MULTIPLE MEASURES AND HIGH STAKE TESTING

Multiple Measures and High Stake Testing



Multiple Measures and High Stake Testing

Introduction

Over 30 years ago, Fiedler (1970, p. 1) examined the hypothesis that leaders with more experience are more effective and subsequently declared that this notion had been ''shot to hell.'' Nonetheless, many organizations continue to place a great deal of emphasis on individuals' previous work experience when selecting managers and leaders, such that individuals with limited experience are less likely to be selected for formal leadership positions. Despite an overwhelming number of studies examining leadership (cf. Yukl and Van Fleet, 1997), surprisingly, few have examined the efficacy of this selection practice. 

From a practical standpoint, if experience continues to be an integral criterion for leader selection, then it is imperative that the relationship between leader experience and effectiveness be determined. Although the face validity of this selection practice is clearly high, early empirical evidence suggested that the actual predictive validity of leader experience might be significantly lower (Fiedler, 1970). In a review of five studies that encompassed 11 different task groups, Fiedler (1970) found the median correlation between years of organizational service (experience) and leadership performance to be .12. More recently, Fiedler again claimed that ''we found no consistent relationship between experience or job tenure and leadership performance'' (Fiedler and Garcia, 1987, p. 41). Subsequently, however, some studies have yielded preliminary support for the experience-effectiveness linkage (Bettin and Kennedy, 1990; Cannella and Rowe, 1995). Thus, the validity of Fiedler's conclusion that experience is not predictive of leader effectiveness is questionable. In an effort to reconcile these seemingly contradictory findings, the aim of the present study is to further explicate and clarify the relationship between leader experience and effectiveness. 

Multiple Measures

Many state and federal regulations now require schools to report multiple measures—multiple measures of student achievement, that is. While we applaud these changes from the old method of using one standardized achievement score to make decisions about how well a school is doing, multiplemeasures of student learning alone are not sufficient for comprehensive school improvement, and, in fact, can be misleading in this regard. Many educators believe that over 50 percent of student achievement results can be explained by other factors. That being true, if we want to change the results we are getting, we have to understand the other 50 percent to know why we are getting the results we are getting. Then we need to change what we do in order to get different results. Any definition of multiple measures should include four major measures of data — not just student learning, but also demographics, perceptions, and school processes. Analyses of demographics, perceptions, student learning, and school processes provide a powerful picture that will help us understand the school's impact on student achievement. When used together, these measures give schools the information they need to improve teaching and learning to get positive results. In the figure that follows, the four major measures are shown as overlapping circles.

The figure illustrates the type of information that one can gain from individual measures and the enhanced levels of analyses that can be gained from the intersections of the measures. One measure by itself gives useful information. Comprehensive measures, used together and over time, provide much richer information. Ultimately, schools need to be able to predict what we must do to meet the needs of all students they have, or will have in the ...