Hi Andrew,

there is a large literature on whether it is reasonable to report standardized effects at all (and concludes that it is not):

Greenland, S., Schlesselman, J. J., & Criqui, M. H. (1986). The fallacy of employing standardized regression coefficients and correlations as measures of effect. American Journal of Epidemiology, 123(2), 203-208.

Kim, J. O. J.-O., & Mueller, C. W. (1976). Standardized and unstandardized coefficients in causal analysis: An expository note. Sociological Methods & Research, 4(4), 423-438. doi:10.1177/004912417600400402

King, G. G. (1986). How not to lie with statistics: Avoiding common mistakes in quantitative political science. American Journal of Political Science, 30(3), 666-687. doi:10.2307/2111095

Besides these more fundamental issues, you can only report standardized effects in a MGA after supporting equal variances - invariance. The reason is simply that standardized effects are a function of the variances of the independent and dependent variable and hence will differ for two groups when they differ in the variances.

Best,

Holger

Morning!

Standardized values should be reported. Think of z-scores and their significance when comparing groups since only significant differences (and not differences in significances) should be reported.

Good luck!

Eugene

Hi Andrew,

in addition to the previous comments:

If one of the groups serves as a "reference", you can standardize effect sizes on the variance in that group (i.e. Glass's Delta), which can sometimes be helpful.

With apologies for citing my own work, this might help:

Article Beyond ANOVA: An Introduction to Structural Equation Models ...

Best,

Heiko

@Holger, @Yevgen, and @Heiko:

Thank you for the input.

Just to clarify: in the event that I am comparing the effects between groups, would I report the unstandardized effects? Conversely, if I am examining the results within groups, would I report the standardized effects?

Also, in AMOS (or perhaps any other program), where would I find the significance of the standardized and unstandardized effects? I presume that the significance for unstandardized effects is found under Estimates > Scalars > Regression Weights. The significance levels for standardized effects are not as apparent.

I am rather new to SEM, so any assistance or clarification would be appreciated.