Week_5_Experimentation & Measurement and scaling Flashcards
What is causality?
For instance: an association between attitude and behavior is evidence of a causal relationship.
Means
1) that a change in one variable will produce a change in another.
2) If two variables are causally linked they should be associated.
3) If association provides evidence of causation, then lack of association suggest an absence of causation.
Before making causal inference, you must make sure 3 factors:
- Concomitant variation - evidence that a strong association exists between an action and an observed outcome.
- The time order of occurrence - causing event must occur before or simultaneously with the effect.
- absence of other possible causal factors - the factor or variable being investigated should be the only possible causal explanation.
Even 3 factors are satisfied, still can not determine the causality exists
What is difference between independent variables and dependent variables.
IV are variables or alternatives that are manipulated; DV are the variables which measure the effect of the independent variables on the test units.
Relationship with test unit and variables.
test unit are individuals or entities response to the independent variables.
What is extraneous variables?
All variables other than independent variables that affect the response of the test units.
Internal validity definition
A experiment that determine whether the independent variable cause the effect of dependent variables
Control of extraneous variables is a necessary condition for establishing internal validity.
External validity definition
A determination of whether the cause-and- effect relationships found in the experiment can be generalized.
What are categories in extraneous variables
History - specific events that are external to the experiment but occur at the same time as the experiment
Maturation - refers to changes in the test units themselves that occur with the passage of time.
Testing effects - caused by experimentation, including main testing effect and interactive testing effect
Instrumentation - changes in the measuring instrument.
Statistical regression - occur when test units with extreme scores move closer to the average score during the course of the experiment.
Selection bias - An extraneous variable attributable to the improper assignment of test units to treatment conditions.
Mortality - loss of test units while the experiment is in progress.（人死了，影响结果）
Treatment Effect (TE)
(O2 - O1) - (O3 - O4), make sure the result is effected only by TE
What do these symbols represent: X, O, R
X = independent variable, treatment, or event, the effects of which are to be determined O = dependent variables on test units or group units R = the random assignment of test units or groups to separate treatments
Ways to control extraneous variables
Randomization - randomly assigning test units to experimental groups.
Matching - matching test units on a set of key back- ground variables before assigning them to the treatment conditions.
Statistical control - measuring the extraneous variables and adjusting for their effects through statistical methods.
Design control - involves using specific experimental designs
Classification of experimental design
Pre-experimental design - one shot case study, not really great.
True experimental - two group including control group and random group.
Quasi-experimental design - Designs that apply part of the procedures of true experimentation but lack full experimental control.
Statistical design - allow for the statistical control and analysis of external variables.
One-shot case study
X, O1 - Single group of test units is exposed to treatment X; A single measurement on the dependent variable is taken (O1). no random assignment of test units (no R). more appropriate for exploratory.
One-group pretest-posttest design
O1 X O2
Static group design
EG: X 01
CG: 02
True Experimental Designs: Pretest-Posttest Control Group Design
EG: R 01 X 02
CG: R 03 04
Quasi-experimental design - time series design
Two kinds: single time and multiple time
01 02 03 04 05 X 06 07 08 09 010
EG : 01 02 03 04 05 X 06 07 08 09 010
CG:01 02 03 04 05 06 07 08 09 010
When is Randomized Block Design from statistical design useful?
Only one major external variable
When is Latin Square Design useful?
When two non-interacting external variables are allowed as well as independent variable are manipulated
When is factorial design useful?
When two or more independent variables.
Differences between Laboratory experiment an field experiments
Laboratory experiment is easier to control than field experiment does. But field experiment is better to get good result
Limitations of Experimentation
- Experiments can be time consuming
- Experiments are often expensive.
- Experiments can be difficult to administer
- Competitors may deliberately contaminate the results of a field experiment.
What is the critical qualities for test market selections?
Be large enough to produce meaningful projections. They should contain at least 2% of the potential actual population.
Be representative.
Be relatively isolated in terms of media and physical distribution.
Have normal historical development in the product class.
Have marketing research and auditing services available.
Not be over-tested.
What is relatively isolated for test market selection?
pick these region like dallas, austin, ei paso
when comparing two products, it is better to rate them than to rank them.
Strongly Disagree 1 2 3 4 5 Strongly Agree
Object and Properties
Adult heads of household
age, gender, awareness of new product
The definition of measurement
The assignment of numbers or other symbols to characteristics of objects according to certain pre- specified rules. (ACUTAL ASSIGNMENT of a number from 1 to 100 to each respondent.)
The definition of scale
The generation of a continuum upon which measured objects are located ( PROCESS of placing the respondents on a continuum with respect to their attitude toward department store)
Two things need to be careful when scaling
Validity (the question you asking can really measurable?)
Reliability (a respondent responds in the same or in a very similar manner to an identical or nearly identical question)
An Illustration of Primary Scales of Measurement (very important)
See week 5_2, slide 11.
Nominal scale and ordinal scale are called as
Non-metric scale, analysis becomes limited
Marketing example for four scales
Nominal: Brand numbers, store types, gender
Ordinal: Preference rankings, market position, social class.
Interval: Attitudes, opinions, index numbers.
Ratio: Age, income, costs, sales, market shares
Zero for interval and ratio scale
Interval: Zero is just another scale position
Zero = the scale point between -1 and 1
What is your attitude toward credit cards -3 -2 -1 0 +1 +2 +3 OR 01234567.
Ratio: Zero has an absolute meaning
Zero = Absence of the property being measured
What is your credit card balance? $________
What is open-ended measurement?
not standardized, each respondent’s response is unique, hard to get data.
Used when conducting exploratory research
Why is better to have categorical measurement?
Sometimes it easier to answer the question, affect people’s response. Be sure that no overlap between each category.
Metric measurement two critical factors
Order and distance.
What are two metric scale?
Natural and synthetic metric scale
When we can use synthetic metric scale?
When there are objective properties( which are physically verifiable characteristics such as age, gender, number of bottles purchased, or last store visited) Subjective properties (which cannot be directly observed because they are mental constructs such as a person’s attitudes, opinions, or intentions)
What is advantage of putting “not applicable/don’t know” instead of putting “middle point”?
Avoid lowering the average of result. So better to use even number.
What is likert scales
- Respondents are asked to indicate their degree or agreement or disagreement on a symmetric agree- disagree scale for each of a series of statements
- A statement is made and the respondent is asked to what degree they agree or disagree with the statement
What is Constant-Sum Comparative-Ratings Scale?
For example: divide 100 points between the following four attributes regarding women’s cosmetics in terms of the relative importance of each attribute to you.
What is stapel scales?
The basis of the Stapel scale format is numerical rather than verbal or visual
– It has numbers that range from a minus end to a corresponding plus end, and typically include “0” as the midpoint.
– The respondent circles the number that best corresponds to his or her feelings on the topic
What is Non-symmetric Synthetic Scales?
The one-way labeled scale is one where the researcher is measuring some construct attribute with the use of labels that restrict the measure to the “positive” side
For example, Bad, good, very good, extremely good
N-point Scale
The n-point scale, meaning a 5-point, 7-point, or 10-point scale format, is a popular choice for researchers measuring constructs on non-symmetric attributes (better use even number)
Unanchored N-point scale
Occasionally, a researcher will opt to not provide the anchors, in which case it will be an unanchored n-point scale (won’t tell respondent which side is the anchore)
When should use non-symmetric scale?
Unless researcher believes there will be very few respondents who will make use of the negative side of a symmetric scale, the researcher should opt for a non-symmetric scale
What are two types of scale technique
Comparative scales and non-comparative scales
What is relative advantages and disadvantages of comparative scales?
- Small differences between stimulus objects can be detected.
- Same known reference points for all respondents.
- Easily understood and can be applied.
- Involve fewer theoretical assumptions.
- Tend to reduce halo or carryover effects from one judgment to another.
- Ordinal nature of the data
- Inability to generalize beyond the stimulus objects scaled
Paired Comparison Scaling
• A respondent is presented with two objects and asked to select one according to some criterion.
• The data obtained are ordinal in nature.
• Paired comparison scaling is the most widely used
comparative scaling technique.
• With n brands, [n(n − 1) /2] paired comparisons are required.
• Under the assumption of transitivity, it is possible to convert paired comparison data to a rank order.
(bilnding taste also is paired comparison
Comparative Scaling Techniques Constant Sum Scaling
- Respondents allocate a constant sum of units, such as 100 points to attributes of a product to reflect their importance.
- If an attribute is unimportant, the respondent assigns it zero points.
- If an attribute is twice as important as some other attribute, it receives twice as many points.
- The sum of all the points is 100. Hence, the name of the scale.
Why scale is so important?
- The choice of the level of measurement for a scale affects which analyses should or should not be performed
- The analysis, in turn, greatly affects what may or may not be said about the property being measured
Examples
– If you wish to calculate an average, you must use a
metric scale
– If you use a categorical scale, you must summarize the results with a percent or frequency distribution
Restriction of international marketing research using scales
In developing countries, the respondents might have difficulty using interval and ratio scales. Consumer preferences in these countries are best measured with ordinal scales.
The primary scales should be matched to the profile of the target respondents.
Ethics and Scale Development
A conscientious market researcher will devote as much time and energy as possible to ensure the reliability and validity of the research throughout the entire process
