All posts by Stéphane Vautier

Projet pour un précis de méthodologie à l’usage des psychologues qui réfléchissent librement

Le temps est venu de contribuer à un débat critique publique sur la méthodologie de la recherche scientifique en psychologie. J’espère attirer l’attention des personnes intéressées par la manière dont on fait de la recherche en psychologie, afin qu’elles puissent commenter en toute liberté les idées qu’il s’agit de préciser et d’organiser dans un livre de méthodologie de la recherche en psychologie.

La discussion publique que permet Internet me semble utile à l’élaboration d’un tel ouvrage. A mes yeux, il ne s’agit pas de réfléchir dans son coin puis de “publier”, mais de formuler des critiques et des propositions déjà été éprouvées par une discussion informelle au sein d’une communauté de personnes qui, à un moment ou à un autre, se trouvent concernées. Je souhaite donner une légitimité tant aux affirmations (par principe contestables) qu’aux questions (par principe plus ou moins bien formulées).

Vous pouvez contribuer à la discussion en proposant des commentaires aux différents “billets” que je vais publier (à commencer par celui-ci). J’assume le rôle de modérateur, c’est-à-dire que je décide de les publier ou de ne pas les publier. (Je ne sais pas exactement ce qui pourrait m’empêcher de publier un commentaire.)

Retest change at the questionnaire item “Love to think up new ways of doing things”

A sample Ω108 of 108 French adults (mean age = 36.69, SD = 15.41, 54 males) took part in a survey which consisted of answering questionnaire items twice. The time lapse between the test and the retest ranged from three to five weeks.

The questionnaire item “J’adore élaborer de nouvelles façons de faire les choses” (Love to think up new ways of doing things) from the International Personality Item Pool (IPIP, Goldberg et al., 2006) was used along with a 5-point Likert scale from “Very Inaccurate” (coded 1) to “Very Accurate” (coded 5). Thus, the descriptive variable XΩ108 -> M(X), where M(X) = {1, 2, 3, 4, 5}2 = {(1, 1), (1, 2), …, (5, 5)}.

Values of the subset {(5, 1), (5, 2), (5, 3)} describe no respondents although 30 respondents chosen the value 5 at the test.

An interpretation is that respondents who chosen the value 5 at the test remembered an example of being engaged in thinking up new ways of doing something with positive feelings (event M), and that when one is able to remember such an experience (test event M∧5), one is able to retrieve a similar memory at the retest, which precludes the judgements “Very Inaccurate”, “Inaccurate”, or “Neutral” (retest event M∧(4∨5)).

For that interpretation to be an explanation, it has to have a falsifier. The test-retest event (M∧5)∧¬[M∧(4∨5)] is a falsifier.

Reference

Goldberg, L. R., Johnson, J. A., Eber, H. W., Hogan, R., Ashton, M. C., Cloninger, C. R., & Gough, H. C. (2006). The International Personality Item Pool and the future of public-domain personality measures. Journal of Personality Research, 40, 84-96.

Retest change at the questionnaire item “Like to solve complex problems”

A sample Ω108 of 108 French adults (mean age = 36.69, SD = 15.41, 54 males) took part in a survey which consisted of answering questionnaire items twice. The time lapse between the test and the retest ranged from three to five weeks.

The questionnaire item “J’aime résoudre des problèmes complexes” (Like to solve complex problems) from the International Personality Item Pool (IPIP, Goldberg et al., 2006) was used along with a 5-point Likert scale from “Very Inaccurate” (coded 1) to “Very Accurate” (coded 5). Thus, the descriptive variable XΩ108 -> M(X), where M(X) = {1, 2, 3, 4, 5}2 = {(1, 1), (1, 2), …, (5, 5)}.

Values of the subset {(5, 1), (5, 2), (5, 3)} describe no respondents although 12 respondents chosen the value 5 at the test.

An interpretation is that respondents who chosen the value 5 at the test remembered an example of being engaged in complex problem solving with positive feelings and that when one is able to remember such an experience (test event M∧5), one is able to retrieve a similar memory at the retest, which precludes the judgements “Very Inaccurate”, “Inaccurate”, or “Neutral” (retest event M∧(4∨5)).

For that interpretation to be an explanation, it has to have a falsifier. The test-retest event (M∧5)∧¬[M∧(4∨5)] is a falsifier.

Reference

Goldberg, L. R., Johnson, J. A., Eber, H. W., Hogan, R., Ashton, M. C., Cloninger, C. R., & Gough, H. C. (2006). The International Personality Item Pool and the future of public-domain personality measures. Journal of Personality Research, 40, 84-96.

General fact

By default, a general fact will refer to a general, descriptive, and non-tautological fact, associated with a possibly multivariate variable X: Ω -> M(X). It is a statement that is true for any element of Ω. A general fact is a negative (or restrictive) statement about at least one descriptive value of M(X): there exists a non-empty subset α of M(X) such that no being of Ω takes a value in α.

Remark: if the set Ω is unknown, the statement “there exists a non-empty subset of M(X) such that no element of Ω takes a value in it” should not be called a fact but a conjecture.