The Durbin Watson in Panel data is not effective, to fixed effect you can see

http://www.benjaminborn.de/pdf/BornBreitung2016_WP.pdf

and to random effects you can see:

https://surface.syr.edu/cgi/viewcontent.cgi?article=1055&context=cpr

Good luck

Thank you so much Dr/ Ahmadi Murjani , Dr/Jorge Ventura

for precious and valuable information,

I will read these articles

MODEL & DATA should be ompatible. There is no need, in fact wrong at bext and unethical at worst, to fix or manikate data so that it could fit the assumption of the model. If the model could not explain the data or if the data violates certain assumptions of the model, it just means that the data and the model are not compatible. An investigator cannot and must not fix the data, but has the option of looking for approriate model to explain the data. If the current available models cannot explain the data, then modify the model or propose a nee one, thus, a new breakthrough is achieved.

DATA MINING is the examining and removal of defects in data, such as incompleteness, etc., but never fixing or removing data in such a manner that is akin to data manipulation to meet model agenda. Objectivity remains a guiding principle in ressearch. If the data fails or the model fails, an investigator reports so accordingly. That in itself has utility in research, eithef as an exception case or pointing out possible break point for breaking new ground in research.

Thank you very much prof. Paul Louangrath for your precious information

I will take in my account