Hi,
I'm new to MLwiN so sincere apologies if these questions are too elementary for this forum. Can you tell me please:
1. What does the error "Explanatory variables are different lengths(offs)" mean which appears sometimes when I try to fit a basic 2-level model?
2. Can I export my caterpillar chart with residuals by rank to, for example, a Word document? And if so, how can I do this?
3. How do I label each point on the caterpillar residual chart with my School IDs, so that they all show up together (e.g. like in SPSS with putting labels on all points on a scatterplot)?
Many thanks for your help.
Lorraine
____________________
Dr Lorraine Gilleece
Research Associate
Educational Research Centre,
St Patrick's College
Dublin, Ireland
____________________
Fitting MLM-beginner questions
-
- Posts: 1354
- Joined: Mon Oct 19, 2009 10:34 am
Re: Fitting MLM-beginner questions
1) This error message usually indicates that not all the variables in your model are the same length. To check this choose:
Data Manipulation->Names
and check that the "n" column for all variables in your model (including cons) has the same value.
2) You can copy most MLwiN windows, including graphs to the clipboard. You can then paste them into any application that accepts images. To do this select:
Edit->Copy
once you have chosen whether you want just the graph, or the window borders too the image will be on the clipboard.
3) If I understand your question correctly the closest method I could think of would be to do the following after producing the caterpillar plot:
Graphs->Customised graph(s)
Click "plot what?"
under "group" choose the residual rank column
Click "other"
Tick "group code"
Click the "Labels" button
Select "on graph"
Click "Done"
Click "Apply"
You may find that you get better results by copying the residuals columns into another package, for example Excel and producing the graphs from there. To do this open the "names" window with:
Data Manipulation->Names
select the columns that you want to copy and then use the "copy" button under the data section.
Data Manipulation->Names
and check that the "n" column for all variables in your model (including cons) has the same value.
2) You can copy most MLwiN windows, including graphs to the clipboard. You can then paste them into any application that accepts images. To do this select:
Edit->Copy
once you have chosen whether you want just the graph, or the window borders too the image will be on the clipboard.
3) If I understand your question correctly the closest method I could think of would be to do the following after producing the caterpillar plot:
Graphs->Customised graph(s)
Click "plot what?"
under "group" choose the residual rank column
Click "other"
Tick "group code"
Click the "Labels" button
Select "on graph"
Click "Done"
Click "Apply"
You may find that you get better results by copying the residuals columns into another package, for example Excel and producing the graphs from there. To do this open the "names" window with:
Data Manipulation->Names
select the columns that you want to copy and then use the "copy" button under the data section.
Re: Fitting MLM-beginner questions
Hi Chris,
Many thanks for your advice!
Lorraine
Many thanks for your advice!
Lorraine
-
- Posts: 1
- Joined: Fri Jan 21, 2011 12:18 am
Re: Fitting MLM-beginner questions
No question is too basic to ask so feel free to ask anything on this forumgilleecl wrote:Hi,
I'm new to MLwiN so sincere apologies if these questions are too elementary for this forum. Can you tell me please:
1. What does the error "Explanatory variables are different lengths(offs)" mean which appears sometimes when I try to fit a basic 2-level model?
Regarding the above quote, this error message usually occurs when running a parametric bootstrap. In such a case the bootstrap results will also be incorrect. This usually occurs because the bootstrap machinery overwrites column 50. usually this problem arises if you have an explanatory variable in column 50.
The common work around this problem is to avoid holding any data in c50 while running a parametric bootstrap.