Core Course for Marketing
Marketing Research
Instructor：Dan Huang
Grading & Requirements
★ Class Participation 36%
★ Project 34%
★ Final Exam 30%
Class Participation: Guidelines
Adhere to Antai Discipline for Class
Turn off cell phones!
No laptops please!
Be prepared! Use study questions.
Prepare with friends.
Develop a voice in class. Be wrong (on occasion!)
Engage with your classmates.
Debate with/build on one another.
Use common information (cases, readings) so we are all on the same page.
Class Participation: Grading
Absent without notification= -2 points
Lack of preparation= -1 point
Absent with notification= 0 points
Silent or just mention a case fact = 1 point
Made substantive contribution= 2 points
Made a very significant contribution= 3 points
Introduction to Marketing Research
Class 1
Petrol War:
CNPC vs Sinopec
↓ ↓
↓ ↓
?
Feb. 2, 2009,7pm
Introduction to Marketing Research
Definition:
Introduction to Marketing Research
Systematical
Objective
Identification
Collection
Analysis
Dissemination
Use of Information
Improve Decision Making
Introduction to Marketing Research
The role of marketing research
Uncontrollable Environmental Factors
Economy
Technology
Laws and regulation
Social and cultural factors
Political factors
Customer Groups
Consumers
Employees
Channel Members
Suppliers
Controllable Marketing Variables
Product
Pricing
Promotion
Distribution
Marketing Research
Marketing Decision Making
Marketing Managers
·Market Segmentation ·Market Programs
·Target Market Selection ·Performance and control
Assessing
Information
Needs
Providing Information
Introduction to Marketing Research
General Business
Environment
Finance &
Accounting
MIS
Investigation
Marketing
Location
M & A
Management and
organization
Behavior
Corporate
Responsibility
Strategic
Planning
Capital
Portfolio
Investment
Portfolio
Asset Pricing
Credit Risk
Cost Analysis
Product
Development
Pricing
Advertisement
Channel
Market segment
Customer
Satisfaction
Data Mining
Technology
Supporting
Computer
system
Staff
satisfaction
Corporate
Culture
Salary
Management
Law、zoology & Ethic
Scope of Marketing Research
Problem identification & problem solving
Keywords： ☆Data ☆Information ☆knowledge ☆Concept ☆proposition ☆Theory
Introduction to Marketing Research
Marketing research cycle
Introduction to Marketing Research
Defining the problem
Developing an approach to the problem
Formulating a research design
Doing field work or collecting data
Preparing and analyzing data
Preparing and presenting the report
Defining new problem
……….
Defining the problem
Introduction to Marketing Research
Tasks Involved
Discussions with Decision Makers
Discussions with Decision Makers
Discussions with Decision Makers
Discussions with Decision Makers
Environmental Context of the Problem
Step 1: Problem Definition
Management-Decision Problem
Marketing Research Problem
Step 2: Approach to the Problem
Analytical Framework and Models
Research Questions and Hypotheses
Specification of Information Needed
Step 3: Research Design
Approach to the problem
Introduction to Marketing Research
Management-Decision Problem vs Marketing Research Problem
☆Ask what the decision ☆Ask what information is
maker needs to do needed and how it
should be obtained
☆Action oriented ☆Information oriented
☆Focuses on symptoms ☆Focuses on the underlying
causes
Marketing Research Problem
Introduction to Marketing Research
Broad Statement
Component
2
Component
1
Component
3
Components of approach to marketing research
1. Analytical framework and models (verbal, Graphical, Mathematical models)
2. Research questions and hypotheses
3. Specification of the information needed
Introduction to Marketing Research
Components of
the marketing research
problem
Research
questions
Hypotheses
Methodology for Marketing Research Ⅰ
Class 2
Methodology for Marketing Research Ⅰ
The purpose of science research: To forecast the behavior or the development of research object in the future. Method of science research: To induce theories from past experience, which can be able to apply to the analog.
Methodology for Marketing Research Ⅰ
The process of scientific method ☆Estimate the knowledge and theories for the problem we want to probe ☆Describe relative concepts and proposition exactly ☆Put forward the theoretical hypothesis ☆Design the method for testing hypothesis ☆Collect experiential data needed ☆Analyse and assess the data ☆Do the explanation for the research problem or advance new problem
Methodology for Marketing Research Ⅰ
Projective techniques（心理投射技术）： ☆Word association（词语联想） ☆Sentence completion（完形测试） ☆Third-person technique（第三人技术） ☆Role playing（角色扮演） ☆Thematic apperception（主题领悟测试）
Methodology for Marketing Research Ⅰ
Scaling： ☆ Nominal scaling（称名度量） ☆ Ordinal scaling（顺序度量） ☆ Interval scaling（等距度量） ☆ Ratio scaling(比率度量)
Reliability
All researcher expect to get the same result even from different tests under same context. The consistency between two tests is called reliability.
Methodology for Marketing Research Ⅰ
Classical True Value Model
X = T + E
Where ：
X：observational value
T：True value
E：Random error
T=єX
Methodology for Marketing Research Ⅰ
Definition of error
E = X - T
Where:
є E =є（X- T）
є E =єX - T=0
Methodology for Marketing Research Ⅰ
Reliability Index
Because：x=t+e
Reliability Index ：
Methodology for Marketing Research Ⅰ
Parallel tests
＊The true values of two tests are same
＊The variance of these two tests are same
To set up the mathematic relation between
and
Methodology for Marketing Research Ⅰ
Coefficient of Reliability
x1 and x2 represent the observational value in two different parallel tests：
x1=t1+e1;
x2=t2+e2
The correlativity of the two observational values is：
Methodology for Marketing Research Ⅰ
Derivation：
Methodology for Marketing Research Ⅰ
Coefficient of stability（稳定系数）
Do the same test twice on the same group of people at different time, in order to get the result of parallel tests.
Methodology for Marketing Research Ⅰ
Coefficient of equivalence (等值系数)
Do the similar tests but in different form on the same group of people at the same time, in order to get the result of parallel tests.
Methodology for Marketing Research Ⅰ
coefficient of stability and equivalence (稳定和等值系数)
Do the same tests but in different form on the same group of people at the different time, in order to get the result of parallel tests.
Methodology for Marketing Research Ⅰ
The reliability of composite score
＊ Spearman Brown Formula
C=A+B+C+D+…+K
Methodology for Marketing Research Ⅰ
Methodology for Marketing Research Ⅰ
Because:
So:
Methodology for Marketing Research Ⅰ
The reliability of composite score
＊Cronbach's α coefficient
Methodology for Marketing Research Ⅰ
Methodology for Marketing Research Ⅰ
While the test is not perfect parallel, it’s lower limit:
1、While the k tests are not perfect parallel, there must be one test (g), whose variance of true value is not less than any other tests, say:
Methodology for Marketing Research Ⅰ
2、If two tests are not perfect parallel, the sum of the variance of their true value is not less than twice of the covariance of the tests, say：
Methodology for Marketing Research Ⅰ
3、The sum of variances of k tests, which are not perfect parallel, is not less than the sum of k(k-1) covariances divided (k-1), say：
Methodology for Marketing Research Ⅰ
Methodology for Marketing Research Ⅰ
reliability of composite score is not less than
Methodology for Marketing Research Ⅰ
The standard error of measurement
Methodology for Marketing Research Ⅰ
Methodology for Marketing Research Ⅰ
Methodology for Marketing Research Ⅱ
Class 3
Methodology for Marketing Research Ⅱ
Estimating the reliability
1、Executing test twice
＊Alternating: Prepare two similar tests and execute them on the same group of people. The interval of tests should be short and the order of tests should be arrange carefully ( to is acceptable for Coefficient of equivalence )
Methodology for Marketing Research Ⅱ
＊Redo: Do the same test twice on the same group of testee at different time, and then calculate the pertinency between the two tests. The key issue is: how belong should we wait to do another test? The interval should be long enough so that the testees have forgotten the test, but it can’t be too long so that the testees have been changed. ( is acceptable for Coefficient of stability)
Methodology for Marketing Research Ⅱ
＊ Alternate redoing: Synthesize the redo and alternating. Do test 1, and wait a period of time and then do the test 2. (coefficient of stability and equivalence will be less than both Coefficient of equivalence and Coefficient of stability)
Methodology for Marketing Research Ⅱ
Estimating the reliability
2、Single test
＊divide：Do single test on testees and then divide the test items into two similar groups (odd-even dividing, divide by the difficulty of the items, random, divide by matching content）
Methodology for Marketing Research Ⅱ
Spearman-Brown Formula
Rulon Formula
D=A-B
Methodology for Marketing Research Ⅱ
＊covariance means：
αCoefficient:
When the tests are
prefect parallel,
α present the of the whole tests，so we can only say 。
Methodology for Marketing Research Ⅱ
Kuder Richardson
Where: pq (p is the proportion of 0- scored items，q is the proportion of 1- scored items) is the variation of item i .
KR20 can only be applied for binary items.
Methodology for Marketing Research Ⅱ
Where： is the mean of test scores，
is the variance of the whole tests.
Methodology for Marketing Research Ⅱ
Hoyt mean:
Where： is the mean of variance which be drawn from the variance analytic table.
is the mean of the variance of the residual which drawn from the same table.
Methodology for Marketing Research Ⅱ
On the reliability of single test
1.αcoefficient can be used as the index of the inner consistency.
2.αcoefficient can be the lower limit of reliability coefficient.
3.αcoefficient is the average of all the reliability coefficients, which are calculated by the Rulon means.
4. A high αcoefficient doesn’t mean the test is single dimension.
Methodology for Marketing Research Ⅱ
Estimate the true value
T’ is the estimated value of true value，X is the value of observation, is the mean of X.
Methodology for Marketing Research Ⅱ
Definition of validation
The consistency between the result of the observation and the purpose of the test is validation.
Methodology for Marketing Research Ⅱ
Content validation（内容效度）
The index used to measure the correlation between the test question and the research object.
Methodology for Marketing Research Ⅱ
The research process of Content Validation
☆Define the domain of the behavior under research
☆Built up a expert team who have experience in this domain
☆Put forward the analytic structure for measuring the behavior
☆Collect the relative data
Methodology for Marketing Research Ⅱ
Index of Content Validation
Where：N is the number of object, is the rate of measure item i against object k. If the measure item i matches the object k, we rate 1, if it is uncertain that whether they match, we rate it 0, if they don’t match, we rate it -1. And is the average rate of item i over all object.
Methodology for Marketing Research Ⅱ
Notes for Content Validation
1. Do it twice in order to keep the inner consistency：
2. Is the character of testees suitable to the research object.
3. Appraise the operational files of the test.
Methodology for Marketing Research Ⅱ
Criterion-related validation （ 准则关联效度 ）
The correlation between the test result and the relative behavior. Predictive validity (预测效度) and concurrent validity (同时效度) are two kind of Criterion-related validation 。
Methodology for Marketing Research Ⅱ
Process of Criterion-related validation research
☆Make sure what criterion behavior and mean of measure you need
☆Make sure the suitable testees
☆Test and record the score of the testees
☆Watch the behavior of testees
☆Count the correlation between score and behavior
Methodology for Marketing Research Ⅱ
Notes for Criterion-related Validation
☆Right criterion
☆Scale of sample
☆Criterion pollution
☆Scope limit
☆The reliability of score and criterion
Methodology for Marketing Research Ⅱ
Validation Coefficient (效度系数)
The correlation between measure score and criterion score.
Determine Coefficient (决定系数)
The square of validation coefficient, which represents how much portion of the measure score difference make by criterion prediction.
Methodology for Marketing Research Ⅱ
We can apply this formula to predict the score of criterion by measure score：
Where: is validation coefficient, and are the standard variance of measure score and criterion score, is the mean value of criterion, is the mean value of measure score.
Methodology for Marketing Research Ⅱ
Construct validation （ 结构效度 ）
The correlation between the artificial structure and real structure. (Such as factors analysis)
Research Design
Class 4
Research Design
Define the problem exactly
☆ Merge and acquire A company
☆ Merge and acquire analogous company
☆ Expand productivity
☆ Expand the capacity of sale
☆ Maximize the value of company
Sample
Research Design
The process of defining problem
☆ Make sure the object of research
☆ Know the background of the problem
☆ Decompose and fix the problem
☆ Put forward the theory framework
☆ Choose the analysis unites
☆ Set up the relative variables set
☆ State the problem and object of the research
Research Design
The form of the research proposal
☆The object of research
☆The scheme of research design
☆Sampling design
☆Data collection
☆Data examination and analysis
☆write the research report
☆Do the research budget and schedule
Research Design
The main issue of defining problem
☆What is your research purpose?
☆What information is in hand?
☆What information do we need for the research?
☆What variables should be measured , and how to measure?
☆Is the information wanted accessable?
☆Is the research feasible?
☆Can the hypothesis be described clearly?
Research Design
The main issue of research design
☆What questions must be answered?
☆Is descriptive research or causal research necessary?
☆Where are we can access the data?
☆Will interview give us the objective answer?
☆Are we restrict by the time budget in collecting data?
☆How to describe the survey issue?
☆How to operate the experiment?
Research Design
The main issue of sampling design
☆Do we need to do sampling？
☆Is the target person able to be defined？
☆Where can we find the data？
☆What precision is need in sampling？
☆Is random sampling necessary？
☆Is stratified random sampling necessary ？
☆How large is the sample？
☆How to choose the way of sampling？
Research Design
The main issue of data collecting
☆Who will collect the data？
☆How much time will it take to collect data？
☆What about the process of data collecting？
☆Does it need to take supervision and management？
Research Design
The main issue of data analysis and appraise
☆How to classify the data？
☆Does it need standard editing and coding？
☆What kind of tabling should we take, Computer or manual work？
☆What questions does it need to answer？
☆How many data does it need to deal with？
☆How to set the appraise criterion？
Research Design
The point of preparing report
☆Who is the object of the report？
☆Does it need policy suggestion？
☆How many time does it need to introduce the report？
☆What about the report form？
Research Design
The point of general appraise
☆How much is the research budget?
☆Is the research interval acceptable？
☆Do we need external supporting？
☆Can the research design achieve the goal？
☆When will the research launch？
Research Design
Sampling Design
☆Confirm the investigation matrix
☆Choose the sampling box
☆Coding
☆Choose the method of sampling
Research Design
Method of Sampling
☆Accidental (probability) sampling (随机抽样,概率抽样)
----Simple random sampling (简单随机抽样)
----Stratified random sampling (分层抽样)
----Systematic sampling (系统抽样)
----Cluster sampling (分群随机抽样)
----Multistage sampling (多阶段抽样)
Research Design
Method of Sampling
☆Non-probability sampling (非随机抽样)
----Convenience sampling (任意抽样)
----Judgement sampling (判断抽样)
----Quota sampling (配额抽样)
----Snowball sampling (雪球抽样)
Research Design
Factors which affect the sampling error
☆The variance between the individual of the matrix
☆The scale of the sample
☆The method of sampling
Research Design
Estimation of the sampling error
----Error of mean by duplicated sampling
Where: is the average error of sampling, is the variance of matrix, which can be replaced by variance of sample, is the number of sample.
Research Design
Estimation of the sampling error
---- Error of mean by non-duplicated sampling
Where: is the number of the matrix, is the adjustment coefficient.
Research Design
Estimation of the sampling error
---- Error of mean by proportion index duplicated sampling
Where: is the proportion index.
Research Design
Estimation of the sampling error
---- Error of mean by proportion index non-duplicated sampling
Research Design
Scale of sample
while duplicated sampling，
So： or
while non-duplicated sampling:
or
Where: t is the probability value.
Research Design
Case：The average weight of product A is 50 gram, the standard variance of the matrix is 2 gram. Now we plan to sample from the products by Simple random sampling (duplicated), the reliability is required to be 95%, the error is not allow great than gram, what should the sample scale be?
Exploratory Research Design
Class 5
Secondary data
Data that have been collected for purposes other than the problem at hand
Exploratory Research Design
The role of secondary data
☆Identify the problem
☆Better understand and define the problem
☆Develop an approach to the problem
☆Formulate an appropriate research design
☆Answer certain research questions and test some hypotheses
☆Interpret primary data with more insight
Exploratory Research Design
Primary data Secondary data
Collection Purpose For the problem at hand For other problems
Collection process Very involved Rapid and easy
Collection cost High Relatively low
Collection time Long Short
Exploratory Research Design
Disadvantages of secondary data
The value of secondary data is typically limited by their degree of fit with the current research problem and by concerns regarding data accuracy.
It would come from the objectives, nature, and methods used to collect secondary data, lacking of accuracy, compatibility of units of measurement, time frame.
Exploratory Research Design
Evaluating secondary data
☆Methodology used to collect data (size and nature of the sample, response rate and quality, questionnaire design and administration, procedures used for field work, data analysis and reporting procedures)
☆Accuracy of the data (stemming from research approach, research design, sampling, data collection, reporting stages of the project )
Exploratory Research Design
Evaluating secondary data (Cont.)
☆When the data were collected
☆The purpose for the study
☆The content of the data (Units of measurement, categories used, relationships examined)
☆How dependable are the data
Exploratory Research Design
Qualitative research
Used to define the problem more precisely, formulate hypotheses, identify or clarify key variables to be investigated in the quantitative phase.
Exploratory Research Design
Exploratory Research Design
Qualitative Research Quantitative Research
Objective To gain a qualitative under- To quantify the data
standing of the underlying results from the sample
reason and motivations to the population of
interest
Sample Small number of non- Large number of
representative cases representative cases
Data collection Unstructured Structured
Data analysis Nonstatistical Statistical
Outcome Develop a richer Recommend a final
understanding course of action
Classification of qualitative research procedures
☆Focus-Group Interviews
☆In-Depth Interviews
Exploratory Research Design
Exploratory Research Design
Characteristics of Focus-Groups
Group size 8 to 12 participants
Group composition Homogeneous, respondents prescreened
Physical setting Relaxed, informal atmosphere
Time duration 1 to 3 hours
Recording Use of audiocassettes and videotapes
Moderator Observational, interpersonal, and
communication skills
Exploratory Research Design
Advantages of Focus-Groups
☆Immediacy and richness of the comments which come from real customers.
☆A wider range of information, insights and ideas than do individual interviews.
☆The comments of one person can trigger unexpected reactions from others.
Exploratory Research Design
Disadvantages of Focus-Groups
☆Group members often speak leads to a tendency for researchers and managers to regard findings as conclusive rather than as exploratory.
☆Focus groups are difficult to moderate.
Applications of Focus Groups
☆Understand consumer perceptions, preferences, and behavior concerning a product category.
☆Obtain impressions of new product concepts.
☆Generate new ideas about older products.
☆Develop creative concepts and copy material for advertisements.
☆Secure price impressions
☆Obtain preliminary consumer reaction to specific marketing programs.
☆Interpret previously obtained quantitative results.
Exploratory Research Design
In-Depth Interviews
Unstructured, direct personal interviews in which a single respondent is probed by a highly skilled interviewer to uncover underlying motivations, beliefs, attitudes, and feelings on a topic.
Exploratory Research Design
Exploratory Research Design
Advantages of In-Depth interviews
☆Uncover deeper insights about underlying motives than focus group.
☆Attribute the responses directly to the respondent unlike focus group is difficult to determine which respondent made a particular response.
☆Result in a free exchange of information.
☆Get at real issues when the topic is complex.
Exploratory Research Design
Disadvantages of In-Depth interviews
☆Skilled interviewers are expensive and difficult to find.
☆The lack of structure makes the results susceptible to the interviewer’s influence.
☆The data obtained are difficult to analyze and interpret.
☆The length of the interview combined with high costs means that only a small number of in-depth interviews can be conducted in a project.
Exploratory Research Design
Applications of In-Depth interviews
☆Detailed probing of the respondent.
☆Discussion of confidential, sensitive, or embarrassing topics.
☆Situations where strong social norms exist and the respondent may be easily swayed by the group’s response.
☆Detailed understanding of complicated behavior.
☆Interviews with professional people.
☆Interviews with a competitor’s customers, who are unlikely to reveal the information in a group setting.
☆Situations where the product consumption experience is sensory in nature.
Characteristic Focus Groups In-Depth Interviews
Group synergy and dynamics + –
Peer pressure/group influence – +
Client involvement + –
Generation of innovative ideas + –
In-depth probing of individuals – +
Uncovering of hidden motives – +
Discussion of sensitive topics – +
Respondents who are competitors – +
Respondents who are professionals – +
Scheduling of respondents – +
Amount of information + –
Bias in moderation and interpretation + –
Cost per respondent + –
Exploratory Research Design
Descriptive Research Design
Class 6
Survey Errors
Descriptive Research Design
Measurement error
Sampling design error
Substitute information error
Answer error
Sampling error
Respondents error
Interviewer error
Random sampling error
Total error
Systematic error
Answer error
Measurement tool error
Process error
Sampling box error
Refused error
Survey Methods
A structured questionnaire given to a sample of a population and designed to elicit specific information from respondents. Survey may be conducted in person, by telephone, through a mailed questionnaire, or electronically via the computer.
Descriptive Research Design
Advantages of survey research
☆Easy
☆Reliability
☆Simplicity
Descriptive Research Design
Disadvantages of survey research
☆Respondents may be unable or unwilling to provide the desired information.
☆Fix-response choices may result in loss of validity for certain types of data.
☆Properly wording questions is not easy.
Descriptive Research Design
Telephone Methods
☆Traditional Telephone Interviews
☆Computer-Assisted Telephone Interviews
Descriptive Research Design
Personal Methods
☆Personal In-Home Interviews
☆Mall Intercept Personal Interviews
☆Computer-Assisted Personal Interviewing
Descriptive Research Design
Descriptive Research Design
Mail Methods
☆Mail Interviews
☆Mail Panels
Descriptive Research Design
Electronic Methods
☆E-mail Surveys
☆Internet Surveys
Improving Survey Response Rates
☆Prior Notification
--Incentives (Prepaid or promised incentive)
--Follow-up
☆Other Facilitators of Response
Descriptive Research Design
Observation Methods
Observation involves recording the behavioral patterns of people as well as data on objects and events in a systematic manner to obtain information about the phenomenon of interest.
Descriptive Research Design
Classification by structured
☆Structured observation
☆Unstructured observation
Descriptive Research Design
Descriptive Research Design
Classification by object
☆Direct observation—observe the behaviors
--Disguised observation
--Undisguised observation
☆Indirect observation—observe the outcome
--Archives
--physical traces
Descriptive Research Design
Classification by intervention
☆Participant observation
☆Non-Participant observation
Descriptive Research Design
Classification by Observation method
☆Personal observation
☆Mechanical observation
Personal observation
☆Mystic guests
☆Content analysis
☆Audit
Descriptive Research Design
Advantage of Observation Methods
☆Don’t require conscious respondent participation
☆Interviewer bias resulting from interaction with the respondent or subjective interpretation of the questionnaire is minimized
☆The errors inherent in self-reported behavior are eliminated
☆Data regarding product preferences or reactions to marketing materials from children or pets can best be collected using observational techniques
☆Best applied to phenomena that occur frequently or are of short duration
Descriptive Research Design
Disadvantage of Observation Methods
☆ Attitudes, motivations, and values are all lost to the observational method
☆Highly personal behaviors are not available for observation
☆Individuals have a tendency to selectively observe only what they want to
☆Can be adopted for only frequent behaviors of short duration
Descriptive Research Design
Causal Research Design-- Experimentation
Class 7
Causal Research Design
Concept of Causality
A causal inference relates to whether a change in one marketing variable produces a change in another variable.
Experimentation
We can control the test condition during the research, in order to manipulate one or several variables, so as to test the hypothesis about the dependent variable.
Causal Research Design
Condition for causality
☆Concomitant Variation
☆ Time order of Occurrence of Variables
☆Absence of Other Possible Causal Factors
Causal Research Design
Definition of Symbols
EG: Experimentative Group
CG: Control Group
X: The exposure of a group to an independent variable, treatment, or event, the effects of which are to be determined.
O: The process of observation or measurement of the dependent variable on the test units of group of units.
R: The random assignment of test units or groups to separate treatments.
M: The respondents are match to EG and CG.
Causal Research Design
Validity in Experimentation
☆Internal Validation:
☆External Validation
Causal Research Design
Threaten to Internal Validation
☆History(历史效应)
☆ Maturation(熟化效应)
☆Testing Effects( 参与试验效应)
☆Instrumentation(工具效应)
☆Statistical Regression(统计回归效应)
☆Selection Bias(选择偏差)
☆Mortality(损耗效应)
☆Cross Effect between Selection and Maturation
(选择—熟化交叉效应)
Causal Research Design
Threaten to External Validation
☆The reaction of the test and interaction (Cross effects)
☆Selection Bias and interaction
☆The reaction of the test
☆Interaction among multi-test
Causal Research Design
Preexperimental Designs
☆One-Shot Case Study: EG X O
☆One-Group Pretest/Posttest Design: EG O1 X O2
☆Static Group Design:
☆Match-Control Group Design:
EG M X O1
CG M O2
Causal Research Design
True Experimental Designs
☆Posttest-Only Control Group Design:
EG R X O1
CG R O2
☆Pretest/Posttest Control Group Design
EG R O1 X O2
CG R O3 O4
☆Intensified Design
EG R O1 X O2
CG R O3 O4
EG R X O5
CG R O6
Causal Research Design
True Experimental Designs (Cont)
☆Solomon Four-groups Design
RG1 Ｏ1 X Ｏ2
RG2 Ｏ3 — Ｏ4
RG3 — X Ｏ5
RG4 — — Ｏ6
Causal Research Design
Quasi-Experimental Designs
☆ Time Series Design
EG O1 O2 O3 O4 X O5 O6 O7 O8
☆ Factorial Design
Causal Research Design
Statistical Experimental Designs
☆Perfect Random Design
EG1 R X1 O1
EG2 R X2 O2
EG3 R X3 O3
☆Inner-Group Random Design
EG1 R X1 O1
CG1 R O2
EG2 R X2 O3
CG2 R O4
Causal Research Design
Statistical Experimental Designs (Cont)
☆Ladin Square Design
☆Cause-Analysis Design
EG1 R X1 (Hi. Ad.、Hi. Price) O1 n = 6
EG2 R X2 (Hi. Ad.、Lo. Price) O2 n = 6
EG3 R X3 (Lo. Ad.、Hi. Price) O3 n = 6
EG4 R X4 (Lo. Ad.、Lo. Price) O4 n = 6
EG5 R X5 (No Ad.、Hi. Price) O5 n = 6
EG6 R X6 (No Ad.、LO. Price) O6 n = 6
1
2
3
4
Store Num.
Producer B
Producer A
Retailer B
Retailer A
21 cents
22 cents
25 cents
26 cents
III
IV
I
II
II
III
IV
I
I
II
III
IV
IV
I
II
III
Causal Research Design
Limit of Experimentation
☆Cost
☆Safety
☆Implemental Difficulty
☆uncertainty to the duration of the test result
Measurement and Scaling
Class 8
Measurement and Scaling
Definition of Measurement
Measurement means assigning number or other symbols to characteristics of objects being measured, according to predetermined rules.
Measurement and Scaling
Definition of Scaling
Scaling is the process of placing research object along a continuum. It can be considered a part of measurement.
Measurement and Scaling
Primary Scales
☆Nominal Scale: Used for classification purposes
☆Ordinal Scale: Used for ranking purposes, it is used to measure relative attitudes, opinions, perceptions,and preferences.
☆Internal Scale:Numerically equal distances on the scale represent equal values in the characteristic being measured.
☆Ratio Scale:Possesses all the properties of the nominal, ordinal, and interval scales.
Measurement and Scaling
Scaling Techniques
☆Comparative scales: can detect the small differences between objects under study.
☆Noncomparative scales: objects are scaled independently of each other, which referred to as monadic or metric scales.
Measurement and Scaling
Comparative Scaling Techniques
☆Paired Comparison Scaling: is useful when the number of brands under consideration is limited to no more than five.
☆Rank Order Scaling: is used to measure preferences among brands as well as among brand attributes.
☆Constant Sum Scaling: allows for fine discrimination among alternatives and does not require too much time.
Measurement and Scaling
Noncomparative Scaling Techniques
☆Continuous Rating Scale: Place a mark on a continuous line which is easy to construct, but only applied in computers, internet, and other technologies.
☆Itemized Rating Scale: has a number of brief descriptions associated with each response category.
Measurement and Scaling
Itemized Rating Scale
☆Likert Scale: Degree of agreement on a 1(strongly disagree) to 5(strongly agree) scale.
☆Semantic Differential Scale: Seven-point scale with bipolar labels.
☆Stapel Scale: Unipolar 10-point scale, -5 to +5, without a neutral point (zero).
Measurement and Scaling
Itemized Rating Scale Decisions
☆Number of Scale Categories
☆Balanced versus Unbalanced Scale
☆Odd or Even Number of Categories
☆Forced or Nonforced Choice
☆Nature and Degree of Verbal Description
☆ Physical Form or Configuration
Questionnaire and Form Design
Class 9
Definition of Questionnaire
A formalized set of questions for obtaining information from respondents.
It has three objectives: translate the information needs into a set of questions that respondents are willing and able to answer, minimize demands imposed on respondents, minimize the response error.
Questionnaire and Form Design
Questionnaire Design Process
☆Specify the objective (s) of the research
☆Formulate the theoretical framework
☆Specify the information needed
☆Specify the type of interviewing method
☆Determine the content of individual questions
☆ Design the questions to overcome the respondent’s inability and unwillingness to answer
Questionnaire and Form Design
Questionnaire Design Process (Cont.)
☆Decide on the question structure
☆Determine the question wording
☆Arrange the questions in proper order
☆Choose the form and layout
☆Pretest the questionnaire
☆Reproduce the questionnaire
Questionnaire and Form Design
Sensitive information
☆Place at the end of questionnaire
☆To be rapport
☆Project technology
☆Anonymity questionnaire
☆ Kinsey technology
☆Random reply
Questionnaire and Form Design
Right order of questionnaire
☆Begin with a simple unthreatening question
☆Range of questions should be logical
☆From common issue to special topic
☆Sensitive or difficult questions should not be place at the beginning of the questionnaire
Questionnaire and Form Design
Data Collection
Class 10
Characteristics of Outstanding Interviewer
☆Work Orientation ☆Teamwork
☆Aptitude ☆Pride
☆ Discipline ☆Third Ear
☆Command ☆Woo
☆Ethics
Data Collection
Data Collection
Field Work Process
☆Selection of field workers
☆Training of field workers
☆Supervision of field workers
☆Validation of field workers
☆Evaluation of field workers
Data Collection
Training of Field Workers
☆Making the initial contact
☆Asking questions
☆Probing
☆Recording answers
☆Terminating the interview
Data Collection
Supervision of Field Workers
☆Quality control and editing
☆ Sample control
☆Control of cheating
☆Central office control
Data Collection
Validation of Field Work
The supervisors call 10 to 25 percent of the respondents to inquire whether the field workers actually conducted the interviews.
Data Collection
Evaluation of Field Workers
☆Quantity
-Cost and Time
-Response rates
☆Quality
-Quality of interviewing
-Quality of Data
Editing and Coding
Class 11
Editing and Coding
Types of Editing
☆Field editing
☆Office editing
Editing and Coding
Request to Editing
☆Consistence
☆Integrity
☆Edit the wrong order answers
☆Help to coding
☆Edit the “I don’t know” answers
Editing and Coding
Coding
Coding is a process of assigning the numbers or anther symbol to different answers, which can make the data can be processed by the computer.
Errors Analysis
Class 12
Errors Analysis
Muti-variables regression
Y=α+β1x1+ β2x2+…+ βnxn+ε
Errors Analysis
Muti-Collinearity(多重共线性问题)
The estimation of the coefficient will be mis-estimated due to the linear-relationship between the variants. In this situation, the β which is the correlation between this two variants. So we can find out the Muti-Collinearity by test the β coefficient.
*
Test of Muti-Collinearity
Regress all the two variants and find out whether there is Muti-Collinearity in regression.
Errors Analysis
*
Sample:
In regression among Y and x1、x2、x3、x4、x5、x6，regress all the two x1、x2、x3、x4、x5、x6 and get the determine coefficients--R1、R2、R3、R4、R5、R6 which is 、、、、、，it’s F value is (Significant)、（N.）、（S.）、（S.）、（N.）、（S.）.
Errors Analysis
*
Heteroskedasticity (异方差)
The error in regression vary by the variable, instead of a constant value.
The estimate is not valid if there is Heteroskedasticity in the regression, even thought it is unbiased.
Errors Analysis
*
Test of Heteroskedasticity
We will get the variance, ε2；and then regress ε2 and all independent variables, the Squares of the independent variables, and the product of independent variables, to find out whether there is Significant relationship between them. And then get the relationship (Zi) between ε2 and xi, so as to we can eliminate the Heteroskedasticity.
Errors Analysis
*
Eliminating the Heteroskedasticity
Divide by Zi in both side of the regression equation, and then we can get the non- Heteroskedasticity regression.
Errors Analysis
*
Auto Correlation (自相关)
Auto Correlation is that there is the significant relation ship betweenεt and εt-1，εt-2，…. Auto Correlation will make the variance of estimate greater and then the least squares estimate will be not invalid.
Errors Analysis
*
Durbin-Watson Test
yt=α+βxt+εt
εt=ρ·εt-1+νt
Errors Analysis
*
Because the difference between ε2t and ε2t-1is less enough we can deem they are not different. So:
Errors Analysis
*
Thus，d=2-2ρ=2(1-ρ)。
-1≤ρ≤1, So, 0≤ d ≤4
We can get the the upper limit (du) and the lower limit (dL) from Durbin-Watson, which is depended on the significance of the d.
Errors Analysis
*
2
4
dL
4-dL
4-du
du
ц
I
Ш
Errors Analysis
*
Three areas of DW test value
1、If d∈(dL,du) or d∈(4-du,4-dL), we can’t make sure there is Auto Correlation.
2、If d∈(du,4-du), there is not Auto Correlation.
3、When d∈(0，dL) or d∈(4-dL,4)，there is positive or negative Auto Correlation.
Errors Analysis
*
Eliminating Auto Correlation
In the regression of
yt=α+βxt+εt ， εt=ρ·εt-1+νt
Multiply yt-1=α+βxt-1+εt-1 by ρ：
ρ yt-1=α ρ +βρ xt-1+ ρεt-1
yt-ρyt-1=α(1-ρ)+β(xt -ρxt-1)+εt- ρεt-1
yt-ρ yt-1=α(1-ρ) +β(xt -ρ xt-1)+ νt
Errors Analysis
Quality Forecasting
Class 13
Quality Forecasting
Delphi Method
Delphi is The Delphi method is a systematic, interactive forecasting method which relies on a panel of independent experts.
The process of Delphi
After each round, a facilitator provides an anonymous summary of the experts’ forecasts from the previous round as well as the reasons they provided for their judgments.
Thus, experts are encouraged to revise their earlier answers in light of the replies of other members of their panel. It is believed that during this process the range of the answers will decrease and the group will converge towards the "correct" answer.
Finally, the process is stopped after a pre-defined stop criterion (. number of rounds, achievement of consensus, stability of results) and the mean or median scores of the final rounds determine the results.
Quality Forecasting
Quality Forecasting
Pros. Of Delphi
The advantage of Delphi method is that ideas can be gathered from group members who are too geographically separated or busy to meet face to face. Its disadvantages are that members are unable to ask questions of one another
The Cons. Of Delphi
Disadvantage of this method is the danger of a forming of an opinion by the group dynamics, in which a perhaps necessary serious estimate deviation is subject to the group obligation. A further disadvantage is that due to several iteration loops for the forming of an opinion the entire estimate expenditure can become quite extensive.
Quality Forecasting
The Brainstorming
Brainstorming is the name given to a situation when a group of people meet to generate new ideas around a specific area of interest.
Using rules which remove inhibitions, people are able to think more freely and move into new areas of thought and so create numerous new ideas and solutions.
The participants shout out ideas as they occur to them and then build on the ideas raised by others. All the ideas are noted down and are not criticized. Only when the brainstorming session is over are the ideas evaluated.
Quality Forecasting
The Pros. Of Brainstorming
An advantage of brainstorming is that the group is more likely to generate creative ideas. Also, brainstorming often helps group members build rapport and cohesion because the process is fun, inventive and sometimes humorous.
Quality Forecasting
The cons. Of Brainstorming
☆Have to listen to others and may spend time repeating their ideas until they get sufficient attention. ☆Going through the protocol, processing and ordering the ideas can become a complex procedure. ☆Advising participants to let others speak without making them feel offended or intimidated can be difficult. ☆Participants with the ability to express their ideas faster and more effective gain the general attention of the group.
☆People are not very skilled at controlling their non-verbal reactions and might influence the creativity of others with their posture, gestures or facial expressions. On the other hand, attempting to control their non-verbal behavior might inhibit their own creativity. ☆More discrete or introvert participants might find it difficult to express their crazy or unorthodox ideas. More discrete or introvert participants might find it difficult to express their crazy or unorthodox ideas.
Quality Forecasting
Quality Forecasting
Method of Analogy
Compare the research object to the analogue, and then judge the characters of the research object by the characters of the analogue.
Research Report and Review
Class 14
*
The structure of research report
1、Title
2、Abstract
3、Body of the text
4、Appendix
Research Report and Review
*
The structure of Title
①Cover，including the title of the report, researcher and the organization, name of client, date of the report submitted.
②Letter to the trustor（Do not need while internal usage）
③The copy of the letter of attorney（Do not need while internal usage）
④catalogue, including the body, tables, figures, appendix etc.
Research Report and Review
*
The structure of abstract
①Main finding, summarize the core information systematically
②Conclusion, interpret the finding above
③Suggestion, give advice to the user of the report against the research objective
Research Report and Review
*
The structure of the body
①Definition of the issue, including the background of the issue, description of the issue, the objective of the research etc.
②Approach of research, including the main methods, such as statistics, case study etc.
③Design of investigation, including type of design, relative information, secondary material collection, first hand material collection, technic of measurement, design of questionnaire, design of sampling, implement of survey.
④Data analysis, introduce the method of analysis.
⑤Result of investigation, sort the data and result.
⑥Limitation and specification.
⑦Conclusion and recommendation.
Research Report and Review
*
The structure of appendix
questionnaires, tables and figures, detail of technic, procedure of analysis and conclusion, and other materials
Research Report and Review
*
What does a good report like
⑴Object orientation
⑵To be objective and real
⑶ To be Logical
⑷Focus on emphases
⑸charts and tables
Research Report and Review
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
