How we did this project

This analysis of food safety and hygiene in Philadelphia restaurants is based on our evaluation of public records of more than 18,000 inspections conducted by the city's Health Department over the last five years.

The department uses a form that has a standardized checklist of 56 items that inspectors use to determine if the restaurant is in or out of compliance. On the second page of the form, the inspectors write more detailed information about the reason for the violations. Inspectors, who are called sanitarians, usually return within 30 days to see if the violations are corrected.

Not all violations are created equal. Some involve minor matters, such as faulty signage. Some go to the core of food-safety practices, such as proper storage of food, presence of rodent or insect infestations or temperature controls for hot and cold food.

The department follows guidelines developed by the Food and Drug Administration after a rash of national outbreaks of food-borne illnesses in the middle of the last decade. The department posts facsimiles of each inspection report on its website.

The department does nothing to help consumers understand the reports. It amounts to a massive jumble of data and bureaucratic narrative that defies analysis by everyday consumers.

For this project, our reporters and analysts based our ratings on the average of all inspections conducted of a restaurant over the last five years.

To accomplish this, we went to experts in the food-service industry and asked them, based on their experience, to assign numerical values to each of the 56 violations, ranging from one to three: one for the least serious, three for the most serious.

Our principal consultant was Andres Marin, a former chef who is professor of culinary arts at Community College of Philadelphia and who also teaches safe food handling for people in the restaurant business. We then vetted Marin's work with three chefs and a former catering executive. They generally agreed with the numerical rankings of the violations.

(We had nine violations with a value of one; 25 with a value of two and 22 with a value of three.)

In the spring, we filed a "Right to Know" request with the Health Department for a data­base version of all their online inspection reports, dating from January 2009 through March of this year. The department turned down our request and has not had any involvement with this project, other than to answer questions we posed.

Using software, the project team scraped the information from the posted reports and converted them into an Excel file, which contained a total of approximately 1.4 million pieces of data. The data included 3,800 "eat-in" restaurants, which is a Health Department classification. It does not include take-out restaurants or food trucks.

For each violation, we assigned the appropriate number, added them up, then divided the total by the number of years the restaurant was inspected. If, for example, a restaurant had compiled 100 points worth of violations over five years, its average would be 20.

Not all restaurants are inspected every year. We accounted for that fact by dividing a restaurant's total points by the actual number of years it was inspected.

We added two other elements to the equation when the city took further actions.

There were 904 cease and desist orders issued during the period examined. These are orders, officially issued by the city's Department of Licenses and Inspections, which require a restaurant to cease operations until what inspectors view as a serious violation is corrected. We added a 10-point penalty for each cease and desist order.

During the same period, the Health Department referred 298 restaurants to the Law Department to get court orders for enforcement. This was done only after the restaurant repeatedly failed to correct violations. We added a 20-point penalty if a restaurant was sued by the city.

It's important to note that restaurants rarely undergo inspections without some violations. The average points for all restaurants in the city was 27. Taking the spread of all the scores as a guide, we created four categories and awarded food-safety stars for each restaurant, based on their average point score.

Those restaurants with average scores of 16 or below did better than the Philadelphia restaurant average and got four stars. Those that got average scores from 17 to 37 met the Philadelphia average and got three stars. Those with average scores of 38 to 58 fell below the Philadelphia average and got two stars. Those with average scores of 59 and above fell far below the average and got one star.

Related stories:

A searchable database of all 3,800 restaurants

An article about food safety in the city's restaurants

A letter from the editor: Why we did this project

CP's points for 20 popular restaurants.

The project team consisted of reporter Ryan Briggs and data analysts Ryan Friel, Scott Lohbauer and Casey Thomas. Tom Ferrick served as senior editor and project coordinator.