h The -index: Friend or foe? Ian Rowlands

advertisement
The h-index: Friend or foe?
Ian Rowlands
Research Services Manager & Bibliometrician
University of Leicester
www.le.ac.uk/library
• Background to the h-index
• Practical considerations
– recall issues
– database scope issues
• Interpreting the h-index
– what does it mean?
– normalising the h-index
• Some pointers for good practice
Background to the h-index
What is the h-index?
Papers
224
A new phenomenon
173
Numbers of published papers on the h-index
145
115
63
31
19
6
2005
2006
2007
Source: Scopus July 2013
2008
2009
2010
2011
2012
What is the h-index?
• A single number that starts with your first citation
and accumulates over your research career
• It starts easy but gets progressively harder
• Summarises your broad research impact
Advantages of the h-index
• Simple: easy to generate and easy to understand
• Valid: correlates well with career achievements and
soft judgments about reputation
• Credible: difficult to game
• Flexible: any set of papers can have an h-index
Practical considerations
Database scope issues
Web of Science
934 unique titles
11,377 joint titles
Scopus
8,342 unique titles
Database scope issues
Source:
Jisc Academic Database
Evaluation Tool
Database scope issues
Web of Science
934 unique titles
11,377 joint titles
Scopus
8,342 unique titles
Database scope issues
Web of Science
934 unique titles
h = 59
Scopus
8,342 unique titles
Database scope issues
Web of Science
934 unique titles
h=65
Scopus
8,342 unique titles
Interpreting the h-index
Interpreting the h-index
• Tom
– 60 papers
– 6,000 citations
– 100 citations per paper (arithmetic mean)
• Harry
– 60 papers
– 6,000 citations
– 100 citations per paper (arithmetic mean)
Interpreting the h-index
• Harry
h-index = 20
– 60 papers
– 6,000 citations
– 100 mean citations per paper
• Tom
– 60 papers
– 6,000 citations
– 100 mean citations per paper
Interpreting the h-index
• Harry
h-index = 20
– 60 papers
– 6,000 citations
– 100 mean citations per paper
• Tom
h-index = 40!
– 60 papers
– 6,000 citations
– 100 mean citations per paper
1,000
Interpreting
the h-index
800
600
Number of citations
400
200
Tom (h=40)
Harry (h=20)
0
1
11
21
Rank order of paper
31
41
51
1,000
Interpreting
the h-index
800
The h-index measures
consistency not absolute
impact.
Number of citations for each paper
600
400
200
Tom (h=40)
Harry (h=20)
0
1
11
21
Rank order of paper
31
41
51
1,000
Interpreting
the h-index
800
The h-index measures
consistency not absolute
impact.
Quite a few Nobel
laureates have fair to
middling h-indexes.
Number of citations for each paper
600
400
200
Tom (h=40)
Harry (h=20)
0
1
11
21
Rank order of paper
31
41
51
Interpreting the h-index
“Physicists are particularly fond of models involving
a minimum of parameters”
The h-index discards the following information:
• papers with citations > h gain no extra credit
• papers with citations < h get no credit at all
and it fails to take account of differences in academic
age between researchers or of citation rates between
disciplines.
Variations on a theme
Correcting h for academic age
m quotient
𝑚 = ℎ/𝑦
where y = years since first published paper (`academic age’)
Correcting h for subject differences
Field
Essential Science Indicators
Citations per paper
after five years
Molecular biology & genetics
23.3
Immunology
18.5
Clinical medicine
10.0
Chemistry
7.8
Materials science
4.6
Computer science
3.1
Source: Juan E Iglesias and Carlos Pecharroman, Scaling the h-index for different
scientific fields, Scientometrics (2007) 73(3) 303-320.
doi: 10.1007/s11192-007-1805-x
Correcting h for subject differences
Field
Essential Science Indicators
Normalisation factor
subject correction
Molecular biology & genetics
0.44
Immunology
0.52
Clinical medicine
0.76
Chemistry
0.92
Materials science
1.36
Computer science
1.75
Source: Juan E Iglesias and Carlos Pecharroman, Scaling the h-index for different
scientific fields, Scientometrics (2007) 73(3) 303-320.
doi: 10.1007/s11192-007-1805-x
Correcting h for subject differences
It follows that a computer scientist
with h=20 is of equivalent merit to a
molecular biologist with h=79
(normalised h=35 in both cases).
Some pointers for good practice
Some dos
• Do take care to check publication lists for
accuracy and completeness: the h-index is not
externally audited
• Do check out Scopus and Web of Science: you may
be falling between the cracks
• Do quote h in context (total numbers of papers
and citations, for example)
• Do remember that h measures consistency rather
than `absolute quality’
Some don’ts
• Don’t compare h across different academic ages:
normalise
• Don’t compare h across different subjects:
normalise
• Don’t use the arithmetic mean for h: use the
geometric mean
• Don’t put too much emphasis on a single number
It's Clare Madge from Geography here.
I spent yesterday listening to the recording of the Elevenses sessions. I was
particularly interested in the H factor one and I had a couple of questions:
1. According to Iris, my H-Index is 10 (Web of Science) and 13 (Scopus). I then
searched it on Google Scholar and it was 20. I guess this must be because I am
a social scientist and the first two searches do not pick up social science
articles? Is there anything I can do about this? (I have checked the points you
mention in the presentation)
2. is there any way of adding my Google scholar H index to iris? It seems rather
unfair to social scientists to not include it.
3. I have also spent about 10 years of my academic career working on a parttime basis. Does the H factor take into account fractional working? If not, this
may potentially be a serious issue for people who work part-time re: job
opportunities, promotion and even ref impact measures.
Download