Dr. Friedman
Study Sheet for Marketing Research
Philosophy of science, Falsificationism (Karl Popper)
focus group interview: phenomenological, exploratory, clinical.
depth interview
qualitative research
experimentation
control group
before and after design
before and after with control group design
after only with control group design
experimental errors--e.g., history, maturation,...
open-ended vs. closed-ended questions
leading and loaded questions
two-in-one questions
nominal, ordinal, interval, ratio scales
computing the mean and standard deviation for rating scale questions.
95% confidence interval estimate for the mean
graphic vs. itemized rating scales
balanced and unbalanced rating scales, comparative vs. noncomparative rating scales.
forced vs. nonforced choice rating scales
attitude scales, Likert scales
semantic differential scales--aggregate and profile analysis.
properties of attitude scales--reliability and validity
test-retest reliability, alternate forms reliability, consistency, Cronbach's coefficient
alpha.
predictive validity, concurrent validity, convergent validity.
Major surveying techniques: mail, telephone, personal interview, mall intercepts, computer
interactive interviewing, e-mail surveys, fax surveys, personally delivered
self-administered qauestionnaires.
non-response bias, response bias
increasing rate of response to mail surveys
random digit dialing, plus-one telephone sampling
interviewer effects
computer searches--DIALOG: ABI/INFORM, MANAGEMENT CONTENTS, NEWSEARCH, .
Important marketing research companies: Nielsen; Arbitron; Information Resources, Inc.;
etc.
Descriptive statistics: Mean, median, mode, standard deviation, variance
Statistical tests:
(1) two-sample t-test--to compare the means of two groups
(2) one-way ANOVA--to compare the means of two or more groups
(3) Pearson correlation-- to see whether two interval-scale measurements are related, e.g.
weight and blood pressure. The correlation coefficient ranges from -1 to +1.
(4) Chi-square test for independence (contingency table or crosstabulation) is used to
determine whether two nominal variables, e.g. sex and region, are independent or related.
Cramer's V statistic (coefficient) ranges from 0 to 1 and is used with a crosstab to
measure strength of the relationship.
Probability Samples--SRS, systematic sample, stratified random sample, cluster sample,
area sample
intraclass (intracluster) correlation coefficient
Nonprobability samples--judgment and convenience, mall intercept
advantages and disadvantages of each type of sample.
Computer Searching on the web: use the following site to find multiple search engines.
http://www.merrydew.demon.co.uk/search.htm
Other good sites are: