Scientific Sentences of Certainty

Scientists value certainty; after all, we are in the business of knowing. But all scientists know that there is enormous uncertainty in much of what we do, that what may seem clear cut is often not, and that there remain open questions for every one that we close and move on from.

Yet, our hankering for certainty is such that our writing often sidesteps uncertainty. We have developed linguistic mannerisms that convey certainty when we introduce the topic under study, in the way we describe our methodology and its limitations, and again when we present our conclusions.

Breaking this down a bit, we often use the very first paragraph to gain the reader’s assent to unfounded assertions by phrasing them as facts. For example: “There is an urgent need to quantify neurobiological measures of depression.” We have long learned that statements in an ontological style work well. The trick is to assert that a descriptive phrase—in the present tense—is actually an intrinsic property, which therefore does not require questioning. Similar maneuvers work well for entire fields of study: “science is…” or “population health is…”, sometimes with citations of others who have said much the same thing, creating a loop and reinforcing what we think we know.

When we describe the underpinnings of our work, we then point to a sufficient number and kind of prior publications as precedents. This allows us to describe our methods, our assessment tools, as “validated” and our interventions as “piloted.” This tactic presents an interesting variation on the use of the present tense noted above; while it may seem different in form, use of past tense terms like “validated” and “piloted” suggest that statements of proof shifted to the present tense sometime in the past—presumably because the case for value has already been settled.

The slippery use of language is part of how we do things in science. But we are rarely certain, even when we try to give that impression.

We interpret results as if they were perfectly reliable. For example, we discuss treatment effects for “people with depression”—not “people who probably have depression” or “people who were diagnosed with depression on our structured interview.” As if these descriptions were identical, although they are not. We offer a veneer of certainty when we use the term “gold standard” to signal the surest possible choice, as in the randomized controlled trial is the gold standard for scientific studies. The use of “gold standard” has put the more accurate “widely accepted” or “best available” out of circulation.

And if we find positive results in our study, we write our conclusions in the present tense. As the AMA manual of style notes, the present tense is often used in discussion sections for “expressing a statement of fact or general truth.” Writing this way implicitly suggests generalizability without ever having to make the case.

Finally, when we do eventually describe the limitations of our methods, we relegate any mention of uncertainty to a list of technical statements that we make sound as if they are unlikely, rare events that we have been forced to include by the journal’s editors, akin to the mandatory warning of potential side effects at the end of any drug advertisement.

Looking to convey authority, we have developed a way of writing that elevates the aura of science. The slippery use of language is part of how we do things in science. But we are rarely certain, even when we try to give that impression. It seems then useful to be careful with our language, to ask whether the work we are describing is indeed as certain as it sounds, and to make sure our language reflects the state of certainty accurately. This may be harder for us to wrap our brains around in the short term, but it will surely be better for science in the long run.

Previously in Observing Science: The Power of Culture