Equal-appearing Interval Research Articles

Construction of a personality inventory through Thurstone scaling of daydreams.

In this study an empirical technique was used to construct a personality inventory. Daydreams were chosen as the medium in which certain personality variables were reflected and the Thurstone scaling method of equal-appearing intervals served as the technique for scaling the daydreams and constructing the inventory. Daydreams were selected from two daydream questionnaires, and this pool of items (279 daydreams) was subjected to a Thurstone-type scaling by expert judges on nine dimensions (Pessimism-Optimism, Moving Toward People-Moving Away From People, Morbid Content-Nonmorbid Content, Self-orientationOther-orientation, Unpleasant-Pleasant, Unrealistic-Realistic, Positive Social Orientation-Negative Social Orientation, Past-Future, Physical Interest-Psychosocial Interest). This scaling resulted in the formulation of a daydream inventory which has nine scales. Reliability and validity of the daydream inventory were then assessed. A test-retest reliability procedure was used with 20 Ss. Only three scales (Past, Unrealistic, and Movement Toward) showed Spearman rho test-retest values equal to or greater than .70. I t was believed that these low reliability values reflected the use of a small and fairly homogeneous sample of students. The validity study atcempted to differentiate two known groups (20 graduate students and 23 prison inmates). The daydream inventory did separate the two groups and with reasonable accuracy on rwo of the nine scales (Past and Morbid). The only other scale that showed any indication of differentiating the groups was the Unrealistic scale. The results of these preliminary studies give some indication of the inventory's potential reliability and validity. The use of expert judges and the wellestablished methodology add evidence for validation, but a more adequate sample is necessary before the reliability and validity of this inventory can be demonstrated empirically.

Spectral noise levels and roughness severity ratings for normal and simulated rough vowels produced by adult females.

Twenty normal-speaking adult females sustained seven-second productions of the vowels /u/, /i/, /Λ/, /a/, and /æ/ first normally and then with simulated vocal roughness at one intensity. Recordings of the vowels were rated for roughness on a five-point equal-appearing intervals scale by 11 trained judges. Each production was also analyzed to produce a 3-Hz bandwidth frequency-by-amplitude acoustic spectrum in which the level of inharmonic, i.e., noise, components was measured in dB SPL. Noise levels, averaged over selected frequency ranges from 100 to 8000 Hz, for each vowel production correlated highly with the median roughness ratings for the productions. High (≥0.97) and significant multiple correlation coefficients were obtained between each vowel’s 100-Hz section noise levels from 100 to 2600 Hz and its median roughness rating.

Significance of Intra- and Intersentence Pause Times in Perceptual Judgments of Oral Reading Rate

The purpose of this investigation was to determine if alterations in intra- and intersentence pause times in an individual's recorded reading of a standard prose passage (with no changes made in speech time) would result in alterations in perceptual judgments of his oral reading rate. Intra- and intersentence pause times were manipulated in one of the following ways: no change; 25% increase; 25% decrease; 50% increase; 50% decrease; 75% increase; 75% decrease. Electronic reproductions of the original tape recording were made; the removal or addition of pause time was accomplished in all cases by the removal or addition of the appropriate percentages of actual tape at each pause point. The original and pause-altered recordings were arranged in random order on a master tape and played to a group of 78 judges for rate evaluations using an equal-appearing intervals scale. Results of listener evaluations indicate that pause time is an important variable in perceptual judgments of oral reading rate. [Research supported by the National Institute of Neurological Diseases and Blindness.]

Anchoring and sequence effects for the category scaling of stuttering severity.

Equal-appearing interval scaling is frequently used to measure stuttering severity. The scale values generated by this procedure are known to be affected by a variety of extraneous conditions. In particular, anchoring and sequence effects are known to result in predictable observer biases. In this paper, these influences are demonstrated for the category scaling of stuttering severity, and suggestions are made for their experimental control.

Construction of Sportsmanship Attitude Scales

Abstract Alternate form sportsmanship attitude scales were developed using the scale-discrimination technique. The initial set of 152 items were descriptions of ethically critical sportsmanship behavior. The initial item set was successively treated with the equal-appearing interval, summated rating, and scale analysis methods for developing attitude scales. The 42 items selected for the final scale form were divided into separate units in order to have alternate scale forms. The two forms were evaluated for reliability, reproducibility, and empirical validity. A correlation coefficient of .86 was found between scores of Form A and Form B for the single test administration. For Form A the coefficient of reproducibility was found to be .81, and a reproducibility value of .86 was found for Form B. Empirical validity coefficients ranging from —.01 to .43 were found between the test scores and behavior ratings.

The relationship between the affective, behavioral, and cognitive components of attitude

It has long been assumed that attitudes have affective, behavioral, and cognitive components. Two hypotheses were derived from this assumption and tested in three correlational studies. Individuals were predicted to show greater consistency of response to attitude scales measuring the same component than to scales measuring different components. The Campbell and Fiske (1959) multitrait-multimethod matrix procedure was used to test this hypothesis. Second, it was hypothesized that the correspondence between verbal attitude scales and nonverbal attitudinal responses should be highest when both are drawn from the same attitude component. Indices of overt behavior were compared with verbal measures of the affective, behavioral, and cognitive components as a test for the second hypothesis. Construction of verbal measures for the three components required development of a procedure for assessing the amount each verbal statement reflected each component. Scales of attitude toward the church were prepared using the methods of equal-appearing intervals, summated ratings, scalogram analysis, and self-rating. Both hypotheses were supported, but the dominant feature was a high intercorrelation between the three components with the uniqueness of each component contributing very little additional variance.

Some Factors Affecting the Size of Q Values for Speech Ratings

Mean Q values for ratings obtained using seven-point equal-appearing interval scales for various speech dimensions were gathered from the speech pathology literature. An inspection of these data suggests that mean Q values may differ with speech dimension being rated, type of speech stimuli, anchoring procedure, and distribution of average scale values for the stimuli. Interpretation of mean Q value is discussed.

Reliability of 'live' ratings of the speech of stutterers.

On two occasions, separated by 4 mo., 8 college-age stutterers gave a short spontaneous speech to and conversed with a group of judges, each of whom rated each S on each of 7 dimensions of speech using the equal-appearing intervals scaling procedure. Ratings 3 mo. later were obtained from the same judges using audio-recordings of the second ‘live’ session. Intraclass correlation procedures were used to estimate the reliability of the ratings. For 4 dimensions of dysfluent speech the mean scale values for all three rating sessions were highly reliable and equally reliable ratings were obtained from the ‘live’ and audio-taped samples. For 3 dimensions of non-stuttered speech, the ratings were not satisfactorily reliable.

Intraclass correlation coefficient as an index of reliability of median scale values for sets of stimuli rated by equal-appearing intervals.

Intraclass correlation coefficient as an index of reliability of median scale values for sets of stimuli rated by equal-appearing intervals.

Equal-appearing interval scales, own-attitude, and experimental instructions: note.

Equal-appearing interval attitude scales were derived from groups that differed in terms of own-attitude position and the experimental instructions under which Ss completed the scaling task. The results indicated that the magnitude of the linear association between the scales was influenced by the instructions presented to the judges.

Correspondence between Mean and Median Scale Values for Sets of Stimuli Scaled by the Method of Equal-Appearing Intervals

Mean scale values were computed for each of five sets of stimuli for which median scale values and the distributions of judges' ratings had been reported. The correlations (Pearson rs) between the mean and the median scale values for each of the five sets of stimuli range from .995 to .999. These high correlations indicate that mean and median scale values rank order sets of stimuli in approximately the same manner. Several implications of this finding are discussed.

Equal-appearing interval scale values and successive interval scale values derived from the same set of ratings.

Both equal-appearing interval scale values and successive interval scale values were computed for each of five sets of ratings on five different kinds of stimuli. The scale values obtained by these two methods rank ordered almost identically the stimuli which were scaled. For all five sets of ratings, the Pearson r for estimating the relationship between the two sets of scale values was equal to or greater than .99. In view of the close relationship between the scale values derived by the two methods, practical considerations would make the method of equal-appearing intervals the preferable one for many purposes.

Ratio-Estimations of Loudness-Intervals

Two loudness-scales have been published: (a) the sone scale,1 and (b) the lambda scale.2 These two scales, based on different scaling procedures, are quite divers. The sone scale, based on direct estimates of the apparent ratios of loudnesses, is a power function of sound-intensity with an exponent of 0.3. The lambda scale, based on both estimates of equalappearing intervals and the apparent ratios of loudnesses, is much less steep and may be approximated by either a logarithmic function or a power function with an exponent of 0.13. Though the lambda scale, founded on both equisectional and fractionational judgments, is a hybrid scale, the form of the scale based primarily on equisectional data is in general agreement with both discriminability and interval scales.3 The present experiment deals with the adequacy of the sone and the lambda scales to predict ratio-estimations of loudness-intervals. Ordinarily, the apparent magnitude of loudness-ratios is studied by ratio-scaling methods, and the apparent magnitude of loudness-intervals by interval scaling-methods.4 That is, the scalings of ratios and of intervals are coupled to two different scaling tasks: that of making ratio-estimations and that of partitioning a segment into equal intervals. By using a ratio-method to study the apparent magnitude of loudness-intervals, the present experiment unties the property scaled from the method of scaling.

Length of Reference Line in Similarity Ratings

Many scaling techniques involve rating procedures in which Ss are asked to place a stimulus on some kind of scale. In a typical task, S is presented with stimuli one at a time and instructed to place each stimulus on a scale which varies from, say, 1 to 7, the former being assigned when the stimulus has the least amount of the attribute being rated and the latter when the stimulus has a great deal of the attribute. Such a scaling technique is actually a variant of the method of equal-appearing intervals. Edwards (1957, Chap. 4) indicates that, in general, the procedure employed in equal-appearing interval scaling tasks appears to have little effect. The research he cites, however, is largely concerned with comparing two rather dissimilar techniques [e.g., sorting (Thurstone & Chave, 1929) and rating scale line (Ballin & Farnsworth, 1941)l. In the present study, ratings by reference to a line were employed and the length of that line was varied. Forty-four male undergraduates were divided randomly into two groups (A and B). Group A rated the similarity of pairs of adjectives on a 7-in. line, Group B on a 3y2-in. line. In each case, the line had seven equal divisions, and the extremes were defined by the phrases very similar and very different. On the basis of independent data, three pairs of adjectives (sad-impetuous, satisfyingdelicate, merry-happy) were selected as being low, medium, and high, respectively, in similarity. The mean racings for each pair of stimuli were calculated from the ratings given by Groups A and B. The low-similarity adjective pair led to a mean rating of 2.68 for each group (z = 0.00). The standard deviations were 1.33 and 1.04 for Groups A and B, respectively. For the medium-similarity pair, the means were 3.64 (s = 1.24) and 3.59 (s = 0.88) for Groups A and B (z = 0.15). In the case of the high-similarity adjective pair, the means were 6.59 and 6.73 (z = -0.77) with standard deviations 0.68 and 0.52. It is clear thar there are no significant differences in the mean similarity ratings; all z values fall well within the critical bounds of &1.96. Thus varying the length of the rating scale line from 345 to 7 in. appears unimportant in similarity ratlngs. The line lengths evaluated are approximately the upper and lower bounds of the scale slzes conveniently used in paperand-pencil rating. The suggestion in the data thar scale length may have an effect on high similarity stimuli but not on low was also tested, but an analysis of variance showed no interaction between similarity and rating-scale length.

Measurement of the Attitudes of Teachers Toward “Teaching as a Profession”

AbstractThis study concerns the initial development of a scale to measure teachers' attitudes toward teaching as a profession.A modification of the W-technique, a combination of the equal-appearing interval and paired-comparison methods, was utilized to construct the scales. Teachers were used to judge the statements in the preliminary stages of scaling. An alternate scale was constructed for correlational purposes and later a revised scale was also developed from the same statements.An estimate of test-retest reliability was obtained. In one college class a correlation coefficient of .92 was obtained between the original and alternate scales. In another class, the test-retest coefficient for the original scale was .99 and .97 between the alternate and original scales.The high correlation gives evidence for the reliability of the original scale and both scales appear ready for further research purposes.

Intimacy-scaled stimuli for use in studies of interpersonal relations.

Psychometric materials for use in research on interpersonal exchange and self-disclosure included 671 statements about various aspects of the self which people might discuss as they form interpersonal relationships with others. Items were scaled for “intimacy” and topical category by judges from two different populations. From equal-appearing interval analysis were obtained intimacy scale values and Q-scores. Pearson product-moment correlations showed high agreement between the two populations. The validation of a priori category nomenclature indicated that placements of 497 out of 671 statements were agreed upon by at least 8 of 16 judges. An additional 20 statements were reliably categorized by judges in other than the originally assigned topic areas. Thus, there was reliable agreement among judges as to statement topic for 77% of the items. Advantages and suggested usages of the item pool were discussed.

STABILITY OF THURSTONE SCALE VALUES AFTER 35 YEARS

Do attitude scales mean the same when used at different times and with different groups? If so, no difference should occur in scale values developed with the equal-appearing interval method among different generations and over several decades. Some indirect evidence is available regarding the constancy of one such instrument, viz., a scale of 24 items focusing on the social issue of capital punishment, using an 11point continuum of favorableness-unfavorableness.' In 1966, 49 upper undergraduate males in a class exercise in a southeastern technical universicy served as judges for deriving median scale values for the 24 statements. Thurstone's ( 6 ) scale values for the same items were generated by graduate srudents at the University of Chicago. Correlation of the values created in 1930 in Chicago and in 1966 in Atlanta by students, judging the favorableness of the statements toward the concept in question, yielded a value of 9 8 . Ferguson ( 2 ) reports reliabilities for these statements to range from .59 to .88; .44 was the split-half estimate for the present data. Apparently, the general favorableness toward capital punishment in 1966 does not differ appreciably from the views of judges of 1930 vintage. The unusually high relationship berween the scale values reflects the stability over time and among groups, at least for individuals in academic settings. This generalization applies to several kinds of attitudes. Hovland and Sherif ( 3 ) , comparing items on the concept of Negro scaled cwo decades apart, report an r of .96. And, Milholland ( 4 ) , though reducing the judges' hypothetical continuum to 9 piles or steps for attitudes toward the law, found product-moment rs of .98 between values on a 1930 scale developed by Katz ( 6 ) and a 37-member graduate class in 1956. Do these reports of comparable data imply no cultural change regarding selected social issues? Might not Ferguson's ( 2 ) suggestion that we compare group mean scores be more appropriate than just reviewing scale values established through the quasi-rational judging process?

THE EFFECT OF VARIABLE PERSPECTIVES ON JUDGMENTS OF OPINION STATEMENTS FOR THURSTONE SCALES: EQUAL-APPEARING INTERVALS.

THE EFFECT OF VARIABLE PERSPECTIVES ON JUDGMENTS OF OPINION STATEMENTS FOR THURSTONE SCALES: EQUAL-APPEARING INTERVALS.

COMMENT ON UPSHAW'S "OWN ATTITUDE AS AN ANCHOR IN EQUAL-APPEARING INTERVALS".

COMMENT ON UPSHAW'S "OWN ATTITUDE AS AN ANCHOR IN EQUAL-APPEARING INTERVALS".

Own attitude as an anchor in equal-appearing intervals.

Own attitude as an anchor in equal-appearing intervals.