Individual Differences Multidimensional Scaling Research Articles

Multidimensional analysis of textiles coated with electroactive polymers for actuators

This paper presents a multidimensional analysis of textiles coated with conductive polymers for actuators. The purpose of using multidimensional scaling analysis is to compare similarities and dissimilarities between conductive textiles obtained through conductive polymeric film deposition to observe the optimal value for electrical resistance and to select an adequate method to achieve conductive fabric using different polymers (polyethylene glycol, polyvinyl alcohol or polyvinylidene fluoride) and metal microparticles (copper or nickel). The multidimensional analysis is based on mapping a series of material properties from a proximity matrix (similarities or dissimilarities) between these properties. Multidimensional scaling allows rebuilding the exact map of the values (within approximately a symmetry or rotation). To fit dissimilarity or similarity matrices for multiple variables into one common space estimating the weight parameters for each variable, the INDSCAL model (individual differences multidimensional scaling) was used.

Multivariate fMRI responses in superior temporal cortex predict visual contributions to, and individual differences in, the intelligibility of noisy speech

Humans have the unique ability to decode the rapid stream of language elements that constitute speech, even when it is contaminated by noise. Two reliable observations about noisy speech perception are that seeing the face of the talker improves intelligibility and the existence of individual differences in the ability to perceive noisy speech. We introduce a multivariate BOLD fMRI measure that explains both observations. In two independent fMRI studies, clear and noisy speech was presented in visual, auditory and audiovisual formats to thirty-seven participants who rated intelligibility. An event-related design was used to sort noisy speech trials by their intelligibility. Individual-differences multidimensional scaling was applied to fMRI response patterns in superior temporal cortex and the dissimilarity between responses to clear speech and noisy (but intelligible) speech was measured. Neural dissimilarity was less for audiovisual speech than auditory-only speech, corresponding to the greater intelligibility of noisy audiovisual speech. Dissimilarity was less in participants with better noisy speech perception, corresponding to individual differences. These relationships held for both single word and entire sentence stimuli, suggesting that they were driven by intelligibility rather than the specific stimuli tested. A neural measure of perceptual intelligibility may aid in the development of strategies for helping those with impaired speech perception.

Bias in Terms of Culture and a Method for Reducing It

Several sources of bias can plague research data and individual assessment. When cultural groups are considered, across or even within countries, it is essential that the constructs assessed and evaluated are as free as possible from any source of bias and specifically from bias caused due to culturally specific characteristics. Employing the Explanations of Unemployment Scale (revised form) for a sample of 1,894 employed and unemployed adults across eight countries (the United States, the United Kingdom, Turkey, Spain, Romania, Poland, Greece, and Brazil), we applied a method based on individual differences multidimensional scaling and principal component analysis to detect item bias in terms of culture and try to eliminate this bias variance from the overall item variance so as to (a) avoid jeopardizing validity levels and (b) arrive at clearer and more meaningful dimensions after adjusting the raw scores by removing the bias part. The results supported our statistical–psychometric intervention as the structure computed for the unadjusted data was enhanced and clarified when the data were adjusted for bias in terms of culture. Finally, implications for individual assessment procedures are discussed, and a method for evaluating the relative impact of bias in terms of culture on the raw assessment scores is also presented.

Processing bimodal stimuli: integrality/separability of color and orientation

We examined how two distinct stimulus features, orientation and color, interact as contributions to global stimulus dissimilarity. Five subjects rated dissimilarity between pairs of bars (N = 30) varying in color (four cardinal hues, plus white) and orientation (six angles at 30° intervals). An exploratory analysis with individual-differences multidimensional scaling (MDS) resulted in a 5D solution, with two dimensions required to accommodate the circular sequence of the angular attribute, and red-green, blue-yellow and achromatic axes for the color attribute. Weights of the orientation subspace relative to the color subspace varied among the subjects, from a 0.32:0.61 ratio to 0.53:0.44, emphasis shifting between color and orientation. In addition to Euclidean metric, we modeled the interaction of color and orientation using Minkowski power metrics across a range of Minkowski exponents p, including the city-block (p = 1), Euclidean (p = 2) and Dominance metric (p → ∞) as special cases. For averaged data, p ~ 1.3 provided the best fit, i.e., intermediate between separable and integral features. For individual subjects, however, the metric exponent varied significantly from p = 0.7 to p = 3.1, indicating a subject-specific rule for combining color and orientation, as in Tversky and Gati's variable-weights model. No relationship was apparent between dimensional weights and individual p exponents. Factors affecting dimensional integrality are discussed, including possible underlying neural mechanisms where the interaction of the low-level vision attributes orientation and color might shift between uncorrelated (p = 1) or correlated (p ≥ 2) forms.

Decoding the neural representation of affective states

Brain activity was monitored while participants viewed picture sets that reflected high or low levels of arousal and positive, neutral, or negative valence. Pictures within a set were presented rapidly in an incidental viewing task while fMRI data were collected. The primary purpose of the study was to determine if multi-voxel pattern analysis could be used within and between participants to predict valence, arousal and combined affective states elicited by pictures based on distributed patterns of whole brain activity. A secondary purpose was to determine if distributed patterns of whole brain activity can be used to derive a lower dimensional representation of affective states consistent with behavioral data. Results demonstrated above chance prediction of valence, arousal and affective states that was robust across a wide range of number of voxels used in prediction. Additionally, individual differences multidimensional scaling based on fMRI data clearly separated valence and arousal levels and was consistent with a circumplex model of affective states.

Constructing Personality Maps, Mapping Personality Constructs: Multidimensional Scaling Recovers the Big Five Factors from Internal and External Structure

This report examines the structure of similarities underlying the lexicon of personality-trait description, when "similarity" is defined and measured in terms of (a) semantic judgment and (b) covariance in actual use. A lexicon of 60 trait adjectives was examined, using several procedures for collecting semantic judgments. Similarity data of both kinds were analyzed with multidimensional scaling (MDS) to provide a parsimonious representation of underlying structure. The convergence between semantic judgments and covariance within trait-attribution data was substantial; both kinds of data evinced the same structure when collected for subsets of adjectives. Canonical correlation was employed to find the number of dimensions shared across MDS solutions. Interpretation of the results was facilitated by individual-differences MDS, which can select an optimal set of underlying dimensions, and at the same time accommodate the differences between data sets that arise when data-collection procedures differ in the relative emphasis they place upon those dimensions. We interpret the small number and shared nature of the dimensions by arguing that the lexicon's structure relates to trait perception rather than personality structure per se, even when probed with trait-attribution covariance.

Color naming and categorization in inherited color vision deficiencies

Dichromatic subjects can name colors accurately, even though they cannot discriminate among red-green hues (Jameson & Hurvich, 1978). This result is attributed to a normative language system that dichromatic observers developed by learning subtle visual cues to compensate for their impoverished color system. The present study used multidimensional scaling techniques to compare color categorization spaces of color-vision deficient (CVD) subjects to those of normal trichromat (NT) subjects, and consensus analysis estimated the normative effect of language on categorization. Subjects sorted 140 Munsell color samples in three different ways: a free sorting task (unlimited number of categories), a constrained sorting task (number of categories limited to eight), and a constrained naming task (limited to eight basic color terms). CVD color categories were comparable to those of NT subjects. For both CVD and NT subjects, a common color categorization space derived from the three tasks was well described by a three-dimensional model, with the first two dimensions corresponding to reddish-greenish and yellowish-bluish axes. However, the third axis, which was associated with an achromatic dimension in NTs, was not identified in the CVD model. Individual differences multidimensional scaling failed to reveal group differences in the sorting tasks. In contrast, the personal color naming spaces of CVD subjects exhibited a relative compression of the yellowish-bluish dimension that is inconsistent with the typical deutan-type color spaces derived from more direct measures of perceptual color judgments. As expected, the highest consensus among CVDs (77%) and NTs (82%) occurred in the naming task. The categorization behaviors studied in this experiment seemed to rely more on learning factors, and may reveal little about CVD perceptual representation of colors.

ConPar: a method for identifying groups of concordant subject proximity matrices for subsequent multidimensional scaling analyses

A common representation of data within the context of multidimensional scaling (MDS) is a collection of symmetric proximity (similarity or dissimilarity) matrices for each of M subjects. There are a number of possible alternatives for analyzing these data, which include: (a) conducting an MDS analysis on a single matrix obtained by pooling (averaging) the M subject matrices, (b) fitting a separate MDS structure for each of the M matrices, or (c) employing an individual differences MDS model. We discuss each of these approaches, and subsequently propose a straightforward new method (CONcordance PARtitioning—ConPar), which can be used to identify groups of individual-subject matrices with concordant proximity structures. This method collapses the three-way data into a subject×subject dissimilarity matrix, which is subsequently clustered using a branch-and-bound algorithm that minimizes partition diameter. Extensive Monte Carlo testing revealed that, when compared to K-means clustering of the proximity data, ConPar generally provided better recovery of the true subject cluster memberships. A demonstration using empirical three-way data is also provided to illustrate the efficacy of the proposed method.

A random effects individual difference multidimensional scaling model

Recent algorithms for nonlinear mixed effects models can also be used, after some modification, to fit individual differences multidimensional scaling models in which the subject weights are random effects. The models we propose here have several advantages over models which do not use random effects. For example, unlike traditional models, the number of parameters does not increase with the number of subjects, and, because the distribution of the subject weights is modeled, generalization of results to the sampled population of subjects is immediate. These models also have some disadvantages. Here, we discuss the proposed random effects model, give a computational algorithm for fitting the model, describe our experiences with this algorithm, and discuss potential generalizations to other multidimensional scaling models.

Color-vision variations represented in an individual-difference vector chart

The aim was to reconstruct a uniform color space that could accommodate normal and abnormal variations in color vision. Color-dissimilarity data were collected from normal trichromats and red-green deficients, using lights varying in spectral composition and luminance. The data were analyzed with an individual-differences multidimensional scaling program and resulted in a 3D group configuration. In addition, differences in saturation required a hemispherical representation for the “color plane,” with a 4th dimension obtained when a constraint of sphericity was included in the algorithm. Subject-specific color spaces were derived from this group configuration by axial compression, preceded by a rotation of the chromatic plane. In the spaces of color deficients, compression along a red-green axis occurred, though the direction of the compression differed for protans and deutans. The extent of the compression reflected the severity of abnormality. Transformations of the subject-specific color spaces — compression extent and the angle of rotation — were portrayed in a polar-coordinate chart. The chart separates normal trichromats from the red-green deficients, distinguishes protans from deutans, and quantifies the deficiency within each class. © 2000 John Wiley & Sons, Inc. Col Res Appl, 26, S230–S234, 2001

Research note: a multidimensional scaling comparison of color metrics for response times and rated dissimilarities.

Individual-differences multidimensional scaling was applied to a set of proximity data for equiluminant lights (Paramei & Cavonius, 1999) to explore any differences between two data collection procedures (rated dissimilarities, and same/different response times [RTs]), as well as between color-normal and abnormal observers. Two conclusions emerged: (1) The pattern of similarities from observers with anomalous color vision can be understood in terms of a compressed color plane (the weighted Euclidean model of individual differences); and (2) there is evidence that the color "plane" is either curved or governed by a non-Euclidean distance function. When color-normal observers are examined in the weighted-Euclidean framework, minor differences emerge between RT and rating data. But the main distinguishing feature of RT data is a pattern of decreasing accuracy for larger color distances. This must be taken into account, since it can itself induce curvature.

Qualitative Hemispheric Differences in Semantic Category Matching

A unilateral category matching task with words as stimuli was employed to investigate semantic processing in the right and left hemispheres (RH, LH). An overall right visual field (RVF)/LH dominance was observed and performances were better than chance, also in the left visual field (LVF)/RH. A qualitative analysis of reaction times with individual differences multidimensional scaling (INDSCAL) revealed that LVF/RH INDSCAL solutions were significantly more differentiated in structure than RVF/LH solutions in terms of number and size of dimensions. These findings support a depth of activation hypothesis of hemispheric processing, with the LH rapidly and focally and the RH slowly and diffusely activating the semantic network.

The perceptual structure of pathologic voice quality.

Although perceptual assessment is included in most protocols for evaluating pathologic voices, a standard set of valid scales for measuring voice quality has never been established. Standardization is important for theory and for clinical acceptance, and also because validation of objective measures of voice depends on valid perceptual measures. The present study used large sets (n = 80) of male and female voices, representing a broad range of diagnoses and vocal severities. Eight experts judged the dissimilarity of each pair of voices, and responses were analyzed using nonmetric individual differences multidimensional scaling. Results indicate that differences between listeners in perceptual strategy are so great that the fundamental assumption of a common perceptual space must be questioned. Because standardization depends on the assumption that listeners are similar, it is concluded that efforts to standardize perceptual labels for voice quality are unlikely to succeed. However, analysis by synthesis may provide an alternate means of modeling quality as a function of both voices and listeners, thus avoiding this problem.

Structure of health risk behavior among high school students.

The authors test the contention of R. Jessor's (1977) problem behavior theory that adolescent health risk behaviors comprise a single behavioral syndrome. Multidimensional scaling (MDS) and cluster analysis are used to analyze data from a statewide survey of high school students' (n = 5,537) health risk behaviors. A classical MDS analysis was calculated to test the dimensionality of the behaviors. All indicators supported a multidimensional model. An individual-differences MDS (INDSCAL) analysis revealed that a 4-dimensional solution best fit the data, with gender and racial-ethnic differences emerging in the relative salience of the dimensions. The INDSCAL dimensional coordinates for each health risk behavior were then submitted to a hierarchical cluster analysis technique. Five behavioral clusters were identified, 1 of which included many of the traditional "problem behaviors," such as smoking, unprotected sexual intercourse, and alcohol consumption. These findings support a multidimensional structure underlying adolescent health risk behavior. Implications for theory and prevention of health risk behaviors are discussed.

Scaling the affective domain of pain: a study of the dimensionality of verbal descriptors

This study evaluated the multidimensional structure of affective verbal descriptors and investigated individual differences in the scaling of the descriptors. Patients with chronic low back pain, chronic headache and rheumatoid arthritis (25 per group) and 25 control subjects, matched for age and sex, made similarity judgements of a set of 12 verbal descriptors, e.g., awful, miserable. They also completed the Coping Strategies Questionnaire, the Hospital Anxiety and Depression Scale, McGill Pain Questionnaire Short Form, and a measure of verbal intelligence. The similarity data were subjected to individual differences multidimensional scaling (ALSCAL) and the resulting subject weights were related to group membership and the questionnaire data. The multidimensional nature of affective descriptors was confirmed. A 4-dimensional solution was identified: the first 3 dimensions were comparable with previous findings and labeled Tolerability, Focus of Attention and Minor Emotional Reaction. The 4th dimension was enigmatic and attributable to the back pain group. There were significant differences between the groups in their weighting of the dimensions and in their self-reported coping strategies. The results are discussed with reference to an earlier study and the degree of consistency across the studies is noted. Implications of the results for the conceptualisation and measurement of the affective domain of pain report are outlined.

A TASK TYPE FOR MEASURING THE REPRESENTATIONAL COMPONENT OF QUANTITATIVE PROFICIENCY

ABSTRACTTwo computer‐based categorization tasks were developed and pilot tested. For Study I, the task asked examinees to sort mathematical word problem stems according to prototypes. Results showed that those who sorted well tended to have higher GRE General Test scores and college grades than did examinees who sorted less proficiently. Examinees generally preferred this task to multiple‐choice items like those found on General Test quantitative section and felt the task was a fairer measure of their ability to succeed in graduate school. For study II, the task involved rating the similarity of item pairs. Both mathematics test developers and students participated, with the results analyzed by individual differences multidimensional scaling. Experts produced more scaleable ratings overall and primarily attended to two dimensions. Students used the same two dimensions with the addition of a third. Students who rated like the experts in terms of the dimensions used tended to have higher admissions test scores than those who used other criteria. Finally, examinees preferred multiple‐choice questions to the rating task and felt that the former was a fairer indicator of their scholastic abilities. The major implication of this work is in identifying a new task type for admissions tests, as well as for instructional assessment products that might help lower scoring examinees localize and remediate problem‐solving difficulties.

Modeling pathologic vocal quality. Part I

Previous studies of pathologic vocal quality have not yielded a standard set of scales for describing disordered voices, possibly due to the limited number of voices studied. The present study used large sets (n=80) of male and female voices, representing a broad range of diagnoses and vocal severities. Eight experts judged the dissimilarity of each pair of voices, and responses were analyzed using nonmetric individual differences multidimensional scaling. Results indicate that differences between listeners in perceptual strategy are so great that the fundamental assumption of a common perceptual space must be questioned. Group spaces revealed a single ‘‘feature’’ (severity of pathology). Spaces for individual listeners were characterized by clusters of voices which seemed structured by ‘‘family resemblances’’: Each voice in a cluster resembled others with rspect to some property, but the same property did not constitute the resemblance for all voice pairs. Further, voices did not appear consistently in the same clusters across listeners. In the absence of a common perceptual space—of a shared sense of how voices sound similar—valid perceptual scales cannot be defined. The results lead to the conclusion that standardization of perceptual labels for voice quality is theoretically impossible.

Three dimensions mental health professionals emphasize when judging levels of functioning of the severely mentally disabled in the community

Ratings of the level of functioning of severely disabled clients collected from staff of mental health agencies, while attractive as outcome measures, lack specificity and, at times, consistency. Three multidimensional scaling (MDS) studies were conducted to identify the dimensions staff emphasize when they compare clients as to their level of functioning. Analogues of clients were created to serve as stimuli, dissimilarity judgments of pairs of stimuli were collected, and an individual differences MDS model was applied in each study to distinguish client variance from the judges' perspectives. Ratings along specified dimensions also were collected and compared to the MDS solutions. One dimension emerged as pre-eminent, but it corresponded to many of the pre-selected dimensions, such as competence with daily living tasks, clarity of thinking, appearance and manner, and overall functioning. The other two dimensions were belligerence and openness to psychological intervention or distress with one's problems. Differences among judges were sizable; nevertheless, considerable support for emphasizing these three dimensions was found upon reviewing the contents of other assessments of functioning. Concentrating on these three dimensions of functioning may promote greater agreement among raters, highlight where functional improvements occur, and expand the coverage of functional differences without markedly increasing the costs of assessment.

Multidimensional scaling of painful electrocutaneous stimulation: INDSCAL dimensions, signal detection theory indices, and the McGill Pain Questionnaire.

Individual-differences multidimensional scaling (INDSCAL) determined the dimensions underlying ratings of electrocutaneous stimuli, which ranged from innocuous levels to individual pain intolerance at each of three frequencies. Twenty-five healthy males made pairwise similarity judgments of these 15 stimuli for the INDSCAL procedure, and then rated each stimulus on nine property scales. Signal detection theory indices, as well as ratings on the McGill Pain Questionnaire (MPQ), were also obtained. A Sensory Magnitude dimension scaled the stimuli from lowest to highest perceived intensity; this dimension was related to sensory, affective, and arousal property scales. A Frequency dimension ordered the stimuli from lowest to highest frequency; this dimension was related to the Fast-Slow property. Compared to the Frequency dimension, the Sensory Magnitude dimension was more salient to subjects who better discriminated among painful stimulus intensities, set a more stoical pain report criterion, and were less apt to endorse frequency-related MPQ descriptors. Thus, variation of physical intensity and frequency elicited complementary dimensions of subjective judgment, which were related to perceptual and attitudinal differences among individuals.

Schemata and smoking relapse

Cognitive schemata were examined in a prospective study of smoking relapse. 100 recent ex-smokers rated the similarity of 44 high-risk situations according to how difficult it would be to avoid smoking in them, and these judgments were analyzed via individual differences multidimensional scaling. As a group, subjects seemed to use two dimensions when judging the similarity of high-risk situations. One, interpreted as a Negative Affect dimension, reflected the degree to which situations are either devoid of positive emotion or charged with negative emotion. The other, interpreted as a Social Factors dimension, reflected the extent to which situations depict social pressure versus relatively solitary activities. Individual differences in emphasis on these group dimensions were significantly associated with subsequent relapse. Similarity judgments of Lapsers had reflected greater emphasis on the Negative Affect dimension and lower emphasis on the Social Factors dimension than did the judgments of 12-month Continuous Abstainers. Results are consistent with prior research linking negative affect smoking to relapse. The long-term predictive validity of the schema measure lends support to the view that scaling procedures are useful for assessing relatively stable aspects of cognitive structure.