Taking a Quantitative Approach to Litigation

Why is Quantitative Analysis Important for Lawyers?

Introduction

• the present use of computational approaches to lawyering using statistics and quantitative approaches
• what do we mean by quantitative approaches?
• case or narrative analysis versus empirical analysis
• predictive coding uses computational approaches to forecast relevance of documents
• BlueJay Legal uses computational approaches to forecast relevance of cases

The Future is Not So Far Out

• "AI" or computational approaches to practice
• IBM Watson and other AI platforms for legal

Statistics 101—Back to High School

Descriptive Statistics—A High School Review

• the importance of knowing the uses of averages, medians, and modes
• your "worst case" and "best case" scenarios: maximums and minimums
• how "skew" should inform uncertainty and forecasting
• example: descriptive statistics to describe wrongful dismissal awards
• example: descriptive statistics to describe Charter exclusion application success
• example: descriptive statistics to describe enforceability of restrictive covenants

Using Statistics to Inform

• using graphs and charts to persuade
• using timeline charts to show chronologies
• example: illustrating potential penalty awards in a regulated industry

Using Decision Trees and Bayesian Logic to Model Outcomes

What is a Decision Tree

• nodes
• branches
• probabilities
• outcomes and payoffs
• putting the concepts together
• example: using a Decision Tree to model a civil litigation claim
• example: using a Decision Tree to model a criminal case
• summary

Forecasting in a Litigation Context

• the distinction between variability and uncertainty
• using scenario analysis to account for uncertainty
• using Monte Carlo simulations to account for uncertainty
• example: using scenario analysis to predict legal budgets
• example: using Monte Carlo simulation to predict legal budgets
• example: combining Decision Trees and Monte Carlo simulation to forecast litigation outcomes

Questions and Concluding Comments