According to the justification hypothesis, non-employed individuals may over-report their level of work limitation, leading to biased census/survey estimates of the prevalence of severe disabilities and the associated labor force participation rate. For researchers studying policies which impact the disabled or elderly (e.g., Supplemental Security Income, Disability Insurance, and Early Retirement), this could lead to significant bias in key parameters of interest. Using the American Community Survey, we examine the potential for both inflated and deflated reported disability status and generate a general index of disability, which can be used to reduce the bias of these self-reports in other studies. We find that at least 4.8 million individuals have left the labor force because of a work-limiting disability, at least four times greater than the impact implied by our replication of previous models. Copyright © 2013 John Wiley & Sons, Ltd.