Why do statistics journals generally have low impact factors, long publication times, and low acceptance rates?

Question

I've been investigating statistics journals to submit my work to (specifically, an application paper). Many journals are specialized, others quite general.

However, I've noticed that top journals in the field (e.g., sub-journals of the Journal of the Royal Statistical Society (RSS)) generally exhibit the above traits. In many cases, the impact factor is less than 2. In certain cases, the time from submission to publication is 1.5-2 years in duration. In the majority of cases, the acceptance rate is ~10% or lower. Some of the most celebrated contributions to statistics have been published in this way. This trend has been previously investigated by van Nierop (2009) for example (Title: Why do statistics journals have low impact factors?). Interestingly, the Journal of the American Statistical Association (JASA) currently enjoys an impact factor of 3.7, but long times between submission and publication. A similar trend exists for RSS Series B: Statistical Methodology.

Long wait times appear to explain why impact factors are so low, but the association is not causal.

For important work, these trends can be discouraging. A solution however exists in the form of preprints to avoid potentially being "scooped".

Could someone weigh in on this? CV has many seasoned statisticians who've likely served on journal editorial boards at some point in their career.

Reference

Van Nierop, E. (2009), Why do statistics journals have low impact factors?. Statistica Neerlandica, 63: 52-62. https://doi.org/10.1111/j.1467-9574.2008.00408.x

Answer 1

In my mind, the low impact factor of journals in the mathematical sciences is primarily due to the fact that the impact factor only takes into account citations within the first two years after publication. Many statistics and mathematics papers are well cited, and have a very long "shelf life" in that they are still cited many years or decades after publication. In contrast, other fields are more "broad" rather than "deep" in that new knowledge broadens what a field knows rather than deepens it, and in those cases even the most important papers typically have a steeper drop-off of the number of citations over time.

In other words, I believe that the issue is not reflective of something specific statistical/mathematical, but reflective of the way the impact factor is computed.

Answer 2

These are two issues. Long turnover times can indicate difficulty to find good reviewers and/or serious scrutiny of the work. That this is so for a statistics journal may indicate that the field is looking for robust, long term contributions.

Impact factors should always be interpreted in the context of a scientific domain. For example medicine has staggeringly high impact factors (with journals way beyond science or nature) making an impact factor that most other academics would consider very high, rather mediocre in that field. Frequently impact factor percentile within a given category is therefore used, where generally q1 (top 25%) within the category is considered a good impact. Perhaps your journal scores high within the stats category, something to check

Answer 3

Why do statistics journals generally have low impact factors

If you flip the question it makes more sense: that is, the real question should be: Why does "impact factor", a metric ostensibly designed with the express purpose of measuring impact, fail to say anything useful about the relative impact of publications in an entire, very large and impactful, scientific discipline?

In other words, the answer is that it's the fault of the metric, not of the discipline of statistics. Impact factor is just a notoriously bad metric (even within the disciplines which have high impact factors). See the very long "Criticism" section of the wikipedia article about impact factors.

... long publication times

Does this actually matter to anyone anymore these days? What matters is the time to acceptance, since that's when you can claim your publication as "in press" or "accepted", and reap any career rewards associated with publication, whether it's pending or has already occurred. As for the actual "time to publication", if you post the preprint version of your papers on arXiv or some other preprint repository, you will have no reason to care about when your papers are published.

... and low acceptance rates?

You talk about acceptance rates at top journals in statistics being 10% or lower. But these are top journals! That means they will have low acceptance rates by design. A journal that indiscriminately publishes much of the content that is submitted to it will simply not become regarded as a top journal. Conversely, in top journals a low acceptance rate is a feature, not a bug; when you are curating a top journal, your goal is to be as selective as you possibly can while filling the quota of pages you aim to publish each year, and this leads naturally to low acceptance rates.

The above is true for any discipline, not just statistics. So while I don't know for sure, I'd guess that the premise of your question is simply false: journals in statistics do not generally have a lower acceptance rate than journals in other disciplines.