School Psychology Review 1996, Vol. 25, No.1. pp. 94-107 Situational Specificity and Generality of Test Behaviors for Samples of Normal and Referred Children Joseph]. Glutting and Eric A. Youngstrom University of Delaware Tom Oakland University of Texas at Austin Marley W Watkins Arizona State University Abstract: The uses of observations generated during testing were examined through (a) a quantitative synthesis of the available research literaLUre. (b) a study conducted with a na- tionally representative sample of children (N . MO. including 71 with additional behavioral information), and (c) a study completed wlth children referred for psychoeducational evalu- ations eN .. 140). Results demonstrated that behavloral and temperament qualities evaluated by test observations are related to children!> pe r formance on the Wechsler Intelligence Scale for C hildren- Third Edition (WISC·UI ; Wechsler. 1991 ). Test observations provided consid- erably less inSight into children's adaptation and adjustmen t outside the test-session environ- me nLIQs f rom the WI SC-lll showed limited prediction of classroom behavior, ind icating that as much as 97% of [he score variation is tndependent. F ind i ngs are examined in light of re- search on t he situationa l specific ity of behavior. Recording test-session behaviors is an im- portant practice that fulfills a variety of assess- ment functiOns. This article describes possible purposes for evaluating test-session behavior and empirically examines three commonly as- sumed reasons for monitoring such activities: (a) to help illuminate the scorability of chil- dren!> responses on formal tests administered during the same test session, (b) to document the degree to which the test environment is con- ducive to optimal performance, and (c) to pro- vide a sa mple of behavior from a controlled set- ting that may then ge neralize and describe chil- dren!> behaviors and attitudes in other situa- tions. A best ev idence synthesis will be used to evaluate these premises, reviewing the results of published stud ies and test manuals, fo llowed by two empirical studies. The two studies will use a new instrument for evaluating children's test behavior with both a large, nationally stratified sample and a smaller sample of children re- ferred for psychoeducational testing. Results will provide evidence to support the value of formally recording test-session behaviors, as well as indicating limitations on how behavioral observations from test sessions generalize to other contexts. The observation and recording of test be- havio rs is routine in the field of individual ap- praisal. For example, students attending gradu- ate programs in school and clinical psychology are taught to evaluate the behaviors children display when responding to items on individu- ally administered tests. Ukewise, textbooks on individual appraisal and intelligence testing en- courage the recording and interpreting of chil- dren's test behaviors (Kamphaus, 1993; Kauf- man, 1994; Sattler, 1988). Test observations completed by trained clini cia ns offer certain presumed advantages over the be ha vior ratings collected by parents and te achers. Specifically, test observations offe r a standardized methodology for comparing a childs behavior to the behavior of others. The uniformity of conditions under which test ob- servations occur render them relatively free of environmental variation (e.g., home or class- room climate) that can interfere with the collec- tion of valid observations. None of the major contexts of child development (e.g., home, school. and community) offers as hlgh a level of professional expertise, observational control, or uniformity of conditions as the context of indi- vidual test-taking. The relative objectivity of examiners is an- other benefit of test observations. Parmts, teachers, and others entrusted to rate childrens behaviors often lack fonnal training in data col- Add ress all co rrespondence concerning [his anicle to J osephl Gl utting. 211 Willard Hall . Depanment of Educational S lud· i es, Universi ty of Delaware. Newark. DE 19716. Electronic mail may be sent via Internet to gluuin g@strauss.udel.edu. 94