How accurate is E-Verify?

January along with a summary that pointed out E-Verify is accurate “almost half of the time.” “While not perfect, it is important to note that E-Verify is much more effective” than the paper forms used by most employers, the report’s executive summary said.

Radnofsky and Miriam Jordan offer a summary of the report’s main findings:

The evaluation found that, overall, E-Verify accurately screened 96% of workers, correctly identifying 93.1% of cases as people allowed to work in the U.S. and 2.9% as unauthorized. Some 3.3% of cases were illegal workers mistaken for legal workers, and 0.7% were individuals who could work in the U.S. but were not initially identified that way.

Here are the relevant sections from the executive summary of the Westat report (emphasis in the original):

The total inaccuracy rate is approximately 4.1 percent. ….[I]t is estimated that in approximately 96.0 percent of E-Verify cases submitted from April through June 2008, the E-Verify finding was consistent with the worker’s true employment-authorization status (an estimated 93.1 percent of all cases were cases in which an authorized worker was found to be employment authorized and 2.9 percent were unauthorized workers who were not found to be employment authorized). The remaining cases (4.1 percent of all cases submitted to E-Verify) received a finding that was inconsistent with the worker’s true employment-authorized status (3.3 percent were unauthorized workers who were found to be work authorized and 0.7 percent were authorized workers not initially found to be employment authorized). This means the total inaccuracy rate (resulting primarily from identity fraud, out-of-date or inaccurate Federal records, and data input errors) is estimated to be approximately 4.1 percent. The corresponding plausible range is estimated as 2.3 percent to 5.7 percent.

The inaccuracy rate for authorized workers is less than 1 percent. The overwhelming percentage of authorized workers are found to be employment authorized (93.1/ (93.1+0.7)) = 99.2 percent), and only 0.8 percent of authorized workers are not initially found to be employment authorized. In other words, the inaccuracy rate for authorized workers is approximately 0.8 percent with a plausible range of 0.6 percent to 1.0 percent.

Due primarily to identity fraud, the inaccuracy rate for unauthorized workers is approximately 54 percent. … [A]pproximately 3.3 percent of all E-Verify findings are for unauthorized workers incorrectly found employment authorized and 2.9 percent of all findings are for unauthorized workers correctly not found employment authorized. Thus, almost half of all unauthorized workers are correctly not found to be employment authorized (2.9/6.2) and just over half are found to be employment authorized (3.3/6.2). Consequently, the inaccuracy rate for unauthorized workers is estimated to be approximately 54 percent with a plausible range of 37 percent to 64 percent. This finding is not surprising, given that since the inception of E-Verify it has been clear that many unauthorized workers obtain employment by committing identity fraud that cannot be detected by E-Verify.

You may need a degree in statistical methods to grasp the report’s heavily numerical presentation, but the quote that caught our eye — and which caught the eye of most of those who reported on it — is highlighted by the report’s authors: “Due primarily to identity fraud, the inaccuracy rate for unauthorized workers is approximately 54 percent.”