Kilpatrick Stockton LLP report that a Cornell Law School study shows very interesting results for appeals.  There is a much greater reversal rate for trials than one might expect.

"Two Cornell Law School professors recently examined civil appeals in the state-court systems. See Theodore Eisenberg & Michael Heise, Plaintiphobia in State Court? An Empirical Study of State Court Trials on Appeal, Cornell Legal Studies Research Paper No. 07-006 (May 2007). Their study used data from 46 of the nation’s 75 most populous counties and included jurisdictions in California, Florida, Georgia, Illinois, Massachusetts, Michigan, North Carolina, New Jersey, New York, Ohio, Pennsylvania, Texas, and Virginia. The study looked at 8038 jury or bench trials and the resulting 549 appeals that were litigated to conclusion on appeal. (Because the study focused on trial verdicts, cases that were disposed of in other ways, e.g., pretrial motion, were not included.)

The principal conclusions of the study – some of which are surprising and counterintuitive to appellate practitioners – include the following:

Of the 8038 trial cases, only 965 (12%) led to an appeal. And of the 965 appeals taken, only 549 (57%) proceeded to decision in the appellate courts; the rest were terminated during the appellate process (e.g., settled or became moot). Of the 965 cases in which an appeal was commenced, only 24 reached the state’s highest court.
The percentage of trial judgments appealed varied considerably by the subject matter of the case. For example, appeals occurred in 30% of employment-contract cases, 26% of products-liability cases, and 18% of fraud cases.
Defendants that lost at trial were slightly more likely to appeal than plaintiffs that lost – 13% vs. 11%. The losing party was a bit more likely to appeal from a bench trial than a jury trial – 15% vs. 11%.
Appellate courts reversed the trial verdict in 32% of the appeals. The reversal rate varies greatly by state – for example, it was 13% in Georgia and 56% in New Jersey. The overall reversal rate also significantly depends on the type of case – it was 32% in fraud cases and 33% in products-liability cases, but 50% in employment-contract cases.
Notably, as between appeals taken by defendants and those by plaintiffs, the results were “starkly asymmetric” (Study at 13) in favor of defendants. In fact, defendants’ appeals were much more likely to be successful than were plaintiffs’ – 42% vs. 22%.
For example, defendants prevailed in 62% of their appeals in employment-contract cases, while plaintiffs won 39% of their appeals. In fraud appeals, the numbers were 39% and 15%, respectively.
Likewise, notwithstanding the appellate deference usually accorded to jury verdicts, appellate courts were more likely to reverse jury verdicts than bench verdicts – 34% vs. 28%.
Accordingly, the study “suggests an appeals court tilt favoring defendants, especially defendants that lost in a jury trial.” Study at 13 (emphasis added"