How did we reach these conclusions?

Univision Data used a statistical method that is common in social sciences but practically unexplored in newsrooms, called logistic regression.

The investigation of Costa Rica’s public defense system was born from a single question: Is there a link between the type of lawyer that defends a criminal case and the probability that a defendant will be convicted?

Answering that question in Latin America is practically impossible given the systemic lack of official data organized for that purpose. The task was more viable in a small Central American country like Costa Rica.

Costa Rica’s judicial branch has an abundance of unorganized public information that allows the first test to be conducted in the region.

For four years, 10 Costa Rican journalists and lawyers built from scratch a database that is sufficiently solid to begin to explore this question. We took all rulings from the Second Circuit Criminal Court in San José from 2004-13 and extracted multiple variables.

The database has partial or complete information on 11,183 crimes, distributed in approximately 8,000 rulings.

For the analysis, Univision Data used a statistical method common in social sciences but practically unexplored in newsrooms, called logistic regression.

This tool estimates the probability of conviction, acquittal and dismissal of defendants when they have a public defender or a private defense attorney by excluding several factors that could influence the outcome of a criminal trial. Among them are type of crime; the number of years of experience of the defense attorneys, judges and prosecutors who participate in a judicial process; the victim’s gender; the defendant’s gender; whether the legal process is normal or expedited (a variable representing cases in which a defendant has overwhelming evidence either for or against them); and whether the defendant is a national or a foreigner.

Following these parameters, the study aims to define the degree of probability of being convicted in a criminal process that is attributable to whether a defendant cannot afford an attorney and must be appointed one by the state, compared with the option of having a private defense attorney.

Using this methodology, the investigation also examined if the type of crime associated with Public Defender’s Office clients explains the possible disadvantage of using public attorneys.

For example, a possible argument to explain the poorer performance of government-appointed public defenders compared to private ones is that people with low incomes who commit crimes against property in the public space, such as robberies and theft, could more easily be convicted. The concentration of these crimes among public defenders is undeniable. According to the database, of all processed crimes against property, 70% of defendants had a public defender.

Our analysis demonstrated that the greater probability of conviction with government-paid attorneys is invariable, regardless of the concentration of certain crimes in this group.

To be clear, while the majority of crimes against property are handled by public defenders, in general, crime categories by type of attorney are very similar, as shown in the following graph.

Crimes by type of attorney

The majority of cases handled by both public and private defense attorneys involve crimes against property, sexual crimes and offenses involving attempts on someone’s life.

Source: Univision Data analysis

We also tested whether the poorer results of public attorneys is explained by the fact that they defend many cases where the defendant has an abundance of evidence against them. On average, 73% of expedited processes that we analyzed corresponded to public defense.

In those processes, the defendant has enough damning evidence that they plead guilty in exchange for a lesser sentence if it is an ordinary trial and if the judge approves. However, when we looked at the probability of conviction excluding the influence of this variable, the disadvantage associated with a public defender was maintained for various crimes.


Limitations of the study

Our results should be interpreted with caution. This investigation does not conclude that public defenders are the direct cause of clients being convicted more frequently than those with private attorneys. Other factors not contemplated by the model could explain, at least in part, this disadvantage.

One example is the fact that public defenders can’t choose which cases they represent, while private defense attorneys can.

The results also can’t be generalized for the entire country, but are confined to the metropolitan area where the study was conducted. In this case, that area is the jurisdiction of the Second Judicial Circuit of San José.

The results do suggest, for the first time, the possibility that the service of public defense in Costa Rica – considered a model in Latin America – does not counterbalance the weight of criminal defendants’ economic status in the administration of justice.

To learn more about our methodology and our proposals for future investigation, see our complete report published in the Second Report on the State of Justice, edited by the State of the Nation Program.