Genetics

A set of Mx genetics examples from the GenomEUtwin project has been translated to Mplus to show expanded capabilities for genetics analysis found in Mplus Version 4. We would like to acknowledge the helpful advice from Irene Rebollo and Danielle Posthuma.

Click on each example to view the output for that setup or download the entire set of examples.

• Continuous Data
  • Univariate
    • Saturated
      • Twins only
        • No sex effects on variance : rawSAT1
        • Sex limitation : rawSAT2
    • Variance components (A, C or D, E - models)
      • Twins only
        • No sex effects on variance components
          • 2 group ACE/AE/CE/E : rawVC1a
          • 2 group ADE/AE/E : rawVC1b
        • Sex limitation
          • 4 group ACE/AE/CE/E : rawVC2a
          • 4 group ACE STANDARDIZED : rawVC2c
          • 4 group ADE/AE/E : rawVC2b
          • (5 or) 6 group ACE/AE/CE/E : rawVC3a
          • (5 or) 6 group ACE + different shared env factors across sexes : rawVC3c
          • (5 or) 6 group ADE/AE/E : rawVC3b
      • Twins and additional siblings
        • No sex effects on variance components
          • ACE/AE/CE/E models : rawVC5a
          • ADE/AE/E models : rawVC5b
    • Variance components and GxE interaction : rawVCmod1
    • Variance components, testing linkage (A, C or D, E, Q - models)
      • Twins only
        • pi-hat approach ACEQ/ACE model : rawVCQ1
        • IBD mixture distribution approach ACEQ/ACE model : rawVCQ2
  • Bivariate
    • Cholesky
      • Twins only
        • Two group ACE (rg rc re) : rawVC4a
  • Multivariate
    • Variance components
      • Common Pathway : compath
• Ordinal Data
  • Univariate
    • Saturated
      • Contingency tables
        • Twins only
          • 2 group : ctSATut2
          • 4 group, sex limitation : ctSATut4
          • 6 group, sex limitation : ctSATut6
      • Raw data
        • Twins only
          • 2 group : ordSATut2
          • 4 group, sex limitation : ordSATut4
          • 6 group, sex limitation : ordSATut6
    • Variance components
      • Contingency tables
        • Twins only
          • No sex effects
            • 2 groups ACE : ctVCut2c
            • 2 groups ADE : ctVCut2d
          • Sex limitation
            • 4 groups ACE : ctVCut4c
            • 4 groups ADE : ctVCut4d
            • 5/6 groups ACE : ctVCut6c
            • 5/6 groups ADE : ctVCut6d
      • Raw data
        • Twins only
          • No sex effects
            • 2 groups ACE : ordVCut2c
            • 2 groups ADE : ordVCut2d
          • Sex limitation
            • 4 groups ACE : ordVCut4c
            • 4 groups ADE : ordVCut4d
            • 5/6 groups ACE : ordVCut6c
            • 5/6 groups ADE : ordVCut6d

The standard errors in these examples are obtained through the robust maximum likelihood asymptotic theory and the delta method. When parameters are near their boundary constraints the asymptotic theory no longer holds. Thus the standard errors for parameters that are near their boundaries should not be considered trustworthy. Mplus will typically produce a warning message.

THE STANDARD ERRORS OF THE MODEL PARAMETER ESTIMATES MAY NOT BE TRUSTWORTHY FOR SOME PARAMETERS DUE TO A NON-POSITIVE DEFINITE FIRST-ORDER DERIVATIVE PRODUCT MATRIX. ...

The message may give even the particular parameter that may have incorrect standard errors. Typically this is the parameter that is approaching its boundary. Mplus may also fix the parameter to its boundary value. Standard errors for parameters approaching boundaries can be obtained by the bootstrap method. Note also that some parameter constraints are implicitly coded in the model. For example a variance component C=Y*Y has an implicit boundary 0 because C>=0.

Back to index of Examples