Playing with science metrics

There are numerous critiques, both online and in the literature (pdf), of the overused H-index and journal impact factor (IF) metrics, particularly when it comes to assessing the quality of recent research.  However, many of these critiques do not include suggestions for how to improve the situation, aside from pointing out that if h-index equals half the square root of total citations, then it is a redundant number.  Over in Economics, they have gone all out to make a fantasy economics league, but we dirt people have no such construction.  Here, then, are a few easily calculated stats that would be an improvement on the status quo.  The can be calculated using Google Scholar, if necessary, assuming anyone knows how to yank their numbers.

COIF: Citations over Impact factor.

This is the number of citations per year a given paper has relative to the impact factor of the journal. Impact factor/2 is the average citations per year of a journal for papers in their first two years of release; subracting that from the citations per year for each given paper gives each paper a score. averaging those for a researcher gives their score. 

This metric puts the particular work of a scientist into perspective relative to others who publish in similar journals. Of course, the COIF from someone who publishes in journals with IF of 20 is not comparable to that of those who publish in papers with IF of two, but if IF is going to be tied to individual researchers despite all admonitions against this practice, then COIF gives a way to interpret it.

I suspect that most young to mid careers scientists will have a positive COIF; citations, at least in geology, tends to accumulate more in later years than in the first two.  However, a declining COIF might mean that one's work is becoming less relevant as time goes by.

Whether an institution wants a person with low COIF and flashy journals, or a high COIF in esoteric publications probably depends on the particular institution, and what their priorities are.  So the COIF might even be useful for determining how well suited people are to various particular institutions.

As an industry person who publishes occasionally, I have few enough papers to be able to calculate this for myself manually and easily (using Google scholar, which probably inflates the numbers by 20%). Anyone with a basic knowledge of programming could probably automate the process, though.

Paper	year	Journal	IF	CPY	COIF
Birch et al.	2007	AJES	1.6	1.8	1.0
Parsons et al.	2008	Am Min	2.0	4.3	3.3
Klemme et al	2008	Geostandards	3.2	2.9	1.3
Parsons et al.	2009	CMP	3.5	3.3	1.6
Aleinikoff et al.	2012	Chem Geol	3.5	7.7	5.9
Magee et al	2014	SIA	1.2	1.0	0.4
Mean				3.5	2.2

SCP: Self citation percentage

What percentage of a paper's citations come from authors of that paper? This is simply The number of times a paper is cited by one or more of its authors divided by the total number of citations. This has been looked into by a number of people in the never ending struggle to interpret citation numbers.  At least some suggest that the number in generally in the twenties, and doesn't have enough variant to be useful, but I find that surprising, as the papers I've published vary quite a bit:

Demonstrating on myself again, it can range from 4% to 100%.

paper	year	cites	sefies	SCP
Parsons et al.	2008	30	6	20%
Aleinikoff et al.	2012	23	1	4%
Parsons et al.	2009	20	7	35%
Klemme et al	2008	20	6	30%
Birch et al.	2007	14	2	14%
Magee et al	2014	1	1	100%
Total		108	23	21%