Absence of evidence is not evidence of absence is an adage which denotes that not having proof that something exists is different from having proof that it doesn’t exist. For example, lack of research about the efficacy of a new medical treatment (i.e., absence of evidence), is different from research showing that the new treatment is ineffective (i.e., evidence of absence).
This concept plays a role in many contexts, so it’s important to understand it, as well as common ways in which it’s misapplied. As such, in the following article you will learn more about why absence of evidence is not evidence of absence, and see what it means for you in practice.
Examples
In medical research, a study that fails to find a statistically significant effect for the difference between the treatment/control groups might claim to have shown that the associated effect doesn’t exist (i.e., that there is evidence of absence). However, the findings might actually reflect absence of evidence instead, for instance if the study’s sample was too small for it to detect the effect at a statistically significant level. Such issues can occur in many types of studies, including clinical trials and meta-analyses, and can lead to the wrongful rejection of useful medical treatments.
Another example of this distinction is that, if a patient gets a negative result from a medical test, then that could be interpreted as evidence of absence for the condition that the test is meant to identify. However, if the test has a high false negative rate, then it might be more reasonable to interpret its result mainly as absence of evidence instead.
In addition, the distinction between absence of evidence and evidence of absence also plays a role in many fields beyond medicine, such as physiology, neurobiology, ecology, environmental science, geoscience, astronomy, linguistics, psychology, anthropology, forensics, law, bioethics, philosophy, and statistics. For example, lack of evidence that two ancient cultures had contact doesn’t necessarily mean that they didn’t have contact, especially if there’s very little evidence about these cultures in general. Similarly, lack of evidence that a person was at a certain crime scene doesn’t necessarily mean that they weren’t at the scene, especially if there’s little evidence available at the crime scene for some reason (like a fire).
Note: This adage originated in an 1887 scholarly paper by Reverend William Wright (who phrased it as “absence of evidence is not evidence”). Its current formulation was coined in an 1895 article by Thomas Sheppard (who attributed it to scientist Dugald Bell), and was popularized in the 1970s by Martin Rees and Carl Sagan (among others).
Absence of evidence vs. evidence of absence evidence
There’s an important distinction between absence of evidence and evidence of absence (or negative evidence):
- Absence of evidence is the lack of evidence on something. For example, this happens when no studies have been conducted on the effectiveness of a certain medical treatment.
- Evidence of absence is evidence showing that something doesn’t exist. For example, this happens when studies show that a certain medical treatment has no effect compared to a placebo. It’s sometimes also called negative evidence, a term that’s also used to refer to evidence which shows that something is false.
This distinction can also be described as the difference between not knowing if there are footsteps in the snow outside because we haven’t looked at the snow (which reflects absence of evidence), compared to knowing that there are no footsteps because we’ve looked at the snow and it’s completely smooth (which reflects evidence of absence).
However, the distinction between these two concepts is sometimes unclear, for example when studies provide inconclusive evidence as to whether an effect exists, which can be interpreted as absence of evidence, evidence of absence, or both. Accordingly, there’s sometimes disagreement regarding what constitutes absence of evidence as opposed to evidence of absence, and the original adage on this has been interpreted in several different ways.
Note: There’s also an important distinction between confirmatory evidence, the absence of which is sometimes wrongly interpreted as meaning that something must be false, and falsifying evidence, the absence of which is sometimes wrongly interpreted as meaning that something must be true.
When absence of evidence is evidence of absence
Absence of evidence is sometimes interpreted as evidence of absence (i.e., as negative evidence), if someone searches for evidence but doesn’t find it. The use of failure to find evidence as evidence in itself is referred to as negative inference, absence-based inference, and inference-from-absence arguments.
For example, if someone searches extensively for information about the types of medals awarded by ancient Romans—by using relevant documents and archeological findings—and finds no evidence that they awarded medals posthumously, then this absence of evidence could be interpreted as evidence of absence.
This is also illustrated in the following classic example from a Sherlock Holmes story:
Inspector Gregory: Is there any point to which you would wish to draw my attention?
Sherlock Holmes: To the curious incident of the dog in the night-time.
Inspector Gregory: The dog did nothing in the night-time.
Sherlock Holmes: That was the curious incident. [Holmes recognized that the dog would have likely barked upon encountering a stranger at night, so their failure to bark means that the perpetrator likely wasn’t a stranger.]
— From “Silver Blaze” in “The Memoirs of Sherlock Holmes” by Arthur Conan Doyle (1893)
Similarly, the absence of evidence of the existence of animals such as unicorns can be interpreted as evidence of absence, since we would expect to have found evidence of their existence by now.
This can be represented through the Bayesian account of confirmation, in which case absence of evidence serves as evidence of absence, since it’s more likely that you haven’t seen any unicorns if none exist than if some exist, so if you haven’t seen any unicorns, that supports the hypothesis that there are none. Nevertheless, absence of evidence often provides only weak evidence of absence, depending on factors such as the quantity and quality of evidence that was gathered.
In addition, it’s possible to draw inferences not only from absence of evidence, but also from paucity of evidence. For example, the point when humans began using fire can be based not only on absence of evidence of the use of fire during a certain period, but also on the there being only limited and low-quality evidence to the contrary, which might be dismissed as an error.
Note: When assessing a study’s null or negative results based on statistical significance, it’s possible to consider various additional factors, such as the sample size (and the associated statistical power), the magnitude of the effect size, and the width of the confidence intervals.
Absence of evidence and arguments from ignorance
Arguments from ignorance suggest that if something hasn’t been proven to be false, then it must be true (and vice versa). This often revolves around using absence of evidence as supposed evidence of absence, where something can be assumed to be true or false based simply on lack of evidence to the contrary. This can be captured through the following logical structure:
We don’t have evidence that P is true, therefore P is false. (and vice versa for the true/false values)
Or:
P isn’t known to be true, therefore it must be false. (and vice versa for the true/false values)
For example, such an argument can be as follows:
There’s no evidence that this treatment is unsafe, therefore it’s safe.
These arguments are often fallacious, but they can also be reasonable in some cases, depending on factors such as what they’re based on and how they’re phrased, though they’re still usually inconclusive. For example, the following could be considered a fallacious argument from ignorance, which mistakenly assumes that absence of evidence is evidence of absence:
We don’t have evidence that P is true (although we haven’t looked for evidence either), therefore P is false.
Conversely, the following is a logically sound argument, involving the same concept:
We don’t have evidence that P is true (after searching extensively, in a systematic and valid way), and we would have expected to find evidence if it is, therefore it’s more likely than not that P is false.
Epistemic closure
Arguments that are based on absence of evidence (especially arguments from ignorance) are sometimes considered a form of defeasible reasoning, meaning that they’re rationally compelling but not deductively valid. This means that although the premises of the argument support its conclusion, the conclusion may later be proven false, based on additional information.
A key question regarding such arguments is whether there’s epistemic closure, which occurs when all the relevant information that exists has been gathered and considered, so the search for evidence is fully complete. In cases where this happens, an argument from ignorance might be considered deductively valid.
To illustrate, consider that we want to know whether our notebook is in one of 5 rooms or not.
- If we haven’t searched any of the rooms, and we say “we don’t have evidence that it’s in the rooms, therefore it’s not there”, we’re assuming that absence of evidence is evidence of absence in a fallacious way.
- If we searched 3 of the rooms, “we don’t have evidence that it’s in the rooms, therefore it’s not there”, our argument is stronger now, since it’s based on an attempt to find evidence. However, it’s still not deductively valid, since there’s no epistemic closure; we haven’t collected all the possible information (i.e., we haven’t searched all the room). This argument is also still fallacious, since we assume that the wallet is necessarily not there, even though we can’t know this for sure. We can make this argument logically sound by instead saying “after checking more than half the rooms, we don’t have evidence the wallet is there, so it’s more likely than not that it isn’t”.
- If we searched 4 of the rooms, the same applies as above, except our evidence is stronger now, and we can be more certain about our conclusions (though still not entirely certain).
- If we searched all 5 rooms, and still haven’t found it, then we collected all the relevant evidence, and reached epistemic closure (assuming our search was perfect). This allows us to conclude with certainty that the wallet isn’t in any of the rooms.
There is also the question of whether you would expect to have certain evidence or not. For example, consider the following argument, which can be considered logically sound and deductively valid:
Premise 1: If it was raining here now I’d see it.
Premise 2: I don’t see that it’s raining here now.
Conclusion: It’s not raining here now.
Conversely, consider the following argument:
Premise 1: If it was raining now on Jupiter then I wouldn’t see it.
Premise 2: I don’t see that it’s raining on Jupiter now.
Conclusion: It’s not raining on Jupiter now.
This argument is fallacious, since the lack of evidence in this case (i.e., not seeing that it’s raining on Jupiter now) doesn’t tell us whether it’s raining or not, since we wouldn’t expect to be able to see it in the first place. When such arguments are used in reality, the first premise is often implicit, as people don’t acknowledge that they wouldn’t expect to have the relevant evidence regarding the claim that they’re making.
In addition, two further issues that can render such arguments fallacious. The first is false claims that epistemic closure has been reached, especially when this is done to discourage or prevent others from trying to collect relevant new evidence. The second is exaggeration of the strength of the relevant evidence, for example when people wrongly state that their search for evidence was flawless.
Related fallacious arguments
The claim that absence of evidence is not evidence of absence is sometimes used in a fallacious manner. For example, if a medical company fails to find benefits to their product after extensive testing, they might say that “absence of evidence is not evidence of absence” to suggest that their evidence of absence is merely absence of evidence, and wrongly imply that their product could be beneficial.
Similar claims are often used by proponents of pseudoscience in an attempt to silence doubters and critics. For example, they may claim that the absence of evidence that their stance is right doesn’t mean that it’s wrong even after extensive search for evidence, and then try to shift the burden of proof to critics, by asking them to provide evidence that disproves the original claim.
This is often done in the context of claims that suffer from a verifiability problem, like providing a negative or disproving something that can’t actually be falsified. A common analogy used to illustrate this is Russell’s teapot, which was proposed by philosopher Bertrand Russell:
“Many orthodox people speak as though it were the business of sceptics to disprove received dogmas rather than of dogmatists to prove them. This is, of course, a mistake.
If I were to suggest that between the Earth and Mars there is a china teapot revolving about the sun in an elliptical orbit, nobody would be able to disprove my assertion provided I were careful to add that the teapot is too small to be revealed even by our most powerful telescopes. But if I were to go on to say that, since my assertion cannot be disproved, it is intolerable presumption on the part of human reason to doubt it, I should rightly be thought to be talking nonsense.”
— From “Is There a God?” (Bertrand Russell, 1952), as cited in “The Collected Papers of Bertrand Russell”, Volume 11 (1997). Note that this idea is sometimes illustrated using similar analogies, like disproving the existence of an undetectable animal (e.g., an invisible dragon or unicorn).
Accounting for this concept
It can be beneficial to account for the difference between absence of evidence and evidence of absence, as well as when and how absence of evidence can be considered evidence of absence. For example, this can help you respond to cases where people make fallacious arguments relating to this concept, and can also help you avoid making such fallacious arguments yourself.
When doing this in relation to specific arguments, you can consider the following:
- Are you observing absence of evidence or evidence or absence? For example, is there no research about a potential effect of something (i.e., absence of evidence), or is there strong research indicating that this effect doesn’t exist (i.e., evidence of absence)?
- How was evidence searched for? For example, was there any attempt to search for it, and if so, what was its quality? You can also consider related factors, such as the quality and quantity of the evidence that was found, and how strong the evidence would need to be to fulfill the burden of proof.
- How is the argument phrased? For example, does it imply certainty, or does it acknowledge any uncertainty involved? Does it acknowledge the relevant burden of proof, or does it attempt to shift it in a fallacious way?
Summary and conclusions
- Absence of evidence is not evidence of absence is an adage which denotes that not having proof that something exists is different from having proof that it doesn’t exist.
- For example, lack of research about the efficacy of a new medical treatment reflects absence of evidence, and is different from research showing that the new treatment is ineffective (which is evidence of absence).
- Absence of evidence can be considered evidence of absence when a thorough search finds limited or no evidence of something. For instance, if thorough archeological digs find no historical artifacts in a certain region, then that can be taken as evidence that humans haven’t lived there.
- When evaluating potential evidence of absence, consider how thorough the search for evidence was, as well as the quality and quantity of evidence that was found (if any).
- This adage is often used fallaciously, like when a company thoroughly tests their product, fails to find benefits to using it, and misrepresents this evidence of absence as absence of evidence to make it seem like the product might be useful.