Screening a promotion claim when bibliometric sources disagree
A state university's screening committee is evaluating Dr R, an Associate Professor with nine years' service, for promotion to Professor under the Career Advancement Scheme. Her research score claim rests on 14 journal papers, 2 books, 3 conference papers, an h-index of 12 on Google Scholar and 7 on Scopus, and an i10-index of 18. Six of the fourteen papers appeared in journals that were on the UGC-CARE list on the date of publication but are absent from the current revised list. Two papers, both in the same journal, were published within eleven weeks of submission, carry a publication fee of ₹38,000 each, and the journal's website lists an editorial board of whom four cannot be traced. One paper has 240 citations in Google Scholar and 4 in Scopus; a departmental colleague is a co-author on nine of the fourteen papers. The committee has Scopus, Web of Science and Google Scholar access but no research analytics budget, and must submit its recommendation this week.
1. How should the committee reconcile the Scopus, Web of Science and Google Scholar indicators?
Read the worked answer
The three sources do not measure the same thing. Google Scholar crawls theses, conference papers, books, preprints and journals outside mainstream indexes, so a higher h-index there (12 against 7) is expected and is not by itself evidence of misconduct - though a wide gap can indicate heavy publication in venues outside curated indexes.
The committee should adopt a rule and apply it uniformly:
- One primary source per claim, named in the minutes with the retrieval date. Scopus is the defensible primary for indexed-journal claims; Web of Science as corroboration.
- Never take the highest number across sources. Selecting Google Scholar's 12 because it is larger is cherry-picking and will not survive appeal.
- Interrogate the 240-versus-4 paper. Check the citing documents: if they come from the same journal or the same small group, the Google Scholar count is inflation, not impact. Check self-citations and citation stacking.
- Assess contribution, not just output. Nine of fourteen papers share one co-author. Ask for a contribution statement and check for salami publication - the same data split across multiple papers.
- Record the limitation that no field-normalised indicator was available, and do not invent a composite index to compensate.
2. What is the defensible treatment of the six delisted-journal papers and the two suspect papers?
Read the worked answer
The question for the six is not 'is this journal listed today' but 'was it a legitimate venue on the date of publication'. UGC-CARE operates through a published list that has been revised in tranches; the relevant fact is the status at the time of publication. The committee should cite the specific list version and date, credit the papers, and note the subsequent delisting without converting it into retrospective punishment. Anything else makes the outcome depend on which week the committee sits.
For the two suspect papers, apply a checklist rather than an impression:
- peer-review turnaround disclosed and plausible;
- article processing charge stated before submission, not solicited after acceptance;
- editorial board verifiable, independent and reachable;
- indexing claims checkable against the index itself;
- valid ISSN and a verifiable publisher address;
- membership of DOAJ or a recognised publishing association, if open access.
An eleven-week turnaround is not disqualifying on its own. The combination - fast turnaround, high fee, unverifiable board - is a red flag. If the checklist fails, the committee should not count those two papers toward the research score, and should say so with reasons. The candidate keeps the publications; the score is what the committee controls.
3. What should the committee minute, and what should it refuse to do?
Read the worked answer
Minute: the primary database and date of access for every indicator; the list version and date relied on for the six papers; the checklist outcome for the two suspect papers with the specific criteria that failed; the contribution statement and any duplicate-publication finding; the research score computed line by line against the regulation's table, with the arithmetic shown.
Refuse to:
- compute a blended index mixing Google Scholar, Scopus and Web of Science;
- reject a candidate on h-index alone - it is career-length- and field-dependent and penalises early-career researchers and slow-moving disciplines;
- count the same output twice, for example a paper and a book chapter drawn from it, or a conference paper later published as an article;
- substitute the committee's own preferred list for the one the regulation names;
- leave the reasoning oral. A decision that cannot be reconstructed from the file will not survive an appeal, an RTI query or an audit.
The committee should also record that the same rubric was applied to every candidate in that sitting. Uniformity of process is the strongest defence available to it, and it is the part most often missing from real screening files.