Which Justices of the California Supreme Court are most likely to be in the majority in civil cases which draw dissents? Or to put it in a more practice-driven way – if you’re expecting a contentious case, which Justices’ votes are the likeliest indication of how the majority might come out? Last time, we reviewed the data for civil cases between 1990 and 2004. Now, let’s bring the data all the way up to the present day – 2005-2018.
In Table 766, we review the data for six of the thirteen Justices who have sat on at least one divided civil case since 2005 – Justices Baxter and Brown, Chief Justice Cantil-Sakauye and Justices Chin, Corrigan and Cuellar. Justice Baxter’s rate was consistently quite high almost across the board, save only 2007 and 2013. In 2007, he was in the sixties, but eight times he was in the seventies (2005 and 2008-2011), eighties (2014), nineties (2006) or at one hundred percent (2012). In her last year on the Court, Justice Brown joined the majority in divided civil cases two-thirds of the time.
In our last post, we showed that Chief Justice George was one of the most reliable predictors of how a divided Court would go, routinely voting with the winning side in eighty percent or more of all cases. Chief Justice George’s successor, Chief Justice Cantil-Sakauye, has continued that trend voting with the majority in every divided case in 2012, 2013 and 2014, in 91.67% in 2016 and in the eighties in 2011 and 2017. She has been below that level only twice in her tenure: 75% in 2015 and in the outlier year of 2018, when there were virtually no divided civil cases, 50%. Justice Chin’s rate was a bit below the Chief Justice’s, but fairly consistent with the early years of his tenure: nine times he was been at seventy percent or more agreement with the majority in a divided civil case – in the seventies in 2005, 2008, and 2010-2011, in the eighties in 2009, 2012, 2014 and 2017, and at 100% in 2006. Only five times has he been below seventy percent – in the sixties in 2007 and 2016, the fifties in 2015 and 2018, and at forty percent in 2013. Justice Corrigan’s numbers are quite similar. Nine times she has been at seventy percent or more – four times in the seventies (2007-2008, 2011 and 2017), three in the eighties (2006, 2009 and 2012) and twice at 100% (2014 and 2018), Only four times has she been below the seventy percent level – twice in the sixties (2013 and 2016) and twice in the fifties (2010 and 2015). Justice Cuellar joined the majority in divided cases seventy-five percent of the time in 2015, 100% in 2016, seventy percent in 2017 and half in 2018.
Next, we address the last seven seats on the Court – Chief Justice George, Justices Kennard, Kruger, Liu, Moreno and Werdegar, and a composite number for the Justices pro tem sitting for a single case where there’s an unfilled vacancy or a recusal.
As we noted above (and last time), Chief Justice George’s majority rate was consistently very high during the years 1990 to 2004. That continued in the final years of his tenure. The Chief Justice was at 80% in 2005 and 77.78% in 2010, but in between those years, he was at 92.86% in 2006, 95.24% in 2007 and 100% in 2008 and 2009. Reviewing the years 1990 to 2004, we showed that Justice Kennard’s rate was fairly low during the earlier part of the period and fairly high during the latter part. That was true again from 2005 to her retirement. Between 2005 and 2009, Justice Kennard was in the forties once (2006), the fifties three times (2050, 2007 and 2009) and the sixties only once (2008). But after that, her rate reached the seventies three times (2010-2011 and 2013) and 100% in 2014. Justice Kruger’s rate of voting in the majority has been slightly lower than expected – seventy-five percent in 2015, 58.33% in 2016 and sixty percent in 2017.
Justice Liu’s rate of agreement has also been comparatively low – 71.43% in 2012 and 2014, 50% in 2015 and 2018, 70% in 2017 and 83.33% in 2016. Justice Moreno’s rate of voting with the majority in divided civil cases was surprisingly low too. In four of the eight years of the period where he participated in divided civil cases, his rate of voting with the majority was between fifty and sixty percent (2006 and 2008-2010). Overall, Justice Werdegar’s agreement rate was fairly high; she was in the seventies four times (2008, 2012, 2015-2016), in the eighties three (2005, 2009 and 2017) and at 100% twice (2011 and 2013). On the other hand, she was between forty and sixty percent four times – in the forties once (2014), in the fifties twice (2006 and 2010) and in the sixties once (2007).
Finally, we have the pro tems. Pro tems generally voted with the majority in the few divided cases in which they sat during these years: 100% in 2006-2007, 2009-2010 and 2012, 75% in 2013, 66.67% in 2011 and 2016-2017 and 50% in 2005 and 2018. The only outlier was 2014, when pro tems joined the majority in only 25% of divided cases.
At the outset of our last post, we wondered whether it was possible to fashion the beginnings of a model for close cases by combining questioning data from the oral arguments with agreement-with-the-majority rates for divided decisions. Does it actually work?
There is insight to be had here, so long as attorneys and clients keep constantly in mind the limitations of all litigation analytics. Say you’re at an oral argument. You count the questions to each side – 25 to the appellant, 21 to the respondent. Justice X, who typically averages around 6 questions per argument, asks the appellant 14 questions and the respondent 9. You consult the data and discover that Justice X votes with the majority in divided civil cases, for the most part, between seventy and eighty percent of the time.
Conclusion: the Court’s decision is likely to be an affirmance – Justice X appears to be voting to affirm, the majority seems to agree (although a close vote seems possible), and Justice X is generally in the majority when the Court is split. And that’s a reasonable prediction, which would be possible to quantify more precisely with a complete dataset.
So what’s the caution? It’s a variation on the principle of reversion to the mean. Try this example (which I’m taking from an old episode of Numbers about a random sniper case): you crumple a piece of paper and shoot it towards your trash can. You miss. But then you do it ten more times, hitting the last seven in a row. What will the next one be? It’s psychologically tempting to say “hit” – hey, you’re on a roll. But you’d be wrong. Although it’s possible to hit seven shots in a row, if you sit there long enough to take a hundred shots – or five hundred – or a thousand – your hit rate will converge very close to 50%. But that doesn’t enable us to predict with certainty what a single random shot out of the series will do.
All litigation analytics are based on datasets of hundreds of thousands of data points and up (the California and Illinois datasets on which our blogs are based are both around a quarter million data points). Across dozens or hundreds of cases, our observations will zero in on the result predicted by the database: the party questioned more heavily will generally lose, the Court will reverse in a certain percentage of its cases, and certain groups of Justices will agree quite often. But that doesn’t mean that if we pluck out a single case out of all those observations, an unexpected result is impossible.
Litigation analytics becomes more omnipresent in the profession every day, and that’s a good thing. But they do have traps for the unwary, and it’s important that lawyers (and clients) understand what the numbers do and don’t mean. Predictive algorithms are coming (and are already here, in some cases), but highly reliable ones will likely have to await further refinements in technology.
Join us back here on Thursday as we turn our attention to the numbers in the Court’s criminal docket.