- Quick - Inexpensive - Efficient - Generally accurate
- Describe (Description) - Learn why? (Explanation)
Method of Communication - Personal Interview - Self-administered survey - Telephone - Mail - Internet Temporal Dimension (Time) - Longitudinal studies (over time) - Cross-sectional studies (a point in time) Types of Question - Structured versus unstructured - Disguised versus undisguised
Characteristics - Direct communication - Questions asked face-to-face Types of Personal Interview - Door-to-door a. Response high if appointments used b. May exclude large groups of people c. Requires “call back” - Mall intercepts a. High refusal rate b. Can demo products Advantages - Opportunity for feedback - Ability to probe for details - Interview can be long - High probability of complete interview - Can use props, visual aids - Higher overall response rate Disadvantages - Interviewer influence - No Anonymity of respondent - Costs
Respondent takes the responsibility for reading and answering the questions.
Types - Mail - Package inserts “Pick-up” surveys - Internet Key Issues - Mail : non-response - Others : self-selection
Advantages Disadvantages - Speed - Low cost relative to personal - More likely to divulge personal info than in personal interview - Easier to obtain access and cooperation - Absence of face-to-face - Refusal rates may be high - Interview must be short - Lack of visuals
- Geographic reach and flexibility - Relatively low cost - Convenient for the respondent - Anonymity can be achieved - More likely to obtain sensitive information - Survey can be longer than phone survey
- Subject to different interpretations by respondents - Questions should be structured - Low response rate - Data collection is slow - Item non-response is high
Least expensive Faster turnarounds possible Quick feedback from customers and/or employees Surveys can be interactive Elimination of researchers More Internet survey software packages available Internet surveys better than website feedback forms Good response rate (e.g., telephone survey response rate has dropped to 30% or less)
Not sure who is really responding Sampling issues a. May not be representative of the population b. No guarantee that demographics are accurate c. Bad e-mail addresses d. Multiple e-mail addresses from the same person e. Inaccessibility of the poor and elderly g. Self-selection bias f. Not considered a mainstream research methodology
Cost per response Speed of return No. of interviews completed Design constraints Risk of interviewer bias Survey control Anonymity of response Mail Low Low Low Medium Low Medium High Telephone Medium High High High Medium High Medium Person-to person High Medium High Low Medium High Low Internet Low High Medium High Medium High Low
- Collection of data at different points in time - Types a. Trend b. Cohort c. Panel - Collects data at a single point in time - Often examines different “sections” of the population
- Structured questions have a limited and known set of choices - Unstructured questions are open-ended
- Disguised questions uses techniques to obtain information without directly asking the question under study.
- Disguised questions used for sensitive studies.
Check Points - Does the questionnaire provide the necessary decision making information?
- Are the measures reliable and valid?
- Does the questionnaire consider the respondent?
Avoiding Common Problems - Avoid complexity - Avoid leading and loaded questions - Be as specific as possible - Avoid double-barreled questions - Avoid making assumptions - Avoid burdensome questions
: Determine Survey Objectives, Resources, and Constraints - Survey objectives: The decision-making information sought through the questionnaire.
: Determine Data Collection Method(s) - Personal Interview - Self-administered survey - Telephone - Mail - Internet
Step 3: Determine Question Response Format
Open-ended questions: Questions that ask respondents to reply in their own words.] Closed-ended questions: Questions that ask respondents to choose from a list of answers.
a. Dichotomous questions: Questions that ask respondents to choose between two answers.
b. Multiple-choice questions: Questions that ask respondents to choose among a list of more than two answers.
c. Scale-response questions: Multiple-choice questions with the choices designed to capture the intensity of the respondents’ answers.
Four general guidelines are useful to bear in mind during the wording and sequencing of each question.
- The wording must be clear.
- Select words so as to avoid biasing the respondent.
- Consider the ability of the respondent to answer the question.
- Consider the willingness of the respondent to answer the question
Shoulds - Question should be focused on a single issue or topic Should-Nots - Should not assume criteria that are not obvious - Question should be brief - Should not be beyond the respondent’s ability or experience - Question should be interpreted the same by all respondents - Should not use a specific example to represent a general case - Question should use respondent’s core vocabulary - Question should be as grammatically simple as possible - Should not ask the respondent to recall specifics when only generalities will be remembered - Should not require respondent to guess a generalization
Questionnaire Design - Layout a. Use white space b. Attractive and easy to follow Questionnaire Order - Screens - Warm-Ups - Transitions - Order of question a. Sensitive questions at the end b. Ask general questions before specific questions c. Be aware of order bias - Complicated and Difficult to Answer Questions - Classification and Demographic Questions
Is the question necessary?
Is the questionnaire too long?
Will the questions provide the desired information to accomplish the research objectives?
Appearances of mail and self-Administered questionnaires Avoid a cluttered look Allow plenty of space for open-ended responses Consider color-coding the questionnaires Instructions printed within the questionnaire should be distinguishable from questions
: Obtain Approval from All Relevant Parties
: Pretest and Revise - Pretest: Trial run of a questionnaire.
: Prepare Final Copy Pre-testing - Trying it out on a small sample, then making changes - critical to having an easily understood questionnaire ► Translation - Best to use back translation - English-foreign language English
Supervisor’s instructions : Written directions to the field service on how to conduct the survey.
Interviewer’s instructions : Written directions to the interviewer on how to conduct the interview.
Call record sheets : Interviewers’ logs listing the number and results of a contact.
Field management companies : Firms that provide support services such as questionnaire formatting, screener writing, and data collection to full-service research companies.
Sampling Method Budget Type of questioning a. Need for stimuli b. Structure Length of questionnaire Incidence rate Time
Systematic errors-(Respondent or administrative error) - Result from some problem with the design or execution of the research. “Mistakes” Random sampling errors - Chance variations in the elements selected for the sample which make the results non representative Types of Respondent error - Non-response error - Response bias • Non-response errors - Differences between the answers obtained from those who responded as compared to the answers which would have been obtained if everyone in the sample responded.
- Includes refusals and not-at-homes - Often those that respond have - stronger feelings than those who do not.
Reducing Non-Response Error - Call-backs and follow-ups - Pre-set appointments or call ahead - Incentives - Cover letters - Prestigious sponsorship
When respondents answer untruthfully, deliberately or not Types of Response Bias - Acquiescence bias - a tendency to agree with all or most questions.
- Extremity bias - a tendency to answer on the extremes - Interviewer bias - the presence of the interviewer influences responses - Auspices bias - the sponsoring agency influences the responses - Social desirability bias a tendency to give a “socially correct” answer
Errors that result from improper administration and execution of the research Types of administrative errors - Data processing errors (Coding and entry) - Sample selection error a. Error in sample design which results in non-representative sample - Interviewer error - Interviewer cheating