Estimation of the standard error and confidence interval of the indirect effect in multiple mediator models
Nancy E.Briggs · OhioLink ETD Center (Ohio Library and Information Network) · 2006
Mediation analysis seeks to go beyond the question whether an independent variable causes a change in a dependent variable.Mediation addresses the question of how that change occurs.Specifically, simple mediation occurs when the effect of a predictor variable on a dependent variable is transmitted through an intervening variable (the mediator).However, with the complicated relationships observed in the social sciences, including more than one mediating variable can provide a more complete picture of the change.Researchers can test the statistical significance of the indirect effects in multiple mediation models.Researchers can obtain an estimate of the standard error of the indirect effects and use them to calculate a test statistic or confidence intervals.Typically, the standard error estimate derived from the multivariate delta method (MDM) is used.However, the distributional assumptions of this method are often violated, especially with small sample sizes.Bootstrapping the standard error and confidence interval has been suggested as a remedy.This simulation study examined the performance of confidence intervals resulting from the MDM, the bootstrap estimate of the standard error, the bootstrap iii percentile method, the bias-corrected method, and the bias-corrected and accelerated method.Simulations were performed to examine the performance of these standard error and confidence interval estimates in models with two mediating variables, varying the sample size, amount of mediation, the relative importance of the two indirect effects, and inclusion of correlated errors of the mediator variables.Results indicated 49 Power, 95% confidence interval, First indirect effect, Populations 1 and 2 .......................................