Reliability
Reliability is a statistical test that measures the ability of an instrument to give consistent results at different times, in the same sample population, in different sample populations, or using different methods. High reliability of results occurs when there is consistency in results obtained from different measurements or instruments. For a researcher to depend upon certain instruments or methods of measurement, it must have the consistency of results at different conditions of the study. Kaplan and Saccuzzo (2009) argue that reliability is a property of an instrument or method of measurement to give consistent results over a period or under varied conditions of research (p.103).
Since research studies seek to obtain valid and credible results, the reliability of an instrument is imperative so that other researchers can be in a position to validate research assertions of any experiment. However, although an instrument can be reliable and give consistent results, there is variability in samples, which gives errors in measurement. As errors in measurement can occur due to inherent property of samples or defect of an instrument, several forms of reliability tests such as test-retest reliability, inter-rater reliability, parallel-forms reliability, and internal consistency reliability are applicable in ascertaining overall reliability. Therefore, how are test-retest reliability and its standard error of measurement applicable in research studies?
Test-retest is a reliability test that measures the reliability of an instrument or method of measurement over a given period. The test-retest is an important reliability test because it measures the consistency of certain property or quality across time. Since confounding variables may cause certain property of measurement to vary across time, test-retest reliability seeks to ascertain if there are significant changes that would invalidate research findings if not considered.
The test-retest reliability presumes that there is no significant change in a given property of measurement over a certain period; therefore, results are reliable even when measured over a long period. According to the American Educational Research Association, the American Psychological Association, and the National Council on Measurement in Education (1999), test-retest reliability is applicable in experiments that have a control group without any treatment to ascertain whether the variability of a property occurs by chance or due to treatment (p.55). Thus, the consistency of a given property has significant value in the assessment of variables.
Test-retest reliability is subject to error of measurement that emanates from the defective measurement of an instrument, the inherent variability of parameters over time, and biases of researchers. The error of measurement is the difference between the theoretical value and the measured value. The test-retest reliability assumes that a given parameter of study is constant across time, and this forms the theoretical value, while the measured value is variable depending on several confounding variables that affect it.
Since the size of errors vary according to types of parameters measured, the standard error of measurement gives a conventional view of errors to enhance statistical interpretation of research findings. According to Harvill (2009), the standard error of measurement is a standard deviation of errors calculated from variability in the theoretical value and the measured value (p.182). Hence, the standard deviation of theoretical value and measured value in test-retest reliability gives a standard error of measurement, which evaluates the consistency of results across time.
Because the credibility of research findings depends on the reliability of an instrument and method of measurement, research design must consider factors that threaten reliability. In the case of the test-retest form of reliability, it requires parameters that do not vary from time to time such as quantitative determination of minerals concentration or test of intelligence in exams.
Internal consistency reliability is also essential in determining reliability within samples of study because it assesses if there are inherent attributes that affect the consistency of the findings. To ascertain the consistency of instruments, parallel forms of reliability is critical as it determines whether research findings vary from one instrument to another. Thus, test-retest reliability, internal consistency reliability, and parallel forms of reliability are critical in designing research.
References
American Educational Research Association, American Psychological Association, & National Council on Measurement in Education. (1999). Standards for Educational and Psychological Testing. Washington, DC: American Educational Research Association.
Harvill, L. (2009). Standard Error of Measurement. Instructional Topics in Educational Measurement, 181-189.
Kaplan, R., & Saccuzzo, D. (2009). Psychological Testing: Principles, Applications, and Issues (7th Ed.). Belmont, CA: Cengage.