Search
Program Calendar
Browse By Day
Browse By Time
Browse By Person
Browse By Room
Browse By Division
Browse By Session Type
Search Tips
Personal Schedule
Sign In
Automated text analysis methods have made possible the classification of large corpora of text by measures such as topic, tone, and ideology. However, even when using dictionary-based methods that require few inputs, many decisions must be made before a measure of any kind is produced from the text. When coding media, the analyst must decide on the universe of sources to sample from, as well as the criteria for selecting articles. If utilizing a supervised learning method, the method of generating training data presents many decisions: the unit of analysis, choice of coders, number of units to code, number of coders per unit, etc. In this paper we consider the choices that automated analysis of media text involves -- using as a running example efforts to measure the tone of newspaper coverage of the economy -- and highlight our key findings throughout.
Pablo Barbera, U of Southern California
Jonathan Nagler, New York U
Ryan McMahon, Pennsylvania State U