US results are based on interviews via landline and cellular telephones conducted in English with 2,303 adults, age 18 or older, living in the continental United States, during August 2011. The German survey is based on phone interviews with 1000 people age 18 or older, conducted between August 24 and September 6, 2011. The German component included landlines only—standard practice due to the lower number of cellphone-only households. For results based on the entire US sample, the margin of error is plus or minus 2%. In the German case the margin of error is 4%. Our results are also weighted to better represent the demographic profiles of the two countries.
US and German surveys used the same question set—though minor cuts were made in the German version to meet the 20-minute time limit accorded most phone surveys.
Standard age brackets used in national surveys differ slightly in the US and Germany: US surveys generally use 50–64 and 65+ brackets. German surveys use 50–69 and 70+. We followed these norms. For the great majority of questions, this introduced no statistically significant differences.
For the handful of questions where we report household income categories, we break the population into rough terciles—low income, medium income, and high income. These brackets describe similar income ranges in the two countries, but are usually reported differently. In the US the top income tercile starts at around $75K per year before taxes; in Germany, at around €30K per year after taxes. At the time of the survey, €1 equaled about $1.40.
Some of the important questions in this study targeted very small groups. Only 4% of the adult German population (N=39), for example, reported using P2P services. Any further breakdown of the behavior of this group is statistically unreliable. We have excluded the majority of these low-sample-size findings, except where they clearly validate wider findings or raise interesting questions for further study. We have indicated when the sample size drops below a reasonable confidence level.
More generally, the survey results show volatility on some topics that, we believe, significantly exceeds the statistical margin of error. Variations on questions about different kinds of file sharing, for example, produced different response rates and—occasionally—statistically significant divergences in interpretation between age groups. Some questions required considerable knowledge of or experience with the practices described, such as “seeding” BitTorrent files or “hiding IP addresses.” We have done our best to clarify these results and signal where we lack confidence in them. The larger point is that although the survey methodology is fairly rigorous, the survey language and the social reality it maps are not.
The surveys were sponsored by the American Assembly, with support from a research award from Google. Both US and German surveys were conducted by Princeton Survey Research Associates International. The design of the surveys and the interpretation of the data are solely the responsibility of the researchers.
Under-Reporting Of Piracy
All surveys contend with the tendency of respondents to answer strategically—to tell the interviewer what they think he or she should hear. This is a particular concern when dealing with activity widely perceived to be unethical or illegal, such as the uploading of music to P2P networks.
There is no consensus about the scale of underreporting of copyright infringement, and our survey provides no means of estimating it. However, an Ofcom / Kantar Media study (2010) of file sharing in the United Kingdom did shed some light on this issue by exploring variations in responses to the same questions asked using different methods, including face-to-face interviews, phone interviews, and e-mail questionnaires. With regard to questions about participation in “illegal downloading,” results ranged from 12% of the population based on the face-to-face method, to 13% based on phone interviews, to 19% based on e-mail solicitation. Weighing all methods, the authors estimate that 15% of the UK population has engaged in “illegal downloading”—implying that phone surveys undershoot actual behavior by a modest amount (in this case, by 2%). This is worth keeping in mind throughout the analysis, especially where sample sizes are small. Germany, for example, consistently shows one of the lowest rates of P2P use in Europe, and our findings are at the low end of reported numbers: 4% of Internet users. Earlier in 2011 the German National Statistics Opinion Survey, using slightly different language via a self-administered mail survey, put the proportion of P2P users at 10% of Internet users. Survey firm GfK, asking a question via phone survey about “illegal downloading” in 2011 found 6% participation. It is safe to assume that such uncertainties run through many of the topics addressed here.