Browse Definitions :


Contributor(s): Ivy Wigmore

Falsifiability is the capacity for some proposition, statement, theory or hypothesis to be proven wrong. That capacity is an essential component of the scientific method and hypothesis testing. In a scientific context, falsifiability is sometimes considered synonymous with testability.

In hypothesis testing, the null hypothesis usually states the contrary of the experimental or alternative hypothesis. The null hypothesis provides the basis of falsifiability, describing what the outcome would demonstrate, should the prediction of the hypothesis not be supported by the study. The researcher's hypothesis might predict, for example, that fewer hours working correlates to lower employee productivity. The null hypothesis would be that fewer hours working is correlated with higher productivity, or that there is no change when employees spend less time at work.

The requirement of falsifiability means that conclusions cannot be drawn from simple observation of a particular phenomenon. The black swan problem is an illustration: If a man lives his life seeing only white swans and never knows that there are any non-white swans, he might assume that all swans are white. For falsifiability, it isn't necessary to know that there are black swans but simply to understand that the statement "All swans are white" would be disproven should a single non-white swan exist.

The Austrian philosopher and scientist Karl Popper (1902-1994) introduced the concept of falsifiability in his writings on the demarcation problem, which explored the difficulty of separating science from pseudo-science.

Watch a brief tutorial about falsifiability:

This was last updated in January 2017

Continue Reading About falsifiability

Join the conversation


Send me notifications when other members comment.

Please create a username to comment.

Your definition "Falsifiability is the capacity for some proposition, statement, theory or hypothesis to be proven wrong" seems to convey the  basic idea , but it is a false impression.  It is true that a falsifiable statement can be proven wrong, but it's not the definition. It misses the most important and basic idea of the definition, which is how it can be proven wrong.  This is made clear  by the fact that you next exemplify this statement in terms of the empirical concept of statistical tests.  Falsifiability has nothing to do with empirical testing, except very indirectly in the definition of the empirical basis (i.e. the set basic statements, also called observation statements).  One key point is that the definition of the empirical basis is purely a convention, not an arbitrary convention, but still a convention, because there is no rigorous procedure to define it. This is well accepted by most, if not all philosophers. It's called the problem of the empirical basis.   For that reason alone, a rigorous definition of falsifiability must accept the set of basic statements as a parameter that is fixed in advance.  Once this set is fixed, then it can be stated rigorously.  For example, consider the statement "All unicorns are white".  This is not falsifiable, but why?  It has the same form as "All swans are white."  The reason is that "There is a  black swan now on this river" is an accepted  basic statement, by convention, whereas  "There is a black unicorn now on  this road" is not a valid basic statement.  One  might think that this is only a technical detail and the  basic idea is still that it can be empirically proven false, say using a statistical test.  This would be bad thinking. The problem with an empirical view on the notion of falsifiability is more than only this technical detail.  Consider the law "No heavy dense object  will reach the ground from a given altitude 10 times faster than another heavy dense object from the same altitude and made with the same material." We are  being very conservative here.  We know that the material does not matter and they will reach the ground at the same time.  This is a very good example of a falsifiable statement, because the story says that it was believed to be false before Galileo, but it is, of course, true.  So, at the time, if someone had said  that this law is falsifiable, nobody would have objected.  Now, the key point about the  notion of falsifiability is that it does not depend upon whether or not the theory is true.  If the law is true, then it is not possible at all to contradict the law. In this example, it is clear that nobody will ever succeed to describe an experiment with two objects that contradicts the law.  So, the definition is not "the capacity ... to be proven wrong" when this capacity is interpreted empirically.  At the end, you say something along these lines:  "For falsifiability, it isn't necessary to know that there are black swans but simply to understand that the statement 'All swans are white' would be disproven should a single non-white swan exist."  Yes, but you only say it at the end, and it is not consistent with your interpretation of the definition in terms of  empirical tests. This statement should appear earlier, because it is an essential part of the basic idea of the definition.  But,  it only help to convey this basic idea.  It does not address the issue of the empirical basis. For example, I could say that "All unicorns are white" is falsifiable,  because  we all understand that the statement 'All unicorns are white' would be disproven should a single non-white unicorn exist."   You see, the basic concepts are missing. It is an insult to the genious of Popper to say that we can convey even only the basic ideas  without mentioning at the least informally, in some way, the notion that some statements such as statements about unicorns are not (valid) basic statements to falsify a law and that a statement is falsifiable if it is contradicted by some (valid) basic statements.
When I say that "All unicorns are white" is not falsifiable, I assume that there is no  convention to accept this basic statement as an empirical statement. Of course, as I said, this is a question of convention. Perhaps, some people will accept as convention that a unicorn is a valid empirical concept. For example, any animal that looks like a horse, but has a horn on his head might be their convention for what is an unicorn. Of course, the magical aspect is missing, so it is unlikely to be a well accepted convention.  Anyway, for the point that I am making here, we simply have to assume that there is no accepted convention on what is a unicorn that is also accepted as empirical.
I meant, I assume that there is no convention to accept a basic statement such as "this unicorn now on the road is  black" as empirical.


File Extensions and File Formats


  • California Consumer Privacy Act (CCPA)

    The California Consumer Privacy Act (CCPA) is legislation in the state of California that supports an individual's right to ...

  • compliance audit

    A compliance audit is a comprehensive review of an organization's adherence to regulatory guidelines.

  • regulatory compliance

    Regulatory compliance is an organization's adherence to laws, regulations, guidelines and specifications relevant to its business...


  • privilege creep

    Privilege creep is the gradual accumulation of access rights beyond what an individual needs to do his job. In IT, a privilege is...

  • BlueKeep (CVE-2019-0708)

    BlueKeep (CVE-2019-0708) is a vulnerability in the Remote Desktop (RDP) protocol that affects Windows 7, Windows XP, Server 2003 ...

  • endpoint detection and response (EDR)

    Endpoint detection and response (EDR) is a category of tools and technology used for protecting computer hardware devices–called ...



  • disaster recovery team

    A disaster recovery team is a group of individuals focused on planning, implementing, maintaining, auditing and testing an ...

  • cloud insurance

    Cloud insurance is any type of financial or data protection obtained by a cloud service provider. 

  • business continuity software

    Business continuity software is an application or suite designed to make business continuity planning/business continuity ...


  • Hadoop as a service (HaaS)

    Hadoop as a service (HaaS), also known as Hadoop in the cloud, is a big data analytics framework that stores and analyzes data in...

  • blockchain storage

    Blockchain storage is a way of saving data in a decentralized network which utilizes the unused hard disk space of users across ...

  • disk mirroring (RAID 1)

    RAID 1 is one of the most common RAID levels and the most reliable. Data is written to two places simultaneously, so if one disk ...