In my testing of statistical significance by bootstrapping, p_value shows that there are no specific indirect effects, but confidence intervals bias corrected 2.5% - 97.5% does not contain the value 0, this indirect effect has significance (followed source I read).
So there was a conflict and I wondered to use which result. Am I missing something?
First, be very aware that bias-corrected bootstrap confidence intervals for mediation analysis is very prone to Type I errors. But there could be some issues with your data in regards to your models here, are there any severe issues with non-normality of your variables?
You may check Chapter 7:
Book A Primer on Partial Least Squares Structural Equation Modeli...
And the case study:
Chapter Chapter 7 Case Study Update Using SmartPLS 4
In the context of SmartPLS 4, when assessing the significance of a specific indirect effect, it is indeed acceptable to use bias-corrected confidence intervals (BCa) as an alternative to p-values. The BCa method computes the confidence intervals for the group-specific estimations of parameters in the PLS path model and is considered to be more robust than the traditional p-value, especially when the parameter estimates distributions do not adhere to symmetry or normality. This approach is particularly recommended for indirect effects, which often do not follow the assumptions of t-distributed parameters. Therefore, if the bias-corrected confidence intervals do not include zero, the specific indirect effect can be considered significant, even if the p-value does not indicate significance at the conventional 0.05 level. It's important to note that while p-values are based on t-distribution testing and make certain distributional assumptions, BCa confidence intervals relax some of these assumptions, providing a potentially more accurate indication of the presence of indirect effects.
In addition to Dr-Zaffar Ahmad Nadaf response, this article may be useful:
Article Asymptotic and Resampling Strategies for Assessing and Compa...
Thank you Prof. Ringle, Prof. Ramayah and Dr. Nadaf for your response.