Data on Disability Determination Services Accuracy

Last updated on May 4, 2017

Download this dataset

This dataset contains fiscal year data and is currently available in the following formats:


We pay disability benefits through two programs: the Social Security disability insurance program and the Supplemental Security Income (SSI) program. An application for disability benefits can be filed online, in the local Social Security office, or by phone. We review the application to make sure some basic requirements for disability benefits are met. If these requirements are met, we send the disability application to the Disability Determination Services (DDS) office in the state in which the applicant resides. Using federal laws, regulations, and Agency policies and procedures, this state agency completes the disability determination for Social Security.

The Social Security Administration (SSA) reviews a sample of initial disability claims prior to a final determination in order to assess the accuracy component of DDS performance. The data contained in this dataset shows the results of this review for fiscal years 2007 and later.


Agency Program Description

The Code of Federal Regulations (20 CFR 404.1640-404.1670) establishes that SSA will conduct a quality assurance review of initial DDS disability determinations. The purpose of the review is to ensure that the states are complying with our regulations and other written guidelines, including meeting and maintaining established national performance standards. These standards help ensure effective and uniform administration of the disability programs and measure whether the performance of the disability determination function by each state agency is acceptable. The performance standards are also designed to improve overall state agency performance in the disability determination process, and to ensure that benefits are made available to all eligible persons in an accurate and efficient manner. See the link below to learn more about SSA’s disability programs-

Dataset Description

This dataset provides accuracy rate figures at the state and national level for SSA disability medical determinations made by the state DDSs. The data covers results from fiscal year 2007 onward. The accuracy rates provided are the combined (weighted) results of the Social Security review of initial-level favorable and unfavorable determinations and reflect cases with a substantive error (i.e., those that result in a changed decision from favorable to unfavorable or unfavorable to favorable), and cases returned to the DDS for additional action and the DDS did not complete requested action(s) timely (about 90 days).

Data Collection Description

We review a random selection of approximately 70 favorable and 70 unfavorable determinations from every state every three months. From this, we are able to project the overall accuracy rate with respect to the determinations made by each DDS.

Data Dictionary

Field A:  Federal Fiscal Year (FY), a 12 month period from October through September.

Field B:  State (with national shown as the total of all states)

Field C:  Accuracy Rate, shown as a percentage of cases

Field D:  Sampling Variability, at the 95 percent confidence level, shown as a plus or minus percentage

Field E:  Comments

Statistical Information

Statistical Methodology:
Accuracy rates represent a mean percentage of the upper and lower limits of a mathematically calculated variability range. The mathematical formula used to establish these limits considers the size of the sample and, indirectly, the number of deficiencies found. The formula used also incorporates a mathematical constant "Z score" calculated to produce a variability range acceptable at a 95.0 percent confidence level.

Sampling variability is expressed as plus or minus percentages and shown (to the second decimal) following each accuracy rate quoted. Thus, we can, for example, be 95.0 percent sure that an accuracy rate of 94.2 percent, based on the review of a sample of the workload, with a sampling variability of +/- 3.19 could be no higher than 97.4 percent and no lower than 91.0 percent if the entire workload were reviewed. We arrive at these results by adding the variability percentage to the accuracy rate to obtain the upper limit and subtracting the variability percentage from the accuracy rate to obtain the lower limit.

As the sample accuracy rate approaches 100 percent, the system rejects a sampling variability that exceeds 100 percent. The accuracy rate of the universe cannot exceed 100 percent. Therefore, in those instances, we manually compute the sampling variability and display an upper and lower limit.

All accuracy percentage rates are weighted to reflect the data as they would appear if the total workload had been reviewed.