How willing is the US Supreme Court to declare itself wrong?

Question

In comments on this answer User Suprcat wrote (in part)

I think a big problem is that Marbury v. Madison has been applied in a way that ignores a major caveat: a court's job is to say what the law is, which means that if the court is properly doing its job what the Court says and what the law is will be one and the same, but that will be because a Court that is doing its job will write a decision consistent with the law, not because the law magically changes to match what the Court says. Courts seem to feel a sense of duty to pretend that all previous decisions have been correct, no matter how insanely twisted the law would have to be to fit all past decisions. Much better would be to recognize that Supreme Court decisions might occasionally be wrong, but that would not prevent them from being binding upon the parties involved, or in some cases providing some level of protection from those would rely upon them.

Historically, how willing has the US Supreme Court been to say that its previous decisions were wrong? How willing are US courts to say that previous discussions are right, even if they are inconsistent with current decision? How much of this attitude comes from the Marbury v. Madison case, in which for the first time the US Supreme Court held an act of Congress void as being without Constitutional authority? How does this compare with this attitudes of courts in other common-law countries?

Answer 1

As for SCOTUS being willing to overrule itself, here is a table of such cases, starting with Hudson v. Guestier 10 U.S. (6 Cr.) 281 (1810) which overturned Rose v. Himely, 8 U.S. (4 Cr.) 241 (1808) up to Ramos v. Louisiana, No. 18-5924 (U.S. Apr. 20, 2020) which overturned Apodaca v. Oregon, 406 U.S. 404 (1972) (plurality opinion) and Johnson v. Louisiana, 406 U.S. 366 (1972) (Powell, J., concurring). This is 234 cases since 1798, slightly above once a year. No idea what the rate is for the UK or Canada.

Answer 2

Less than 1% of the time

From this article:

In my forthcoming book, “Constitutional Precedent in Supreme Court Reasoning,” I point out that from 1789 to 2020 there were 25,544 Supreme Court opinions and judgments after oral arguments. The court has reversed its own constitutional precedents only 145 times – barely one-half of one percent.

However, most of those reversals have been in the last 100 years. Partly this is because you can’t overturn precedent until you have a precedent and you can’t have a precedent until an issue has been litigated for the first time. However, the court is a creature of its time and there has been more social and technological change in the last 100 years than the previous 250 and so more opportunities for historical precedent to fall out of step with community expectations. Because the precedent is arguably wrong more people are willing to argue it and the court is more likely to want to hear those arguments creating more opportunity for the precedent to be overturned or, at the least, narrowed in scope.

This, of course,leads to an issue of semantics. I haven’t read the above book nor all 145 decisions that reversed “constitutional precedents” and certainly not the 25,000+ that didn’t. So:

• I don’t know if the author is excluding cases that overturn common law as opposed to constitutional precedents. He probably is. However, if the US Supreme Court is like the Australian High Court, they are far more likely to revisit a constitutional decision than a common law one. This is on the basis that the latter can be changed by statute while the former can’t - if the legislature doesn’t agree with the courts decision on common law, they can change it but “the people” are stuck with decisions on constitutional law. While judges have the power to “find” (make) law in Common Law systems most are uncomfortable exercising it as it isn’t democratic.
• The numbers quoted relate to reversals, courts will often distinguish cases on the facts which don’t meet that requirement. In some cases the distinguishing is so broad that the facts of the original case become virtually impossible to replicate. That is, the original precedent has no practical application but it hasn’t been reversed.

Answer 3

Building on the fascinating data set shared by user6726, we can see that there are some noticeable patterns in how often the Supreme Court has been willing to reverse prior decisions. Since it's a broad question covering many different eras of American law, it might help add some context.

---

First, let's start with frequency: is there any pattern to how often the Supreme Court decides to overrule previous decisions? A few things stand out immediately: changes tend to cluster together and the overall rate increased dramatically since WW2. It also seems to coincide with major wars quite often (though that doesn't necessarily mean they are related).

You'll notice that the last 20 years appear to be lower. We can explain this by looking at the distribution of the time gaps - how often are they overruling older decisions compared to more recent ones? This can be shown using a cumulative distribution function. The blue curve shows what fraction of cases were overruled in less than X years (e.g. we can see that both reach ~0.5 within 20 years, so half of cases take less than that).

We can also just plot the overruling date vs the original date directly. This essentially combines the information in the previous two plots. The important things here are the density of points and the vertical distance from the black line (the farther below, the longer it took to overrule).

---

There were two things that really stood out to me:

1. Most overruled decisions are recent. This honestly surprised me, I assumed that the list would be dominated by archaic laws from a different era. It turns out that, according to this data set, 50% of overruled decisions were made within the previous 20 years and only 15-20% were older than 50 years.

2. Change comes in (increasingly frequent) bursts. We can visibly see several clusters, such right after the Civil War, during both World Wars, and throughout the 60s/70s. The 80s and 90s show a noticeable decline, but the frequency is still much higher than anything pre-WW2.

Based on this, it seems like the Supreme Court since 1940 has been far more willing to overrule earlier decisions (often relatively recent ones). These changes tend to come in bursts that appear to coincide with transformative events (like major wars and cultural shifts), although that becomes harder to tell from 1960s onward due to the much higher baseline rate.

Answer 4

I thought I'd tackle the comparative law portion of your question and look especially at Canada and the UK. I'll caveat that I'm only looking at explicit breaking of precedent, there's always room for semantic arguments in some cases whether a precedent has been broken in practice, or narrowed to the point of being wholly inapplicable.

The numerical analysis is for the 38.5 years between July 1966 – December 2004 for all jurisdictions for fair comparison. The start date is due to the UK House of Lords changing its stance on overturning its own precedent; the end date is a practical consideration of my Canadian source ending at that date.

Note that my methods aren't exactly 100% accurate, one could argue over the inclusion/exclusion of some cases, but this was more an interesting bit of research prompted by an interesting question.

UK - about 0.6%

In the UK, the House of Lords for much of its history stuck firmly to its own precedent as a rule. The outlook changed in 1966 when they issued the Practice Statement in July 1966, whose key part is as follows:

Their Lordships nevertheless recognise that too rigid adherence to precedent may lead to injustice in a particular case and also unduly restrict the proper development of the law. They propose therefore, to modify their present practice and, while treating former decisions of this house as normally binding, to depart from a previous decision when it appears right to do so.

That criterion is quite vague, but this was expanded upon as quoted in A v Hoare [2008] UKHL 6 at para. 20:

I find the reasoning compelling and therefore consider that Stubbings was wrongly decided. But that is not in itself a ground for departing from it. The Practice Statement (Judicial Precedent) [1966] 1 WLR 1234 was intended, as Lord Reid said in R v National Insurance Comrs, Ex pp Hudson [1972] AC 944, 966, to be applied only in a small number of cases in which previous decisions of the House were "thought to be impeding the proper development of the law or to have led to results which were unjust or contrary to public policy." Lord Reid also observed, at p 966:

"It is notorious that where an existing decision is disapproved but cannot be overruled courts tend to distinguish it on inadequate grounds. I do not think that they act wrongly in so doing: they are adopting the less bad of the only alternatives open to them. But this is bound to lead to uncertainty for no one can say in advance whether in a particular case the court will or will not feel bound to follow the old unsatisfactory decision. On balance it seems to me that overruling such a decision will promote and not impair the certainty of the law."

According to the above linked Wikipedia article, the House of Lords has indeed been fairly restrained, explicitly invoking the Practice Statement only 21 times up to the replacement of its judiciary functions by the UK Supreme Court in October 2009.

However, following the citation reveals a table of 21 cases where not all of them are marked as an explicit overruling. Sampling a few of the cases, I'm inclined not to count the four that are labelled "Declined to follow" and the unique DPP v Camplin case which doesn't seem to even mention the Practice Statement. Removing those, and the two which occur post-2004 leaves 14 cases.

From a rough and quick count of House of Lords cases from 1997-2008, they rendered an average of 62 judgments annually. Extrapolating that to the whole timeframe results in approximately 2390 cases, for a precedent breaking rate of approximately 0.6%, or about once every 2.7 years.

The UK Supreme Court inherited the Practice Statement when it replaced the House of Lords' judicial functions. I'm unfamiliar with how to search the UK case law efficiently, but there's no reason I can think of that the ratio would change drastically in the last decade. The most recent overruling of precedent I could find is [2020] UKSC 47 at para. 253, though a much shorter, simpler and digestible example is [2016] UKSC 9.

Canada - about 0.9%

In Canada, choosing the start date for analysis is slightly more arbitrary, but I'll note 1966 isn't far off from important distinctions in Canada's judicial history. Until 1949, decisions of the Supreme Court of Canada (SCC) were appealable to the UK Judicial Committee of the Privy Council (JCPC). Cases that had already commenced retained the possibility of appeal to the JCPC, with the last such case heard in 1959. Furthermore, the earliest case I can find where a majority clearly confirms the court's ability to overturn its own decisions is Binus v. R. from 1967 (though such ability was not exercised in that case, and a 1961 case is cited by my source below as a departure from precedent).

Overall, the philosophy of the SCC in overturning its own decisions is similar to the UK's. A non-exhaustive list of factors is elaborated on in Dickson CJ's dissent in R. v. Bernard, [1988] 2 S.C.R. 833 (not specifically disagreed with by the majority on this point & frequently cited in future cases). Three of the four factors are not too dissimilar from UK case law: The decision is excessively harsh/unjust for the criminally accused, attenuation of a decision has already rendered it very narrow, and the decision itself creates excessive legal uncertainty which the doctrine of stare decicis is meant to prevent in the first place.

The fourth factor is specifically Canadian: Bringing previous common law more in line with the constitutional Canadian Charter of Rights and Freedoms introduced in 1982. It's difficult to account for whether introduction of the Charter would increase or decrease the court's likelihood of overturning precedent, especially since my analysis only looks at 16 pre-Charter years, but it's something to keep in mind. While the court has cautioned itself against overturning precedent favourable to the accused, I've also managed to find two cases where due to Charter influence, the court overturned precedent that was unfavourable to the accused when it arguably did not need to.

I've managed to locate Taking precedents seriously (Mark Chalmers, 2005) which provides in Appendix B an equivalent list for Canadian overturned precedents. It lists 44 cases, though for the purposes of this question I'm inclined to remove the ones marked as overturned for Charter reasons: these are highly likely to have been overturned due to the change in Constitution, meaning the original precedent was correct according to the Constitution at the time. Removing those and the single pre-1966 decision leaves 32 cases.

Note that I've decided to retain both instances of the SCC overruling its own pre-1949 decisions when it wasn't the final court of appeal, and instances of it overruling JCPC decisions which are today equivalent to SCC decisions. This is because in both instances, those decisions became the highest possible legal interpretive authority once appeal to the JCPC was removed.

From 2007-2017, the SCC averaged 70 judgments a year. Extrapolating that to the timeframe gives about 2700 cases for a precedent breaking rate of 0.9%, or about once every 1.2 years.

While I'm unable to list all post-2004 cases, I'd just like to note that the two most recent overturnings of precedent I'm aware of have also been the most wide-reaching in my opinion: Vavilov in 2019, overhauling judicial review of administrative decisions, and Jordan in 2016, setting presumptive time limits for the right to a speedy trial.

US - about 3.7%

For comparative purposes, I'll briefly run the US numbers in the same approximate fashion for my chosen timeframe.

Per the table cited by @user6726's answer, the US Supreme Court has overruled itself a comparatively astonishing 108 times in my chosen timeframe. Per Ballotpedia, between 2007-2019, the US Supreme court decided an average of 76 cases per year. Extrapolating out to the timeframe gives approximately 2930 cases, for a precedent breaking rate of about 3.7%, or about once every 0.4 years (once every 4 and a half months).

Comment

While my methods were quite approximate and per @Gumbercules's answer my timeframe happens to line up with the US Supreme Court's most active overturning time period, the disparity is so large that I believe it's safe to conclude that in the recent history starting from 1966 when the UK House of Lords allowed itself to overrule its own precedent, the US uses this ability more than either the Canadian or UK top courts.