By Prof. Bernard Chao, Academic Advisor to the Civil Jury Project
Federal Rule of Evidence 407 and its many state counterparts prohibit plaintiffs from introducing evidence of subsequent remedial measures to show that the defendant is to blame. The rule is intended to prevent jurors from judging a defendant’s conduct using hindsight bias. Not surprisingly, plaintiff attorneys often try to take advantage of the rule’s numerous exceptions to introduce evidence of remedial measures for other purposes (e.g. to prove feasibility or to impeach a witness). Of course, their actual hope is that juries will use the evidence for the impermissible purpose — to show that the defendant’s original actions were wrong. My student, Kylie Santos, and I decided to test whether this assumption about jury decision-making was right. I presented a draft of our results to the NYU Civil Jury Academic Roundtable this past spring and share some of the highlights here.
For each experiment, we took a personal injury case and created multiple different scenarios. In the basic case, there was no evidence of subsequent remedial measures. Then in another version, we added that evidence. In still two more versions, we layered on limiting instructions. The first limiting instruction was a simple jury instruction that said that jurors could use the evidence of subsequent remedial measure for one purpose, but not another. The second limiting instruction went on to explain that the rule encourages people to fix problems.In the first experiment, the plaintiff fell down a steep staircase in her apartment building and sued the landlord for negligence. The evidence of subsequent remedial measure was a redesigned staircase that included a landing. In the second experiment, the plaintiff was injured snowboarding. He sued the snowboard manufacturer alleged that the snowboard was defective. The manufacturer then redesigned the snowboard.
For each scenario, we made a PowerPoint Presentation with narration. A judge introduced the case. Then the plaintiff’s attorney presented the highlights of the plaintiff’s case. Next, the defendant’s attorney presented defendants’ arguments. Finally, the judge returned and instructed mock jurors on the law. Videos were then made for each scenario. They lasted between approximately fifteen to seventeen minutes. We then recruited mock jurors online paying them a few dollars. Over one thousand seven hundred subjects watched one of our scenarios and rendered verdicts on liability, and if applicable, contributory negligence and damages.
As expected, evidence of subsequent remedial measures helped plaintiffs win more often in both our experiments. But surprisingly, our results also suggested that introducing this evidence may lower damages in some cases, thereby counteracting the increased liability findings. In our second experiment, we found that introducing evidence of subsequent remedial measures lowered damages (from $310k to $259k) and increased rates of contributory infringement (from 31% to 46%).
To be clear, both findings were just short of reaching statistical significance. 
But when we combined the two numbers and looked at net damages (the amount of damages the plaintiff would actually recover after discounting for contributory negligence), the findings were significant. 
These findings are consistent with what we know about juries’ decision-making on damages. Other studies have shown that juries award higher levels of damages when they view the underlying conduct as more blameworthy. By taking remedial measures, the defendant is probably looking less blameworthy. Thus, we see the opposite effect; damages decrease. A quick word of caution: we did not observe the same effects in our first experiment. It may be that fixing stairs did not make the defendant as morally upstanding as redesigning a snowboard.
The upshot of our study is that evidence of subsequent remedial measures helps plaintiffs on liability but may hurt them on damages in some cases. Thus, if liability is hotly contested, plaintiff should try to introduce evidence of subsequent remedial measures. Obviously, the defendant should try to do the opposite and exclude that evidence. On the other hand, if liability is straightforward and the real dispute lies with damages, the parties may want to switch tactics. Plaintiffs may want to steer clear of presenting evidence of subsequent remedial measure to avoid lower damage awards. On the flip side, defendants may want to introduce that same evidence to show what good citizens they are. Of course, in most cases, the parties care about both liability and damages. In these cases, parties are going to have to prioritize. Do they want a little more help on the liability side and risk hurting their damages case or is the opposite true?
If you want to learn more about the details of our experiments, the full paper, How Evidence of Subsequent Remedial Measures Matters, will be published in the Missouri Law Review. A draft is available at this link: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3374022##
There are some interesting findings about limiting instructions too. The short is answer is that they may work if you use the more detailed instruction that explains the basis for the instruction!
 When social scientists analyze data, if there is less than a 5% chance that the distribution could be found randomly (p=.05), they say the results our statistically significant. Our finding on raw damages and contributory negligence were at the p=.076 and p=.11 levels respectively.
 Here, the p value was .04 making the findings statistically significant.
Prof. Bernard Chao
Professor at the University of Denver Sturm College of Law, Co-Director of the Denver Empirical Justice Institute and a principal of Hugo Analytics LLC, a consulting firm that performs online experiments for trial attorneys.