Survey Research Questionnaire construction Types of surveys Wording Ordering Layout/format Types of surveys Self administered Interview Strengths and weaknesses
Question Wording Ask questions R’s can answer Make questions clear and unambiguous Make questions relevant to topic Avoid negative items Avoid biased items, leading questions Avoid double-barreled questions Short items usually better
Question Ordering Goal: To get R to answer the questions Self-Administered Begin with interesting, non-threatening items Less interesting middle to end Most personal information at the end Interview Begin with least threatening items Most personal items in the middle Finish with less sensitive items
Format/ Layout Neat, uncluttered Easy for R to follow One question per line Answers marked in boxes or circles: ( ) [ ]
Contingency Questions Apply to only some of the respondents Set off with boxes and arrows or use instructions to skip questions: Do you have children under 18 at home? No Yes If yes: please list ages_____________ _______________________________
Alternative format for contingency questions 4. Do you have children under 18 at home? No please skip to question 15 Yes please go on to question 5 5. How old is your youngest child? 6. What child care arrangements do you use while at work? (Check all that apply)….. 7………
Matrix Questions Set of questions with identical answer choices Example: Strongly agree Agree Don’t know Dis-agree Strongly Disagree Husbands and wives should be equally responsible for financial support of their families. 5 4 3 2 1 A woman should take care of her family first, then her career. Preschool children may be harmed if their mothers work. A father can care for young children just as well as a mother can.
Using matrix questions Easy to follow Efficient use of space May foster “response set” – solution is to alternate points of view occasionally Answer choices should be appropriate for all items
Self-Administered Questionnaire Mail, e-mail, web page, or handed out May not produce high response rate May be good for sensitive topics Less costly than interviews
Insuring quality Use probability sample Precede with cover letter. Include letter of explanation Monitor returns, use follow-ups if needed Always pay return postage!
Follow-ups on mail surveys Maintain confidentiality – use ID numbers Keep list of returns After reasonable time, send another questionnaire May follow up with phone call Improves response rate May allow conclusions about nonrespondents
Interviews Questions read by researcher Answers recorded by researcher May be face-to-face or telephone Higher response rate than mail surveys More accurate responses – interviewer may ask for clarification Better for more complex topics Interviewer can observe and record without asking directly (but is it ethical?)
Two main rules for interviewers Get R’s confidence: Good listener, non-judgmental, non-intimidating, appear interested in R’s opinions, dress appropriately, may be same gender as R Don’t affect the results: Don’t volunteer own opinion, don’t appear as authority or expert, don’t judge R.
Interviewer preparation Know purpose of study, sponsor, researcher Know how much to reveal to R Questionnaire must include interviewer instructions Follow wording and instructions exactly. Record responses exactly. Use probes only as instructed. Interviewers should be supervised.
Example of interviewer instructions: http://www.macalester.edu/~dolan/POLI30/interviewerinstructions.html
Telephone Interviews Use for short, uncomplicated topics Saves money and time May precede with cover letter – avoid being seen as “nuisance call” Raises sampling issues – cell phones, unlisted numbers, people without phones
Computer Assisted Survey Research: Computer Assisted Telephone Interviewing (CATI) Like regular telephone interview, but interviewer types respondent’s answers directly into computer Computer Assisted Personal Interviewing (CAPI) Like CATI, except used in face-to-face interview Computer Assisted Self Interviewing (CASI) Research worker brings a computer to the respondent’s home Respondent answers on the computer Computerized Self-Administered Questionnaire (CSAQ) The respondent receives electronic copy of the questionnaire Respondent returns the data file Also: TDE (touch-tone data entry), VR (voice recognition)
Advantages and Disadvantages of Surveys Good for describing large populations; statistical inference Can handle large data sets and samples Standardization = easy analysis Can have multiple indicators of concepts Relatively inexpensive Disadvantages Can be overstandardized; categories may be artificial May create opinion