FAQs
FAQ's regarding to purchasing SuperMix
 
  1. What is SuperMix? Should I use it for my analysis?
  2. What are the differences between SuperMix and HLM? What methods are used by SuperMix for estimation?
  3. What operating systems are required to run SuperMix models?
  4. How can I purchase SuperMix?

1. What is SuperMix? Should I use it for my analysis?

SuperMix is a statistical software package, that deals with mixed-effects models, also known as multilevel, hierarchical, or random-effects models. It handles the following types of outcome variables: continuous, count, ordinal, binary and nominal. These models can be used for the analysis of cross-sectional hierarchical data and longitudinal data, where each individual may be measured on and at a different number of occasions.

You may look at the SuperMix Examples to see the capability of the software. If you still can not decide, we suggest you to download either the Student Edition of SuperMix or the 15-day Trial Edition of SuperMix.

2. What are the differences between SuperMix and HLM? What methods are used by SuperMix for estimation?

Currently, if the outcome variable is count, ordinal, binary or nominal, SuperMix uses maximum likelihood estimates of the model parameters via numerical quadrature (integration). HLM uses approximate methods to obtain parameter estimates and also does not produce a deviance statistic for comparing nested models. 
In the case of continuous outcomes, both HLM and SuperMix use ML and yield identical estimates. SuperMix, however, also allows users to impose constraints on the covariance matrices of the random effects. HLM offers both restricted and full ML. SMIX is full ML all the way.

3. What operating system is required for running SuperMix models?

SuperMix can be run on either Windows XP or Vista.

4. How can I purchase SuperMix?

To purchase SuperMix, please read relevant information on the ordering section of our web site.

 
FAQ's regarding to using SuperMix
 
  1. How can I get started with SuperMix?
  2. Should I sort the data before running a model?
  3. Why can't I get SuperMix output?
  4. Other notices for SuperMix.

1. How can I get started with SuperMix?

Do you need guidance with setting up your SuperMix analysis? We suggest checking the online help file, which is accessed via the Contents option of the SuperMix Help menu.

If you are not a SuperMix 1.1 user yet and interested to learn more about the program capabilities, see: SuperMix 1 WebHelp file.

2.Should I sort the data before running a model?

Before running a model, the data must be sorted by level 2 ID (level-2 model) and in the case of a level-3 model, IDs should be sorted in the sequence (level-3; level-2).

3. Why can't I get SuperMix output?

Step 1
Check version number: SuperMix 1.1, September 2008. Contact SuperMix tech support if your version number and/or date are different. 
Check operating system.

Step 2
(1) .inp and .out files are stored in a folder with an extremely long name or non-English characters
(2) folder is write-protected
(3) running from a USB stick with too little space available for temp files.

4. Other notices:

  • When running SuperMix using Optimization Method: maximum posterior (Advanced tab)
    The following message is written to the output file:
        **** Convergence was not attained. Results below may not be reliable.
        **** Increasing the maximum number of iterations may solve the problem.
    Ignore this message if the actual number of iterations is smaller than the value specified in the Number of Iterations: text box (Configuration tab).
    This problem will be corrected in the first free upgrade of SuperMix.
  • Contact tech support (This email address is being protected from spambots. You need JavaScript enabled to view it.) to report any bugs .
  • There is a bug for mixrega.exe, contact tech support for the update.
  • Data-based graphs with missing values. 
    Currently, -9 is assumed to be a missing value code when doing data-based graphs. If -9 is a valid value for some variables, then one can change the default to the actual missing value code or to -999999 using the Settings menu when the graph is displayed.

 

  • Residual plots.
    Residual plots are not available for all model types. Currently, it is only available for the following cases: 
    (1) Normal continuous outcomes.
    (2) Distributions that use the log- link function.