State of the First Amendment:A survey of public attitudes

Sunday, May 30, 1999

State of the First Amendment
1999 Survey
  • Overview
  • Foreword
  • Analysis
  • Questionnaire
  • Methodology
  • News advisory

  • The First Amendment Center at Vanderbilt University and the Center for Survey Research and Analysis at the University of Connecticut jointly developed this general public survey of attitudes on the First Amendment. At the University of Connecticut, Professor Ken Dautrich directed the project. Dr. Larry McGill of the Media Studies Center and Paul K. McMasters, First Amendment Ombudsman for The Freedom Forum, aided in developing the questionnaire. Kenneth A. Paulson, executive director of the First Amendment Center, provided overall direction for the project.

    Telephone questionnaires were pre-tested with 30 respondents. The pre-test was used to ensure that questions were understood by respondents and response categories were appropriate.

    Sample design

    The University of Connecticut follows procedures in sampling and data processing that are designed to minimize error in the results. For the sampling procedure, we utilized a variation of random-digit dialing-working residential “blocks” were identified with the aid of
    published directories. These exchanges were chosen in a modified stratified procedure based on the proportion of the theoretical universe residing in the geographic area covered by each published directory. Thus, in general, if 10% of the universe lives in the area covered by a
    directory, 10% of the exchanges will be chosen from that area.

    The universe for the First Amendment project was the adult non-institutionalized population of the contiguous 48 states who were 18 years of age and older. The geographic distribution in sampling was based on estimates of the distribution derived from the census
    figures for towns.

    Once “working blocks” were identified, one telephone number was generated at random for each block. A household was given five distinct opportunities to be contacted before a substitution was made for it.

    Once it had been determined that the household did, in fact, contain an eligible respondent, a random selection-unbiased by age or sex among the eligible respondents-was made. If that person was not the one who answered the telephone, an eligible respondent was called to the phone

    “Household” was defined as a dwelling where at least one adult 18 years of age resided. Such institutions as college dormitories, prisons and the like were omitted.


    All interviewing for this project was conducted at the University of Connecticut’s telephone center. Interviewing was conducted by telephone between Feb. 26 and March 24, 1999, using a Computer Assisted Telephone Interviewing (CATI) system. The CATI system utilizes computerized questionnaires, thereby reducing the amount of human error in the survey process.

    The telephone interviews took place in the evenings on weekdays, on Saturday mornings and afternoons and on Sunday afternoons and evenings. This schedule avoided the potential for bias caused by selecting people who were at home only at certain times. If a given telephone number did not result in an interview-for whatever reason-a substitution was made for it from within the same working block (which functioned as a single member “cluster”). This meant that one person’s not being at home, for example, did not keep his or her cluster from coming into the survey.

    Sampling Error

    A total of 1,001 interviews were conducted with a national scientific sample of adults 18 years of age or older. Sampling error for a sample of this size is ±3% at the 95% level of confidence. Sampling error for subgroups (e.g. men, women, etc.) is larger.

    The size of sampling errors depends largely on the number of interviews. The following table may be used in estimating the sampling error of any percentage in the report. The computed allowances have taken into account the effect of the sample design upon sampling error. They may be interpreted as indicating that the range (plus or minus the figure shown) within the results of repeated sampling in the same time period could be expected to vary 95% of the time, assuming the same sampling procedure, the same interviewers and the same questionnaire were used.

    The table is used in the following manner: If a reported percentage is 33% for a group that included 1000 respondents, go first to the row headed “percentages near 30″ and go across to the column headed “1000.” The number at this point is 4. This means that the 33% obtained in the sample are subject to a sampling error of plus or minus 4 points. Another way of saying it is that, very likely (95 times out of 100), the average of repeated sampling would be somewhere between 29% and 37%, with the most likely figure being 33%.

    Sample size

    Over Time Comparison

    Some of the questions in this survey are repeated from questions administered in the 1997 First Amendment study sponsored by The Freedom Forum. These serve as time-line comparisons to track changes in opinion.