George-Ezzelle, C. E., & Skaggs, G. (2004, April). Examining the validity of GED tests scores with scheduling and setting accommodations . Paper presented at the annual meeting of the National Council on Measurement in Education (NCME), San Diego, CA. https://eric.ed.gov/?id=ED510061

Participants experienced one of the following accommodations/accommodation packages: (a) extended- time only, (b) extended-time and private room only, or (c) extended-time, private room, and supervised breaks only.

An extant nationwide (U.S.) dataset used for the study came from the 2002 examination cycle of the Tests of General Educational Development (GED). A sample (n=1,251) of test takers who requested and received accommodations constituted the experimental group, and 1500 randomly selected individuals who took the test without accommodations served as the reference group.

The dependent variable was scores on the 2002 administration of the GED (extant assessment data). The findings were described in terms of the content areas of English language arts/reading, science, social studies, and English language arts/writing; the math test results were not examined.

Overall, participants in the reference group had higher mean raw scores than participants receiving accommodations.
DIF analyses were performed both on individual test items, and on subtest forms. Eleven items were flagged for DIF across all test forms, and specialists examined the items to determine possible reasons the items might favor the indicated group. For several items, the specialists were unable to determine what characteristic of the item might have provided one group with an advantage. For other items, it appeared that the accommodated group might have benefited from extra time to understand the item (DIF in favor of accommodated group) or that extra time caused the accommodated group to second-guess their initial response (DIF in favor of the non-accommodated group).
Examination of DIF across subtests revealed similar patterns; DIF was found in several areas on the subtests in different subjects, but specialists had a difficult time determining why the items might have favored one group over the other. The one consistent finding showed DIF in favor of the reference group on the Mechanics subtest in the language arts and writing subject tests, leading specialists to hypothesize that non-accommodated candidates may be more accustomed to unassisted editing in capitalization, punctuation, and spelling.
The researchers concluded that the GED appeared to be valid for both groups. Limitations were discussed, and future research directions were noted.