Suchergebnisse

This paper describes issues in developing a scientifically based system for relating the abilities of people required to perform jobs to the nature of the task characteristics of these jobs. Conceptual and methodological issues in classifying tasks are discussed as are the objectives of task taxonomies and criteria for their evaluation. Alternative taxonomic approaches are described. A particular approach to classifying tasks in terms of ability requirements is described in more detail and evaluated according to a number of criteria and applications. These latter include encouraging utility in job analysis, test development, setting performance standards, and classifying jobs into families.

This report reviews and discusses a number of the methodological questions relating to the application of an experimental-correlational approach to the problem of assessing complex performance. The basic point of departure is the specification of the requirements for a task taxonomy and an analysis of the value of factor analytic investigations in combination with experimental methods in providing the framework for such a taxonomy. The way in which this approach has been applied in the past and the expected benefits of its successful implementation are discussed. It is concluded that experimental-correlational studies offer considerable promise in attacking complex performance but that a more extensive research program is needed. The general outlines of such a program is described.

Two hundred and four Ss practiced the components of a complex multidimensional compensatory pursuit task, singly and in combination. These components involved discrete display-control relationships. The total task, which was practiced last, requires an integration of these components; that is, the S must operate the multiple controls in order to minimize error indications on all displays simultaneously. The problems investigated were (a) the extent to which performance on task components, individually practiced, is predictive of subsequent total task performance; (b) the extent to which practice on combinations of components is predictive of total task performance; (c) the interrelationships among component performances; and (d) the relative contribution of various component performances to total and subtask performances. The resulting correlational and multiple correlational analyses provide some tentative principles of part-whole task relationships relevant to the understanding of skilled performance.

Measures of helicopter pilot proficiency were obtained on several hundred student pilots in the Primary and Basic training phases. Measures were based on students' performance on 75 and 76 tasks (items), for the Primary and Basic phases, respectively. Intercorrelations of tasks in each phase were subjected to factor analysis. The 12 factor rotation solutions were presented in detail for each phase, and the 18 and 24 factor rotations solutions were described briefly. In almost all cases the same tasks (e.g., RPM; Altitude) tended to cluster together across different maneuvers. The factors are interpreted in terms of the operations performed for each task, and the theoretical and practical implications of the findings are discussed.

The Complex Coordination Test was used to examine component-total task relationships when component tasks are practiced in different orders. Sixty subjects practiced (two trials of two minutes each) on six component tasks and on the total task, practicing these in various orders. Part-tasks were three single-level (one rudder and two stick control) tasks and three double-level (one stick-stick and two rudder-stick) tasks. Correlational analyses were performed to determine component-total task relationships and component to total task predictability. It was found that order of part-task practice affects total task proficiency. Also, part-task scores can be combined in a statistically independent manner to yield prediction scores on a more complex task only when a specific double-level task is combined with a specific single-level task. It was also found that observed and predicted total task scores agree better when double-level tasks are practiced before rather than after the total task. Finally, prior practice on double-level tasks led to better total task proficiency than did prior practice on single-level tasks.