A note on questionnaire design and adminstration Dr D.K. Nyahunzvi Midlands State University 2012
Questionnaire design Need to be well-thought out Influences response rate and quality of data collected Test the questionnaire’s validity first Make questionnaire acceptable to target population Decide on mode of adminstration
People respond favourably to issues that are relevant to them Refer back to your objectives Questions must be directly related to the research question Others add to the flow of the questioning
Wording of the questions has a direct bearing on responses given Stick to the 3s: short, simple and specific questions More difficult questions are problematic Pay particular attention to the wording f the question wil either produce an inaccurate response or more likely the respondent wil give up and fail to complete the questionnaire
Avoid assumptions in your questions Time memory bias Can use software packages to calculate the readability of your questions Avoid assumptions in your questions Time memory bias Instead break into do you visit the dent. For check ups. If yes. How often do you go for chekp. If unavoidable use time-scales eg 6-12 months or put it in some context whe elections happeneed e.g How often do you visit a dentist people find it difficult to recall esp. more than 6months
Take note of order of questions Studies have shown better response rates If general precedes specific (sensitive) questions Decide on either open/closed or a mixture
Design Layout Should facilitate data coding and analysis Capture attention Make people interested in completing the questionnaire Avoid cramming too many questions on a page
Divide into sections for readability Print on good quality paper to grab attention Instructions should be at the beginning and not throuthout to enable easy navigation
Pre-pilot testing Increases confidence about the specificity of your questions Test layout to ensure people can navigate easily Testing is usually overlooked in the researcher’s eagerness to start collecting the data that your questions are unambiguous, appropriate and acceptable
Test the validity Test replicability Test acceptability does it measure what it purports to replicability-the reproducibility and consistency of an instrument Face and factual validity and criterion validity; face validity-after the interview to find out if the rresp. Has given his real opininions Test ? On 2 separate times-use Kappa for caytegorical data and Spearma Correlation Coefficient for continuous dtaa Internal consistency of ?=ask the ? More than one way=see responses Ask how did you fid the ?
Editing To identify and eliminate errors made by respondents e.g. clerical Check data for completeness-is there any answer for every question Cross check missing answers from other sections of the survey/ Contact respondents again
Check accuracy of responses-could be respondents were careless If these errors are not picked up in the editing process; reduces validity Check for uniformity-respondents followed instructions and questions uniformly
Coding/data reduction Is the process of converting data into meaningful categories to facilitate analysis Preferable to build your coding scheme into your questions Discuss the coding scheme with supervisor/statistics person Use software packages to facilitate entering data directly from your questionnaire
Data cleaning Once data is in database, it must be cleaned Identifying inconsistencies and outliers Is a laborious process, there is a strong temptation to leave this stage Cleaning removes data entry errors Simple cross tabulations can be used to indentify nonsensical responses
Data Analysis Refer back to your objectives to keep your analysis focused Many types of analysis: Descriptive-describe the distribution and range of responses to each variable Also involves examining data for skewness
Bivariate analysis use simple cross-tabs. to identify tends and possible associations between 2 variables Multivariate analysis/regression analysis techs. to test the effect of one variable on an outcome whilst controlling the other
Hints on maximising returns Postal questionnaires Include stamped addressed envelope Use multiple reminders to request returns Stress importance of the benefits to the target group e.g. minority that is struggling to be heard Provide interim data from returns to non-returns to keep them engaged in the research
Follow-up in a personalised way or phone call Tailor follow-up requests to individuals Indicates there are personally known and are important to the research Avoid blanket generalised letters
Address features of the questions ease of completion, sensitivity of questions, time to spend answering length of questionnaire
Understand the nature of sample for effective targeting strategies Adequate pilot testing Sending subjects information in advance about the study Ask subjects prior to sending if they wish to participate