Does the bootstrap mediation effect depend on the p value?

When testing the mediation effect in bootstrap, you do not need to look at the p value; when using the Bootstrap method to test the mediation effect, it is judged based on whether the interval "BootLLCI, BootULCI" contains 0. If it does not contain 0, The mediating effect is significant, and if it contains 0, the mediating effect is not significant.

The operating environment of this tutorial: Windows 7 system, DELL G3 computer

Does the mediation effect of bootstrap depend on the p value?

Use The Bootstrap method is used to test the mediation effect. It is not judged by the P value, but by whether the interval (BootLLCI, BootULCI) contains 0. If it does not contain 0, the mediating effect is significant; if it contains 0, it is not significant.

In the case given, the value of the mediating effect (indirect effect) is 0.1969, which is significant. The total effect of the independent variable on the dependent variable is 0.9373, which means that the mediating variable mediates 21 % effect (0.1969/0.9373), which is an incomplete mediation.

At the same time, it should be noted that the above numbers are non-standardized effect sizes. SPSS only gives the value of the standardized mediation effect, but does not give the standardized effect value of the total effect and direct effect.

In fact, the results are similar when using standardized or unstandardized effect sizes to calculate the proportion of mediating effects.

The most widely used application of Bootstrap is to test the mediation effect.

Other methods of mediating effects include:

·Most commonly used: stepwise test of regression coefficients (stepwise method) (Baron & Kenny, 1986)

Step 1: Test X →Y, which is c, is significant. If it is not significant, don’t do it.

Step 2: Test whether X →M, M →Y is significant. They must all be significant. One of them is not significant Don't do

Step 3: If the above are significant and c' is smaller than c, then it is a partial intermediary. If c' is not significant, it is completely mediated. This situation is relatively rare.

·Sobel method:

The test power is higher than the stepwise test, but assuming that a*b obeys the normal distribution, even if a and b are both normal distributions, their product is usually not positive Stateful,

So Sobel has limitations

Advantages of Bootstrap: no normal distribution is required, higher sensitivity (more likely to produce significant results)

Bootstrao To test the mediation effect, take the Process plug-in in SPSS as an example:

Step 2: Set parameters.

Select the dependent variable, independent variable, and control variable from [Variables]. These variables will form your regression equation.

Select 4 for Model number. This is the model number for mediation analysis. If you select anything else, an error will be reported.

Number of bootstrap samples is the bootstrap sample size mentioned earlier. The default is 5000, usually between 1000-5000, and 1000 or 5000 is usually filled in. The sample size of Bootstrap is different, and the data that comes out is slightly different. different.

At the same time, check save bootstrap estimates and bootstrap inference for model coefficients.

Click [Options] in the upper right corner, check show total effect model (display total effect model), and click Continue

Finally click [OK] to get the operation result

The operation result:

1. The mediating variable is the regression result of the outcome variable

2. Take the dependent variable as the outcome variable. At this time, the impact of MV on DV and the direct impact of IV on DV can be obtained

3. Total effect model: This is the natural effect without the mediating variable MV. The overall effect of variable IV on the dependent variable DV; that is, all the effects of the independent variable on the dependent variable before being mediated. After mediation, the effect of the independent variable on the dependent variable will be divided into direct effects on the dependent variable (direct effect) and The two parts of the indirect effect on the dependent variable (indirect effect)

Here comes the key point! ! ! ! The most intuitive mediation effect is here! ! !

4. Results of mediation effect test.

Related recommendations: bootstrap tutorial

The above is the detailed content of Does the bootstrap mediation effect depend on the p value?. For more information, please follow other related articles on the PHP Chinese website!