Descriptive and Causal Research

advertisement
Descriptive Research
Marketing Research – MKTG 446
Angela D’Auria Stanton, Ph.D.
Descriptive Research

Descriptive research (often referred to as survey
research) is aimed at characterizing phenomena and
identifying association among selected
variables. Descriptive research can be used to:




describe the characteristics of certain groups
estimate the proportion of people in a specified population who
behave in a certain way
make specific predictions
Some specific examples of descriptive studies include:



The proportion of companies that have an electronic storefront
Target customer's evaluation of key product attributes
Measuring employee satisfaction
Classification of Descriptive
Studies
Longitudinal
Cross-Sectional
 the true panel
 the single measures over time


Longitudinal Surveys



field studies/surveys
Provides a series of pictures that, when pierced together, provide a
movie of the situation the changes that are occurring.
The sample remains relatively constant through time and the sample
members are measured repeatedly.
Cross-Sectional Studies


Provides a snapshot of the variables of interest at a single point in time.
The sample is typically selected to be representative of some known
universe.
Longitudinal Data – Advantages
Consumer or Business Panel
 Allows turnover analysis
 Allows collection of a great deal
more classification information
from respondents
 Allows longer and more exacting
interviews
 Produces fewer error in reporting
past behavior because of natural
forgetting
 Produces fewer interviewerinterviewee interaction error
Single Measures Over Time:
The Tracking Study

Tends to produce more
representative samples of
population of interest

Produces fewer errors due to
respondent's behavior being
affected by the measurement
task

Allows the investigation of a
great many relationships
An Example of a Panel Design
Number of Households in Panel Purchasing Each Brand of Detergent
TABLE 1
Brand Purchased
At Time Period 1
At Time Period 2
A
200
250
B
300
270
C
350
330
D
150
150
Total
1,000
1,000
TABLE 2A – Turnover Analysis
At Time Period 2
At Time
Period 1
T1 Total
Bought A
Bought B
Bought C
Bought D
Bought A
175
25
0
0
200
Bought B
0
225
50
25
300
Bought C
0
0
280
70
350
Bought D
75
20
0
55
150
T2 Total
250
270
330
150
1,000
Another Way of Looking at It
Loyalty Analysis
At Time Period 2
T1
Total
At Time
Period 1
Bought A
Bought B
Bought C
Bought D
Bought A
87.5%
12.5%
0.0%
0.0%
100%
Bought B
0.0%
75.0%
16.7%
8.3%
100%
Bought C
0.0%
0.0%
80.0%
20.0%
100%
Bought D
50.0%
13.3%
0.0%
36.7%
100%
Survey Research Methods

Personal Interview

Telephone

Self-Administered

Online (computer administered)
Survey Research Methods

Personal Interviews (or Person
Administered Surveys) – requires the
presence of a trained human interviewer
who asks questions and records the
subject’s answers
In-home or In-office
 Executive interview
 Mall intercept
 Purchase intercept

Personal Interviews
Advantages








By physically being there, the
interviewer may persuade the person
to supply answers (probably the
highest response rate).
Information on the situation may be
observed, without asking.
Best for getting response from
specific, identified person.
Versatility of questioning methods (and
flexibility in sequencing) and use of
visual materials.
Allows for probing of open-ended
questions and clarification of
ambiguous questions.
Long questionnaires may be used
successfully under interviewer's
urging.
If the respondent is having trouble
understanding, interviewer may notice
and remedy this.
Selection of sample members can be
more precise.
Disadvantages









Expensive and time intensive.
Generally narrow distribution.
People may be reluctant to talk with
strangers.
Often difficult to identify individuals to
include in the sampling frame.
Interviewer's presence, mannerisms
and inflections may bias responses.
Respondents know that they can be
identified, which may inhibit their
willingness to give information.
Great difficulty in trying to supervise
and control field interviewers.
Staffing with capable interviewers,
especially when conducted in distant
places.
May inaccurately record respondent’s
answers
Survey Research Methods

Telephone Interviews – personal interviews conducted
via telephone technology typically from a central location
Advantages








Relatively low cost
Wide distribution eliminates distance
obstacle
Callbacks
Avoids personal travel to interview
Rapid coverage of even widely
scattered sample.
Interviewers can be closely supervised
Less interview bias due to anonymity
Allows easy use of computer support
(CATI/CAI)
Disadvantages








Representative sampling frame difficult
to establish due to unlisted numbers
and increased use of cell phones
Inability to observe a respondent
Limitation to audio materials
Difficulty of conducting long interviews;
no way to prevent hang-ups
More difficult to establish rapport over
the telephone than in-person
Difficult to determine that appropriate
respondent is being interviewed
Restrictions on types of data collected
Misperceptions and “sugging”
Telephone Interviews
Results of First Dialing Attempts
Result
No answer
Busy
Out of service
No eligible person
Business
At home:
Refusal
Completion
Total
Number of Dialings
Probability of Occurrence
89,829
5,299
52,632
75,285
10,578
25,465
3,707
21,758
259,088
.347
.020
.203
.291
.041
.098
.014 (.146)*
.084 (.854)
1.000
* Probability of occurrence given eligible individual is at home

What to do about lack of listed numbers & cell
phones?



Plus-one dialing
Random digit dialing
Systematic random digit dialing
Survey Research Methods

Self-Administered – the respondent reads the survey questions
and record his/her own answers without the presence of an
interviewer. Typically done via mail, fax or drop-off.
Advantages








Sampling frame easily developed
when mailing lists are available
Respondent reads and answers
questions without interviewer influence
May respond whenever convenient
and without pressure
Any visual materials can be used
Relatively less expensive
More confidential information may be
divulged
Anonymity of respondents easier to
achieve
Eliminates the need for an interviewer
Disadvantages







A majority may not respond, and those who do
may not be typical
Significant time lag between the time the
survey is mailed and when returned
Nothing can be learned except what is written
on the questionnaire
The apparent low cost becomes relatively high
when response is poor
Questions may be misunderstood or skipped
A person may read the entire form before
answering any questions, so later questions
can influence answers to earlier ones.
Significant problems in "pass along" effect.
Survey Research Methods

Online – typically conducted via e-mail or the Internet
Advantages








Survey setup and execution can be
done very quickly
Visual materials can be used
Respondent responds without being
influenced by the interviewer
May respond whenever convenient
and without pressure
Typically the least expensive
Can permit the respondent to be
interrupted and later resume where
he/she left off
Eliminates the cost of the interviewer.
Permits real-time data cleaning
Disadvantages








Response rates are becoming an issue
May be confused with spam.
Will not be able to reach people without
access or desire to use the Internet
Nothing can be learned except what is
written on the questionnaire
No one present to stimulate replies or offer
instructions.
Potential for "pass along" effect.
Respondent frustration if questionnaire
forces response.
Issues with projectability of the sample
Selecting a Survey Method:
Factors to Consider
Situation Characteristics
Budget
 Time frame
 Quality requirements (e.g.
completeness of data, generalizability,
of the results, etc.)

Task Characteristics




Difficulty of the task
Stimuli need to elicit response
Amount of information needed
Research topic sensitivity
Respondent Characteristics



Diversity
Incidence rate
Respondent participation
Errors (or Bias) Affecting
Survey Research
Total Error
=
Sampling Error
(difference between the
actual sample results and the true population results)
+
Non-Sampling (Systematic) Error
Non-Sampling Errors

Respondent Errors

Non-response error

Response Bias
Deliberate falsification (social desirability error,
auspices error, hostility, yea and nay-saying)
 Unconscious misrepresentation (faulty recollection,
fatigue, acquiescence error, extremity bias)

Non-Sampling Errors

Measurement/Research Design
Errors





Construct development error
Scale measurement error (inappropriate questions,
scale attributes or scale point descriptors)
Survey instrument error (improper sequence, length,
poor or no instructions, etc.)
Data analysis error (use of wrong analytical
technique, etc.)
Misinterpretation error (making the wrong inference,
using only a selected portion of the study results)
Non-Sampling Errors

Problem Definition Errors


Misinterpreting the true nature of the problem
situation
Administrative Errors



Data processing errors (coding, data entry or editing)
Interviewer error (cheating, recording error,
misinterpretation, carelessness
Sample design error (sample selection error, sampling
frame error, specifying the wrong population, etc.)
Download