Premium
Analysis in man‐task system behavior studies
Author(s) -
Loslever Pierre
Publication year - 1993
Publication title -
behavioral science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.371
H-Index - 45
eISSN - 1099-1743
pISSN - 0005-7940
DOI - 10.1002/bs.3830380203
Subject(s) - computer science , set (abstract data type) , preprocessor , task (project management) , scale (ratio) , a priori and a posteriori , artificial intelligence , sample (material) , data mining , transformation (genetics) , level of measurement , machine learning , mathematics , statistics , philosophy , chemistry , physics , biochemistry , epistemology , chromatography , quantum mechanics , gene , programming language , management , economics
Studies of man‐task system behavior start with the obtaining of behavioral data. The recorded variables are generally numerous, can be objective and/or subjective and can be classified by their scale types. The next stage is to describe the behavior through a preprocessing technique in such a way that all variables have the same scale type. To achieve this goal, the qualitative scale is chosen but to lose as little information as possible from a quantitative to qualitative scale transformation, fuzzy categories are considered. The next stage is to analyze the resulting data set. A data set is considered through an observation X category table and studied using either the simple or multiple correspondence factor analysis. These methods yield both mathematical and descriptive behavior patterns. Their advantages are to not consider a priori too much constrained mathematical hypotheses and too much synthesized indicators (computed over the subject sample, for example). The general data analysis procedure (data characterizing/correspondence analysis) is described for four different data set types: judgments using quantitative scales, multidimensional signals, viewer behavior using eye movements, and operator behavior in tracking task.