mutiple imputation mixed effect model predicted value

[Queran Lin]

Dear members, I am analyzing a cross-sectional big dataset includes serval CVDs and demographic data and want to plot a plot like the attachment. Our dataset is post-imputated, Hence, I am using a code: mi estimate: “mixed sbpa age if medication!=1||country:, reml emonly” However after this model, It seems “predict plot1, fitted” doesn’t work as usual. It can use if I drop the mi estimate commend. I wanna plot this graph with 95%CI fitted in the line and without the scartter of true value. Could you please kindly give me some advice? Many Thanks for your help!

[Rich Goldstein]

there is no attachment - please read the FAQ; also, see

help mi predict

[Weiwen Ng]

... I wanna plot this graph with 95%CI fitted in the line and without the scartter of true value. Could you please kindly give me some advice? Many Thanks for your help!

When you ask for a 95% CI and a fitted line, I wonder if you should be using the margins command instead. margins doesn't normally work with mi estimate. You'd want the user-written command mimrgns (by our own Daniel Klein, available on the SSC site, type ssc install mimrgns).

Of course, maybe this isn't what you want - I'm just guessing. If you want to post an attachment, do read the FAQ. If you couldn't successfully upload a PNG file to the forum, it can help to upload your graphic to a site like Imgur, then post the link here.

[Queran Lin]

Thanks for replying to me guys! here are the acctachments! what I really want to plot is the first plot! But I can't get it when I use mi estimate. Could anyone advice on this?

I have tried in the subdataset by using

mixed sbpa age if stroke==1&gender==1&medication!=1||country:, reml emonly

and then I used

predict plot, xb
twoway (lfitci age plot)

However, What I got was a straight line which doesn't make sensce. I manually plot a plot by calculating each mean of blood pressure in age group and get the attachment2 which is a curve instead of a straight line. Could you please to advice on this? how to get the curve like the figure 1 after mi estimate ? Many thanks!

[Queran Lin]

This is a dataset with 145 million datapoints BTW. Thanks!

[Queran Lin]

Clyde Schechter can you please to take a look at this? I am sorry that I am new to here! Now I know how to used this forum! Many Thanks!