Privacy Polls and Public Opinion

Accurately measuring public opinion is critical given the range of groups that use polling data for decision-making. How public opinion is measured, that is, the way the questions are asked and the way results are reported, may have an influence over groups that some critics find problematic . For example, one study reported that political leaders assume citizens are more conservative about innovations in the justice system than citizens report themselves to be (Flanagan, 1997). Additionally, the media coverage of events can affect results. Opinions on crime during intensive media coverage of events such as the O.J. Simpson trial are likely to be different than during times when events are not covered (Flanagan, 1997).

The way in which an issue is presented can influence the public's response. For example, two separate polls conducted at about the same time in the year 2000 examined whether the government should pass laws on Internet privacy. The results of the two polls differed depending on the way the poll questions were worded. The first poll, conducted by Harris in association with Business Week magazine, gave respondents three choices about government involvement and asked which would be best at this stage of the Internet's development. The choice reading, 'government should pass laws now for how personal information can be collected and used on the Internet' was selected by 57% of respondents, while choices with less government involvement (not take action now or recommend privacy standards) were selected by far fewer people (15% and 21%, respectively). The second poll, sponsored by the Democratic Leadership Council (200), asked respondents which statement was closer to their view on the best way to protect privacy of personal medical and financial records. Almost two third selected the viewpoint saying, 'give individuals more personal control over who sees their records,' while only 29% said 'pass strong federal restrictions.' Thus, it is possible that legislative decisions on privacy could vary based on which poll was assessed by policy makers .

In order to determine how privacy is characterized in public opinion polls, 32 polls by reputable sources (including polling centers such as Harris and Gallup, news media such as ABC and Fox, and academic organizations like the Pew Center) conducted over the past ten years were analyzed (Table 9-1). Together, these polls resulted in a total of 76 questions that addressed the issue of online privacy.

Table 9-1: Summary of privacy polls: 1996-2002
Polling Group	Year Polls Done	Total Questions
1st Amendment Center	2002	2
ABC	2000, 2001	3
ACT	2001	1
Atlanta Journal Constitution	2002	2
BW/Harris	2000	4
CMOR	2001	1
Consumers Union	2002	1
Direct Marketing	1996	1
Forrester	1999,2001	2
Fox	2000	7
Gallup	2000, 2001	5
GVU	1997, 1998	3
Harris	1994,1999, 2000, 2001, 2002, 2003	15
IPPSR	2002	1
Markle	2001	1
New York Times	2002	1
PANCS	2002	1
Pew	2000,2001	9
UCLA	2000	1
USA Today	2000	1
Yankee	2001	1
Zogby	2002	1

These questions were analyzed based on two assessments: breadth and direction. Breadth has to do with the scope or generality of the issue being studied - it is the concept that is provided for assessment highly specific to an issue or generalized. In this case, the polls' questions about online privacy were compared to the five-dimensional framework described above to see how well the polls reflected the complex, contextual nature of online privacy. Direction characterizes the response, whether the respondent is for or against a concept (such as statements about practices which are an invasion of privacy), and polls tend to focus on responses that are preponderantly favorable or unfavorable.

The questions were then divided into three general categories of polling questions to better understand how the polls were studying privacy. The first group includes questions that defined and assessed online privacy. The second group addresses information sensitivity, that is, the types of information that are collected online and the final group addressed questions of what should or should not be done to protect online privacy.

How Polls Define Privacy

The first group, definitions of privacy, includes 34 total questions (45% of total) that asked in general terms about respondents' opinions with privacy on the Internet. In this group, seven questions defined privacy online in just that way: privacy on the Internet. This question is quite broad in breadth, and the directionally this question tended to be measured in terms of a level of concern, with polls reporting that a majority of Americans were concerned and that seeming high percentages were very concerned . The majority of the questions in this category, though, provided specific (if limited) definitions of what privacy online entailed. Eighteen survey questions asked specifically about control over information collection and usage. Four surveys defined privacy as the ability to control the collection of information about oneself, and eight surveys defined privacy as the ability to personally control who that gets information. Six questions asked about the practice of Web sites sharing information, and directionally this was assessed by levels of concern and comfort , as well as perceptions of whether Web sites sharing information was a violation of privacy. Again, a wide range of perceptions were found. These two questions capture the awareness and usage dimensions of the five-dimensional framework. These questions tended to be measured in terms of how important it was for individuals to have that control, although some questions presented the direction in terms of the level of risk in control was not available.

Four questions defined privacy in terms of information that can be stolen, and directionally queried respondents about how worried or how much at risk they felt for such an activity. This type of question was the only one in this category that set up a possible negative bias for the response, that is, it framed the question in a negative sense (information could be stolen) rather than a positive sense (information is safe). Finally, one question asked about the 'right of privacy' and directionally queried about the degree to which it was essential (resulting in high levels of essentialness). Four questions asked about government access and monitoring of information, and directionally measured whether respondents supported or opposed such activities. These last two questions are somewhat analogous to the relationship dimension of the privacy framework.

Polls and Information Sensitivity

The second category of questions examined the type of information collected, and included 21 questions (28% of all questions). This section is most analogous to the 'sensitivity' dimension in the five-dimensional framework. Seven questions asked about the collection of financial records and credit card information, directionally assessing level of concern, with a majority of Americans concerned about such information. Five questions asked about Web tracking, that is, that an individual's movements around the Internet can be monitored and collected. Directionally, the questions measured levels of comfort and perceptions that such activity is a violation, with an overwhelming majority uncomfortable with such a practice and a range indicating that it is a violation. Four questions examined levels of agreements or assessment of the risk of online buying as being a threat to privacy. A broad range of levels of agreement and perceptions of risk were reported. One questioned asked about cookies being an invasion of privacy, with a low percentage of respondents agreeing.

Two questions asked about collection of health records, directionally measured using level of concern, which ranged from 50% to two- thirds of respondents indicating concern. One questioned asked about information available via a published directory (such as home address or phone number), again using level of concern as a directional measurement, with about half of respondents indicating a concern. One question asked about Social Security numbers , with three-fourths of respondents very concerned about the availability of that information.

Polls and Privacy Protection

A final group of 21 questions (27%) asked about how individuals' online information should be protected. Three types of questions addressed government involvement. Eight questions measured a level of agreement that government should pass laws to protect online privacy, with a range of agreement from 57% to 87%. Six questions asked whether current laws protect consumers, with a level of disagreement from 38-63%. One question asked whether respondents agreed that violators of laws should be disciplined, and an overwhelming majority agreed. Three types of questions addressed non-legislative ways to address this problem. Four questions assessed opt-in policies, with a high level of support for such policies. Similarly, one statement assessed level of agreement as to whether Web sites should disclose policies (resulting in a high level of agreement) and one statement assessed levels of agreement to a statement that consumer education was better than law (resulting in a high level of agreement).