The Reality of Disability Survey Nonresponse

In a recent post, we highlighted some of the problems that may be encountered with collecting and analyzing the disability status of employees.  In particular, there are data collection problems with relying upon a voluntary self-identification method and allowing an option to not self-identify. The post was concerned with the different disability utilization percentages that could be calculated, depending on how the individuals who do not self-identify are treated in the analysis. We thought further elaboration on the topic of survey completion percentages was warranted, as there is substantial academic research on the topic. For those interested in the specifics of the research that has been done, refer to our brief white paper on survey nonresponse. In summary, the research suggests the following:

• Survey nonresponse is a reality that will affect the disability data collected using OFCCP’s self-identification form. (Remember, contractors are only obligated to invite employees to participate and would be in violation of OFCCP’s guidance if they mandated self-identification to increase response rates. More on that later.)
• Nonresponse may bias the results of utilization analyses, such that the percentages calculated will not reflect the actual workforce percentages.
• There are available strategies to increase response rates, but substantial nonresponse is typical even if such strategies are employed.

One obvious solution to combat nonresponse is to mandate completion of the form. However, OFCCP has made it clear in a recent meeting that contractors cannot make completion of the form mandatory. We understand OFCCP’s position due to the sensitivity of the data and the voluntary nature of self-identification; however, OFCCP must acknowledge the analytic and interpretation problems that will arise with the collection of data using the prescribed form and format.

Eliminating the option to not self-identify would also increase the amount of useable data, but this also not an available strategy. Again, we understand why the option is there, but it certainly complicates the utilization analysis. The question is, what does OFCCP expect contractors to do with the proportion of the workforce that either (a) does not complete the form or (b) chooses the option to not self-identify? (Note: In applicant analyses it is a common practice for both contractors and OFCCP to remove those individuals that did not self-identify from the analysis. Why would “unknowns” be included in a utilization analysis?)

OFCCP has stated that all employees are to be included in the utilization analysis and that those who do not self-identify (either by not completing the form or choosing the option to not self-identify) should be treated as not disabled. As we pointed out in our prior post, this poses considerable problems with analysis and interpretation and will lead to utilization percentages that are below the actual percentage of disabled employees in the workforce. According to OFCCP’s stated policy, if less than 7% of your workforce participates in the survey, it will be impossible to meet OFCCP’s utilization goal. Further, OFCCP has made clear in statements from multiple representatives that the only drivers of low response rates are lack of trust in the organization or lack of knowledge about the survey. The conclusion they have drawn is that as long as employees trust that the organization has their best interests in mind and know about the voluntary self-identification form, low response rates will not be a problem. As we outline in our white paper, such a perspective is not consistent with the research literature…nor is it consistent with the understanding of those who have endeavored to conduct research involving a voluntary survey.

By Kayo Sady, Ph.D., Senior Consultant, David Cohen, M.S., President, and Emilee Tison, Ph.D., Consultant at DCI Consulting Group