c**********5
发帖数: 653 | 1 it is dietary dataset. DATA has been collected at baseline ,6m ,12m.
per each id , there are 3-5 repeated measurement.
id visit outcome( chicken,turkey,ham,fish)
1 baseline chicken
1 baseline turkey
1 baseline ham
1 6m turky
1 6m turky
1 6m ham
1 6m chicken
1 12m turky
1 12m fish
1 12m ham
1 12m chicken
PI would like to see if there is breakfast dietary change over time?
What kind of model do you suggest?
How I can capture the correlation?
Thanks | Y****a
发帖数: 243 | 2 这跟longitudinal没啥关系吧，two-way contigency table，chi-square test不就解
决了？
还是我没搞懂你到底想干什么？ | p*******i
发帖数: 1181 | 3 如果是两两时间比较（比如都比较Change from baseline) 楼上说的很正确， 或者如
果是小样本的话用 Fisher Exact test 也不错~
如果你是要overall的话 试着把后面每个时间点的dietary换成 1 if same as
baseline, 0 otherwise. 然后用Logistic Regression， 看看time是不是significant | y*****w
发帖数: 1350 | 4 This could be a split-plot design. Treat DIET as the whole unit and TIME (
baseline, 6 mo, 12 mo) as the sub-units. Use the count (frequency) of each
diet within each id at each time point as the outcome variable. I’ve made
up a data as follows, assuming that at each time point an id could have any
combinations of diets but the total number of meals is always 4 (e.g. 2
meals of chicken, 0 meal of turkey, 2 meals of ham, 0 meal of fish for ID
#2 at baseline):
/*
coding of diet - chicken 1
turkey 2
ham 3
fish 4
*/
data tmp;
input diet id time yesno 3.;
datalines;
1 1 0 3
1 1 6 1
1 1 12 1
1 2 0 2
1 2 6 0
1 2 12 2
1 3 0 2
1 3 6 0
1 3 12 1
1 4 0 1
1 4 6 0
1 4 12 2
1 5 0 1
1 5 6 1
1 5 12 2
2 1 0 0
2 1 6 2
2 1 12 1
2 2 0 0
2 2 6 1
2 2 12 2
2 3 0 0
2 3 6 0
2 3 12 2
2 4 0 0
2 4 6 4
2 4 12 2
2 5 0 1
2 5 6 0
2 5 12 0
3 1 0 0
3 1 6 1
3 1 12 0
3 2 0 2
3 2 6 0
3 2 12 0
3 3 0 0
3 3 6 2
3 3 12 1
3 4 0 2
3 4 6 0
3 4 12 0
3 5 0 1
3 5 6 3
3 5 12 0
4 1 0 1
4 1 6 0
4 1 12 2
4 2 0 0
4 2 6 3
4 2 12 0
4 3 0 2
4 3 6 2
4 3 12 0
4 4 0 1
4 4 6 0
4 4 12 0
4 5 0 1
4 5 6 0
4 5 12 2
;
run;
The mixed model is as follows:
proc mixed data=tmp cl;
class diet id time;
model yesno=diet time diet*time / ddfm=kr;
random id(diet);
run;
Part of the output:
Type 3 Tests of Fixed Effects
Num Den
Effect DF DF F Value Pr > F
diet 3 48 0.59 0.6229
time 2 48 0.00 1.0000
diet*time 6 48 2.30 0.0493
Just look at the effect of diet*time interaction term. The p-value is 0.049,
indicating that time had a significant effect on the meal frequencies
across the diet types, i.e. the diet changes over time. Note that the p-
value for time is always 1.0 because the total number of meals is set up
as 4 for each time point.
【在 c**********5 的大作中提到】
: it is dietary dataset. DATA has been collected at baseline ,6m ,12m.
: per each id , there are 3-5 repeated measurement.
: id visit outcome( chicken,turkey,ham,fish)
: 1 baseline chicken
: 1 baseline turkey
: 1 baseline ham
: 1 6m turky
: 1 6m turky
: 1 6m ham
: 1 6m chicken
|
|
