Comment manipuler son facteur d’impact

Ou : de l’observation des effets des mesures sur les comportements qu’elles sont censées mesurer, vu du côté des revues.

Je ne me lasserai pas de le répéter : les indicateurs bibliométriques ne portent pas sur des particules passives mais sur des êtres humains (et des revues gérées par des êtres humains). On peut de plus penser qu’en tant que chercheurs, ils ont parfois quelques idées, voire quelque expertise, quant à la manière d’opérer des mesures, aux effets possibles de l’observation sur la réalité observée, ou encore à la notion d’effet pervers ou à la possibilité de détourner une décision venue d’en haut à son propre profit (rayez la mention inutile, ou ajoutez-en d’autres, selon votre spécialité). Bref, l’usage des indicateurs bibliométrique incite à modifier ses comportements de publication, de citation, etc. pour obtenir un meilleur score, et c’est vrai tant des revues ou des institutions que des individus. On pourrait y trouver une forme de consolation, mais c’est évidemment plutôt inquiétant tant à cause de l’énergie ainsi perdue que, plus fondamentalement, de l’influence que cela peut avoir à terme sur les choix de sujets de recherche ou de manière de les traiter.

En tout cas, un correspondant qui a souhaité rester anonyme nous propose un très bel exemple de tout cela dans le cas d’une revue d’histoire – de bonne facture, dit-il – publiée par un éditeur important. Il ne s’agit là nullement, de toute façon, de dénoncer une revue précise, d’abord parce que, nous le percevons tous plus ou moins clairement, ce type de discussion a lieu (ou aura bientôt lieu) partout – combien de revues de grands éditeurs universitaires, y compris maintenant en histoire, affichent leur facteur d’impact sur la page d’accueil de leur site -, ensuite parce que cette revue prend au moins la peine d’en discuter clairement, de mettre à plat ces pratiques et de tenter de séparer l’éventuel bon grain (ou le grain acceptable…) de l’ivraie. En outre, l’introduction souligne que la revue n’aurait pas à se poser de telles questions si les bibliothécaires, mais aussi les chercheurs qui cherchent où proposer leurs textes, n’avaient pas déjà adapté leurs comportements. Ce texte nous fait aussi découvrir encore de nouvelles particularités des indicateurs, par exemple  l’effet de la taille des articles ou de l’ordre dans la table des matières… (si tout cela n’avait pas des enjeux très concrets, ce serait passionnant en termes d’observation des pratiques de lecture des chercheurs). Et il rappelle incidemment quelques problèmes particulièrement graves pour les SHS, notamment le fait qu’on ne compte habituellement que les citations dans les deux ans qui suivent la publication de l’article, ce qui est, pour nous, très court (mais qu’on peut apparemment arriver à tourner en partie, vous verrez comment).

A noter qu’outre les citations suivantes, le document comparait un certain nombre d’indices entre la revue et celles perçues comme ses principales concurrentes. (je ne traduis pas, partant du principe que tous nos lecteurs lisent l’anglais, mais si ce n’est pas le cas ou si des sigles méritent une explication, vous me dites…)

« Citation metrics are becoming increasingly important to all journals in the humanities. While most who have studied the way that these metrics are compiled will conclude that they are an inaccurate and sometimes misleading there is plenty of anecdotal evidence that librarians are basing subscription decisions, and funding bodies are basing funding decisions, on these rankings. From year to year we see more evidence that subscription performance and the quantity of quality submissions are linked to metric performance […]

Improving Impact Factor

There are many well rehearsed and oft employed methods for manipulating the Impact Factor of a journal. Many do not improve the genuine quality of the journal; some will actively detract from it of lead the journal away from its genuine academic purpose. Many of these strategies outlined below will not be suitable, and would not be recommended without caution, but I include them all in order to give a clear and complete sense of how the journal’s content, structure and schedule affect its Impact Factor:

The type of articles published will make a difference:

Publish articles in rapidly growing and/or new subject areas. Although a new area of interest might be small in terms of practitioners, these ‘early-adaptors’ will tend to cite a larger proportion of the total corpus of literature than in the case of well-established subjects.

Publishing controversial articles can generate increased debate in the community, resulting in more citing articles being written to discuss or refute the article.

Publish articles by authors with a ‘highly-cited’ track record.

There is some statistical evidence from the BMJ that publishing longer articles will lead to increased citation per page.

Publish more citable material – articles, reviews, review articles, surveys – and less non-citable material – editorials, letters, bibliographies – as the latter take up value space in the journal that could be given to attracting citations. ‘Non-citable’ articles are not counted in the IF denominator so there will be no improvement in this respect.

Reduce the number of articles published, focusing only on articles which will achieve high impact, improving the metric-measured ‘quality’.

Volume make-up also makes a significant difference to impact factor:

Hot, topical articles should be placed at the beginning of an issue (thus appearing at the top of a table of contents). Readers tend to scan and note articles appearing near the top of the list and there is clear evidence from within the industry that articles at the top of TOCs are read more, and articles that are read more tend to be cited more.

Hot, topical articles should be placed in the first issue of the year. When the IF target window opens articles which were published in early journal issues will have already been published for up to a year. For humanities journals, for which the two-year citation window is too short, the additional year of exposure can add significantly to the number of citations received.

Moving to an incremental publishing model – wherein articles are copyedited, typeset, granted a DOI and published online as soon as they are accepted, instead of waiting for an issue (then repaginated and bound in issues when appropriate) – works in the same way as above to give articles citable exposure before their official publication date. The greater majority of STM journals operate this workflow, increasing numbers of humanities journals do too, increasing the speed of production. The option is available to [initiales de la revue] should you wish to consider it.

The simplest way to increase an impact factor, or any other citation based metric, is to generally increase the number of citations the journal receives. There are a couple of controversial methods employed to achieve this:

Encourage editorial board members and authors to cite articles in the journal in their own publications outside of the journal.

Encourage authors to cite other articles within the Review in their article for the Review – this used to be employed regularly by editors who wished to manipulate their impact factor but has led to a number of journals being dropped from the JCR for unacceptable manipulation which can be clearly identified by a rapidly rising self-citation number. This method is discouraged. »




Improving Impact Factor
There are many well rehearsed and oft employed methods for manipulating the Impact Factor of a journal. Many do not improve the genuine quality of the journal; some will actively detract from it of lead the journal away from its genuine academic purpose. Many of these strategies outlined below will not be suitable, and would not be recommended without caution, but I include them all in order to give a clear and complete sense of how the journal’s content, structure and schedule affect its Impact Factor:
The type of articles published will make a difference:
Publish articles in rapidly growing and/or new subject areas. Although a new area of interest might be small in terms of practitioners, these ‘early-adaptors’ will tend to cite a larger proportion of the total corpus of literature than in the case of well-established subjects.
Publishing controversial articles can generate increased debate in the community, resulting in more citing articles being written to discuss or refute the article.
Publish articles by authors with a ‘highly-cited’ track record.
There is some statistical evidence from the BMJ that publishing longer articles will lead to increased citation per page.
Publish more citable material – articles, reviews, review articles, surveys – and less non-citable material – editorials, letters, bibliographies – as the latter take up value space in the journal that could be given to attracting citations. ‘Non-citable’ articles are not counted in the IF denominator so there will be no improvement in this respect.
Reduce the number of articles published, focusing only on articles which will achieve high impact, improving the metric-measured ‘quality’.

Vous aimerez aussi...

1 réponse

  1. Gaïa dit :

    Des chercheurs en médecine ont également noté l’émergence d’un éventail de comportements socialement indésirables associés à la montée en puissance des indicateurs bibliométriques.
    A lire ici : http://rachelgliese.wordpress.com/2009/12/21/les-pathologies-du-h-index/

Laisser un commentaire

Votre adresse de messagerie ne sera pas publiée. Les champs obligatoires sont indiqués avec *