this:

... confession typically trumps physical evidence ....
Drizin & Leo, The Problem of False Confessions in the Post-DNA World (2004)(114 pp., examining 125 proven false confession cases):

This Article proceeds as follows:

Part I discusses, from a historical perspective, the study of wrongful convictions and the prominent role that false confessions have played in such studies. Part I also discusses the development of DNA testing and its role in renewing interest in the study of wrongful convictions.

Part II highlights the connection between police interrogation methods and false confessions, focusing principally on the social psychology of false confessions and research on the causes and consequences of false confessions.

Part III discusses the methodology used to compile and analyze the false confessions that make up this Article’s cohort, defines critical terms, and discusses the limitations of the data.

Part IV sets forth the quantitative findings gleaned from the cohort.

Part V takes a more qualitative approach to the data set, highlighting some of the common themes and trends that emerge from the cohort cases and describing illustrative cases in some detail.

Finally, Part VI concludes this Article with several policy recommendations suggested by the aforementioned findings, and highlights some recent positive developments which suggest that reforms designed to reduce the frequency of false confessions may stand a better chance of being implemented now than ever before.
Which is why no good lawyer will allow a client to be interviewed directly by police, whether that client is an accused, a focus person, a person of interest or a "mere" witness. Recall that Martha Stewart started off a potential witness in an investigation focused on someone else.

Regards

Mike