Judgment of counseling process: Reliability, agreement, and error.

Abstract

The purpose of this article is to present a solution in meeting the problem of denning and measuring agreement of judges and raters as used in psychological research. Because reliabilities and correlations are measures of consistency rather than agreement, a chi-square analysis is presented and probabilities are discussed to reveal the degree of agreement or disagreement. By using this method, the researcher can predetermine his own criterion and compare the frequencies of agreements with expected probabilities.

The Justification of Beliefs in Young Adults: A Longitudinal Study

“…For 1977 ratings, the proportion of agree ment in the total group was 0.87. After cor rection for chance agreement [Lawlis andLu. 1972: Tinsley andWeiss, 1975], the figure was 0.81 (p < 0.05).…”

Cardiovascular Correlates of the Psychotherapeutic Process

“…For ordinal and interval scales they suggest Lawlis and Lu's [1972] chi square and T index. Using a defi nition of agreement as all raters within one point of each other, we found percent agreements calculated on an interview by interview basis, to range between 65 and 100%.…”

Evaluating the performance of Automated Text Scoring systems

“…7 Substituting P e in equation (1) gives (P a − 1/g)/(1 − 1/g), which is better suited when one or both of the marginals are free to vary. When the grades are not uniformly distributed, P e may be higher than 1/g; nevertheless, it can be a useful lower limit for P e (Lawlis and Lu, 1972). Note that in the example in Table 1, BP would be the same for both ats\gs 1 and ats\gs 2 , (0.8 − 0.5)/(1 − 0.5) = 0.6, and thus effectively remains unaltered under the effects of trait prevalence.…”