Market Research MT Class 8

In-person interviews

pros: direct customer contact, good response rate, can show product, cons: expensive, safety issues

telephone interviews

pros: good segmentation, customization of questionnaire, probe customer, cons: cost, low response rates, short surveys, no showing product

Mail surveys

pros: not too expensive, targeted mailing list, longer questionnaires, cons: low response rates, manual data entry

online/text surveys

pros: consistent appearance, cheap for large sample, inputting and coding is efficient, cons: low-effort and low-attention responses, low response rate, sampling bias

survey method determinants

desired sampling precision (standard deviation), budget, respondent reaction needs, quality of data, questionnaire length, incidence rate, questionnaire structure, time to complete

3 error types

sampling, measurement, population specification

sampling error (random error)

difference between population and sample due to chance variation, reduce by increasing sample size

systematic error (bias)

difference due to intentional/unintentional mistakes, reduce with careful design/implementation

population specification (systematic error)

incorrect definition of popuatlion (too narrow/wide/different - theoretical population), incomplete/inaccurate sampling frame (study population/sampling frame), sampling procedure inappropriate/incorrectly applied (actual sample)

Measurement errors

surrogate info, non-response bias, interviewer bias, response bias, processing error

surrogate info error

difference between data needed and data sought (ask which big chains people prefer when only small chains wanted)

non-response bias

difference between respondents and non-respondents (low-income participants skip income questions)

interviewer bias

interviewer accidentally influences (“since current options are good/bad…”)

response bias

due to intentional/unintentional info misrepresentation (embarrassed to admit or mis-estimate)

processing error

errors in data recording, transcription, etc. (enter wrong response in excel)

Response error examples

ambiguous words (never, occasionally, etc), loaded questions — counteract with counter-biasing (give statistic that it’s normal)

Questionnaire mechanics

Skip patterns, branching, piping, coding, proof-read

Skip patterns

sequence of questions based on respondents’ answers

Branching

simple branching (skips some questions), dynamic branching (different question sets offered)

Piping

integrates responses from one area into another area

Coding

grouping and assigning numeric codes to various question responses

Proof read

Make sure all mechanics are followed before publishing!