How to use bootstrap for mediation analysis in spss

In the analysis process of mediating effects and moderating effects, there are two main ideas, one is manifest variables, and the other is latent variable structural equation model. The corresponding software is also divided into two categories, one is SPSS, SAS and other software based on manifest variable path analysis model, and the other is structural equation modeling software such as Lisrel, Amos, Mplus and so on based on latent variable model. Because SPSS is simpler to operate, many scholars are interested in how to use SPSS to analyze mediation effects and moderating effect models. They mainly use Process plug-ins or bootstrap to perform mediation analysis.

Related recommendations: "Bootstrap Introductory Tutorial"

SPSS uses the sequential regression method to test the mediating effect.
First test the regression of X-Y and analyze the total effect;
Then test the regression of M——Regression of Y, test b parameter (regression coefficient of M) and c' parameter (regression coefficient of X).
If a and b are both significant, then the mediating effect exists.
To use bootstrap, just select the bootstrap option in the regression analysis. You can set the number of samplings yourself. Usually, the sampling is at least 1000 times. At this time, when you analyze the significance of the a and b parameters, you will not look at the original significance. Instead of looking at the bootstrap confidence interval, it is significant if the confidence interval does not cover 0.
The bootstrap sampling function requires a newer spss version.

The above is the detailed content of How to use bootstrap for mediation analysis in spss. For more information, please follow other related articles on the PHP Chinese website!