog ispitivanja možda mogao biti i "logaritam"

offset variable and exposure variable. An offset variable is a variable
that is to appear on the right-hand side of a model with coefficient
one:

y_j = offset_j + b_0 + b_1x_j + ...

In the above, b_0 and b_1 are to be estimated. The offset is not
constant. Offset variables are often included to account for the
amount of exposure. Consider a model where the number of events observed over a period is the length of the period multiplied by the number of events expected in a unit of time:

n_j = T_j e(X_j)

When we take logs, this becomes

log(n_j) = log(T_j) + log{e(X_j)}

ln(T_j) is an offset variable in this model.

When the log of a variable is an offset variable, the variable is
said to be an exposure variable. In the above, T_j is an exposure
variable.



--------------------------------------------------
Note added at 1 hr (2009-03-28 17:20:11 GMT)
--------------------------------------------------

U nekim kontekstima spominje se da se kao offset varijabla uzima
log(time) ili time, pa me to navodi na mogućnost da se može uzeti logaritam vremena ili nominalno vrijeme