Big Data Potential for Personal Injury Cases

Personal injury lawsuits are often prematurely settled for lower awards than they are deserved. The reasons are often monetary: law firms can be limited in pursuing cases due to inability to commit resources when the outcome is uncertain. Add to this, more uncertainty created when lawyers must rely on personal experience with judges, venues, and case types when deciding whether to take a case to trial or settle.

Big data has the potential to change all of that. Statistical analysis of case characteristics can now drive down the level of uncertainty in case outcomes. Reducing uncertainty will encourage more small-to-medium sized law firms to take personal injury cases to trial when they would have otherwise sought a settlement.

Data Available from Case Warehousing Sites

Several categorical and numerical features can be obtained directly from case information warehousing sites. Predictors may include such factors as location, court (state or federal), injury type, plaintiff sex and age, defendant attorney’s previous trial history, and Judge’s previous trial history. The list of possible predictors is unlimited and are tailorable to any specific case.

Applying text-mining techniques to the available data from case warehouses can generate a pertinent set of features. For example, text mining could be used to discover facts about the case such as the type of evidence presented, plaintiff occupation, victim activity at time of injury, details of the injury, and content of expert testimony. This type of data could be used either as case predictors or to help build case strategy and/or case themes – depending on the specific goals of the analysis and the nature of the case compared to available data.

Building from Case Data

Data mining would be the primary tool used to analyze a corpus of records on case outcomes.

Since different individuals input each case data, the language used across the corpus will vary significantly. This presents challenges when building features form the unstructured text. For example, multiple words or phrases will be used to represent similar information, so it is necessary to identify distinct words and phrases with similar meanings.

In many reference systems, it’s useful to group words and phrases that are synonyms. A good database will hold definitions and semantic relations between thousands of words and terms – distinguishing between nouns, verbs, adjectives and adverbs and relating concepts in so-called synonym sets. It may also be necessary to expand the database to include medical and legal terms to fit the needs of this venture.

Another approach available in common text mining tools is finding associations for a given term, which is a further form of count-based evaluation methods. This is especially helpful when analyzing a text for a specific purpose – for example, extracting associations of the term “expert” from the case data.

Predicting Case Outcome from Regression Modeling

Once text mining is complete and a data set is built using available past case data, a model can be trained using a variety of regression techniques. If predictors (corresponding to those described above) are known for a new case, then type (verdict or settlement) and amount (of monetary award) can be predicted with some level of certainty.

Further studies can be undertaken to evaluate the impact of varying approaches to a case on potential outcomes. For example, the relative benefit of adding one or more of a particular type of expert witness could be explored.

Added Benefit of Virtual Focus Groups

Focus groups can be extremely helpful in gaining potential juror reactions to an array of aspects of the case. A small research group cannot ‘predict’ a certain verdict in your case, but it will be invaluable in helping determine which values, themes, hot buttons, and sensibilities will resonate most in the jury room. Virtual focus groups have also proven to be very cost effective.

What do jurors want in the way of persuasive evidence? Brainstorm with them about the evidence that they’ve used to come to their conclusions, and what additional evidence they would need to change their minds.

You can gain great insight on such issues as:

  • What will a jury think of a particular witness(es)?
  • What evidence would have the most favorable impact on your case?
  • What themes, arguments and language resonates most effectively with potential jurors?

To be effective, a focus group must be constructed of savvy individuals. Look for a cross-section of individuals from the jurisdiction, including possible influencers who can exert influence on jury deliberations. Members need to be fully engaged in the process. Members also need to understand their role is not partisan: they are there at the behest of both parties in a dispute headed for the trial.

The litigating sides are blind to what the focus group might think of the case, but it is hoped that the collective wisdom of the group will offer a strategy for your most advantageous course of action – whether seeking resolution or continuing to trial. Stress to the group that their impressions and conclusions are important to the ‘fair and impartial’ process. This makes the participants feel they are key players in an important process.