Error in Classical Testing and The Generalization Theories.
A test construction refers to the activities involved in developing and evaluating a psychological function test. Testing of a mental function involves three processes, namely, specification of the construct of interest, a decision on the test format, and choosing the method to be applied in the process. However, decision making skill constitutes diagnosis, description of expertise, and the recovery of the procedure to be used, and identification of the methodology to be applied (Parekh, 2015).
The psychological functioning test is vulnerable to errors that can originate from the individual carrying out the analysis, the examinee, or the methodology and process involved in the testing. The results obtained from the functional and psychological testing are crucial in determining the performance and abilities of the examinees. Errors and their sources ignite a series of problems in test construction, which are estimated mostly to originate from the test items and constructs involved (Traub, 1997). Therefore, it is necessary to identify and isolate these errors to boost the validity and efficiency of the psychological functioning test. This paper expounds on how mistakes can be identified and explained in the classical testing theory and the generalization theory.
- i) Error in classical testing theory
The classical testing theory is simply a model that expounds on how measurement errors can influence the scores observed during psychometric testing. The classical testing theory conceptualizes; the measurement error recognition, description of failure as a random variable, and the introduction of correlation and its indexing.The classical testing theory assumes that each person being tested has an observed score and one correct score. In other words, it implies that each person being tested has an exact test score. The Classical True Test Score describes how errors of measurement can influence the observed score. However, the classical testing theory aids in the determination of the reliability and characteristics of measurement instruments. Moreover, the classical testing theory introduces three types of test scores, namely; the correct score, error score, and test scores.
However, the classical testing theory assumes that error scores and test scores are not correlated. The method further expounds that error scores of the examinees are zero and that the error scores are on two parallel tests and not related. Each examinee has an observable rating that could be available in the absence of errors involved in the measurement. Generally, these visible scores could have a certain degree of variance from the individual’s actual ability. This difference could originate from the imperfection of the instruments used in the psychometric function testing. According to the classical testing theory, the difference observed between the correct score and the observable score is considered to yield from errors sourced from the measurement.
The classical testing theory relies on the assumption that error is a variable which randomly and generally distributed throughout the testing. The method further stipulates that the individual score obtained is the correct score, and therefore it should not change even on the repetition of the same testing procedure. (Traub, 1997). However, according to the classical testing theory, this individual’s correct score is influenced by a certain degree of error, which can either be negative or positive.
According to the classical testing theory, the standard deviation of random errors for each individual involved in the test highlights the intensity of measurement incurred during the testing process.Classical Test Theory analyses are done on the entire test as opposed to each item. Item statistics apply to only that group of test-takers on that specific collection of objects. Furthermore, the classical testing theory assumes that the distribution of random errors remains constant in all the individuals taking the psychometric test (Crocker & Algina, 2008). Therefore, the classical testing theory uses the standard deviation of errors as its essential error measurement.
Practically, standard deviation and test reliability of the score observed are applied in the estimation of standard error involved in the measurement. According to this theory, the smaller the standard error engaged in analysis, the higher the degree of accuracy in the score obtained. The high degree of precision means that the observed score is more close to the correct score obtained. Finally, the standard deviation of errors has used the creation of intervals of confidence in a specific observed score.
- ii) Error in generalizability theory
The second theory is the generalization theory, which is psychometrical and bases its statistical sampling on a partitioning of scores and their respective underlying sources of variation. The generalization theory also provides a framework for identifying and estimating several errors of variance (Allen, & Yen, 1979). However, the generalization theory further initiates the division of deviation of observed scores into three main sections. The first division is the variance, which originates from personal and measurement error. In contrast, the second division is the variance caused as a result of man, and lastly, the deviation generated from the interaction of the effects of several sources.
The generalization theory further addresses thesurvivor study. According to this study, a survivor is viewed as the object to be measured, and errors in the measurement form the unexpected inconsistencies caused by raters and interviews (Allen, & Yen, 1979). Precisely, the errors in measurement are attributed to errors incurred during the sampling of interviewers, theirinteractions, evaluators, and other sources that can be unspecified.
For instance, according to the generalization theory, error in measurement associated with interviewers could represent that overall the 20 interviews, one interviewer elicited a response which indicates higher levels impairment compared to their interviewers. However, the generalization theory reinforced by statistical procedures upholds the simultaneous variance estimation caused by a single source regarded as a variance component (Kean &Reilly). Furthermore, the standard error of components of variance demands attribution during the examination of the various parts. Elevated standard errors signal the instability in the variance components estimation. When the value of a negative estimate is higher than its standard error, then that component of variance should not be treated as zero according to the generalization theory.
References
Allen, M. J., & Yen, W. M. (1979). Introduction to Measurement Theory. Long Grove, Illinois.
Crocker, L., & Algina, J. (2008). Introduction to classical and modern test theory Mason. OH:
Kean, J., & Reilly, J. (2014). Item Response Theory. In Statistics (pp. 195-197). Hammond.
Parekh, B. (2015). Item Response Theory. Psychometrics.
Traub, R. E. (1997). Classical Test Theory in Historical Perspective. Educational Measurement: Issues and Practice.