310 likes | 449 Views
Evaluative Bibliometrics: Citation based Research Indicators. Bo Jarneving Bibliometrifunktionen GUB 2011. Variables for citation based performance indicators. Document type articles, reviews, letters, notes Observation period Citation interval Self-citations With or without
E N D
Evaluative Bibliometrics: Citation based Research Indicators Bo Jarneving Bibliometrifunktionen GUB 2011
Variables for citation based performance indicators • Document type articles, reviews, letters, notes • Observation period Citation interval • Self-citations With or without • Subject classification Generally based on ISI ”Journal Subject Categories” Classification of papers published in multidisciplinary journals • Counting method decisions How to count co-authored papers? How to count when there are multiple author affiliations? How to count when there are multiple subject categories?
Observation period • Observation period • 2004-2009 • Variable or fixed citation interval (window)
What is a self citation? Schubert A, Glanzel W, Braun T (1989) Scientometric Datafiles - A Comprehensive Set of Indicators 0n 2649 Journals and 96 Countries In All Major Science Fields and Subfields 1981-1985. Glanzel W, Thijs B, Schubert A, Debackere K(2009) Subfield-specific normalized relative indicators and a new generation of relational charts: Methodological foundations illustrated on the assessment of institutional research performance.
Journal Subject Category • A proxy for subject field • Sometimes multiple assignments of journals to categories • Based on citation data and ”manual methods” • A total of 173 categories in the science volume of Journal Citation Report (JCR) • Alphabetical order – no hierarchy
Classification of multidisciplinary papers • Reversing pathological neural activity using targeted plasticity • NATURE, 2011, V470, N7332, P101-U114
Counting method: fractional counts or whole counts • Whole counts • Every author, institution, department and country is assigned one full paper • Fractional counts • A paper is divided up over its authors • If a paper has 3 authors, every author is assigned 1/3 of the paper. • If 2 of these authors are associated with institution A and 1 with institution B,A recieves2/3 of the paper and B 1/3.
Counting methods: fractional counts or whole counts When more than one institutional affilation of an author: • Whole counts • One author associated with two institutions – each insitution recieves one full paper • Fractional counts • One author associated with two institutions – each insitution recieves one half of a paper
Counting methods: fractional counting of papersExample 1: 1 paper, 1 category, 2 authors, 3 institutions (A-C) 1c 1/4 1/2 Agets Bg Cg 1/4 1/2 1/2
Counting methods: fractional counting of papersExemple 2: 1 paper, 2 categories (1, 2), 2 authors, 3 institutions (A-C) 1/4 1/4 1/8 1/2 1/4 1/8 Category2 Agets Bgets Cgets Category 1 Agets Bgets Cgets 1c 1/4 1/4 1/2 1/4 1/8 1/8
Counting methods: fractional counting of citations.Example 3: 1 paper, 1 category, 2 authors, 3 institutions (A-C).The paper has recieved 10 citations (c). ¼ x 10 1c 1/2 Arecieves 2.5 c Brecieves 2.5 c Crecieves 5 c ¼ x 10 1/2 ½ x 10
Counting methods: fractional counting of citationsExample 4: 1 paper, 2 categories (1, 2), 2 authors, 3 institutions (A-C). The paper has recieved 10 citations (c) ¼ x 10 ¼ 1/8 x 10 ½ ¼ Category 2 Arecieves Brecieves Crecieves Category 1 Arecieves Brecieves Crecieves 1/8 x 10 1c ¼ x 10 ¼ ½ ¼ 1/8 x 10 1/8 x 10
Counting methods: whole counts of citationsExample 5: 1 paper, 2 categories (1, 2), 2 authors, 3 institutions (A-C). The paper has recieved 10 citations 1 x 10 1 x 10 1 x 10 Kategori 2 Ar Br Cr Kategori 1 Ar Br Cr 1 x 10 1c 1 x 10 1 x 10 1 x 10 1 x 10 1 x 10
Normalization of citation counts • Observed value: institutional (departmental) citation counts • Expected value: approximated world averages • based on Web of Science citation data
Normalization of citation counts The expected value (reference value) • Define groups on basis of the institutional publication pattern. Group by: • Publication date (year) • Journal Subject Category • Document type • For each such group, compute the Field Citation Score (FCS) • Based on citations between papers indexed in WoS • Approximations of world-citation-averages for each group
Normalization of citation counts: defining groups • Example 1: The department (group) publishes in journals assigned to one journal subject category only (neurosciences). • Brain Research 4 articles 2004 • Brain Research 1 letter 2004 • Journal of Neuroscience 3 articles 2002 • Journal of Neuroscience 1 review 2003 • All authors are from one department • Gather all articles published in journals assigned to neurosciences and published 2004 • Gather all letters published in journals assigned to neurosciences and published 2004 • Gather all articles published in journals assigned to neurosciences and published 2002 • Gather all reviews published in journals assigned to neurosciences and published 2003
Normalization of citation counts: reference values For each group (1- 4) compute the world-average number of citations (Field Citation Score [FCS]): • Neurosciences-articles 2004 FCS = 4.2 • Neurosciences-letters 2004 FCS = 1.0 • Neurosciences-articles 2002 FCS = 3.8 • Neurosciences-reviews 2003 FCS = 8.5
Normalization of citation counts: averages • CPP = Citation Per Publication (department) • FCSm = Mean Field Citation Score • CPP/FCSm = the Crown Indicator • Compute CPP: (3 x 7 + 1 x 6 + 2 x 5 + 1 x 4 + 2 x 1) / (3+1+2+1+2) = 4.78 • Compute FCSm: (4 x 4.2 + 1 x 1.0 + 1 x 8.8 + 3 x 3.8) / (4 + 1 + 1 + 3) = 4.22 • Compute CPP/FCSm: 4.78 / 4.22 = 1.13
Normalization of citation counts: defining groups • Example 2: The department (group) publishes in journals assigned to more than one journal subject category. • Brain Research 4 articles 2004 • Brain Research 1 letter 2004 • Journal of Neuroscience 3 articles 2002 • Neuroimage 1 article 2003 Four classifications: • Gather all articles published in journals assigned to neurosciencesand published 2003 • Gather all articles published in journals assigned to neuroimagingand published 2003 • Gather all articles published in journals assigned to Radiologyand published 2003 • Gather all articles published in journals assigned to Nuclear Medicine & MedicalImagingand published 2003
Normalization of citation counts: reference values • For each group (1- 7) compute the world-average number of citations (Field Citation Score [FCS]): • Neurosciences-articles 2004 FCS = 4.2 • Neurosciences-letters 2004 FCS = 1.0 • Neurosciences-articles 2002 FCS = 3.8 • Neurosciences-articles 2003 FCS = 5.5 • Neuroimaging-articles 2003 FCS = 3.2 • Radiology-articles 2003 FCS = 2.8 • Nuclear Medicine & MedicalImaging articles 2003 FCS = 4.6
Normalization of citation counts: averages • CPP = Citation Per Publication (department) • FCSm = Mean Field Citation Score • CPP/FCSm = the Crown Indicator • Compute CPP: (2 x 7 + 1 x 6 + 2 x 5 + 1 x 4 + 2 x 1 + ¼ x 7 + ¼ x 7 +¼ x 7+ ¼ x 7) / (2+1+2+1+2+¼ +¼ +¼ + ¼) = 4.78 • Compute FCSm: (4 x 4.2 + 1 x 1.0 + 3 x 3.8 + ¼ x 5.5 + ¼ x 3.2 + ¼ x 2.8 + ¼x 4.6) /(4 + 1 + 3 + ¼ + ¼ +¼ + ¼) = 3.69 • Compute CPP/FCSm: 4.78 / 3.69 = 1.30
Normalization of citation counts: paper by paper • Normalization on paper level: 10.48 / 9 = 1.16 • Ratio of averages vs. average of ratios
Total performance indicator • The average impact times the total number of papers: • Where ci is the citation score for paper i, ei the expected number of citations for paper i and N the total number of papers for a department
A real-life example • TI- Engineering of a novel Saccharomyces cerevisiae wine strain with a respiratory phenotype at high external glucose concentrations • JN - Applied and Environmental Microbiology • SC- Biotechnology & Applied Microbiology; Microbiology • DT- article • PY- 2005 • AU- Cecilia Henricsson (Institutionen för kemi- och bioteknik, Chalmers) Maria de Jesus-Ferreira (-) Kristina Hedfalk (Institutionen för kemi) Karin Elbing (-) Christer Larsson (Institutionen för kemi- och bioteknik, Molekylär bioteknik, Chalmers) Joakim Norbeck (Institutionen för kemi- och bioteknik, Molekylär bioteknik, Chalmers) Roslyn Bill (-) Stefan Hohmann (Institutionen för cell- och molekylärbiologi) Lena Gustafsson (Institutionen för kemi- och bioteknik, Molekylär bioteknik, Chalmers)
A real-life example Bibliometric data: • FCS or = field citation score (reference value), self-citations excluded • C = observed citations, self-citations excluded • Observation period 2004-2009 • A citation window of 5 years • Fractionalized author counts • Fractionalized subject category counts • Journal subject categories: • Biotechnology & Applied Microbiology • Microbiology
A real-life example Computing the reference value: • Find all papers published in journals assigned to Biotechnology & Applied Microbiology and/or Microbiology during 2005-2009. • Count all citation to these papers • Fractionalize in case of multiple journal subject category assignments • Compute FCS or : • for Biotechnology & Applied Microbiology , and • for Microbiology. • N = 247 (journals)
How to affect the average field-normalized citation score? • Publish in journals covered by Thomsen ISI (SCOPUS) • Publish in high esteem journals • Use a standard departmental name • Use a standard institutional name