Measuring Media Diet in a High-Choice Environment - Testing the List-Frequency Technique

ABSTRACT

How to measure exposure to information in the media is highly disputed, primarily due to the difficulties of obtaining accurate self-reports. The growing supply of outlets and proliferation of information sources have added an additional level of complexity to these problems. Reflecting on old and new approaches for measuring exposure to political information, it is argued that both the specific source and the frequency of exposure must be taken into account. The validity of this so-called “list-frequency technique” is tested using a two-wave panel survey as well as a split sample experiment from the survey pre-test to enable comparison with the “list technique.” The results support the list-frequency technique in being a good solution, since it provides the same aggregate estimates of media use as the already validated list technique, and may give more detailed effect estimates and increase the explained variance when predicting political knowledge.

Notes

¹ We are aware that these lists differ in their target level, since the lists for newspapers, radio, and websites are aimed at the channel level, while the list for television is aimed at the program level. However, individual TV news programs are directly comparable to newspapers and websites, since all of these individual sources contain several news items. Likewise, we could have included individual radio programs to our list for this media type, but exposure to news and political information in the radio was not the main focus of our data collection. Further, radio channels often have a specific focus on either news or music and are thus likely to differentiate more than TV channels. Therefore, we aimed the radio list at the channel level to keep it as short as possible.

² Dilliplane et al. (2013) used fixed effects regression models to test the predictive validity of the program list technique. It was not feasible to use the same approach in this study, since the specific questions used to measure the dependent variable, current affairs knowledge, varied between the two waves.