By Ken Kajihiro
Data analytics is becoming more and more prevalent within the legal profession; especially, within the litigation field. Legal analytics is on the rise to provide litigators with a winning argument. Current legal analytics software utilizes artificial intelligence to determine the success rate of previous arguments before particular judges in accordance with the user’s search inquiry. But data analytics can be used for much more than just analyzing a judge’s decision tendencies.
“Data analytics is the science of analyzing raw data in order to make conclusions about that information.” Data analytics can be used in conjunction with factual evidence and the common law to formulate legal arguments. In short, stick to data the opposing counsel is not contesting – preferably data provided and conceded by opposing counsel. Using opposing counsel’s own data against them will result in a data analytics conclusion and legal argument that will be difficult for opposing counsel to rebut.
In 1968, the California Supreme Court stated in People v. Collins that “[m]athematics . . . while assisting the trier of fact in the search for truth, must not cast a spell over [them].” In Collins, using data analytics, the prosecution sought to establish that there was an overwhelming probability that the defendants were guilty because they matched certain descriptions of the suspects provided by the victim and a witness.
In dispute, however, were the very descriptions of the suspects; the victim’s description did not include a ponytail, whereas the witness’s description included a ponytail. Despite the ponytail description difference, the prosecution’s data analytics included that the female defendant had a ponytail.
In addition, the prosecution’s data analytics included various factors such as the probability that a man has a beard and the probability that a man has a moustache; however, the prosecution did not take into account overlapping categories: the probability that a man has both a beard and a moustache. These discrepancies skewed the raw data; thus, the prosecution’s argument fell flat on its face because the data analytics conclusion was nothing more than a fallacious blunder.
So, what is the solution? The solution is to use data the opposing counsel is not contesting – preferably data provided and conceded by opposing counsel. In a simple example, suppose in a False Claims Act case, Corporation ABC initially reported only $4,000 worth of government contracts; whereas, the government’s audit report found that the reported amount was supposed to be $20,000. In response, opposing counsel for Corporation ABC states that the reported amount was supposed to be $10,000. With this, Corporation ABC motions for summary judgment.
In this hypothetical situation, it is important to note what has occurred. Although, the amount that was supposed to be reported is in dispute (government says $20,000; Corporation ABC says $10,000), opposing counsel and Corporation ABC has conceded that at least $10,000 was supposed to be reported. Therefore, because both the initial reported amount of $4,000 and opposing counsel’s conceded amount of $10,000 are undisputed, these two amounts can be used in a data analysis.
In our False Claims Act example, using data analytics, we can conclude that Corporation ABC had failed to report at least $6,000 ($10,000 – $4,000) of what was supposed to be reported. Or, that Corporation ABC failed to report at least 60% ($6,000 / $10,000) of what was supposed to be reported. Or, that Corporation ABC only reported at the most 40% ($4,000 / $10,000) of what was supposed to be reported. Remember, these data analytics conclusions are based off of undisputed or opposing counsel provided and conceded data; thus, these data analytics conclusions will be difficult for opposing counsel to rebut.
The next step is to apply our data analytics conclusions to caselaw. In United States (ex rel. Liotine) v. CDW Government, Inc., the court denied the defendant’s motion for summary judgment. The court rationalized that genuine issues of material fact existed because a deposed witness stated that approximately 10% of the time at least one item on the invoice or order was not reported to the government.
In the case at hand, using opposing counsel’s provided and conceded data, we have concluded that Corporation ABC failed to report at least 60% of what was supposed to be reported. Comparing Corporation ABC’s 60% with Liotine’s 10% is monstrous. Again, opposing counsel will have a difficult time rebutting the data analytics conclusion and legal argument because opposing counsel provided and conceded the data. Therefore, a court is most likely to deny the defendant’s motion for summary judgment.
Overall, although, this is a simple example, data analytics can be used to formulate legal arguments. Do not make the mistake in Collins. Just remember to stick to data the opposing counsel is not contesting – preferably data provided and conceded by opposing counsel.
 How Lawyers Use AI to Win Before It Begins, JD Supra (June 26, 2019), https://www.jdsupra.com/legalnews/how-lawyers-use-ai-to-win-before-it-90005.
 Jake Frankenfield, Data Analytics, Investopedia (July 1, 2020), https://www.investopedia.com/terms/d/data-analytics.asp.
 People v. Collins, 68 Cal. 2d 319, 320 (1968).
 Id. at 325.
 Id. at 321.
 Id. at 325.
 Id. at 328-29.
 See id. at 332.
 31 U.S.C. § 3729 (2020).
 United States (ex rel. Liotine) v. CDW Government, Inc., No. 05-33-DRH, 2012 WL 2807040, at *12 (S.D. Ill. July 10, 2012) (considering summary judgment for overcharges on freight and unpaid Industrial Funding Fee).
 Id. Important to note is that the witness stated that approximately 10% of the time at least one item on the invoice or order was not reported to the government. Translated, this means that 10% of the time, the invoices or orders contained an error; whereas, if we were to consider each item irrespective of invoice or order, the percentage of error would be much less than 10%. Thus, the genuine issue of material fact threshold is much lower than that of 10%.
 See id.
 See Collins, 68 Cal. 2d at 332.
Image Source: https://www.information-management.com/list/22-top-vendors-for-data-analytics-software