In this area, you will find resources including methods on our media content analysis, definitions of variables, a sample of our music coding sheet, and a comprehensive list of the top 30 songs of each year, 1950-2008, from Billboard’s “Hot 100” year-end charts.


This portion of our website is under development and this page currently contains only placeholder content. Look for music findings in the future.


Music Coder Training

Our coders are undergraduates from the University of Pennsylvania – primarily majoring in Communication at the Annenberg School – who are recruited for their exceptional academic performance, attention to detail, and interest in the project. Another advantage is that the age of these coders closely represents that of the target adolescent population to whom the content is addressed.

Coding is conducted on a line-by-line basis, coders read through a song’s lyrics for content while simultaneously listening to the song to ensure that context informs the coder. New coders (trainees) undergo a rigorous training regimen to familiarize them with the data collection procedures as well as to ensure inter-coder reliability. We use a Krippendorff’s alpha 0.7 as our threshold to determine reliability among coders.

After studying our music code book, the trainees code a practice song in order to become familiar with the coding procedures. Once they are comfortable with the coding instrument, they progress through the next two training phases. During the second phase, coders listen to several songs along with the provided “gold standard” answers in order to learn to interpret our code book. The “gold standard” answers were created by a group of expert coders who have been consistently reliable with each other in the past. These expert coders were asked to read through lyrics for each song and decide on the appropriate coding answers which were then deemed definitive for the training of new coders. At the third phase, coders make their own judgments while listening to the songs before being presented with the standardized responses so that they may reflect on their decisions. After coding multiple songs in this fashion, the students move to the final training phase of coding several songs that expose them to rare content. Inter-coder reliability is tested for all content measures and further training is conducted if necessary. Once satisfactory reliability has been established for each trainee, he or she codes from the project’s actual sample.

Coding Instrument

Considerations on a Web-Based Survey Software Platform:
CHAMP’s Choice for Its Content Analysis


This code book area provides descriptions and definitions of variables coded for the music phase of the CHAMP project. These variables measure the frequency, explicitness, and other characteristics of the recorded behaviors.


The CHAMP Data (Coding of Health and Media Project) is a quantitative content analysis project that has coded a sample of top 30 single songs from U.S. Billboard’s “Hot 100” year-end charts from 1950-2008.

This list consists of 1,794 songs from 1950-2007 coded for the CHAMP project.