A C++ Program for Calculating the Maximal Split-half Reliability of a Test

碩士 === 明新科技大學 === 工業工程與管理研究所 === 97 === Reliability is the stability and precision of test results, or it is the similarity between the results of something that is tested twice or more. This research aims to study the split-half reliabilities of a test and to design a C++ program which can quic...

Full description

Bibliographic Details
Main Authors: Chun-Yen Chuang, 莊俊彥
Other Authors: 馮文俊
Format: Others
Language:zh-TW
Published: 2009
Online Access:http://ndltd.ncl.edu.tw/handle/34273562227464997511
Description
Summary:碩士 === 明新科技大學 === 工業工程與管理研究所 === 97 === Reliability is the stability and precision of test results, or it is the similarity between the results of something that is tested twice or more. This research aims to study the split-half reliabilities of a test and to design a C++ program which can quickly acquire the maximal split-half reliability of a test. Although Guttman(1945) have proven that the maximal split-half reliability of a test was the most appropriate estimate of the population reliability, however, in comparison with provided automatically by most statistical packages, the calculation for acquiring the maximal split-half reliability of a test is difficult and time-consuming, and almost all test users use rather than the maximal split-half reliability to estimate population reliability. This research provides a C++ program for calculating the maximal split-half reliability of a test. It is easily applied and very fast in calculation. Applying this C++ program to the responses of 3494 students, it took about 5.7, 16.8, and 66.4 minutes to obtain the maximal split-half reliability of the test with 20, 22, and 24 items, respectively.