Stat Computing > MLwiN > FAQ				Help the Stat Consulting Group by giving a gift
Loading

MLwiN FAQ
How do I create predicted values?

After creating and estimating a model, we can create predicted values using the Predictions option from Model menu.

Let's look at Table 2.1 Part 2, an example from Chapter 2 of Multilevel Analysis Techniques and Applications by Joop Hox. It uses the data file popular.ws and you can download it from here. The model is the following:

The result of the estimation using IGLS is:

Now we can create different type of predicted values from the model.

Predicting Using the Full Model

For predicted value from the random slopes model we just estimated, select Predictions from Model menu. In the Predictions window, click on the word variables and choose Include all explanatory variables. Click on the level 1 error term to remove it from the prediction equation. The last thing to do is to specify the variable to output the prediction. We can choose an unused variable, for example, c11 and change its name if we want. Now we can click on Calc to calculate the prediction.

Predicting Using Only the Fixed Effect Part of the Model

If we only interested in predicted values using the fixed effect part of the model, we can simply click on these level 2 error terms to remove them from the prediction calculation.

Predicting Using  Some Parts of the Model

Let's say for some reason, we only want to calculate the predicted values using some parts of the model, we can easily do so by select the terms we want to include. For example, we want to calculate the predicted values for cases where sex = 0.

UCLA Researchers are invited to our Statistical Consulting Services
We recommend others to our list of Other Resources for Statistical Computing Help
These pages are Copyrighted (c) by UCLA Academic Technology Services