Executive Summary:
In May 2009, the Future of Privacy Forum launched a research initiative to examine new methods for communicating with users about online advertising and privacy. This study assessed the communication efficacy of behavioral advertising disclosures based on icons and short disclosures placed near webpage advertisements as an alternative to providing transparency and choice via traditional online privacy notices.
The study employed an internet panel to assess the communication efficacy of behavioral advertising disclosures on the web. Disclosures were tested at 2 levels: (1) a level-1 disclosure comprising of a symbol and a short (2-3 word) phrase that was placed above an ad on the webpage (e.g., “interest based ads” or “ad choice”), and (2) a longer (18-20 word) level-2 disclosure that was revealed when surfers moved their cursor over the level-1 disclosure. This level-2 disclosure was intended to inform people that (a) information about their visit to this site is used to select ads they see here and elsewhere, and (b) to provide a link to “opt out” of this type of advertising.
The study involved five phases or sections:
• Part A: Exposure to webpage with level-l disclosure. Measure recall of level-1 disclosure;
• Part B: Re-exposure to webpage with level-l disclosure. Measure comprehension of level-l disclosure;
• Part C: Re-exposure to webpage with level-2 disclosure. Measure comprehension of level-2 disclosure;
• Parts D & E: Participation in 13 online activities and attitudes toward online behavioral advertising and privacy.
The sample size was 2,604 U.S. adults. The majority of the study participants were active internet users. The average respondent participated in 9 of 13 online activities, with approximately half or more of the sample participating in 12 activities. Only 2% had not participated in any activities. More than half of the sample spent at least 15 hours online/week.
Concern for privacy was measured by nine items using a 5-point scale (5=strongly agree). Average for this scale was 3.88. We also asked whether people had taken any of eight steps to protect their privacy online. The average was 4 steps, with approximately 40% of the participants engaging in 7 of 8 activities. Only 8% had never taken any of these steps.
We measured comfort with OBA with and without two key fair information practices (transparency and choice). Without transparency and choice, only 24% are comfortable with OBA. When transparency and choice are offered, 40% are comfortable with OBA. Approximately 30% are neutral about OBA with or without transparency and choice. Transparency and choice increase comfort for people who are most active online and engage in more privacy protective behaviors.
We tested 14 level-1 disclosures: 2 symbols (“Power I” and “Asterisk Man”) combined with seven phrases (Interest based ads, Custom ads, AdChoice, Your choice, Your info and ads, Why did I get this ad?, and a control phrase, Sponsor ads). We tested one level-2 disclosure which provided transparency and choice.
We measured comprehension of the level-1 disclosure with two multiple-part questions (Q8 and Q9). Comprehension of the level-2 disclosure was measured with one multiple-part question (Q11). Results show that for the level-1 disclosures, two of the seven tested phrases (“Why did I get this ad?,” and “Interest based ads”) generally do the best on comprehension. The remaining four phrases don’t do as well as the top two, but still outperform the control phrase (“Sponsor ads”). Also, while the phrase “Adchoice” did not perform as well as the top two on comprehension, it was in some cases less likely to generate agreement with decoy statements (that were unrelated to key communication objectives) than the top phrases. It is important to note, however, that while there were differences in the communication effectiveness of the six different level-1 phrases we tested, in an absolute sense it is not clear that they communicate well enough without additional support. Thus, consumer education will be needed to improve their communication effectiveness over time.
The testing also showed a very slight advantage for the asterisk man icon on some of the comprehension measures (specifically, Q8). Finally, the level-2 disclosure appears to communicate the key issues effectively.
To view the full report click here.