Most Corrupt U.S. States – and What Data Mining Can Do About It

A study in the May/June 2014 issue of Public Administration Review has identified the most (and least) corrupt American states. The report, The Impact of Public Officials’ Corruption on the Size and Allocation of U.S. State Spending, lists the most corrupt states as Mississippi, Louisiana, Tennessee, Illinois, Pennsylvania, Alabama, Alaska, South Dakota, Kentucky and Florida. The least corrupt were identified as Oregon, Washington, Minnesota, Nebraska, Iowa, Vermont, Utah, New Hampshire, Colorado and Kansas.

The major caveat to the study:  Public corruption was identified by counts of convictions for corruption at the federal level. This ignores convictions in lower courts and ongoing public corruption that has yet to be caught. While these aspects, of course, can’t be quantified because they’re unknown, it would be interesting to extrapolate the study results to the larger population.

We have seen a number of public corruption scandals in the BKD Forensics practice over the years, from local entities to state and university organizations. Even in cases where actual criminal public corruption wasn’t found, numerous instances of “conflicts of interest” were identified; these often act as the beginning of more egregious—and sometimes illegal—activities.

What Do We Do Now?

Part of the effort to detect conflicts of interest and corruption includes proactive data mining, text analytics and analysis of relationship networks, or “buddy networks.”

Data mining identifies anomalous data patterns that could indicate inappropriate relationships among individuals inside and outside the public entity. These could include bribes, kickbacks, bid-rigging and other quid pro quo activities.

Text analytics enhances data mining by adding email communications, documents and social media to the analysis. Emails are particularly powerful, as they indicate what’s being discussed as well as the emotional tone of the communications. Overall emotional disposition of an individual or group can be detected before a single email is read.

Relationship network analysis combines findings from publicly available information, interviews, investigator notes, financial data and email/social media to construct a map of how individuals and entities are related. In public corruption cases, links among the various actors can be obscured by several layers or degrees of separation. This is often intentional.

Public corruption is often a conspiracy-level event, with complex structure and a large number of actors and moving parts. Forensic technologies like those discussed above are an effective means of detecting these scandals-to-be and bringing them to light.


Lanny has experience in computer forensics and electronic data discovery assisting attorneys in litigation and disputes by uncovering electronic data to be admitted into evidence. He performs forensic image copying of computer media, as well as mining, analyzing and reporting on the recovered data.

Lanny Morrow – who has written posts on BKD Forensics.

Leave a Reply

Your email address will not be published. Required fields are marked *