ǂ"ǂPeer review" in bibliometrijski indikatorji

magistrsko delo

Robert Matoh (Author), Primož Južnič (Mentor)

Abstract

Bibliometrijske metode in metode ekspertnega ocenjevanja že nekaj časa veljajo za dve prevladujoči skupini metod na področju vrednotenja znanstvenih dosežkov. Čeprav sta si po svoji naravi dokaj različni, gre od njiju po drugi strani pričakovati določene skupne značilnosti oz. rezultate, saj nenazadnje poskušata meriti isto, torej znanstveno kvaliteto. Da bi ugotovili, kakšen odnos med njima dejansko obstoji, podrobneje analiziramo obe metodi, na podlagi katerih nato oblikujemo nekaj osnovnih predpostavk na to temo. V naslednji fazi skušamo ta odnos preveriti še s pomočjo statistične analize rezultatov zadnjega javnega razpisa za sofinanciranje raziskovalnih projektov. Izkazalo se je, da najmanjše razlike med bibliometrijskimi in ekspertnimi ocenami obstajajo na področjih naravoslovja in humanistike; medtem ko je analiza na mikro nivoju pokazala na najvišjo stopnjo povezanosti med kazalcema "normirano število čistih citatov" in "ocena izjemnih raziskovalnih dosežkov vodje projekta". Sodeč po naknadno opravljeni analizi, bi pridobljeni koeficienti lahko bili ugodnejši, če bi recenzenti pokazali večjo mero strinjanja o kvaliteti predlogov projektov. To je hkrati glavna pomanjkljivost obravnavanega sistema ekspertnega ocenjevanja, zato velja premislek o nekaterih predlogih za omejitev tega pojava, kot so število uporabljenih recenzentov na prijavo, način njihovega vrednotenja, način dodeljevanja recenzentov itd. Z manjšo izjemo na področju tehniških ved, smo za prijavitelje z odobrenim financiranjem ugotovili tudi, da njihovi bibliometrijski kazalci v povprečju niso signifikantno boljši od kazalcev zavrnjenih prijaviteljev.

Keywords

bibliometrija;ekspertno ocenjevanje;zanesljivost;raziskovalni projekti;financiranje;razpisi;znanstveno raziskovanje;ocenjevanje;vrednotenje;Slovenija;magistrske naloge;

Data

Language:	Slovenian
Year of publishing:	2012
Typology:	2.09 - Master's Thesis
Organization:	UL FF - Faculty of Arts
Publisher:	[R. Matoh]
UDC:	001:02(043.2)
COBISS:	49832802
Views:	887
Downloads:	390
Average score:	0 (0 votes)
Metadata:

Other data

Secondary language:	English
Secondary abstract:	For some time now, bibliometric methods and methods of expert evaluation are considered as two prevailing groups of methods in the area of scientific performance evaluation. Even though significant differences consist between them in their nature, some sort of common results and characteristics can be expected, since they both attempt to measure the same thing, i.e. scientific quality. To determine the actual relationship, we thoroughly examine both methods, and create some basic assumptions considering the issue. In the next phase, we look upon this relationship by performing statistical analysis on results of the last call for research project proposals. It was revealed, that the lowest differences between bibliometric and expert ratings exist in the areas of natural sciences and humanities, wheareas the analysis at micro level showed, that the highest level of connection exists between indicators "normalised number of pure citations" and "assessment of projectʹs leader outstanding research achievements". Referring to subsequent analysis, those coefficients could have been more favourable, if reviewers had shown greater level of agreement about quality of research project proposals. At the same time this is the main blemish of peer review system under observation, so reflection about some propositions for reducing this phenomenon is welcome, such as number of assigned reviewers per grant, ways of its evaluation, ways of assigning reviewers etc. With small exception in the field of technical sciences, we also found out, that the applicants, whose research proposals were eventually approved, on average didn`t have significantly better bibliometric indicators from those applicants, who didnʹt succed in receiving a grant.
Type (COBISS):	Master's thesis/paper
Thesis comment:	Univ. v Ljubljani, Filozofska fak., Oddelek za bibliotekarstvo, informacijsko znanost in knjigarstvo
Pages:	78 str.
ID:	8890215