The opposite of the decline effect?

The “decline effect” refers to scientific effects that appear to decrease in magnitude as more studies are conducted. For instance, early studies of some phenomenon might report large differences between treatment and control means, but later studies report small differences or no difference. Decline effects might arise because of publication bias, regression to the mean, and changes over time in true effect sizes. Reviews in ecology and other fields suggest that decline effects are common (for ecology see Jennions & Møller 2002, Barto & Rillig 2012).

Question: what should you call the opposite of a decline effect? And do you know of any examples?

No, I’m not being contrarian for the sake of being contrarian, I have a reason for asking. As I said in my talk at #ESA2020, I have reason to think that, in ecology, decline effects are no more common than, um, whatever the opposite of a decline effect is. So I want to know about any other reports of the opposite of a decline effect. Whether anecdotal, or from systematic reviews.

Just off the top of my head, it seem like heath effects of air pollution might be an example? I’m no expert, all I know is what I read in the news, but recent estimates seem to suggest that air pollution is worse for us than we used to think it was.

9 thoughts on “The opposite of the decline effect?

  1. The examples that pop to mind (tobacco and cancer, standardized testing, climate change) may be more to do with when effects become recognised/popularised outside of the field, rather than how primary literature has reported on effect sizes. I wonder how these things themselves (scientists and public perception) influence one another with regards to both decline effects and the opposite.

    Another thing is the evolving fashionability of research. People like Turing, and Belousov+Zhabotinsky gave some pretty compelling evidence for nontrivial spatial and temporal dynamics in simple chemical systems. But it was decades later before this was popular enough for experiments to really show these things clearly (and the jury is still out on how ubiquitous they are in nature).

    • One difference between climate change and the tobacco/cancer example is that climate change is a case where the world changed. Antibiotics would be another example. Our estimates of the effect size of antibiotics on bacterial growth presumably have declined over time because antibiotics really have gotten less effective over time. That strikes me as importantly different from cases in which our effect size estimates change over time even though the world isn’t changing.

      I don’t know the standardized testing example. Are you thinking of the Flynn effect?

      • I’m not familiar enough with the literature, but (anecdotally) I have friends who told me that a sizeable chunk of US education literature in/before the 70s showed better results in organising schools around standardized testing, whereas many things since have pointed in the opposite direction, with increasing fervour (if not effect sizes!)

        I think this may be similar to the antibiotic example though as the literature itself changes policies, which in turn influences subsequent studies. I don’t think any of these are “clean” examples of what you are looking for.

    • I was waiting for “jokes based on words that rhyme with ‘decline'” to start rolling in. I was getting worried it would never happen. 🙂

      “beeline effect” (can’t think of a good joke for that one)
      “undermine effect” (that one’s synonymous with the decline effect)
      “sublime effect” (bit of a loose rhyme, that one, but surely someone can make a good joke of it anyway)
      “pantomime effect” (“THE TRUE EFFECT SIZE IS BEHIND YOU!”)

    • Oh boy, there goes my afternoon:

      “fine effect” (that’s what all authors think their effect size estimates are–fine)
      “malign effect” (any effect size estimate biased by demonic intervention)
      “divine effect” (effect size estimates in creationism research. Or effect size estimates biased by divine intervention.))
      “supine effect” (a really large recline effect)
      “Quine effect” (“two dogmas of effect size”)
      “wine effect” (the jokes practically write themselves)
      “whine effect” (the jokes definitely write themselves)
      “resign effect” (I’m not touching that one)

  2. Pingback: Friday links: free online textbook on ecology with R, blogging lives again (?), and more | Dynamic Ecology

Leave a Comment

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.