Preditions and Standard Errors in r

是 否 +2 论坛币

k人 参与回答

经管之家送您一份

应届毕业生专属福利!

求职就业群

赵安豆老师微信：zhaoandou666

经管之家联合CDA

送您一个全额奖学金名额~ !

立即领取

感谢您参与论坛问题回答

经管之家送您两个论坛币！

+2 论坛币

I have been using MLWin for a while and am used to being able to put out predicted values and their associated standard errors.  I have recently started using r but am struggling to replicate this data.  Any suggestions?

To put my problem into context, I have fitted a three level linear model with a constant and one other independent variable.  I am only modelling the random effects for the constant.  I am interested in the level 3 and level 2 predictions so that I can graph my outcomes for the units at level 3 and level 2.  I know that in R I can get the level 2 predictions for the model using the fixed() command but I don't seem to be able get the associated se, nor can I get the same figure for level 3.  Reading threads on this matter in the R community seems to imply that I am being niaive in my wish but I am confused as to why I can do this in one software package but not in the other.

扫码加我 拉你入群

请注明：姓名-公司-职位

以便审核进群资格，未注明则拒绝

分享0 收藏0 回帖

关键词：Standard Edition editio dition Errors interested recently standard problem started

On http://www.stats.ox.ac.uk/~snijders/mlbook.htm, you can find (in addition to other things) r scripts for the examples in Snijders & Bosker (2012). This covers various basic things you could do in multilevel analysis. For starting, see http://www.stats.ox.ac.uk/~snijders/ch45.r

Best regards,
Tom Snijders

There is no fixed() argument associated with linear models in R. If you're
using fixed() with an object of class "mer" and it is returning something,
can you show what that is returning? Nothing should be happening.

If your goal is to get the predictions of the random effects, then you
need to use the ranef() function for an object of class "mer". There is an
argument in that function "postVar" which returns the variances of the
conditional means as an attribute.

See the help page using ?ranef for examples.