A procedure for estimating reliability in a factor analytic context, when reliability of the extracted factors is not an emphasis, is identified. The procedure is an extension of Dressel's work and might be applied in attitude measurement. It assesses how homogeneous the weighted original item responses are, when they are scored for pattern congruity. Four data sets generated to simulate different sample sizes, numbers of variables, numbers of variables per factor, and proportions of randomly generated error, were analyzed with alpha factor analysis followed by rotation to the varimax criterion. Subsequently, both types of consistency estimates were calculated. Results indicate that the two indices are sensitive to different data characteristics. The alpha coefficients are primarily sensitive to the variance each factor explains, as reflected by the number and magnitude of the salient pattern coefficients associated with each factor. The Dressel coefficients, however, are sensitive to the proportion of error embedded in the original data, and are less sensitive to the number of salient items associated with each factor. (RL)