David Charles Truitt
gradworks.umi.com/37/30/373038
The purpose of this study was to develop a valid and reliable measure and attendant rating scale for the assessment of the sight-reading skills of classical guitarists at the beginning of collegiate study. This study sought to answer the following questions:
1. What musical components and their ranges of values should be measured in a sight-reading assessment for entry-level collegiate classical guitarists?
2. Can a valid and reliable sight-reading assessment for classical guitarists be constructed?
3. Can a rating scale or other scoring mechanism be constructed that would be easy to use and promote a high degree of objectivity and reliability?
Tentative standards were drawn from data acquired by means of an electronic survey questionnaire sent to collegiate guitar teachers throughout North America (N = 1,193) which yielded 241 responses. Musical elements that emerged from the survey results were validated by comparison with specific works from the guitar’s literature that were recommended by survey respondents as appropriate for testing sight-reading skills at the college-entry level, as well as with published sample sight-reading exams. The emergent elements were constructed into six short musical exams and a rating sheet for each, loosely patterned after the scoring mechanism of the Watkins-Farnum Performance Scale. These exams constituted the pilot test, which was administered to 14 student volunteers by their college guitar instructors. Internal reliability coefficients were obtained using several methods, all of which were strong (generally in the .82 to .98 range). Interrater reliability coefficients were also strong (.91 to .99). Based on analysis of the pilot results and suggestions from teachers, one of the original assessments was slightly modified, a new assessment was constructed, and rating sheets developed. The same procedures used in the pilot test were followed in the field test, which was administered to 28 volunteer students. Internal reliability using several methods resulted in coefficients generally in the .90 to .96 range, and interrater
reliabilities ranged from .87 to .98.
