REVIEW: Questionnaires in ecology: a review of past use and recommendations for best practice

Authors


Piran C. L. White, Environment Department, University of York, Heslington, York YO10 5 DD, UK (fax + 44 1904 432998; e-mail PCLW1@york.ac.uk).

Summary

  • 1Questionnaires, or social surveys, are used increasingly as a means of collecting data in ecology. We present a critical review of their use and give recommendations for good practice.
  • 2We searched for papers in which questionnaires were used in 57 ecological academic journals, published over the period 1991–2003 inclusive. This provided a total sample size of 168 questionnaires from 127 papers published in 22 academic journals.
  • 3Most questionnaires were carried out in North America and western Europe, and addressed species-level issues, principally focusing on mammals. The majority were concerned with impacts of species and/or their conservation, and just under half with human–wildlife interactions.
  • 4Postal survey was the method used most frequently to carry out the questionnaires, followed by in-person interviews. Some questionnaires were conducted by telephone, and none was web-based.
  • 5Most questionnaires were concerned with obtaining factual information or perceptions of facts. Ground-truthing (independent verification of the facts) was carried out in less than 10% of questionnaires.
  • 6The mean (± SE) sample size (number of respondents) per questionnaire was 1422 ± 261 and the average (± SE) response rate was 63 ± 3%. These figures varied widely depending on the methods used to conduct the questionnaire.
  • 7The analysis of data was mostly descriptive. Simple univariate methods were the most frequently used statistical tools, and data from a third of questionnaires were not subjected to any analysis beyond simple descriptions of the results.
  • 8Synthesis and applications. We provide recommendations for best practice in the future use of questionnaires in ecology, as follows: (i) the definition of the target population, any hypotheses to be tested and procedures for the selection of participants should be clearly documented; (ii) questionnaires should be piloted prior to their use; (iii) the sample size should be sufficient for the statistical analysis; (iv) the rationale for the choice of survey method should be clearly stated; (v) the number of non-respondents should be minimized; (vi) the question and answer format should be kept as simple as possible; (vii) the structure of the questionnaire and the data emerging from it should be unambiguously shown in any publication; (viii) bias arising from non-response should be quantified; (ix) the accuracy of data should be assessed by ground-truthing where relevant; (x) the analysis of potentially interrelated data should be done by means of modelling. Researchers should also consider whether alternative, interpretative methods, such as in-depth interviews or participatory approaches, may be more appropriate, for example where the focus is on elucidating motivations or perceptions rather than testing factual hypotheses.

Ancillary