Archives mensuelles : octobre 2013

Joel Michell on psychometricians

… the fact that psychometricians, from the founding of their discipline, studiously turned away from investigating whether the attributes they aspired to measure really are quantitative means that their discipline is a pathological science (Michell, 2000) and that their standing as scientists is deeply compromised. Scientists who care more about appearing to be quantitative and the advantages that might accrue from that appearance, than they do about investigating fundamental scientific issues, put expedience before the truth. In this, they do not conform to the values of science and elevate non-scientific interests over those values, thereby threatening to bring science as a whole into disrepute. If the attributes that psychometricians aspire to measure are heterogeneous orders then psychometrics, as it exists at present, is fatally flawed and destined to join astrology, alchemy, and phrenology in the dustbin of history. (p. 7)

Michell, J. (2012). Alfred Binet and the concept of heterogeneous orders. Frontiers in Psychology, 3, Article 261.

Alfred Binet et Th. Simon sur la méthode scientifique

On nous a quelquefois reproché d’être avec aveuglement les ennemis de la théorie et de l’a priori. C’est un reproche injuste. Nous admettons les discussions de théorie, avant les recherches expérimentales, pour les préparer, et après, pour les interpréter ; ce que nous repoussons de toutes nos forces, ce sont les discussions théoriques qui veulent remplacer l’exploration des faits, ou qui s’établissent sur des faits obscurs, équivoques, légendaires, qu’on va recueillir dans des lectures, car c’est là ce que certaines gens appellent observer : c’est lire. L’idéal de la méthode scientifique doit être, à notre avis, une collaboration de la théorie et de l’expérimentation, collaboration bien précisée dans la formule suivante : une méditation prolongée sur des faits recueillis de première main. (Note 1)

Binet, A., & Simon, T. (1907). Le développement de l’intelligence chez les enfants. L’Année Psychologique, 14, 1-94.

Aaro Toomela on neo-Galtonian methodology

Studies where results are limited to only group differences and/or correlations between variables cannot go beyond probabilisitc prediction. The nature of a relationship or a difference between groups cannot be understood unless all exceptions are understood. So, for example, all reports of sex differences, race differences, or differences between cultures should not be attributed to sex, race, or culture, respectively, unless the exceptions are addressed. (p. 13)

Toomela, A. (2007). Culture of science: Strange history of the methodological thinking in psychology. Integrative Psychological and Behavioral Science, 41, 6-20.

James T. Lamiell on the confusion between persons and variables

On the so-called « frequentist » interpretation of probabilistic knowledge, it is knowledge tied inextricably to the consideration of a series of instances of whatever it is about which knowledge is being claimed. For example, to speack of having established empirically that the probability of attendance by « highly extraverted » persons at the next upcoming party is 0.7 is to say that, within some sufficiently large sample of persons, 70% of those who were found to be « highly extraverted » attended the next upcoming party and 30% of them did not. Such a finding, it should be noted, could serve as the basis for a testable–which is to say empirically falsifiable–proposition in a replication study.

However, to assert on the basis of findings of this sort that the probability is 0.7 that some particular « highly extraverted » individual will attend the next upcoming party is to express a knowledge claim that is not empirically testable. As a claim to knowledge, such a statement would be disconfirmed neither by the finding that the individual in question in fact attended the next upcoming party nor by the finding that s/he did not attend, and those are the only two empirical possibilities. Note, moreover, that a knowledge claim of this sort is equally incorrigible for each and every individual case considered as such. The very essence of the frequentist view of probabilistic knowledge generated through population-level studies is that such knowledge quite literally applies to populations and not to individual cases within those populations (cf. Bakan, 1955). In short, we must understand that such knowledge is, quite literally, knowledge of no one. (p. 9)

Lamiell, J. T. (2013). On psychology’s struggle for existence: Some reflections on Wundt’s 1913 essay a century on. Journal of Theoretical and Philosophical Psychology, 33, 205-215.

Paul Feyerabend sur les faits bruts

… la science ne connaît pas un seul « fait brut » …. les « faits » qui entrent dans nos connaissances sont déjà considérés sous un certain angle, et sont, par conséquent, essentiellement spéculatifs. (p. 15)

Feyerabend, P. (1979). Contre la méthode : esquisse d’une théorie anarchiste de la connaissance. Paris : Seuil.

Voir aussi Danziger (1990), Hanson (1958/2010), Popper (1963/2006), Searle (1995).