Questionnaire Construction

contingency questions

-a survey questions intended for only some respondents, determined by their response -follow up question is the contingency question

precoding and data processing

-afterwards, questionnaires are coded into computer using numbers -can cut data entry costs by precoding response -reserve room for large number of open ended questions

"mask" sensitive questions

-arrange critical questions randomly in set of unrelated questions -sometimes fine to leave sensitive questions for the end

double-barreled questions

-asking about several issues at once -"Do you use condoms to avoid STD's?"

questionnaire introduction

-can be spoken or read to respondents, or a cover letter in case of mail surveys -must convince respondent to cooperate -clear and simple statement of purpose -must disarm fears or suspiciousness about motives of researcher

leading questions

-do not bias questions and lead respondents -don't associate questions with an influential group -questions should be neutral

question order

-in general, order from general to specific -depending on population, may want to start or end with demographic questions -questions about same topic should be linked together -yes/no questions v.s. Likert scale -the number of points in the Likert scale

matrix questions

-number of questions with same response options and instructions

length and fatigue

-questionnaire should be short -items simple -15 minutes or less time frame -never ask anything research design does not require

vague questions

-questions should be concrete and explicit -no ambiguity

long list of alternatives

-respondents deal best with small number of judgements (1-5, 1-10)

complex instructions

-some questions require example before reply -instructions should be no more than 3-4 sentences

"strange" response requirements

-some response measures may not be familiar to a person, so provide examples and instructions -"although a little hard, choose what fits best"

closed ended questions

-survey questions in which the respondent is asked to select an answer from among a list provided by the researcher -easier and cheaper -restrictive

coding

-the process of assigning meaning (numerical or verbal) to survey answers

social desirability

-try to phrase questions so as to avoid demand characteristics

general questionnaire format

-uncluttered -1 question per line -consistent format

question wording

-vocabulary and style (simplest language possible, avoid jargon or slang, not too long sentences, 12 year old should understand)

designing questions for survey research

1. types of questions -quality depends on the types of questions asked, the way questions are constructed, and the question format that will be used 2. question construction 3. question strategies and format

tunnel

asks series of questions organized similarly

funnel

begins with broad and open questions then narrow closed ended questions -best with sensitive material

inverted funnel

first narrow closed ended question then open ended question -course evals -long questionnaires

open ended questions

questions for which the respondent is asked to provide his/her own answers -harder to handle -rich source of information -should be used sparingly