M****e
发帖数: 178 | 1 To compare the effect of some treatment on the occurrence of fainting in
patients
patients divided into two groups,
group A receive the treatment and group B does not
record whether fainting occurs within a fixed time period (0 or 1)
three month later
group B receive the treatment and group A does not
record whether fainting occurs within a fixed time period (0 or 1)
Suppose no time effect or carry-over effect
How to analyze the treatment effect: does treatment decrease the probability
of fainti |
T*******I
发帖数: 5138 | 2 Are you a statistician? or not?
probability
【在 M****e 的大作中提到】
: To compare the effect of some treatment on the occurrence of fainting in
: patients
: patients divided into two groups,
: group A receive the treatment and group B does not
: record whether fainting occurs within a fixed time period (0 or 1)
: three month later
: group B receive the treatment and group A does not
: record whether fainting occurs within a fixed time period (0 or 1)
: Suppose no time effect or carry-over effect
: How to analyze the treatment effect: does treatment decrease the probability
|
s*****r
发帖数: 790 | 3 you must be one, and a great one.
【在 T*******I 的大作中提到】
: Are you a statistician? or not?
:
: probability
|
s*****r
发帖数: 790 | 4 my 2c:
this is a cross-over design, and you are assuming no time(period) effect/
carry-over effect, so basically, it is equivalent to the scenario that you
observe all patients for 3 month without treatment, then with treatment. the
order does not matter. Or you can consider the sequence/order effect. your
data look like:
y_ijk, i=1,2 for sequence, j=1:n for patients, k=1,2 for period,
you can model the data like this:
y_ijk ~ bern(p_ijk)
logit(p_ijk)=b0+bi+b_j+b_t*I(treatment)
b_i is the period
【在 M****e 的大作中提到】
: To compare the effect of some treatment on the occurrence of fainting in
: patients
: patients divided into two groups,
: group A receive the treatment and group B does not
: record whether fainting occurs within a fixed time period (0 or 1)
: three month later
: group B receive the treatment and group A does not
: record whether fainting occurs within a fixed time period (0 or 1)
: Suppose no time effect or carry-over effect
: How to analyze the treatment effect: does treatment decrease the probability
|
b*******r
发帖数: 152 | 5 this is a standard matched pairs data - use mcnemar's test. proc freq; table
.../AGREE; |
M****e
发帖数: 178 | 6 Is this design common to generate this type of data? why bother having two
groups and two phases? Coul dit just be before treatment and after treatment?
table
【在 b*******r 的大作中提到】
: this is a standard matched pairs data - use mcnemar's test. proc freq; table
: .../AGREE;
|
