Skip to content
Science, Maths & Technology

Statistically significant: The US Supreme Court takes a view

Updated Tuesday 12th April 2011

A lawsuit over a cold remedy has forced the US Supreme Court into deciding what might be statistically significant.

The rulings of the US Supreme Court aren't the place most people would start looking for a discussion of what the phrase "statistically significant" might mean. But in March 2011, America's highest court of justice made a decision that involved exactly that.

Contemplation of Justice: A figure outside the US Supreme Court Creative commons image Icon Clearly Ambiguous under CC-BY licence under Creative-Commons license
Contemplation of Justice: A figure outside the US Supreme Court building

The case involved a case originally brought by investors in a company called Matrixx Initiatives Inc., who manufacture a range of over-the-counter medicaments including one called Zicam Cold Remedy.

The investors alleged that Matrixx had failed to disclose material information to them, when it had received reports that people who had taken Zicam Cold Remedy had suffered a loss of sense of smell. When these reports did become public, the share price fell, thus (allegedly) losing the investors money.

People do lose their sense of smell sometimes, for reasons that have nothing to do with a drug. Matrixx argued that the number of cases of loss of sense of smell in people who had taken Zicam was not significantly higher, in the statistical sense, than the number one would expect to see if the drug had no effect on sense of smell.

For that and other reasons, Matrixx argued that they had no obligation to disclose the reports to their investors. The case depended to a major extent, therefore, on the idea of statistical significance testing.

What's all that about? I've given a brief explanation in another OpenLearn article, Confusing terms in statistics, so I won't repeat all that, but briefly the point of doing a significance test is to see whether it is plausible that some feature in data can be explained by chance alone.

In the Matrixx case, perhaps they did get more cases of loss of sense of smell than one might have expected if the remedy had no effect on sense of smell, but maybe this could have been due entirely to chance – people who were going to lose their sense of smell anyway might just have happened to take the cold remedy.

If, after performing a statistical significance test, the conclusion was that the data are consistent with this "it's all just chance" explanation, one would report that the results were "not statistically significant".

Matrixx did claim this, but went further and argued that, because they were not statistically significant, they did not have to be disclosed.

An interesting feature of the Matrixx case is that arguments to the court were not submitted only by lawyers, investment experts, and so on, but also by Deirdre McCloskey and Steve Ziliak. They are economists and statisticians who are well known for publishing a book called The Cult of Statistical Significance, which, as you might guess from the title, is not complimentary about the way that significance testing has been used.

McCloskey and Ziliak submitted an amicus brief, or more precisely an amicus curiae brief (from Latin words meaning "friend of the court"). This is a submission to the court from a person or organisation that is not one of the parties involved in the case, but which is intended in some way to throw light on the case.

They did not argue wholesale against the use of the ideas of statistical significance. Instead, they argued as follows:

If a feature in data is not statistically significant, it's reasonable to take that to mean that the feature might plausibly be due solely to the workings of chance.

But just because it might be due to chance, that doesn't mean it is due to chance. That's only one possible explanation.

There will be others, and thus a decision on whether some piece of information is relevant can't be based on statistical significance alone. The court agreed with this line in their judgment. They said quite clearly that "Matrixx's premise that statistical significance is the only reliable indication of causation is flawed."

Most, if not all, statisticians would agree with them. Statistics students often get annoyed with what seems like excessive pedantry from their teachers. The statisticians will insist that one does not accept the hypothesis that some data feature is due to chance alone.

One merely fails to reject such a hypothesis, if the results are not statistically significant.

The idea here is that "accept" sounds too much as if one is saying, "Yes, we know it's just chance". "Fail to reject" allows the possibility that it might just be chance, but allows other possibilities too.

There's reason in our pedantry. To quote another cliché, absence of proof isn't the same as proof of absence, and a result that's not statistically significant is absence of proof that something is happening that goes beyond mere chance.

A couple of aspects of this discussion that might interest you, depending on your background:

First, what is really going on with statistical experts like Ziliak and McCloskey rubbishing significance tests? Surely they are a key and fundamental aspect of the way statistics is done?

Well, not everyone would be quite as critical as Ziliak and McCloskey, but actually, many statisticians would be very critical of the unthinking and routine use of significance testing that goes on in many areas of enquiry. (If you understand a little of the technical ideas behind significance testing, you might enjoy reading another criticism from a slightly different angle, Mindless Statistics, by Gerd Gigerenzer, the well-known author on risk. [The article opens as a pdf.])

Second, some people find it surprising at first sight that the law should be concerned with ideas and methods of statistics. But if you think about it, it's not really surprising at all.

Both statistics and the law involve dealing with uncertainty – admittedly often with different goals in sight, but uncertainty is there as an inescapable aspect in both. Gary Slapper, Professor of Law at the Open University, has written about statistics and the law

Statisticians and lawyers would not always agree on what should be done about the uncertainty, but perhaps that's inevitable – the law has to come to a clear decision on every case, however much uncertainty there is, whereas we statisticians can sometimes have the luxury of saying that there's simply not enough evidence to decide.

Further reading

Discover how to study statistics or law with The Open University.

If you're interested in the legal details, the briefs submitted to the Supreme Court in the Matrixx case (including the McCloskey and Ziliak brief) are online, as are a pdf transcript of the oral argument, and the  "opinion" (i.e. judgment), again as a pdf.


For further information, take a look at our frequently asked questions which may give you the support you need.

Have a question?

Other content you may like

Sacred and Secular: Section 2 - Finding legal judgments and legal commentary Creative commons image Icon Gordon Giles under CC-BY-ND-2.5 licence under Creative-Commons license article icon

Society, Politics & Law 

Sacred and Secular: Section 2 - Finding legal judgments and legal commentary

Judgments are given by judges sitting in different types of courts and these courts exists in different types of legal system. It is useful to understand this context in order to find and analyse judgments. This section explores those differences before moving on to identify where the judgments can be found.  

Naked before the court?: The reality of human rights Creative commons image Icon mrrobertwade under CC-BY-NC-SA licence under Creative-Commons license audio icon

Society, Politics & Law 

Naked before the court?: The reality of human rights

Appearing naked before a judge, and voting while in prison: How far should respect for human rights go?

15 mins
Do crowds behave like fluids? Copyrighted image Icon Copyright: BBC article icon

Science, Maths & Technology 

Do crowds behave like fluids?

It used to be believed that crowds behaved like fluids - until Keith Still proved otherwise.

More confusing terms in statistics Copyrighted image Icon Copyright: Sunnycatty | article icon

Science, Maths & Technology 

More confusing terms in statistics

Kevin McConway explains why statisticians use everyday words for not-so-everyday concepts - and how to translate what they really mean.

Why don't statistics reveal when sports matches have been fixed? Creative commons image Icon Voo De Mar under Creative Commons BY-SA 4.0 license article icon

Science, Maths & Technology 

Why don't statistics reveal when sports matches have been fixed?

Sport is awash in data - so you'd assume it'd be easy to tell if a game has been rigged on the whim of a shady gambling syndicate. It's not that easy, though - and as David Lucy explains, it's the same problem that stops us being able to spot rogue doctors and nurses...

When should you use a stacked area chart? Copyright free image Icon Copyright free: wokandapix article icon

Science, Maths & Technology 

When should you use a stacked area chart?

... or why you should never use stacked area charts, according to Dr Drang.

The Code: Grandmaster of Mosaics activity icon

Science, Maths & Technology 

The Code: Grandmaster of Mosaics

Deep in the dusty pages of an ancient tome is a legend centuries old that only you can overcome. Do you have the ability to identify symmetries and unlock the mosaic in this clever game?


Science, Maths & Technology 

Working with diagrams

Working with diagrams is essential for students of science, technology, engineering and mathematics. This free course is packed with practical activities and tips which make learning from and with diagrams more enjoyable and rewarding. One part of this course deals with the reading of diagrams and the other part with the drawing of diagrams.

Free course
8 hrs
Exploring data: Graphs and numerical summaries Copyrighted image Icon Copyright: Used with permission free course icon Level 1 icon

Science, Maths & Technology 

Exploring data: Graphs and numerical summaries

This free course, Exploring data: graphs and numerical summaries, will introduce you to a number of ways of representing data graphically and of summarising data numerically. You will learn the uses for pie charts, bar charts, histograms and scatterplots. You will also be introduced to various ways of summarising data and methods for assessing location and dispersion.

Free course
20 hrs