BULLET Placing a Value on Public Library Services
Glen E. Holt, Donald Elliott, and Amonia Moore

Inevitably the calculation of direct benefits had to involve an extensive user survey. The preparation for and carrying out of the survey involved many steps.

Organization of the Telephone Survey

In preparation for undertaking a telephone survey, the researcher team turned to the St. Louis Public Library Automation System (SLPLS) cardholder database. On the day that staff pulled the cardholder sample, it had more than 72,000 "active cardholders" (i.e. those who had used their cards within the past 12 months) residing at 39,444 unique addresses.8 Each address was considered a separate household or institution that received library services. From that database, Technology Services staff pulled a random sample of the names of 2,350 “general” cardholders, 400 “teacher” cardholders and 100 “business” cardholders. If a youth card was selected, the survey addressed the parent(s) or guardian(s) of the cardholder.

Then Dr. Holt wrote an invitation letter to all 2,850 persons selected for inclusion in the survey. The letter asked if the person were willing to be interviewed in a telephone survey about library services. In the letter, Holt gave certain assurances: First, that the study was intended to help improve SLPL's ability to serve users better. Second, that a grant from the Public Library Association - not local tax funds - was paying for the nationally significant study. Third, that for completing the telephone survey, the library would provide a choice of premiums: a Friends of the Library tote bag, a set of Friends’ coffee mugs, or a choice of one-of-two SLPL commemorative posters.

Finally, the letter stressed finding a convenient time for library representatives to talk to the user. The mail-back sheet asked the user to check the correctness of the personal name and address, list a correct phone number and choose various weekday evening times (Monday through Thursday) when the user would be available to take a call from a library representative.

16 percent of general users receiving invitations agreed to participate in the survey. 83 percent of invited teachers and 86 percent of invited businesses agreed to participate. Trained interviewers of the SIUE Center for Regional Research and Development Services delivered the telephone interviews. All had participated previously in phone surveys, and all were coached specifically for administering the SLPL-CBA telephone survey.

If the surveyors did not reach the interview subject on the first call, three more attempts were made to contact the user. With this follow up, interviewers completed 322 “general” interviews, 75 “teacher” interviews and 25 “business” interviews. Measured against those who had agreed to be interviewed, the “response rate for telephone interview participation” (i.e. completed interviews) proved to be 87 percent for “general” users, 83.3 percent for “teacher” users and 86.2 percent for “business” users. Checking and improving the survey's statistical reliability

As with any new survey, this one had to be checked for statistical reliability. This was a special concern due to the number of general users that failed to accept the invitation to participate in the survey. Standard statistical (Chi-square) tests, however, demonstrated the overall statistical validity of the survey sample relative to the known characteristics of the cardholder population.

To test further the validity of the sample of general users, however, members of the research team also checked for non-response bias by race and income level—characteristics not available from the cardholder database but which could be inferred from census tract data which SLPL uses regularly as part of its services planning. Not surprisingly, in the sample, whites were over-represented and African-Americans were under-represented, while higher income levels were over-represented and lower-incomes were under-represented. Moreover, those from the outer ring (generally higher income) zip codes of the city and the library's out-of-district suburban users were over-represented while inner-ring (generally lower-income) city dwellers were under-represented. The research team mathematically adjusted the survey results to correct for these biases.

After the adjustments, the weighted sample matched the presumed race, income level and geography demographics of the SLPL user population as a whole. Reassuringly, the valuation estimates from the weighted sample varied little from the valuation estimates from the original data, suggesting in one more way the validity of the survey.

Back to previous section Table of Contents Forward to next section