The instrument was sent to over 9000 addresses on two different subscription lists for White House electronic publications: a list for general documents (6800 addresses) and a list for health care bulletins (2250). This instrument was also distributed through regular usenet channels and sent to a short list of known regular redistributors of the documents, including operators of commercial networks, university gophers and government agencies. These agents were were asked to pass along the survey to their audiences, and obviously distribution of the survey was constrained by the extent to which they did so.
1592 people responded to the survey, including 810 and 167 listed on the two mentioned subscription lists respectively. While these numbers are relatively high percents of the subscription lists, we believe the total number of respondents is a much lower percent of primary recipients of the documents. Recipients of the documents through means other than these lists were less likely to receive and return these documents, because most downstream agents did not respond to our request to send out or post the survey. At several sites, incorrect or confusing directions for obtaining the survey were initially posted, and considerable time past before they were corrected. Note even when correct instructions were posted, recipients were being asked to send mail to an unfamiliar address and use commands unfamiliar to many of them. This may have kept some people from trying to get the survey and led others to abandon the effort when they received an error message. In addition, people without email facilities who get the documents by downloading them from bulletin boards could not take the survey, and there is some evidence of many such recipients. In short, then, the responses best represent the direct subscribers and do not directly represent non-email capable recipients.