Dear Mr. James R Knaub, this is a very interesting and practical question.

Many people consider "impact factor of the journal" is very important and the high "impact factor of the journal" really brings the authors many admirable “benefits” because present people accept the works of the authors. This is good and it is great. But it is not that lucky for some pioneers, the works of these authors may not be accepted by present people and these authors may not even have any chance to have any of their papers published on any of journals with "impact factor", but the pioneering works still going because it is the right time for the “metabolism” of some fields of human science. It is OK and it is understandable that fruits are more attractive than the process. This is human science, this is human and this is researchers’ life.

Yours, Geng

Dear James,

I do not think the impact factor of a journal will have any substantial impact on society. But it is not only statistics journal that have low impact factors but mathematics journals have also low impact factors. For instance the Journal of American Mathematical Society has impact factor of 3.06 compared to New Eng J of Medicine with impact factor of 55.87 looks like measurements taken in two different planets.

The impact factor is just a numerical quotient  of the total number of citations in the Journal Citation Reports and the total number of articles published in the past two years of that particular journal. From this,  journals with low IF signify how low the number of citations are for the mere reason that people who produce results and publish there do much independently of their own with out much referencing to what have been done in the last two years in their fields that appear in that journal unlike journals of medicine where referencing is one of the way of promoting self and and group and gaining funds - that is just the culture of the field.

As you have said, either statisticians or mathematicians, applied mathematicians should participate with people from other  fields in order their papers appear in journals of high IF or they have to change behaviors to cooperate, make references of  works in journals they  intend to publish. 

But above all I prefer to down play the importance of IF and we have to campaign to that effect, but the dissemination of results and knowledge to impact readers and society. Think the difference RG makes compared to other high IF journals, it is those articles that are posted here in RG that are  read more and down loaded more than articles that are in journals of high IF or even in journals where in most cases we have to buy an article to read. 

Dear Jim,

this question is very interesting, and how we calculate the "impact factor" itself is a statistical question, that seems to be quite ironic. My though is quite simple, if it is too obvious to you, dear reader, I would like to apologize first.

The value of impact factor is largely based on how many times articles in a certain journal is cited. However, take the fully-robust White standard errors as an example, every research that uses statistics must use this method as well, but how many of them would cite White's original, i.e., "White, Halbert (1980), "A Heteroskedasticity-Consistent Covariance Matrix Estimator and a Direct Test for Heteroskedasticity", Econometrica 48 (4): 817–838, doi:10.2307/1912934"? I believe if everyone do so, the impact factor of Econometrica would be sky-high.

What I would like to say is that statistics, this discipline itself, has been a "tool" for all other discipline, which means there is no need to cite article which developed our tools nowadays.  Imagine that, if we really require so, that is like we are required that every paper which is involved with Calculus must cite the Newton's, Leibniz's and Riemann's the very first original ones, that would make "References" part of an article totally too multifarious and unreadable.

I agree that right now our way to compute "impact factor" is biased. Luckily, it would leave us, who study statistics, more goals could be pursued.

Sincerely. Kuan-Wei

Dear all,

I have another guess, there might be that some disciplines like Biology, Medicine, Electronic Engineering or Computer Science have much more cross-citations. I do not mean those fields are "easier", but based on one single article published, more relevant articles could be published than, say, theoretical Mathematics, because these disciplines heavily based on experiments. If one researcher has done an experiment, many other researchers could push their own research one step further and then to be published. This kind of cross-citations would naturally increase the impact factors of journals in these disciplines.

I have not yet check if there is data supporting this guess. How do you think about this explanation? I believe the empirical work of this topic could be a very interesting and worthy to be published in our field, i.e., Statistics. Maybe we could start with the ratio of cross-citation of journals in different disciplines.

Sincerely, Kuan-Wei

From Gravity, Brian Clegg, St. Martin's Press, 2012, he notes a translation of Roger Bacon's Opus Majus.  The translation was by Robert Belle Burke, University of Philadelphia Press, 1928, reprinted by Kessinger Publishing, 1942, and on page 116 of that translation, Clegg found something I think to be of interest to our thread here. 

Roger Bacon (circa 1214 - 1294) was a friar, constrained by the church, and at a time when ancient Greek philosophy still held sway, so his ideas of counting on scientific observation were unconventional, partly hidden, and ground breaking. 

Of relevance to our thread, on page 24 in Gravity, Clegg quotes from the above noted translation of Opus Majus:

"Of these sciences the gate and the key is mathematics ... he who is ignorant of [mathematics] cannot know the other sciences nor the affairs of the world ...." 

Very interesting question, James.

Let me just add to what others have said: statistics is not the only discipline where IF tends to be low. In a field like linguistics, most journals have IF below 1, and even the most prestigious ones tend to hover between 1 and 2. The morale is that IF can't really be meaningfully compared across disciplines.

In a sense, there is a bias against many fields (not just statistics) inherent in the IF system. IF as a measure of anything is highly underspecified, until you know something about the overall publishing tradition of a field (preferences for books, journals, conference papers), the ageing of knowledge in the field (in e.g. linguistics you can still find references to very old research that is still relevant), and the degree of fragmentation into "schools" which obviously impacts citation practices. Even with this information in mind, there is the obvious risk that journals / publishers are gaming the system by manipulating the denominator of the IF fraction.

As researchers, I think the best thing we can do with IF is to ignore it. Obviously, that's not the same as saying that all journals are equally good. But I think researchers are fully capable of making that judgement without the aid of a fairly arbitrary IF score.

In my humble opinion its a contest b/w theory and application. Finding a new statistical technique isn't really news outside academia. It is probably too abstruse for people to understand. On the other hand medical journals have high IFs because the research published in them can be shown to affect us all in some way. A cross b/w the two like bio-statistics/healthcare big data is probably a good compromise?

Like most statistics, the IF is a summary measure and generally speaking an indicator of impact, prestige, and influence that journals have in the real world where we do our work. It is one statistic and naturally promotion committees look at numerous factors in evaluating researchers' productivity and their impact in the field. In the medical and scientific research environment, the IF is talked about constantly among researchers because it does reflect, to some degree, the amount of influence a journal has on the research world and to an extent on the public and society. As a statistician myself, I do appreciate its limitations and the fact that it is not intended to be used as an objective comparison between different fields. However truth be told, 99% of researchers in most all fields talk about Impact Factors and certainly try hard to publish in journals with the highest IFs. Not a perfect statistic, but what is? 

David, et.al. -

For those statisticians in the "real world," who hold positions that require innovation, in fast paced situations, journal involvement is often a luxury.  If you have to document your advances less formally, that is what you do.  (A friend, and at the time colleague a couple of years ago, was the one who suggested I load papers on ResearchGate to keep them readily available for use.)  But for statisticians who might not be too busy for a substantial amount of journal writing, if they are writing for statistics journals rather than, say, medical journals, the impact factors are still exceedingly low for statistics journals.  A statistician who helps out on a medical journal article with 9 other people is not in the same position as one or two statisticians coming up with a way to problem solve or present other practical, or theoretical statistics papers.  A medical article given 25 points to 10 people, generates 250 points. A statistics article given 0.95 points for 2 people generates 1.9 points.  That's 250.0 compared to 1.9, and the discrepancy is often greater. 

Yes, no 'statistic' is perfect, but ...

Cheers - Jim

Hi Jim, 

You have a major point here and I do agree with you. It is not automatically clear how much contribution individual people have in terms of innovation or creativity on any given paper. If you have written a statistics paper by yourself (or for that matter an economics paper or biochemistry paper or mathematics paper or psychology paper) then it stands to reason that the ideas are your own and if it's a new method or something novel in the field then all the more reason to feel satisfied that you have contributed in finding a new road or potentially a new discovery. However it does not necessarily follow that most of the times this is the case. On the other hand, you may be on a paper with 3 or 5 or even 9 other people and may in fact have contributed a large amount of the work and creativity, it just happens to be the case that in clinical and often basic science work more authors are listed and included than in statistical articles due to a higher incidence of piggy-backing. The only point I am trying to make is that there is quite a lot of variation in citations between fields (and there are several reasons for this) and that the number of authors may have or may not influence how much an individual has contributed to the actual study or project or publication. To my knowledge, the IF is not something "given" to individuals, not exactly. IT represents the "impact" based on citations that a particular journal carries. People have tried to develop an individual impact factor in terms of their work being cited, such as the h-index, but I am not sure how much this is recognized in various fields.

Regards,

-David