Psy~holo~icalReports, 1991, 68, 1379-1386. @ Psychological Reports 1991 FACTOR ANALYTIC STUDY OF THE AIKEN AND HAGE MEASURES OF PERCEIVED ORGANIZATIONAL STRUCTURE AND TECHNOLOGY' LYNN E. MILLER AND RICHARD M. WEISS Ln Salle Universily Unioersio of Delaware Summary.-Much work on orgaruzational structure has been based on a set of scales developed by Aiken and I-lage to measure centralization, Formalization, and technology, yet the scales have been subjected to limited psychometric testing. This paper presents Factor analyses of the scales and makes recommendations for their revi- sion and use. Studies of organizational structure have fallen into one of two camps, those that measure elements such as centralization, formalization, and tech- nology on the basis of employees' perceptions and those that use more ob- jective measures (e.g., based on company documents). Repeatedly, research has shown that objective and perceptual assessments of structure do not cor- relate very highly. Rather than dismissing subjective measures as simply unreliable surrogates for objective structure measures, Sathe (1978) suggested that the two approaches may be assessing different aspects of structure, with objective measures tapping designed structure and subjective measures pro- viding evidence about the emergent structure. For example, objective meas- ures of formalization may assess the extent of written rules, whereas percep- tual measures may assess the extent to which employees actually pay atten- tion to the rules. In 1980 Dewar, Whetten, and Boje attempted to provide a psychomet- ric examination of the best-known measures of perceived structure, those of Aiken and Hage (1966, 1968). After examining coefficient alpha lower re- liability bounds and patterns of intercorrelations among items within and across the scales, Dewar, et al. tentatively concluded that several scales lacked face, convergent, and/or discriminant validity . Factor analysis likely would have produced more conclusive prescriptions regarding how the scales should be revised, but Dewar, et al. argued that they had too small a sample for that analysis, having averaged the responses of all employees from the same organization. Because the scales are administered to individuals, factor analysis with the individual rather than the organization as the unit of analysis actually may be more appropriate. That is, items that load on the same factor should do so because individual respondents answer them similarly. O n the other 'Correspondence should be addressed to Lynn E. Miller, Management Department, La S d e University, Philadelphia, PA 19141.