At the end of an experiment or observation design which uses a system based on cameras, markers and computers (and perhaps other measurement devices), one has to cope with a very large and complex data base. This data base has both a multivariate aspect (with tens of kinematic and kinetic variables) and a multifactor aspect (with individual and time factors, at the very least). Given these two aspects, this study investigates motion data by first using an exploratory multivariate and multifactor approach with two analyses: 1 / space windowing without any time windowing; 2 / space windowing with time windowing introduced. In both cases, the membership values are studied using Multiple Correspondence Analysis (MCA). To show this point of view, we consider an experimental study aimed at comparing two lifting modes: isokinetic and free. The experimental design considers 13?individuals with repeated measures: for each mode, three levels are considered (three speeds and three loads). For each of the 78?experimental situations, a signal with 232?components is computed from the measurement of markers? positions, and foot and hand forces. The methodology for investigating such a complex data base contains several stages: 1 / fuzzy space windowing (without time windowing) combined with an investigation of a)?kinematic variables, b)?kinetic variables and c)?the most important variables from the two previous analyses; and 2 / fuzzy space-time windowing of these variables. In the four analyses, the initial data are changed into average membership values. These are placed in a table where the rows are linked to individuals and trials, and the columns correspond to the space windows of the considered variables. Finally, the resulting table is investigated using?MCA. Following the statistical methodology and results, the discussion performs a critical analysis of our approach in terms of complexity, subjectivity and place occupation. This is compared with a more classic approach, i.e.?based on the pair " extrema and/or summarizing indicators/inference monovariate method ".