...that is the question.
The article below by Jonathan D. Glater in the New York Times, today, describes how the wrong choice is often made. They don't teach you in law school how to handicap trials, it says, which may or may not be correct.
What Glater and the Times also don't do, is explain how to handicap a trial.
There is such a thing as a Bayesian analysis, a statistical concept, in which one measures the anticipated cost of a trial in terms of hours of preparation, cost of paying expert witnesses such as doctors, engineers, economists, and multiplying by a percentage representing the chances of success vs. disappointment at variouis levels, say 50-50 or 60-40, and using that as a guide towards settlement with a view towards minimizing risk and maximizing value received. There are lots of variables that tend to make the process subjective, including the various assumptions factored in, but the analysis offers a guide for further speculating. Care to roll the dice? This is what lawyers say about going to trial. Guess who pays when the risk is life in prison vs. making a deal for years in a criminal cases that goes down the toilet.
It's a lot harder to evaluate the chances you have of winning a war when you are in the heat of battle than in reading about another general's war before the diplomats settle the thing. In law, the problem is that the lawyer wears two hats, that of the general and that of the diplomat. One minute you're fighting with a passion and the next you're all smiles because you're either negotiating outcome or need a continuance because one of your main witnesses has come down with the flu. Which is why it doesn't pay to get too upset with your adversary, for in a minute you may need his cooperation as a friend. So never tell him/her what you really felt like calling him after that last dirty move s/he made.
I'm surprised that neither Glater nor the Times mentioned it, but its their newspaper, isn't it.
Here's a chart: