Data Collection: Primary Data
Data Collection: Primary Data
Data Collection: Primary Data
The varieties of primary data to be collected are nearly endless. We present the most popular
categories used in marketing research.
One type of primary data of great interest to marketers is a consumer’s demographic and
socioeconomic characteristics, such as age, education, occupation, marital status, gender, income, or social
class. These variables are used to cross-classify data to help make sense of consumer’s responses. For
example, suppose we are interested in people’s attitudes about the environment. We might suspect, test, and
find that attitudes toward green marketing are related to the respondents’ level of education. Similarly,
marketers frequently ask whether the consumption of particular products (e.g., SUVs, disposable diapers,
vacation golf packages) are related to a person’s (or family’s) age, education, income, and so on.
Demographic variables may seem simple (that is, they would not capture preferences), so it may seem
risky to make such generalizations about people. However, consider two stable examples of the usefulness of
demographics in predicting consumer preferences and behaviors: (1) 18-to-49-year-old-men tend to be
interested in sports, and it is this segment that makes sports TV programming highly profitable; (2) most of the
disposable income in the U.S comes from people 40 years old and older. Such demographic information (e.g.,
age and gender) and socioeconomic characteristics (wealth and discretionary funds) are often used to
delineate market segments.
Demographic and socioeconomic characteristics represent attributes of people. Some of these
attributes, such as a respondents’ age, gender, and level of formal education, can be readily verified. Some,
such as social class or cultural traditions can only be approximated because they are relative and not absolute
measures of a person’s standing in society. Income represents an intermediate degree of difficulty in
verifiability-even though a person’s income is an actual quantity, ascertaining the amount sometimes process
to be difficult.
Another type of primary data of interest to marketers is the customer’s psychological and lifestyle
characteristics in the form of personality traits, activities, logical, and values. Personality refers to the typical
patterns of behaviour that an individual exhibits-the attributes, traits, and mannerism that distinguish one
person from another. We often characterize people by the personality traits they display, e.g., aggressiveness,
dominance, friendliness, or sociability.¹ The argument is that personality affects consumer’s choice of stores or
products (e.g., a consumer concerned with animal testing may purchase body lotions only at the Body Shop) or
an individual’s response to an in-store and or point-of-purchase display (e.g., appealing to consumers who are
“impulsive” or “variety seekers”), just as it is believed, say, that extroverts are successful salespeople.
Lifestyle or psychographic analysis rests on the premise that the firm can plan more effective strategies
to reach its target market if it knows more about those customers in terms of how they live, what interests
them, and what they like. For example, the financial services market uses demographics (e.g., income, age)
and comfort-level with risk to understand potential investors. It’s easy to imagine pitching different investment
portfolios to different segment of people. It’s also easy to see why just knowing a client’s income isn’t sufficient-
just because two households share an income in their lifestyles or psychographic profiles, and so they would
find different investment packages attractive.
Much psychographic research has been focused on developing a number of statements that reflect a
person’s activities, interests, and opinions (AIO) and consumption behaviour (see Table 8.1). The statements
include such things as “I like to watch football games on television.” “I like stamp collecting.” “I am very
interested in national politics.” Marketing research firms collect large samples of data on purchases as well as
omnibus lifestyle patterns. The goal is to identify groups of consumers who have similar lifestyles and therefore
are likely to behave similarly toward the product or service. For example, Research Realities 8.1 provides
descriptions of travellers who share the goal of visiting friends or relatives, but who differ in family composition
(children or not), length of trip (short or long), and the activities they seek (visiting beaches or theme parks).
Segment descriptions help the marketer determine which groups to target, and how to formulate their
advertising.
It is inefficient for marketers to develop new psychographics or AIO inventories for every different
category, each product requiring new data collection and analysis. As an alternative “Value and Lifestyles”
research (VALS) avoids these problems by creating a more general psychographic framework that can be
used for a variety of products.² Research Realities 8.2 shows the basis framework of VALS, and Table 8.2
illustrates example of six international segments that have been identified by applying values surveys to
teenagers.
The term attitude used to refer to an individual’s “preference, inclination, views, or feelings toward some
phenomenon,” and opinions are “verbal expressions” of those attitudes. We treat the terms interchangeably as
representing a person’s ideas, convictions, or liking with respect to a specific object or idea.
Attitudes are pervasive because it is generally thought that they are related to behaviour-when
consumers like a product, they are more inclined to buy it than when they do not like it, and when they like one
brand more than another, they tend to buy the preferred brand. Attitudes thus are said to be the forerunners of
behaviour.
Accordingly, marketers are interested in people’s attitudes toward a product (e.g., “Do you like soft
drinks?”), their attitudes with respect to specific brands (“Coke, Pepsi, or 7-Up?”), and their attitudes toward
specific features of the brands (“diet, caffeine-free, cherry?”). Attitude I so important in marketing that we
devote an entire chapter shortly to various attitude measures.
Awareness and knowledge refer to what respondents know about some object. For instance, an
important problem in marketing is determining the effectiveness of ads in such media as TV, radio, magazine,
billboard, and Web banners. One measure of effectiveness is the product awareness generated by the ad,
using one of the three approaches described in Table 8.3. All three tests of memory-unaided recall, aided
recall, and recognition-are aimed at assessing the respondent’s awareness and knowledge. The three
measures reflect differences in the extent to which consumers have cognitively processed (in depth and detail
or just superficially) the ad. It is thought that consumers have retained more knowledge from the ad when they
state the brand in an unaided recall test (“What products and brands do you remember seeing ads for?”)
compared to a recall test where they have been given hints (“Do you remember recently seeing ads for
PC’s?”), and that both of these show superior knowledge and retention over simple recognition (“Do you
remember seeing this ad for Dell?”).
One of the common indices used to measure the short-term success and impact of an ad is day-after
recall (or DAR), which is a phone survey made the day following the airing of a new ad (e.g., the day after the
Super Bowl). The DAR scores are compared to the ad agency’s database of such indices to project sales by
using other recent ads that had achieved similar DAR scores as benchmarks.
Increasingly, advertising researchers are exploring the idea that consumers do not have to explicitly
remember an ad for that ad to nevertheless have an impact on their behaviour. For example, after airing an ad
for Reebok, the researcher might choose to use “implicit” or indirect tests of memory. Rather than asking, “Do
you remember any recent ads for athletic shoes or Reeboks?” the researcher might instead ask consumers to
list brand names of sneakers, then first choice set of sporting shoes, shoes affiliated with athletes, and so on,
to assess the number of times the Reebok brand name appears. Researchers have even asked questions as
oblique as, “Name all the brands of any kind of product that start with it” to see how often Reebok would
appear, along with names such as Reese’s, Roles, and Ramada. The assumption in these tests is that if
Reebok appears disproportionately more than it should (based on market shares), the ad was successful in
bringing the Reebok brand name to mind.
Intentions
A person’s intentions refer to the individual’s anticipated or planned future behaviour. Marketers are
primarily interested in people’s intentions with regard to purchasing behaviour. The Survey Research Center at
the University of Michigan (see isr.umich.edu/src) regularly conducts surveys for the Federal Reserve Board to
determine the financial condition of consumers. The center phones a sample of 500 households monthly,
asking 50 core questions about consumer confidence and buying intentions for big-ticket items such as
appliances, automobiles, and homes during the next few months.
In marketing intentions are often gathered by asking respondents to indicate which of the following best
describes their plans with respect to a new product or service:
The number of people who answer that they “definitely would buy” or “probably would buy” are often
combined into a “top 2 box” to indicate likely reaction to the new product or service.
It is true that behavioural intentions do not predict behaviour perfectly; a disparity often exists between
what people say they are going to do and what they actually do. However, they are fairly correlated. See
Research Realities 8.3 for a demonstration.
The prediction of behaviours by intentions is not perfect, but sometimes behavioural data are
expensive, difficult, or even impossible to obtain. For example, if Doritos were to create a new spicy salsa-
flavor chip as a line extension, by definition no purchase data would exist because the snack food would not
have been available yet for purchase. In the absence of purchase data, consumer intentions are as close to
actual behaviors as marketers can get. To help compensate for the data for the bias that intentions data are
likely to contain (based on their past experience).
Purchase intentions are most often used when studying the purchase of commodities requiring large
outlays, such as an automobile for a family or plant and equipment for a business. The general assumption is
that the larger the dollar expenditure, the more planning necessary, and the greater the correlation between
anticipated actual behaviour.
Motivation
Consumers have motives and drives (hunger, thirst, shelter, and sex) and wants (e.g., hunger needs
give rise to wanting a good steak dinner). A marketing researcher’s interest to motives typically involves
determining why people behave as they do. If we understand the motives behind a person’s behaviour, we
understand the behaviour better and in turn, are in a better position to influence future behaviour.
Behaviour
Behaviour concerns what costumers have done or are doing. Usually in marketing, this means
purchase and usage behaviour. Behaviour is a physical activity that takes place under specific circumstances,
at a particular time, and involves one or more participants. Table 8.4 is a checklist of the key dimensions of
behaviour. There are many facts to each dimension. For example, the “where of purchase” may be specified
according to the kind of store, the location of the store by broad geographic area or specific address, the size
of the store, etc.
Behaviour data are becoming increasingly available through various technologies and are very
important to marketers in customer relationship management. The most prevalent behavioural data are
scanner data, as described in Chapter 7. A different source of behaviour data is online click-streams, which
capture personal profile data, click-stream trails, and records of ads and responses. Many marketers feel more
confident predicting customers’ future behaviour, rather than from surveys of likely future behavioral intentions;
others point out that attitudinal and motivational data supplement behavioral data, giving the marketer an
understanding of why consumers behave the way they do. Clearly this information is complementary and it
would enrich the marketer’s understanding to the both.
Primary data can be collected in several ways (see figure 8.1). The primary decision is whether to use
communication or observation techniques. The first involves a questionnaire or survey, oral or written.
Observation does not involve questioning. Rather, facts or behaviors are recorded. The observer may be a
person (e.g., a researcher conducting a pantry audit) or some mechanical device (namely scanner or the
Web).
Communication and observation each has its own advantages and disadvantage that can be classified
according to several dimensions:
The communication method of data collection has the general advantages of versatility, speed, and
lower cost, whereas observational data are typically more objective and accurate.
Versatility
Versatility refers to a technique’s ability to collect information on the different types of primary data of
interest to marketers. A respondent’s demographic and socioeconomic characteristics and lifestyle, attitudes
and opinions, awareness and knowledge, intentions, motivations underlying actions, and even behaviors
may all be obtained by the communication method (e.g., interview or survey). All we need to do is ask
(problems of accuracy may result, and we discuss this issue later).
Observation is more limited. It is best suited for obtaining information about behaviour and certain
demographic or socioeconomic characteristics. Among behaviors, note that that we are limited to observing
present behavior; we cannot observe past behavior, nor intentions for future behavior. Among demographic or
socioeconomic characteristics, some can be observed but with less accuracy. A person’s age and income, for
example, might be inferred by closely examining persons clothing and purchasing behavior.
The speed and cost advantages of the communication method are closely intertwined (time is money!).
Communications often a fasters means of data collections, because researchers are not forced to wait for
events to occur as they are with the observation method. For example, observer checking for the brand
purchased most frequently in one of several appliance categories might have to wait a long time to make any
observation at all. Much of the time the observer would be idle. Such idleness is expensive, given that the
worker is probably compensated on an hourly rather than a per-contact basis.
There are always exceptions. Observation is faster and costs less than communication in the purchase
of consumer package goods. Scanners allow many more purchases to be recorded and at less cost than if
purchasers were questioned about what they bought.
Typically data are more accurate by observation because this method is independent of the
respondent’s unwillingness or inability to provide the information desired. For example, respondents are often
reluctant to cooperate when their replies would be embarrassing or place them in an unfavourable light. Since
observation allows the recording of behaviors as it occurs, it does not depend on the respondent memory or
mood in reporting what occurred.
Observation typically produces more objective data. The interview represent a social interaction
situation. Thus, the replies of the person being question are conditioned by the individual’s perception of the
interview. With observation, the consumer’s perceptions play less of a role. Sometimes people are not even
aware that they are being observed, thus removing the opportunity for them to tell the interview what they think
the interview wants to hear or to give socially acceptable responses.
Communication Methods
Choosing the communication method of data collection implies several decision. Should we administer
questionnaire be e-mail, online by mail, over the telephone, in person at a shopping mall, or via fax? Should
the purpose of the study be disguised? About method of administration has implications regarding the degree
of structure that must be imposed on the questionnaire.
Figure 8.1 summarizes the decisions that a researcher must make when collecting data, including issue
of structure and disguise. Structure is the degree of standardization imposed on the questionnaire. In a highly
structure questionnaire, the question to be asked and the responses permitted the respondents are
predetermined (e.g., “Circle a number from 1 to 7). In an unstructured questionnaire, the question to be asked
are only loosely determined, and respondents are free to answer in their own words (e.g., “Tell me how your
family feels about ice cream”).
Disguise is the amount of knowledge about purpose of a study communicated to a respondent. An
undisguised questions posed, whereas a disguised questionnaire attempts to hide the purpose of the study.
For example, if ford wished to determine its customers’ satisfaction they might worry that a cover letter from
Ford and questions all about Ford cars and trucks would bias the respondents’ answer favourably toward Ford,
since the survey’s purpose is clear. If Ford wanted more objective data, it might go through an outside
marketing research agency, and ask its drivers about Ford, GM, and Honda cars. In this scenario, the target of
the research is less clear, and it could be expected that the customer would answer more truthfully.
Structured-Undisguised Questionnaires
Structured-undisguised questionnaires are most commonly used in marketing research. Questions are
presented with exactly the same wording, in the same order, to all respondents. The reason for standardizing
questions is to ensure that all respondents are replying to the same question. If one interviewer asks, “Do you
drink orange juice?” and another ask, “Does your family use frozen orange juice?” the replies would not be
comparable.
Typically the responses are also standardized. Consider the following question regarding a person’s
attitude toward legislation about pollution, and note the fixed-alternative responses.
Do you feel the United States needs more or less environment legislation?
Needs more
Needs less
Neither more nor less
No opinion
Unstructured-Undisguised Questionnaires
In unstructured-undisguised questionnaires, the purpose of the study is clear but responses to the
question are open-ended. Interviews are the primary method. For example, “how do you feel about
environmental issues and the need for more legislation?” is delivered the same to every respondents to talk
freely about their attitudes. After the initial question, the initial question, the interview becomes very
unstructured as the interviewer probes more deeply, following whatever direction the respondent interviewer
probes more deeply, following whatever direction the respondent leads. Figure 8.2 classifies a number of
communication-based marketing research techniques by their degree of structure and disguise. Interviews and
open-ended questions on surveys are popular methods.
Structured-Disguised Questionnaires
Structured-disguised questionnaires are the least used in marketing research. They were developed in
an attempt to reveal subconscious motives and attitudes yet to do so in a structured manner to facilitate
coding.
One proposition holds that consumers, knowledge and memory are conditioned by their attitudes.
Individuals selectively perceive and retain ideas that are consistent with their own beliefs. Thus, if we want
information about people’s attitudes when a direct question might produce a based answer, we can simply ask
them what they know. For example, democratic voters would be expected to know more about Democratic
candidates than those intending t voted Republic. This argument is consistent with what we know about the
operation of selective cognitive processes. Thus, if we want to know someone’s attitude about the
environment, but we don’t want them to simply respond in a socially acceptable way, we could ask people what
they know rather than how they feel.
However, one question is whether a high level of awareness is indicative of favorable or an
unfavourable attitude. That is, whether this measure of knowledge can also be interpreted as a measure of the
person’s attitude relies on an assumption about the consumer’s cognitive system.
Questionnaires can also be classified by the method used to administer them. The main methods are
personal interview, telephone, mail, fax, and e-mail and online surveys.
A personal interview is a direct, face-t-face conversation between the interviewer and the respondent or
interviewee. The interviewer asks the questions and records the respondent’s answer, either while the
interview is in progress or immediately afterward. The interview can take place in a home or an office, but it
usually is in a central location like a shopping mall, where shoppers are stopped (or intercepted, hence the
term “mall intercept”) and asked to participate.
The telephone interview is usually a cold call with a few questions (no longer than 2-3 minutes). A mail
questionnaire involves mailing the questionnaire to designated respondents with an accompanying cover letter.
The respondents complete the questionnaire at their leisure and their replies back to the research organization.
Faxed surveys operate like mail survey (except of course that they are faxed) and work well for business-to-
business research.
E-mail survey are one of two types:
The questions of the market research study are embedded in the text of the e-mail itself.
The e-mail directs the recipient to a web site to take the survey.
Each has its pros and cons: Replying to embedded e-mail surveys is extremely quick, but Web surveys can
looks sharp-very professional in that hypertext and graphics can improve both the appearance and quality
control of the survey.
A number of variations and combinations are possible. Questionnaire for a “mail” administration may be
attached to products or printed in magazines. Questionnaires in a personal interview maybe self-administered,
perhaps in the interviewer’s presence, in case the respondents seek clarification from the interviewer.
Alternatively, the respondents might complete the questionnaire in private and drop it in the mail to return to
the research organization.
Often the different modalities are mixed to enhance sample cooperation. For example, marketing
researchers seeking the opinions of busy business managers or physicians typically employ an introductory
letter, e-mail , or phone call asking for their help in the study, and after this pre-notification, the survey is sent
(e.g., faxed) to their place of business.
Each of these methods of communication possesses some advantages and disadvantage. These
include sampling control, information control, and administrative control.
Sampling Control
Sampling control concerns the researcher’s ability to find a designated respondent and to get desired
cooperation from that respondent. A sampling frame is the list of population elements from which the sample is
drawn. With the telephone method, online phone books serve as the sampling frame, from which respondents
are selected randomly, but they are inadequate because they do not include household with unlisted numbers
or those without phones. For example, about a third of the 90 million U.S telephone household have unlisted
numbers. Unlisted households tend to be younger, urban, non-white, more mobile, and either very high or very
low income, with the majority residing in Californian cities. “Do not call” registries have also wreaked havoc,
with nearly 75% of Americans blocking telemarketing calls, even though marketing research calls serve a
different purpose.
Some researchers attempt to overcome the sampling bias of unlisted numbers by using random digit
dealing (RDD). Numbers are randomly generated and called through automatic dialling at a central
interviewing facility. This procedure allows geographically wide coverage. In plus-one sampling, a probability
sample of phone numbers is selected from the telephone directory and a single, randomly determined digit is
added to each selected.
It is conceptually difficult, but partially possible, to achieve sampling control for the administration of
questions using personal interviews. For some select population (e.g., doctors, architects, businesses), a list of
members may be available in trade directories. Consumer studies frequently use mall intercepts, in which
interviews stop shoppers in a mall and ask if they would be willing to participate in a research study. Those
who agree are taken to the firm’s interviewing facility in the mall (a small, rented office), where the interview is
conducted. Mall intercepts pose two issues; 1) Most people shop at malls, but some do not. 2) A person’s
chances of being asked to participate depend on the frequency with which he or she shops there and the time
spent in the mall.
Getting Cooperation
Directing the survey to a specific respondent is one thing; getting a response from that individual is
another. In this respect, the personal interview affords the most sample control, e.g., the respondent’s identity
is known and there is a little opportunity for anyone else to reply. The rate of refusal-to-participate is also
typically lower than with phone interview or mail questionnaires.
Usually the principle holds that the more personal the appeal, the more difficult it is for a respondent to
say no: Malls are face-to-face and phone solicitations are person-to-person. Mall is the least personal, most
anonymous channel, and many mall surveys end up in recycling bins, unless the topic is inherently interesting
to the consumer, or perhaps if some incentive is offered to complete the survey. We say more about factors
later.
Telephone methods suffer from “not-at-homes”; one-third of calls result in a “no-answer”, but this
number surprisingly stable even in the face of the growth of answering machines and caller ID. Research firms
generally can still get through to talk to a live consumer, by being persistent (by making more contacts). See
Research Realities 8.4 for key ingredients to enhancing participation in phone surveys.
Information Control
The different methods of data collection also vary in the type of questions that can be asked, and in the
amount and accuracy of the information that can be obtained from respondents. The personal interview can be
conducted using almost any form of questionnaire; structured or unstructured, disguised or undisguised. The
interaction allows the interviewer to present pictures or examples of ads, list of words, scales, and so on. The
consumer can taste new flavours of cookies or smell new colognes. Visual aids can be used with mail
questionnaires, but not with telephone surveys.
Personal interviews allow the use of open-ended questions that require extensive probes. Written
questionnaires do not lend themselves to such questions. Telephone interviews can incorporate open-ended
questions, but not nearly the same extent as in-person interviews, mostly because phone surveys need to be
brief so as not to be discontinued by an increasingly bored or irritated respondent midway through the survey.
Personal interviews also allow the automatic, contingent sequencing of questions; for example, if the
answer to question 4 is positive, the interviewer may be instructed to proceed to ask questions 5 and 6,
whereas if the answer had been negative, the interviewer asks questions 7 and 8. Automatic sequencing is
also possible with telephone interviews, especially when the interviewer is reading from a computer screen and
the question skipping is programmed.
Skipping questions is not advised for mail questionnaires, because respondents will get confused and
make errors. Web surveys can easily be programmed to contain skip patterns.
Currently, there are two applications of computer-aided interviewing (CAI) software:
Telephone surveys, in which each interviewer has a personal computer from which to ask questions
(see Table 8.5 for Computerized Adaptive Telephone Interviewing, or CATI).
In-person interviews, such as mall intercepts, in which the interviewer transports a laptop computer to
the interview site and uses it to interview the respondent, or places the computer in front of the
respondent and lets the respondent answer questions as they appear on the screen.
One of the most important advantages of computer-assisted interviewing is its level of information
control. The computer displays each question exactly as the researcher intended (and programmed). It
shows only the questions and information that the respondent should see and it displays the next
question only when an acceptable answer to the current one is entered. In addition, you get:
Personalization of the questions. During the course of the interview, the computer can use answers to
previous responses (name of spouse, cars owned, favourite brand) to customize the wording of
subsequent questions, e.g., “When your wife, Ann, goes shopping, does she usually use the Toyota or
the Buick?” Such personalised questions are thought to enhance rapport and provide for higher-quality
interviews.
Customized questionnaires. Key information elicited in the interview can be used to tailor the
questionnaire for each respondent. For example, if it’s clear the respondent is a novice, they won’t be
asked about certain technical product attributes.
After the data collected, preliminary results are available at a moment’s notice, because the data are
already stored. The marketing manager does not have the 2-3 week delay caused by coding and data
entry that happens when questionnaires are completed and entered by hand.
However, even computers have limitations. They cannot win over respondents with social chitchat or
explain questions that are misunderstood. Computers are incapable of recognizing fuzzy or superficial
answers, and they cannot prod respondents to elaborate on their answers.
Administrative Control
One of the greatest advantages of Web surveys is that they provide the marketing researcher
the quickest turn around. Half of e-mail surveys are typically completed and returned the same day they
were sent (see Research Realities 8.5). For phone surveys, a number of calls can be made from a
central exchange in a short period, perhaps 15 to 20 per hour, per interviewer if the questionnaire is
short. It can take two weeks to achieve that same response rate using the postal service.
Figure 8.3 summarizes the logistics comparisons between these major methods of
administering questionnaires. Table 8.8 elaborates on the overall strengths and weaknesses of these
methods and Figure 8.4 illustrates their frequency in practice.
Exercises
1. How do personally administered questionnaires, telephone surveys, mail, fax, e-mail, or Web
surveys differ with respect to the following?
a. Sampling control
b. Information control
c. Administrative control
2. What is a disguised questionnaire? What is a structured questionnaire?
3. What is the rationale for employing unstructured-disguised stimuli?
4. What are the general advantages and disadvantages associated with obtaining information by
questioning or by observation?
5. What are the advantages and disadvantages of structured-undisguised questionnaires? Of
unstructured-undisguised questionnaires?
6. What types of primary data interest marketing researchers most?