13.07.2015 Views

Consciousness-Based Education - Maharishi University of ...

Consciousness-Based Education - Maharishi University of ...

Consciousness-Based Education - Maharishi University of ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

consciousness-based education and governmentother diagnostic tests suggested normality), confirming the statisticaladequacy <strong>of</strong> this regression model.LISREL Analysis. In order to assess the influences <strong>of</strong> the <strong>Maharishi</strong> Effecton the multivariate system <strong>of</strong> the quality <strong>of</strong> life variables, the method<strong>of</strong> analysis <strong>of</strong> covariance structures was employed (e.g., Jöreskog &Srböm, 1979; Long 1983), using the LISREL VI program (Jöreskog &Srböm, 1986). The covariance structure model combines the approaches<strong>of</strong> factor analysis and structural equation causal modeling to assess theimpact <strong>of</strong> independent variables on a set <strong>of</strong> latent variables underlyinga group <strong>of</strong> observed variables. The term “covariance structure model”derives from the fact that the covariance matrix <strong>of</strong> the dependent andindependent variables is modeled in this methodology.The first model considered was that <strong>of</strong> a single general model factorunderlying all 11 dependent variables, with an influence <strong>of</strong> the independentvariable (<strong>Maharishi</strong> Effect Index) upon that factor. There wasa significant positive effect <strong>of</strong> the independent variable upon the quality<strong>of</strong> life factor, (t = 3.42, p < .005) 4 , and the coefficient <strong>of</strong> determinationfor the structural equation was .96. However, the single-factor measurementmodel must be assumed to be fundamentally wrong, since anegative coefficient <strong>of</strong> determination for the factor model was calculatedby the program (Joreskog & Sorbom, 1986, p. 1.36). The overallfit <strong>of</strong> the model to the data was not adequate (X 2 (54) = 94.5, p = .0005).In order to find a more adequate factor model, a two-factor modelwas posited in which the first factor consisted <strong>of</strong> each <strong>of</strong> the variablesthat had loaded substantially on the general factor noted earlier, and thesecond factor consisted <strong>of</strong> the other variables. The automatic modificationprocedure <strong>of</strong> LISREL was used to allow a variable to load on theother factor as well if that would significantly improve the fit <strong>of</strong> themeasurement model. Although this approach to the factor equation isexploratory rather than confirmatory, this was felt to be justified sincethe primary hypothesis being tested concerned the structural equationpart <strong>of</strong> the model (the effect <strong>of</strong> the independent variable) rather thanthe measurement model. Two <strong>of</strong> the variables from the initial generalfactor also loaded on the other factor due to the automatic modificationprocedure. The first factor, which is referred to as the general factorsince it was derived from the initial single-factor model, consisted <strong>of</strong>554

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!