2.研究设计

目录

Research Methods

Qualitative Research in Exploratory Research

Focus Group Study

Depth Interview

Projective Techniques

Causality

Controlled Experiment in Causal Research

Mediation Vs Moderation

Validity of Experiment

---

Research Methods

Objective of Research	Characteristics	When to use	Types
Exploratory 탐색적	tentative less structured/flexible small number of respondents	가설 창출 不知道有哪些变量 insights, ideas lack of knowledge	– Literature search: secondary data – Experience survey/ interview: expert leads discussion – Case study: case reflecting abrupt changes   Other qualitative research [Direct/Nondisguised] – Focus group interview (FGI) – Depth interview [Indirect/Disguised] – Projective techniques
Descriptive 기술적	conclusive	가설 검증 已经知道有哪些重要的变量 summary measure (quantify) provide a snapshot of market env.	In terms of data type – Survey (communication) – Observation   In terms of repetition – Longitudinal (panel) 同一样本，over time – Cross-sectional     single~, also called 'sample survey research'    multiple~   In terms of output – Frequency distribution (dummy tables) – Relationship among variables (ANOVA, CrossTab, Regression, ...) – Data reduction (factors, MDS, ...)
Causal 인과적	conclusive	가설 검증 managerial decision->market outcome?	Controlled Experiment – Laboratory experiment – Field experiment   Uncontrolled Experiment – Natural experiment

 

Qualitative Research in Exploratory Research

Focus Group Study

 

Objectives	– Generate new ideas, exploring new  product concepts; – Uncovering customer attitudes,  “language of consumers” – Refining hypothesis.
Size	8‐12 people
Time	1~3 hrs
Process	• Moderator ensures relevant topics are covered. • Group interaction, open discussions. • Relatively homogeneous groups; Multiple  groups to get heterogeneity in perspective. • Research report summarizes “what was said”; careful in interpreting the results.
Advantage	• Quick information  • Stimulate new ideas • Interactive • Relatively small total cost (but cost/respondent is high)
Disadvantage	• Recruiting problems. • Hard to quantify results. • Small sample size. • Can't generalize to target population.

Depth Interview

Technique
Laddering	• product characteristics -> user characteristics • tap into the consumer's network of meanings.
Hidden Issue Questioning	• socially shared  values×  personal “sore spots”√ • general lifestyles×  deeply  felt personal concerns √
Symbolic Analysis	• comparing with their opposites. – non‐usage of the product – attributes of an imaginary “non‐product,” – opposite types of products

Projective Techniques

Respondents are  asked to interpret the behavior of others.  

Respondents indirectly project their own motivations, beliefs,  attitudes, or feelings into the  situation.

Word Association	• respondents are presented with a list  of words, one at a time, and asked to respond to each with the first  word that comes to mind. • test  words, are interspersed throughout the list which also contains some  neutral, or filler words to disguise the purpose of the study.   analyzed by calculating: (1) frequency (word) (2) response time   (3) # of no response within a reasonable time (have emotional involvement so high that it blocks a response)
Completion Technique	• Sentence completion • Story completion
Construction Technique	• Picture Response:  to describe a series of pictures of ordinary as well as unusual  events.   • Cartoon Tests:  to indicate what one cartoon character might say in  response to the comments of another character.  simpler to administer and analyze than picture response.
Expressive Technique	presented with a verbal  or visual situation and asked to relate the feelings and attitudes of  other people to the situation. • Role playing • Third‐person technique
Pros	• elicit responses that subjects would be unwilling or unable to give if they knew the purpose of the study. (↑validity of response) • Helpful when the issues to be addressed are personal, sensitive, or subject to strong social norms. • Helpful when underlying motivations, beliefs, and attitudes are operating at a subconscious level.
Cons	• Typical disadvantages of unstructured direct techniques, but to a greater extent. • Require highly‐trained interviewers and skilled interpreters • a serious risk of interpretation bias. • expensive. • May require respondents to engage in unusual behavior. (↓generalizability)

Causality

Concept:

• X would be one of several causes of Y

• Y may not occur in a deterministic fashion, may be probabilistic (X↑， % of Y↑)

• Can never prove, but only infer

 

Evidence:

1. Concomitant variation

           – If X changes, then Y also changes (together)

1. Time order of occurence

           – X occurs before Y

1. Elimination of alternative explanation

 

Controlled Experiment in Causal Research

Most important components

• Control: eliminate the intervention of other variables

• Randomization: manipulation of IV should be random to subjects

 

• Treatment Group: subject exposed to manipulation(experimentation)

• Control Group: part of sample that sees no change in IV

 

Multiple IV

• Why? To investigate interaction effect

 

Suppose we have  X1, X2

• Two main effects: 

(1) effect of X1 on Y

(2) effect of X2 on Y

 – main effects can be studied in two separate  experiments

 

• Interaction effect : when effects of  one IV depend on the level of other IVs

Eg. DV: Sales, X1: price (LP, HP), X2: ads (LA, MA,  HA)

Main effect->看average

 

Mediation Vs Moderation

매개효과 vs 조절효과

 

Validity of Experiment

• Internal validity: elimination of other  possible causes.

   – Goal: draw valid conclusion on the study group

   – Lowered by: external factors  (financial crisis); maturation of subjects (agents more experienced), and  so on.

   – Lab experiment > Field experiment.

• External validity: generalizability

   – Goal: make valid generalization to a larger population of interest

   – Lowered by: artificial situation;  inappropriate sample; inappropriate time, and so on.

   – Field experiment > Lab experiment:

 

Major threats to Validity (Extraneous Variables)

• History: extraneous factor occurs  between two measurements

• Maturation: changes in subjects with  the passage of time (older, experienced, tired, …)

• Testing effects: effects due to  subjects’ knowledge of the purpose of study of being tested

• Instrumentation: changes in measuring  instrument

   Consider an experiment in which dollar sales are being measured before and after exposure to an in-store display  (treatment). If there is a nonexperimental price change between O1 and O2, this results in a change in  instrumentation because dollar sales will be measured using different unit prices.

• Statistical regression (a.k.a.  regression to the mean) – extreme people move closer to average during the course  of experiment

   In the advertising experiment, suppose that some respondents had either very favorable or very unfavorable attitudes. People with extreme attitudes have more room for change, so variation is moren likely. This has a confounding effect on the experimental results, because the observed effect (change in attitude)   may be attributable to statistical regression rather than to the treatment (test commercial).

• Selection bias : failure in  randomization

  For example, consider a merchandising experiment in which two different merchandising displays (old and new) are assigned to different department stores.  The stores in the two groups may not be equivalent to begin with. They may vary with respect to a  key characteristic, such as store size. Store size is likely to affect sales regardless of which  merchandising display was assigned to a store.