College and university · 1999 · 25 citations · 0 references
EducationEducational AssessmentEducational MeasurementPerformance MeasurementStandardized TestingTest ConstructionAchievement TestingAssessment ValidityScore ReportingStudent Achievement AnalysisStudent AssessmentTest ValidityAccuracyAssessment Accuracy AnalysisGraduate Record ExaminationsStandardized Test ScoresStudent Learning OutcomesStudent Progress EvaluationStudent Score ReportsTest Score Validity
Abstract The validity and accuracy Yates of self-reported Graduate Record Examination (GRE) standardized test scores are examined using a national sample of 1,408 college students. The data yielded high correlations between actual and student-reported scores, and high accuracy rates for different groups of students (e.g., gender, race, and family income) and for students at different types of institutions. Approximately 88 percent of the student-reported scores are within forty points of the actual score. Differences in accuracy rates are associated with prior student performance on standardized tests and with the selectivity of their undergraduate college. Guidelines for the use of student-reported scores are offered. The author concludes that student-reported test scores are as useful as actual scores, Demonstrating that student outcomes are associated with educational programs requires the use of measures or indicators of student development, growth, change or learning. Thus, the National Education Goals Panel has made a case for the use of short-term ... measures of progress as proxies for more direct measures of educational outcomes (Ewell and Jones 1993). Institutional assessment studies often use college grades, self-reports of change or development, and standardized test scores. College and self-reports are used ostensibly because educational and institutional researchers find them easier to obtain. Surveys such as the Student Information Form (Cooperative Institutional Research Program 1985) and the College Student Experiences questionnaire (Pace 1983) are designed to provide a variety of student-reported indicators of educational outcomes and factual information. The surveys include self-reports of growth or development, behavioral and psychological measures, and items of demographic information, academic and career achievement, and finances. While standardized tests could be used as indicators of cognitive development (Anaya 1992; Anaya 1999), obtaining the data can generally involve substantial expenses and administrative efforts. Obtaining student-reported test scores, along with other factual information through surveys, may be a viable alternative for institutional research and college impact studies. Baird (1976) reviewed the research conducted between 1937 and 1975 on the accuracy of self-reports of factual data. Educational research during this period had examined the accuracy and validity of self-reports of participation in extracurricular activities, awards received, courses taken, and course received. This early research reveals a remarkable degree of consistency in the validity and accuracy rates for a variety of factual academic data reported by students. The correlations between selfreported and school-reported ranged from .74 to .96. Smaller correlations were observed for lower achieving students, specific academic areas (as opposed to overall grade point averages), and for men. The percent of students reporting high school in exact agreement with their school's reports ranged from 71 percent to 92 percent. The lower accuracy rates were found amongst the lower achieving students and those who applied to more selective institutions. The accuracy rates of self-reports of and courses taken in high school were consistently higher than those for co-curricular data. The accuracy rates and the validity for self-reports appear to be affected by several factors. First, the variation in correlations and accuracy rates in the studies reviewed by Baird (1976) reflected differences in definitions. That is, for example, accuracy can be defined as 'exact agreement' with the actual grade, or as falling within a specific range from the actual grade. The validity and accuracy rates are possibly underestimated given that school reports contain errors, thus self reported can be as useful as school-reported grades (Baird 1976). Second, some variation is also associated with certain groups of students and for some types of institutions. …