how much are citations increasing year-on-year?

Question

Increasing citations of an individual or a journal across years is usually taken as an indication that this work is more impactful or gathering more interest. However, I suspect the number of journals, publications, and citations are increasing faster than the amount of publishing scientists, which means that increasing citations may not mean increasing interest. Therefore, is there a way to guess at the amount that citations would have increased over a single year, let's say 2018 compared to 2019, just as a function of increased citation and not more interest in the work relative to the previous year? One simplistic approach would be to multiply the 2019 citations by the proportion of 2018 over 2019 publications (in that field, or overall if that's not available). I welcome other ways to think about this type of question.

Answer 1

The number of citations generated per year is the product of the number of articles produced in a given year times the average number of references per article. Additionally, when it comes to observed citation metrics in tools like Google Scholar, Scopus, and Web of Science, the relevant indicators are the number of indexed articles and their average number of citations per article.

Over time, the number of articles per year has grown as publishing has become easier and as the number of publishing scientists in the world has grown. Reference lists have got longer. And many citation indexing databases have become more inclusive, especially Google Scholar.

Journal citation indicators generally focus on some variant of citations in a given period divided by number of articles published in a period (e.g., impact factor and its variants). On this metric, publishing more articles does not improve the metric, because it increases both the numerator and the denominator. Increasing the number of references per article in a given field will increase impact factor. Likewise, capturing more of the literature in the citation indexing database will also increase measured impact factor. You can see this when you compare Web Of Science Impact Factors with the Scopus Cite Scores. Scopus is more inclusive and tends to return higher values.

Author citation metrics are more complex. There are a range of metrics that get discussed. Prominent ones include h-index, total career citations, citations per year in recent years. If we focus on citations per year, this is the multiple of the number of papers an author has and the average number of citations each paper is receiving. Putting aside factors that make one author more impactful than others, the average number of citations per paper is driven by similar factors as what influences journal citation indicators (i.e., number of references per article, breadth of coverage of the citation indexing database). We then have the issue of whether authors are publishing more articles per year. My understanding is that they are publishing more per year and that the number of co-authors is also increasing substantially over time (see @anyon's comment: https://arxiv.org/abs/1704.05150).

I also think that several factors have made publishing in well-established journals more impactful over time in terms of citations. In short, as the number of lower-tier publications has expanded, this has increased the relative impact of the more established publications. So, for example, if a journal was in the past ranked 50th out of a 100, it might now be ranked 60th out of 200. Furthermore, a few of these low to mid-tier journals publish a huge number of articles per year (e.g., outlets like Frontiers, Scientific Reports, PlosOne). You could add to this PrePrint servers and the like. In summary, if you are the kind of academic who consistently publishes in top-tier journals, then the relative citation impact of those outlets has also gone up. This can be seen in the rise in impact factors over time.

Answer 2

I suspect the number of journals, publications, and citations are increasing faster than the amount of publishing scientists

This may be right to some small degree, but in the end, the number of publications can only grow faster than the number of publishing scientists if the average publishing scientist is writing more papers per year than they did in the past. It is possible that that number has increased by a factor of 1.5 or 2 over the past 50 years, but no more than that.

Likewise, the number of citations given is just the number of publications published times the average number of references a paper has. The average number of citations a paper then receives is the number of citations given divided by the number of publications -- that is, the average number of citations a paper receives equals the average number of references a paper has. (That's really only true assuming that references are given in the same year a paper is published; but it's approximately true anyway.) So the question then comes down to whether, over the past 50 years, the average number of references in a paper has increased. Again, it may have, by a factor of 1.5 or 2, but not substantially more than that, and consequently the average number of citations a paper receives will also not have risen substantially.

In other words, I think that the question posed is interesting, but that the conclusions drawn there are not correct.