The Use of Multitrait–Multimethod Data for Detecting Nonlinear Relationships: The Case of Psychological Climate and Job Satisfaction

Previous monomethod research has found mixed support for nonlinear effects between certain job characteristics (e.g., job autonomy, job complexity) and outcome variables (e.g., job performance). We hypothesized that these weak nonlinear findings may be due to the prevalence of monomethod research that can lead to a lack of complete measurement and/or introduce common methods variance, either of which may mask the true shape of relationships. Using hierarchical regression analyses and a multitrait–multimethod (MTMM) research design, we found strikingly different results between monomethod and MTMM data when considering the relationships between three psychological climate variables and job satisfaction. While the monomethod results mirrored earlier inconclusive findings, the MTMM data indicated that nonlinear equations explained significantly more of the relationship between all three climate dimensions and job satisfaction. These results suggest that the use of MTMM data allows one to more effectively test for nonlinear effects. Furthermore, these nonlinear results suggest that the format of employee questionnaires should probably change from asking how much an employee has of certain constructs to asking does the employee want more or less of these constructs.     

Improving the metric quality of questionnaire data

A procedure is proposed whereby questionnaire data, which is usually ordinal in nature and often error-ridden, may be transformed to reduce the error variance in the data and to improve the metric properties of the individual variables. The technique is suggested by a result of Eckart and Young. The properties of the method are investigated by means of a Monte Carlo study. Various matrices were generated representing the usual concept of true scores. These matrices were distorted using two levels of random errors and two kinds of categorization. The distorted matrices were in turn transformed by the proposed methods and compared to the true scores. In all cases an overall measure of similarity reveals the transformed matrices are better approximations to the true scores than the untransformed data. Some properties of the transformation are discussed and some possible applications of the general technique are suggested.     

Generalizability of scores influenced by multiple sources of variance

Generalizability theory concerns the adequacy with which a universe score can be inferred from a set of observations. In this paper the theory is applied to a universe in which observations are classifiable according to two independent variable aspects of the measuring procedure. Several types of universe scores are developed and the variance components ascertained for each type. The composition of expected observed-score variance and the adequacy of inference to a particular type of universe score is a function of the procedure used in gathering data. A generalizability study provides estimates of variance components which can be used in designing an efficient procedure for a particular decision purpose.This study was conducted under Grant M-1839 from the National Institute of Mental Health while the authors were on the staff of the University of Illinois. Dr. Rajaratnam shared responsibility for the technical report of July, 1961 on which this paper is based. The present revision was made subsequent to her death in 1963. The present addresses of the other authors are: Goldine C. Gleser, Department of Psychiatry, Central Clinic, Cincinatti, 29, Ohio; Lee J. Cronbach, School of Education, Stanford University.     

Correcting for unreliability in nonlinear models of attitude change

The paper opens with a brief discussion of the problems in testing nonlinear models of attitude change. The regression artifacts produced by unreliability are shown in both the linear and nonlinear case. Classical solutions for the linear case are quickly reviewed. A new solution to the linear case is presented and applied to the nonlinear case. It is shown to work well under a broad set of conditions. Regression artifacts in bivariate regression are then discussed. If the predictors are independent, then the univariate correction procedure can be applied to each predictor separately. But if the predictors are correlated, a joint correction procedure must be used. One such procedure is defined and shown to work perfectly in the case of linear regression and reasonably well in a broad set of conditions in which the regression is nonlinear.     

Varisim: A new machine method for orthogonal rotation

The derivative of Kaiser's Varimax criterion, if set to zero, yields a set of equations which are quite similar to those obtained for a least-squares problem of the Procrustes type. This similarity suggested an iterative technique for orthogonal rotation, dubbed Varisim, which was programmed for the IBM 7094 in FORTRAN. An empirical comparison between Varimax and Varisim, which was based on a number of data sets taken from the literature yielded three major results so far: (i) Varisim is slower than Varimax, roughly by a factor of 3, (ii) Varisim yields factors which in general contribute more evenly to the common test variance than Varimax factors, and which (iii) line up more closely with oblique configurations obtained with Binormamin than Varimax factors.This paper is based on parts of a thesis submitted to the Graduate College of the University of Illinois in partial fulfillment of the requirements for a degree in Psychology. Some of its contents were read at the joint meeting of the Psychonomic and the Psychometric Society, Niagara Falls, Ontario, 1964. Most of the computer-bound work was carried out while the author was employed by the Statistical Service Unit, University of Illinois; it was written up while the author held a Postdoctoral Fellowship at the Psychometric Laboratory of the University of North Carolina. Although space limitations forbid more complete acknowledgments, I am bound to record my deep gratitude to my advisor, Prof. H. F. Kaiser.     

Cognitive complexity and the similarity-attraction paradigm

The study investigated the role of cognitive complexity as a moderator of the similarity-attraction relationship described by Byrne (1971). Subjects conducted face-to-face interviews with confederates who played roles as job applicants. Similarity-dissimilarity was manipulated by the confederate roles, as well as by the information given the subjects in accordance with the procedures described by Byrne (1971). The situation was viewed as one in which judgments were made about a complex, multidimensional stimulus. Information received about another person in an interview setting was viewed as consisting of different dimensions to be assessed by the perceiver. The hypothesis was tested and confirmed that cognitively complex judges were more likely than simple judges to perceive and evaluate similarity/dissimilarity in others. Thus, complex judges evaluated similar applicants more positively than dissimilar applicants, while no differences were found for simple judges. Cognitive complexity thus appears to moderate the “Law of Attraction” described by Byrne (1971).     

A distance-based variety of nonlinear multivariate data analysis, including weights for objects and variables

In the distance approach to nonlinear multivariate data analysis the focus is on the optimal representation of the relationships between the objects in the analysis. In this paper two methods are presented for including weights in distance-based nonlinear multivariate data analysis. In the first method, weights are assigned to the objects while the second method is concerned with differential weighting of groups of variables. When each analysis variable defines a group the latter method becomes a variable weighting method. For objects the weights are assumed to be given; for groups of variables they may be given, or estimated. These weighting schemes can also be combined and have several important applications. For example, they make it possible to perform efficient analyses of large data sets, to use the distance-based variety of nonlinear multivariate data analysis as an addition to loglinear analysis of multiway contingency tables, and to do stability studies of the solutions by applying the bootstrap on the objects or the variables in the analysis. These and other applications are discussed, and an efficient algorithm is proposed to minimize the corresponding loss function.This study is funded by The Netherlands Organization for Scientific Research (NWO) by grant nr. 030-56403 for the PIONEER project Subject Oriented Multivariate Analysis to the third author.     

Consistent Small-Sample Variances for Six Gamma-Family Measures of Ordinal Association

Gamma-family measures are bivariate ordinal correlation measures that form a family because they all reduce to Goodman and Kruskal's gamma in the absence of ties (1954) Goodman, L. A. and Kruskal, W. H. 1954. Measures of association for cross classifications.. Journal of the American Statistical Association, 49: 732–764. [Taylor &; Francis Online], [Web of Science ®] , [Google Scholar]. For several gamma-family indices, more than one variance estimator has been introduced. In previous research, the “consistent” variance estimator described by Cliff and colleagues produced more accurate confidence intervals (CIs) than 3 other types of variances. However, the Cliff consistent (CC) variance is available for only 4 of 10 gamma-family measures. In the present research, a CC variance is derived for the remaining 6 gamma-family measures, and CIs constructed with the CC variance are compared with CIs constructed using other available variance estimators. The CIs are evaluated and compared in simulations and illustrated with scores on the Disinhibition and Avoidance scales of the Schedule for Nonadaptive and Adaptive Personality. Computer code for calculating all 10 gamma-family measures and their CC variances (using the R program) is available on a Web site (http://www.artsci.wustl.edu/~cwoods/CCvarR.html).     

The Moderating Role of Intergroup Contact in Race Composition,Perceived Similarity,and Applicant Attraction Relationships

Purpose

This study investigated the moderating effect of intergroup contact on the relationship between the race composition of organizational representatives, perceived similarity, and minority applicant attraction.

Design/Methodology/Approach

344 minority Malaysian-Chinese university students read a job advertisement that varied the racial composition of organizational representatives (100 % Malay or 50 % Malay–50 % Chinese or 100 % Chinese). Of these participants, 161 were Malaysian-Chinese in Malaysia (high intergroup contact location) and 183 were Malaysian-Chinese in Australia (low intergroup contact location). After reading the advertisement, participants responded to a series of scale items (e.g., perceived surface-level similarity, perceived deep-level similarity, and applicant attraction).

Findings

Results showed that the effect of race composition on attraction was stronger for minority participants in Australia than for minority participants in Malaysia. Perceived deep-level similarity mediated this moderated relationship.

Implications

The study findings suggest that organizations should include minority representatives in their recruitment advertising to attract minority applicants, particularly to attract minorities in locations with few opportunities for intergroup contact.

Originality/Value

By testing the mediating effects of perceived surface-level and deep-level similarity, this study contributes to our understanding of the mechanism linking the interaction between race composition and location with applicant attraction.

     

Determinants of perceived role conflict among women in bangladesh

The purpose of the present study was to find important determinants of perceived role conflict (PRC) by building a regression model. The paper examined the joint linear influences of wife's age, her income, her time for household chores and outside job, her female role perception (FRP), husband's income, his time for household work and outside job, total number of children, child-density, age of youngest child, and age of oldest unmarried girl to the variance of PRC of women. It also examined relative contribution of each of the independent variables controlling the influences of other independent variables to the variance of the PRC. The subjects were 140 women (homemakers—63; working women—77) from Dhaka City. One demographic and work characteristics questionnaire, one FRP Scale (Ilyas & Zaman, 1988), and one PRC Scale (Begun & Tasneem, 1984) were administered by two female interviewers visiting the household of the subjects. Regression analysis indicated that 27.12% of the variance (R ² change = .2712) of PRC was explained by all the independent variables operating jointly. However, FRP alone explained 20.61% variance of the PRC. Standardized beta indicated that one standard deviation unit change in FRP introduced greatest change ( = -.512) in PRC and one unit change in husband's time for outside job, the least change ( = -.003). Part and partial correlation coefficient also indicated FRP is related negatively to PRC. Thus, FRP is the only independent variable contributing significantly to the variance of the PRC.     

Perceptions of similarity between striving tasks and the generality of task preferences

Ninth graders rated their degree of liking for each of 12 striving tasks and, on another occasion, rated their perceptions of the similarity between each pair of tasks. Analyses of variance revealed strong PersonX Task interactions, suggesting the possibility of individual differences in task preference generality. A derived measure of task preference generality-specificity related significantly to an overall measure of perceived task similarity—generalizers saw the tasks as being more similar than did specifiers. Investigation of individual differences in degree of cross-situational generality was suggested as a way to integrate trait and situational theories of behavior.Data collection and initial data analyses for this study were conducted under the auspices of the Institute for Juvenile Research, Chicago, Illinois. Gratitude is expressed for the cooperation of the staffs and students at St. Michaels and Immaculata high schools, Chicago; for useful discussions of theoretical issues with Mark Oberlander; and for much assistance with data analysis provided by Arthur Kendall.     

The effects of depression and cognitive vulnerability to depression on judgments of similarity between self and other

The present research was conducted to investigate the effects of mild levels of depression, and cognitive vulnerability to depression, on people's perceptions of their similarity to others. Depression level was assessed using the Beck Depression Inventory, and cognitive vulnerability was measured with the Dysfunctional Attitude Scale. Subjects rated their similarity to others on a 7-point scale and also generated a list of attributes thought to typify the average other person. In terms of similarity judgments, individuals perceived themselves to be less similar to others as depression level increased. Furthermore, and also as predicted, only individuals scoring high on the vulnerability measure exhibited this pattern. Given the consistently positive view of others expressed by all subjects, this latter finding suggests that vulnerable individuals saw themselves as increasingly distinctive with respect to their own negative attributes, as depression level increased. This social comparison interpretation is consistent with previous research indicating that vulnerable individuals change from a focus on self-referent positive traits when nondepressed to a focus on negative traits when depressed. This pattern is also identified as a possible contributor to the social isolation and interpersonal difficulties characteristic of depression.     

Job burnout and job leaving in public school teachers: Implications for stress management

This study addressed the questions of whether selected teacher and organizational variables contributed to (a) significant amounts of variance in teachers' scores on three components of burnout, and (b) teachers' intentions to leave special education teaching. Study participants (N =490) included a sample of respondents to a survey of all of the 1096 special education teachers in Hawaii. Instruments used in this study included the Maslach Burnout Inventory, the Porter Need Satisfaction Questionnaire, and the Special Education Teacher Survey. Results of multiple regression analyses indicated that for the burnout components, Emotional Exhaustion and Depersonalization, relatively large and significant amounts of variance could be accounted for by a set of predictor variables. For the third burnout component, Personal Accomplishment, the variance accounted for by a set of predictor variables was significant but small. Stepwise discriminant function analysis was used to distinguish between teachers who stated they did or did not intend to leave special education teaching. Eight variables were identified as predictors of Intention to leave special education teaching, correctly classifying 73% of teachers into the appropriate Yes-Leaving and No-Leaving groups.This research was conducted in partial fulfillment of requirements for the Doctoral dissertation by the first author.     

The dimensionality of the flanker compatibility effect: A psychophysiological analysis

The psychophysiological approach was used to evaluate the effects of feature similarity and intrinsic response mapping on the flanker compatibility effect. Symbol (e.g., < > < and < / <) and letter arrays (e.g., HCH and SCS) were used. Results showed that delays in stimulus evaluation and both peripheral and central response competition contributed to the compatibility effect, with the contribution of these processes depending on feature similarity and the intrinsic response mapping of the stimuli. For letter stimuli, the difference in the size of the compatibility effect for similar and dissimilar arrays could be accounted for in terms of stimulus evaluation. For symbol arrays, differences in size of the compatibility effect could be accounted for by response competition. Thus, symbol and letter arrays do not appear to be processed differently; what is different is the degree to which stimulus and response-related processes are affected by incompatibility.     

Determining the bias and variance of a deterministic finger-tracking algorithm

Finger tracking has the potential to expand haptic research and applications, as eye tracking has done in vision research. In research applications, it is desirable to know the bias and variance associated with a finger-tracking method. However, assessing the bias and variance of a deterministic method is not straightforward. Multiple measurements of the same finger position data will not produce different results, implying zero variance. Here, we present a method of assessing deterministic finger-tracking variance and bias through comparison to a non-deterministic measure. A proof-of-concept is presented using a video-based finger-tracking algorithm developed for the specific purpose of tracking participant fingers during a psychological research study. The algorithm uses ridge detection on videos of the participant’s hand, and estimates the location of the right index fingertip. The algorithm was evaluated using data from four participants, who explored tactile maps using only their right index finger and all right-hand fingers. The algorithm identified the index fingertip in 99.78 % of one-finger video frames and 97.55 % of five-finger video frames. Although the algorithm produced slightly biased and more dispersed estimates relative to a human coder, these differences (x=0.08 cm, y=0.04 cm) and standard deviations (σ_x=0.16 cm, σ_y=0.21 cm) were small compared to the size of a fingertip (1.5–2.0 cm). Some example finger-tracking results are provided where corrections are made using the bias and variance estimates.     

The Performance Appraisal Milieu: A Multilevel Analysis of Context Effects in Performance Ratings

Purpose

The purpose of this study was to take an inductive approach in examining the extent to which organizational contexts represent significant sources of variance in supervisor performance ratings, and to explore various factors that may explain contextual rating variability.

Design/Methodology/Approach

Using archival field performance rating data from a large state law enforcement organization, we used a multilevel modeling approach to partition the variance in ratings due to ratees, raters, as well as rating contexts.

Findings

Results suggest that much of what may often be interpreted as idiosyncratic rater variance, may actually reflect systematic rating variability across contexts. In addition, performance-related and non-performance factors including contextual rating tendencies accounted for significant rating variability.

Implications

Supervisor ratings represent the most common approach for measuring job performance, and understanding the nature and sources of rating variability is important for research and practice. Given the many uses of performance rating data, our findings suggest that continuing to identify contextual sources of variability is particularly important for addressing criterion problems, and improving ratings as a form of performance measurement.

Originality/Value

Numerous performance appraisal models suggest the importance of context; however, previous research had not partitioned the variance in supervisor ratings due to omnibus context effects in organizational settings. The use of a multilevel modeling approach allowed the examination of contextual influences, while controlling for ratee and rater characteristics.

     

Models and fictions in science

Non-actual model systems discussed in scientific theories are compared to fictions in literature. This comparison may help with the understanding of similarity relations between models and real-world target systems. The ontological problems surrounding fictions in science may be particularly difficult, however. A comparison is also made to ontological problems that arise in the philosophy of mathematics.

Multicomponent models for the retention of numbers

Summary Several multicomponent models of memory (Bower, 1967) are presented and applied to the retention of three-digit numbers. Their application is based on the notion that both the digits making up a number stimulus and the position in which they appear can be interpreted as the components of a memory vector. The models are fitted to the data obtained in a Peterson-type experiment in which a single three-digit number was shown per trial. During the retention interval the subject engaged in an arithmetical task for either 13.6, 19.6, or 25.6 s. Following this, the subject attempted recall. Confidence ratings were also obtained. Twenty-four different types of correctly and incorrectly recalled responses were scored. Their frequency distributions were best predicted at each retention interval by a dual encoding model which relies on the assumption that stimuli are stored both component-wise and by means of a single unit code encoding the entire stimulus number. It is also shown that the confidence ratings may be successfully predicted from estimates of the expected number of components retained. About 80% of the rating variance is predicted by a parameter-free procedure.The original experimental study was funded by Grant Ey 4/3, Deutsche Forschungsgemeinschaft, Bad Godesberg. Further developments were made possible by Grant Sch 350/1 to R.S. from the same agency. Part of the theoretical work was done while D.V. was an NSF postdoctoral fellow at the Department of Psychology of New York University, New York. Free computer time was generously supplied by the Courant Institute of Applied Mathematics. We are grateful to Dietrich Albert, Stephanie Kelter, Micha Razel, and Paul Vitz for stimulating discussions.     

Representation of structure in similarity data: Problems and prospects

Conclusion After struggling with the problem of representing structure in similarity data for over 20 years, I find that a number of challenging problems still remain to be overcome—even in the simplest case of the analysis of a single symmetric matrix of similarity estimates. At the same time, I am more optimistic than ever that efforts directed toward surmounting the remaining difficulties will reap both methodological and substantive benefits. The methodological benefits that I forsee include both an improved efficiency and a deeper understanding of discovery methods of data analysis. And the substantive benefits should follow, through the greater leverage that such methods will provide for the study of complex empirical phenomena—perhaps particularly those characteristic of the human mind.Presidential address delivered before the annual meeting of the Psychometric Society at Stanford University on March 28, 1974 (in a somewhat revised form necessarily, for purposes of publication, with more words and fewer pictures). Much of the work surveyed in this paper was supported by the Bell Telephone Laboratories, during my eight years as a member of the technical staff there, and by grants (principally, GS-1302, GS-2283, and GB-31971X) from the National Science Foundation, during my subsequent eight years at Harvard and at Stanford. I wish to acknowledge, too, the important contributions to this work made by my many students and associates during these 16 years including, particularly, Doug Carroll, Jih-Jie Chang, Steve Johnson, and Joe Kruskal (at the Bell Laboratories) and Phipps Arabie, Glen Crawford, and Jim Cunningham (at Stanford). Finally, I am indebted to Sherry Huntsberger for suggestions and extensive help in connection with the preparation of the paper itself, and to Bert Green for useful editorial comments.