Marketing Research

1
Marketing Research

• Dr. David M. Andrus
• Exam 1
• Lecture 6

2
Themes of My Presentation

• Sources of Error in Research Designs
• Random Sampling Error
• NonSampling Error
• Response Error
• Nonresponse Error

3
Potential Sources of Error in Research Designs
Total Error
Random Sampling
Nonsampling Error
Response Error
Nonresponse Error
4
Potential Sources of Error in Research Designs
Response Error
Researcher Errors
Interviewer Errors
Respondent Errors
-Surrogate Information -Measurement -Population
Definition -Sampling Frame -Data Analysis
-Inability error -Unwillingness error
-Questioning error -Recording error -Cheating
error
5
Field Work/Data Collection Process
Selection of Field Workers
Training of Field Workers
Supervision of Field Workers
Validation of Field Workers
Evaluation of Field Workers
6
Errors Related to Field Work
Interviewer Errors
Nonresponse Errors
Refusals Not-at-Homes
Respondent Questioning Recording Cheating Selectio
n Error Error Error
7
Methods for Handling Field Errors

• Match background characteristics of interviewer
  and respondent
• Make interviewer instructions clear and written
• Conduct practice training sessions
• Examine interviewers understanding of the
  studys purpose and procedures
• Have interviewers complete the questionnaire and
  examine the replies to see if there is any
  relationship.
• Verify a sample of each interviewers interviews.

8
Methods for Handling Nonresponse Errors

• Not-At-Homes
• Have interviewers make advance appointments.
• Call back at another time on a different day.
• Attempt to contact designated respondent using
  another approach.

9
Methods for Handling Nonresponse Errors

• Refusals
• Attempt to convince respondent of the value of
  research and importance of their participation.
• Provide advance notice, set a time limit to
  complete the questionnaire and identify the
  sponsor.
• Guarantee anonymity and include only vital
  questions.
• Provide incentive for participation.
• Try to get a foot-in-the-door.
• Use personalized cover letters and follow-up
  contact.
• Adjust the results to account for nonresponse.
• Give the questionnaire a professional appearance
  and use short questions.
• More expensive mailing methods decrease refusals

10
Methods for Handling Noncoverage Errors

• Improve basic sampling frame using other sources.
• Select sample to reduce ineligibles on a list.
• Adjust the results by appropriately weighting the
  subsample results.

11
Methods for Handling Office Errors

• Use field edit to detect the most glaring
  omissions and inaccuracies in data.
• Use a second edit to decide how data collection
  instruments containing incomplete answers and
  obviously wrong answers are to be handled.
• Use closed questions to simplify the coding

12
Methods for Handling Office Errors

• When open-ended questions with multiple coders
  are used, divide the task by questions.
• Have each coder code a sample of the others
  work to ensure a consistent set of coding.
• Follow established conventions when coding the
  data for computer analysis
• Prepare a codebook that lists the codes for each
  variable and the categories included.
• Use appropriate methods to analyze the data.

13
Improving Response Rates
Reducing Refusals
Reducing Not-At-Homes
Prior Notification
Incentives
Follow-up
Prior Notification
Motivating Respondents
Follow-up
Other Facilitators
Questionnaire Design and Administration
Questionnaire Design and Administration
Other Facilitators
Motivating Respondents
Incentives
Callbacks
Callbacks
14
Potential Survey Problems

• Total Error Difference between the true
  information being sought and the collected
  information resulting from the measurement
  process.
• Systematic Error Caused by a constant bias in
  the design or implementation of the measurement
  instrument.
• Random Error Caused by inconsistency in
  respondents or their circumstances.
• Collected Information True information
  Systematic error Random error

15
Pre-Sampling Errors

• Population Definition Error Results when the
  population is wrongly defined.
• Sample Selection Error Occurs if we select a
  sample that is not representative of the
  population.
• Measurement Instrument Error Occurs when the
  survey contains questions or statements that bias
  respondents or make it difficult for the
  researcher to clearly understand intended
  responses.

16
Present-Sampling Errors

• Procedure Error When sample subjects are
  improperly selected.
• Nonresponse Error When a high percentage of
  respondents do not participate.
• Response Error When respondents answer
  particular questions or statements incorrectly.
• Interviewer Error Whenever a person administers
  a survey, there is a chance that the interviewer
  will influence responses by providing additional
  information.

17
Post-Sampling Errors

• Information-Processing Error Caused by mistakes
  in coding or inputting data into a computer for
  analysis.

18
Selecting A Data-Collection Method

• Budget Allocation
• Time Availability
• Demographic Composition of Targeted Respondents
• Communication Vehicle Availability
• Need for Product or Service Demonstration
• Survey Content and Length
• Survey reliability, validity, and
  generalizability
• Survey Structure
• Desired Response Rate
• Desired Sampling Precision

19
Response Rates

• Mail surveys usually produce lower response rates
  than telephone surveys.
• Telephone surveys usually produce lower response
  rates than personal interview surveys.
• Well designed surveys can produce response rates
  of 70 (mail), 75 telephone, and 80 (personal
  interviews).