Transcript Measurement
Descriptive Research &
Questionnaire Design
Descriptive Research
Survey versus Observation
Survey
Primary data collection method based on communication with
a representative sample of individuals
Observation
Systematic process of recording behavioral patterns of people,
objects, and occurrences without questioning or
communicating with them
Involves recording -- someone (something) has to do the recording
Deals with behaviors -- can’t observe attitudes, emotions, etc.
Broad-based -- may observe, people, objects, occurrences
Survey Research
Methods of Administration
Personal Interview -- face-to-face communication
between an interviewer and respondent
Telephone Interview -- information is gathered
over the telephone
Mail Questionnaire -- self-administered survey
sent via the mail
Electronic Questionnaire -- survey distributed and
gathered via e-mail or the Internet
Survey Research
Personal interviews
Advantages
Probing ability
Longer survey instruments
Completeness
Use visual aids
Disadvantages
Interviewer bias
Not anonymous
Expense
Ability to set up specific times for call backs
Survey Research
Telephone interviews
Advantages
Efficiency
Probing ability
Ability to have “pure random samples”
Ability to set up specific times for call backs
Direct data entry (CATI)
Disadvantages
Limited reach (not all have telephones)
Cell phone issues
Annoyance
Limited duration
No visuals
Survey Research
Mail questionnaires
Advantages
Low in cost
Respondent convenience & anonymity
No interviewer bias
Longer surveys
Disadvantages
Lack of speed
Low response rates (usually below 50%; often much lower)
No interviewer
Cost can get high
Survey Research
Electronic Questionnaires
E-mail surveys
Convenient
Limited reach
Best used in conjunction with another method (i.e., mail or
telephone)
Web-based surveys
Convenient
Limited reach
Some visual aids may be used
Best application:
In conjunction with another method
Panels
Primary Data
When to use communication versus
observation?
When data collection needs to be versatile
Have a wide range of data needs
Want to know needs, ideas, descriptions of customers,
for example
Able to gather by “communicating” with a respondent
Some data cannot be observed (e.g., attitudes)
Primary Data
When to use observation versus
communication?
When a respondent’s willingness to provide
information is low
When the potential for interviewer bias is high
Certain types of data points can only be observed
(e.g., behaviors)
Survey Research
Cross-Sectional versus Longitudinal Designs
Cross-sectional -- snapshot of population at one
point in time
Longitudinal -- fixed sample of population
elements repeatedly measured over time
Allows for tracking of certain changes over time (e.g.,
brand switching, brand loyalty)
Observation-Based Research
Nature of observation
No interaction
Limited non-response do to unwillingness to participate
Little social desirability
If they do not know – they can’t act differently
Characteristics of observation
Sometimes people do not know they are being observed
Objectivity
Accuracy
IKEA research
Questionnaire Design
What is a questionnaire?
Formalized schedule for collecting data from
respondents
Outlines information to be gathered
Key criteria of a good questionnaire
Relevance to the problem at hand
Accuracy in terms of its measures
Questionnaire Design
Components of a Questionnaire
ID data
Request for cooperation
Instructions
Information sought
Classification data
Questionnaire Design
Procedure for developing a questionnaire
More of an art than a science (We will cover the
following steps)
Specify the information to be sought
Determine the type of questionnaire and method of
administration
Determine content of individual questions
Determine response form to each question
Determine specific wording of each question
Questionnaire Design
Procedure for developing a questionnaire
(cont’d)
Determine question sequence
Determine physical characteristics of the
questionnaire
Re-examine the above steps and revise (if
necessary)
Pretest questionnaire and revise
Questionnaire Design
Information sought
Guided by the problem definition and the
hypotheses
Be careful to focus on the problem at hand and
the necessary hypotheses
“Interesting” information will only make the questionnaire
longer which will likely contribute to non-sampling error
Questionnaire Design
Type of questionnaire and method of
administration
Depends on the problem at hand
Structured versus unstructured depends on the problem
definition
Questionnaire Design
Determine content of individual questions
Is a question necessary?
Are several questions necessary, or just one?
Do typical respondents have the necessary
information to answer the question?
Example -- Opinion Metallic Metals Act of 1947
If they don’t they will still answer it
Questionnaire Design
Determine content of individual questions
(cont’d)
Will respondents provide the information?
Options for sensitive questions
Hide the question
State the question using terms that suggest the behavior or
attitude in question is not unusual
Phase the questions in terms of others
State the question with response categories (easy to check
a response box)
Randomized Response Model
Questionnaire Design
Randomized Response Model
Procedure
At random, respondent receives questionnaire
containing either SQ or IQ
We know the probability of “Yes” answer to IQ is 1/12 =
.08
Use the following formula to calculate the true
percentage “Yes” answers to either SQ or IQ
Example – Assume number of “Yes” answers (to the IQ
and SQ) was .20
Questionnaire Design
Questions that respondents may not answer accurately (i.e.,
sensitive questions)
Avoid?
Use Randomized Response Model
Example:
Issue to be known – Have you ever been convicted of a felony?
Sensitive Question (SQ):
Have you ever been convicted of a felony?
_____ Yes
_____ No
Innocuous Question (IQ): (But must know probability of the answer)
Is your birthday in January?
_____ Yes
_____ No
Questionnaire Design
Form of the response
Open-ended
Fixed alternative
Dichotomous
Yes/No
Multi-chotomous
Strongly Agree
Agree
Disagree
Strongly Disagree
Questionnaire Design
Question wording
Simple words
Avoid ambiguous words
Avoid leading questions
Avoid implicit alternatives
Avoid estimates
Avoid double-barreled questions
Questionnaire Design
Question Sequence
Simple & interesting opening questions
Funnel approach
Start broad and get narrow
Difficult or sensitive questions late
Classification information last
Questionnaire Design
Determine physical characteristics
Securing response – this is most important
Facilitate handling and control
Re-examination and revision
Pretests