An Attitude Toward Blindness Questionnaire (ATBQ) was developed to assess individual progress through a Veterans Administration blind rehabilitation program. The instrument is also meant to measure attitudes toward blindness for blind persons, rehabilitation workers, and a "naive" group without contact with the blind. A method of assessing model fit by a Rasch Rating Scale Model was used to make group comparisons for any chosen demographic stratification without re-calibration for every set of comparisons. A conventional analysis of group and item distribution was undertaken, with an alternative analysis further recognizing that group differences must also exist within the residuals from the original overall analysis. The residuals resulted from the difference between observed and modeled responses, and were scaled to yield a standardized residual. The analysis calibrated items to uncover peculiar individual response patterns from a combined sample of workers and blind persons, facilitating the identification of specific rehabilitation needs. Estimate plot, matrix techniques and graphs for comparing group residuals, and the uses of negative residual distributions in the calibrations are described. (CM)