|
Size: 677
Comment: Initial commit
|
← Revision 9 as of 2025-12-19 21:52:52 ⇥
Size: 852
Comment: Fixed link
|
| Deletions are marked like this. | Additions are marked like this. |
| Line 15: | Line 15: |
| {{attachment:flow.svg}} | {{attachment:path.png}} |
| Line 19: | Line 19: |
| ''t = β,,2,,*β,,3,, / SE,,2,3,, = β,,2,,*β,,3,, / (sqrt(β,,2,,^2^*σ,,3,,^2^ + β,,3,,^2^*σ,,2,,^2^))'' | {{attachment:formula.svg}} |
| Line 21: | Line 21: |
| The current practice is to instead use a bootstrapping method to estimate the variance of ''β,,2,,*β,,3,''. | The current practice is to instead use a bootstrapping method to estimate the [[Statistics/Variance|variance]] of ''β,,2,,β,,3,,''. The calculated test statistic is then compared to a [[Statistics/TestStatistic|critical value]] from either a [[Statistics/NormalDistribution#Standard_Normal_Distribution|Z]] or [[Statistics/StudentsTDistribution|t distribution]]. |
Sobel Test
The Sobel test is a test for mediation.
Contents
Description
The hypothesis is formulated such that the relationship between an independent and dependent variable is an indirect effect through a mediator variable.
In these pictured terms, the Sobel test statistic is calculated as:
The current practice is to instead use a bootstrapping method to estimate the variance of β2β3.
The calculated test statistic is then compared to a critical value from either a Z or t distribution.