Menu
Home
Log in / Register
 
Home arrow Environment arrow Research Methods in Anthropology: Qualitative and Quantitative Approaches
Source

STEPS IN DILLMAN'S METHOD

1. Professionalism: Mailed questionnaires must look thoroughly professional. Jaded, hard-bitten, oversurveyed people simply don’t respond to amateurish work. Print questionnaires in booklets on standard-size paper: 8.5" X 11" in the United States and analogous size (like A4) in the rest of the world. Dillman recommends using white paper. Fox et al. (1988) reported good results with light green paper, but Beebe et al. (2007), in a controlled field test, found that white paper and smaller booklet size had the best response rate.

You must be thinking: “Controlled tests of paper color?” Absolutely. It’s because social scientists have done their homework on these little things that a response rate of 70% is achievable—provided you’re willing to spend the time and money it takes to look after all the little things. Read on and you’ll see how small-but-important those ‘‘little things’’ are.

2. Front and back covers: Don’t put any questions on either the front or back covers of the booklet. The front cover should contain a title that provokes the respondent’s interest, the name and address of the survey's sponsor, and some kind of eye-catching graphic design or photo. By ‘‘provoking interest,’’ I don’t mean “threatening.” A title like ‘‘The Greenville Air Quality Survey’’ is fine. ‘‘Polluted Air Is Killing Us’’ isn’t.

Be careful in the use of photos—they contain an enormous amount of information, and you never know how respondents will interpret the information. If a respondent thinks a photo contains an editorial message (for or against some pet political position), then the survey booklet goes straight into the trash. If you don't have an appropriate photo, then use a graphic design.

The back cover should have eye-catching photos or designs, and a brief note thanking the respondent and inviting open-ended comments about the questionnaire. Nothing else.

3. Question order: Be sure that the first question is directly related to the topic of the study (as determined from the title on the front of the booklet); that it is interesting and easy to answer; and that it is nonthreatening. Once someone starts a questionnaire or an interview, they are likely to finish it. Introduce threatening questions well into the instrument, but don't cluster them all together.

Put general socioeconomic and demographic questions at the end of a questionnaire. These seemingly innocuous questions are threatening to many respondents who fear being identified (Sudman and Bradburn 1982). Once someone has filled out a questionnaire, they are unlikely to balk at stating their age, income, religion, occupation, etc.

4. Formatting: Mailed surveys have to look good and be easily readable or they get tossed out. Use bolded letters for instructions to respondents and plain text questions themselves and line answers up vertically rather than horizontally, if possible.

Q26. During the past five years how much better or worse has Kinshasa become as a place to live?

  • ? A lot better
  • ? Somewhat better
  • ? No change
  • ? Somewhat worse
  • ? A lot worse
  • ? Not sure

Never allow a question to break at the end of a page and continue on another page.

5. Length: Keep mailed questionnaires down to 12 pages, with no more than 125 questions—that’s three sheets of 11" X 17" folded in half and printed on both sides. Beyond that, response rates drop (Dillman 1978).

It is tempting to save printing and mailing costs and to try to get more questions into a few pages by reducing the amount of white space in a self-administered questionnaire. Don’t do it. Respondents are never fooled into thinking that a thin-but-crowded questionnaire is anything other than what it seems to be: a long questionnaire that has been forced into fewer pages and is going to be hard to work through.

Use lots of open space in building schedules for personal interviews, too. Artificially short, crowded instruments only result in interviewers missing items and possibly in annoying respondents (imagine yourself sitting for 15 minutes in an interview before the interviewer flips the first page of an interview schedule).

6. The cover letter: A one-page cover letter should explain, in the briefest possible terms, the nature of the study, how the respondent was selected, who should fill out the questionnaire (the respondent or the members of the household), who is funding the survey, and why it is important for the respondent to send back the questionnaire. (‘‘Your response to this questionnaire is very important. We need your response because. . . .’’)

The one thing that increases response rate more than any other is university sponsorship (Fox et al. 1988). University sponsorship, though, is not enough. If you want a response rate that is not subject to bias, be sure to address the cover letter directly and personally to the respondent—no ‘‘Dear Respondent’’ allowed, unless you only have addresses, without names—and sign it using a blue ballpoint pen. Ballpoints make an indentation that respondents can see—yes, some people do hold those letters up to the light to check. This marks the letter as having been individually signed. In Japan, Jussaume and Yamada (1990) signed all their letters with an inkan, or personal seal, and they wrote the address by hand on the envelope to show that they were serious.

The cover letter must guarantee confidentiality and must explain the presence of an identification number (if there is one) on the questionnaire. Some survey topics are so sensitive that respondents will balk at seeing an identification number on the questionnaire, even if you guarantee anonymity. In this case, Fowler (1984) recommends eliminat?ing the identification number (thus making the questionnaire truly anonymous) and telling the respondents that they simply cannot be identified.

If you do this, enclose a printed postcard with the respondent’s name on it and ask the respondent to mail back the postcard separately from the questionnaire. Explain that this will notify you that the respondent has sent in the questionnaire so that you won’t have to send the respondent any reminders later on. Fowler (1984) found that people hardly ever send back the postcard without also sending back the questionnaire.

  • 7. Packaging: Package the questionnaire, cover letter, and reply envelope and postcard in another envelope for mailing to the respondent. Print the respondent’s name and address on the mailing envelope. Avoid mailing labels, unless the envelope is too big to fit in your printer. Use first-class postage on the mailing envelope and on the reply envelope. Some people respond better to real stamps, especially bright commemorative stamps, than to metered—even first-class metered—postage (Hensley 1974).
  • 8. Incentives: What about incentives to complete a survey? Here, the research is unambiguous: money talks and a prepaid incentive works better than a promise of one (Church 1993; Dillman et al. 2009:275; Warriner et al. 1996). Mizes et al. (1984) found that offering respondents $1 to complete and return a questionnaire resulted in significantly increased returns, but offering respondents $5 did not produce a sufficiently greater return to warrant using this tactic. In 1984, $5 was close to the value of many respondents’ time for filling out a questionnaire. This makes responding to a survey more like a strictly economic exchange and, as Dillman pointed out, makes it easier for people to turn down (1978:16).

In other words, despite inflation (the $5 in 1984 would be about $11 now) there is a Goldilocks solution to the problem of how much money to send people as an incentive to fill out and return a survey. If you send people too much money or too little, they may throw the survey away. If you send them just the right amount, they are likely to fill out the survey and return it. Today, except for special populations (like physicians) incentives between $1 and $5 are the norm in the United States.

9. Contact and follow-up: This is crucial. Send a letter, by first-class mail, to each respondent explaining the survey and informing the respondent that a questionnaire will be coming along soon. You can send people an e-mail message to tell them that a letter is coming, or you can follow up with an e-mail to ask if people have questions, but don’t skimp on sending real invitation letters through the mail (Converse et al. 2008). The pre-notice letter should arrive just a few days to a week before the actual survey shows up. A postcard thank-you/reminder to all potential respondents should arrive a week after sending out the questionnaire. The card thanks the recipient if they’ve already sent the survey back (and it’s crossing the postcard in the mail) and reminds them to fill out the survey if they haven’t yet done so. Don’t wait until the response rate drops before sending out reminders. Some people hold on to a questionnaire for a while before deciding to fill it out or throw it away. A reminder after 1 week stimulates response among this segment of respondents (Dillman, Smyth, and Christian 2009:250).

Send a second cover letter and questionnaire to everyone who has not responded 3 weeks later, and include another copy of the questionnaire. There is no incentive sent with this packet and the tone of the letter is more urgent. For example: ‘‘We really need everyone’s cooperation in order to make sure that the result represents people in your area.’’ Finally, 4 or 5 weeks after the postcard, send another cover letter and copy of the questionnaire, but this time send it by courier (FedEx, UPS) or special delivery from the post office.

 
Source
Found a mistake? Please highlight the word and press Shift + Enter  
< Prev   CONTENTS   Next >
 
Subjects
Accounting
Business & Finance
Communication
Computer Science
Economics
Education
Engineering
Environment
Geography
Health
History
Language & Literature
Law
Management
Marketing
Mathematics
Political science
Philosophy
Psychology
Religion
Sociology
Travel