You are on page 1of 17

PETER N.

GOLDER*

Several researchers have advocated historical or longitudinal ap-


proaches to study marketing phenomena. Although some have applied
this approach, more often it has been overlooked or denigrated. The au-
thor argues that historical method is capable of producing scientific
knowledge that is currently useful, rather than simply a remembrance of
the past. The author presents a complete description of the historical
method, so researchers can use this article as a guide when applying this
method. The value of the method is illustrated by examining the prevailing
finding in the marketing literature that market shares are stable over time.
Although this finding is considered an empirical generalization, an analy-
sis of more than 650 brands in 100 categories raises doubts about the
longevity of market share stability.

Historical Method in Marketing Research with


New Evidence on Long-Term Market
Share Stability

Poetry is more scientific than history. have argued for its use to provide a better understanding of
—Aristotle marketing’s past (Fullerton 1987; Pollay 1985; Savitt 1980).
History is bunk. Still others, representing a more traditional research per-
—Henry Ford spective, have recognized that historical analysis could be
useful for studying strategic issues (e.g., Aaker and Day
Several researchers have advocated using historical or 1986; Day and Wensley 1983; Prahalad 1995).
longitudinal approaches to study marketing phenomena However, other researchers have characterized historical
(Aaker and Day 1986; Lieberman and Montgomery 1988; method as unscientific (Bonoma 1985; Calder and Tybout
Prahalad 1995; Zaltman 1997). Recently, some studies have 1987; Nevett 1991). Bonoma (1985) argues that stories have
used archival data (e.g., Bayus, Jain, and Rao 1997; Golder similar data integrity and more generalizability than archival
and Tellis 1993; Low and Fullerton 1994; Tellis and Golder research. Calder and Tybout (1987) believe that historians
1996). However, no article in the marketing literature fully generate interpretive knowledge and therefore consider it
describes the historical method as it has been developed by unscientific. Even Nevett (1991), who calls for the use of
historians or characterizes it as a means of generating scien- historical analysis, proposes that it is useful for marketing
tific knowledge. The purpose of this article is to present a practitioners rather than researchers and specifically con-
comprehensive description of the historical method and trasts it with scientific analysis. The common belief that his-
demonstrate its usefulness for generating knowledge about torical research is unscientific may explain the limited num-
marketing phenomena. ber of articles that apply this method in marketing.1
In marketing, most proponents of historical method come However, this belief is surprising, given that many histori-
from an interpretivist or relativist perspective (Hudson and ans view their discipline quite differently. Even economists
Ozanne 1988; Sherry 1991; Smith and Lux 1993). Others value the historical method, as is evidenced by the awarding
of the 1993 Nobel Prize in economics to Robert Fogel and
Douglass North for their historical analyses of economic
*Peter N. Golder is Associate Professor of Marketing, Stern School of
Business, New York University (e-mail: pgolder@stern.nyu.edu). The con-
and institutional change.
tributions of Dave Stewart and Gerry Tellis were most valuable in the early This article differs from previous articles on historical
stages of this project. The comments of Bob Shoemaker, Sudhir analysis in marketing and consumer behavior. Savitt (1980)
Karunakaran, and the three anonymous JMR reviewers were helpful in im-
proving the article. The author gratefully acknowledges the data collection
efforts of Kathy Eng, Liesel Johnson, Amy Leo, and Edward Zelmanowitz. 1Several historians have presented their research on marketing topics in
To interact with colleagues on specific articles in this issue, see “Feedback” journals such as Business History Review and Journal of Economic History
on the JMR Web site at www.ama.org/pubs/jmr. and books. Examples include Cusumano, Mylonadis, and Rosenbloom
(1992), Olmstead and Rhode (1995), and Tedlow (1996).

Journal of Marketing Research


Vol. XXXVII (May 2000), 156–172 156
Historical Method 157

addresses the importance of studying marketing’s past rather as well as the recent past that is not readily available from
than providing a description of traditional historical method. conventional data sources.
His article references two works on the historical method, Historical method has been defined by one researcher as
whereas this article presents a synthesis of nearly 30 refer- a process of critically examining and analyzing the records
ences. Savitt’s (1980, p. 54) position is “that historical re- and survivals of the past (Gottschalk 1969). The definition I
search in marketing must conform in every way possible to use is the process of collecting, verifying, interpreting, and
experimental control as applied in the experimental presenting evidence from the past. I fully elaborate the ele-
method,” including preparing hypotheses before collecting ments of this definition in a subsequent section in which the
data. This view is not common among historians, because historical method is presented.
they do not control the conditions under which their data are
OBJECTIVES OF HISTORICAL RESEARCH
created. Also, historians seek to minimize potential biases
while collecting and analyzing data rather than focus on spe- Like scientists and unlike fictive artists, historians ac-
cific hypotheses. Finally, Savitt recommends conducting cept a primary obligation to check their assertions
historical research for looking at the past rather than using against evidence or data about a world that is or once
was, “out there,” evidence that is open to public
an understanding of the past to make better decisions today. scrutiny and criticism by reexamination of the respec-
More recently, Nevett (1991, p. 20) proposes using his- tive records. (Hexter 1971, p. 13)
torical analysis “as a supplement to the techniques normally
employed in marketing practice.” However, Nevett believes [The first law of the historian] is never to dare utter an
that the historical method is “impressionistic and intuitive” untruth and the second, never to suppress anything true.
(p. 22) and specifically differentiates it from scientific (Handlin 1979, p. 414)
analysis. Also, his article does not present a detailed de-
scription of the historical method. Historians seek to develop accurate descriptions of social
Finally, Smith and Lux (1993) argue that historical re- phenomena on the basis of a careful consideration of all rel-
search is useful for explaining changes in consumer behav- evant and available data. They strive to understand events in
ior over time. They develop a “process of doing historical their full context (Elton 1967). Although few historians ex-
research not previously explicated, either by historians or by pect to find laws, they do seek generalizations while recog-
consumer researchers” (p. 599). Although their approach is nizing that interesting and useful generalizations admit ex-
useful, it does not include all the steps in the traditional his- ceptions (Elton 1967; McCullagh 1984). Generalizations in
torical method. Moreover, they consider their approach “a history are guided by the following principles: They are
qualitative interpretive method” (p. 595), which differs from founded on all facts and contradict none, they are plausible
the way many historians view their own method. and do not contradict the accepted laws of nature, they are
Although these three articles are useful, their orientation capable of disproof and verification, and they are as simple
differs from that of the present one. This article has three as possible (Shafer 1974).
primary objectives that contribute to the marketing literature The nature of proof in history is often compared to the le-
in unique ways: gal model in which events are established beyond a reason-
able doubt (Fogel and Elton 1983; Salvemini 1939; Shafer
•to describe how historians view their own research method and 1974). Similar to the five percent statistical significance
its objectives, rule, knowledge is accepted when it is highly likely rather
•to present a succinct description of the historical method that than certainly true.
can be used as a reference for marketing researchers applying Historians try to understand the causes of events and to
this method, and assess their relative importance (Gottschalk 1969;
•to demonstrate the usefulness of this method by reexamining
Salvemini 1939). They look beyond immediate causes to de-
the empirical generalization that market shares are stable over
prolonged periods. termine underlying ones as well, because immediate causes
may be just precipitating events (Gottschalk 1969; Shafer
The historical method may appear to be limited in the 1974). If an underlying cause makes an event highly proba-
type of questions it can address. However, several market- ble, immediate causes are less important (McCullagh 1984).
ing researchers have used archival or longitudinal data. Historians seek to generate knowledge that is falsifiable,
During the 1990s, 19 articles with these data were published thus satisfying one important criterion of scientific knowl-
in major marketing journals (see Table 1).2 However, only a edge (Calder and Tybout 1987; Fogel and Elton 1983). This
few of these articles specifically applied the historical falsification process is common among historians. As
method. Although many people believe that the historical Marwick (1970, p. 212) notes, “Should [a researcher’s]
method is useful only for gaining an understanding of pre- work not be based on a thorough scholarship, should it not
vious events, only one article adopted this perspective. The be offered in good faith, the ever-ready police battalions of
more useful applications of historical method are cases in the historical guild will soon club his pretensions into jelly.”
which an understanding of previous events provides insights The research objectives of most historians are similar to
into the current marketing environment. Furthermore, this those of many marketing researchers. In addition, historians
method’s principles are useful for evaluating and validating have developed their own rigorous method of collecting and
all secondary data. Finally, the historical method provides evaluating data. Most historians believe that applying this
an approach to uncover and verify data from the distant past method generates scientific knowledge (Cochrane 1929;
Gottschalk 1969; Iggers 1979; Marwick 1970; Salvemini
2These articles are from Journal of Marketing, Journal of Marketing 1939). Although the historical method has proved useful in
Research, and Marketing Science. Journal of Consumer Research was not a few marketing studies, it could be used in many more,
included, because some consider its scope broader than marketing. specifically those with longitudinal or archival data.
158 JOURNAL OF MARKETING RESEARCH, MAY 2000

Table 1
MARKETING RESEARCH USING ARCHIVAL DATA

Research Area Insight from Application Reference

Market entry Market pioneers have higher failure rates and lower market shares than has been found in previous Golder and Tellis (1993)
research.
Initial competitive positioning and media coverage drive long-term performance. Green, Barclay, and Ryans
(1995)
Speeding a new product to market to capture a first-mover advantage is misguided if product Bayus, Jain, and Rao (1997)
performance is poor or the size of the potential market is overestimated.
New products Fear of cannibalization inhibits incumbent innovation. Ghemawat (1991)
Impact of next-generation products drives declining prices for consumer durables. Bayus (1992)
Marketing has a limited role in new product development in an engineering-driven culture. Workman (1993)
Line extensions of strong parent brands and symbolic parent brands are more successful. Reddy, Holak, and Bhat
(1994)
New product success in the United States and China is based on appropriate resources and expertise. Calantone, Schmidt, and
Song (1996)
Sales of really new consumer durables have a dramatic increase or takeoff in sales many years after Golder and Tellis (1997)
introduction.
History of Early philosophical foundation of marketing based on German historical school of economics. Jones and Monieson (1990)
marketing
New concept of marketing focusing on relationships is developed on the basis of historical Webster (1992)
development of marketing.
Future of brand management is evaluated on the basis of its historical development. Low and Fullerton (1994)
Other Identification of factors characterizing buyer behavior process in high-technology markets. Weiss and Heide (1993)
Multidecade analysis of the grocery channel concludes that retailers have not gained profitability at Messinger and Narasimhan
the expense of manufacturers. (1995)
Celebrity endorsers have a positive impact on stock returns. Agrawal and Kamakura
(1995)
Longitudinal analysis of optometry market supports the conclusion that some consumers are satisfied Parker (1995)
with low quality at high prices.
Firms with competitor-oriented objectives are less profitable and less likely to survive than those that Armstrong and Collopy
are profit oriented. (1996)
Framework for how to incorporate marketing strategy with environmental consciousness. Menon and Menon (1997)
Determination of the forces that foster long-term business-to-business relationships. Keep, Hollander, and
Dickinson (1998)

HISTORICAL METHOD torians needed a precise approach to analyzing data


History is, to be sure, scientific in method. (Gottschalk (Marwick 1970).
1969, p. 8) Historical method is broken down into five stages: (1) se-
lect a topic and collect evidence, (2) critically evaluate the
What characterizes the scientific approach is a highly
sources of the evidence, (3) critically evaluate the evidence,
critical attitude toward our theories. (Popper 1968, p. 94) (4) analyze and interpret the evidence, and (5) present the
evidence and conclusions. Table 2 lists the elements of each
There is striking unanimity among historians regarding stage, and the next sections describe these elements.
what constitutes historical method (Gottschalk 1969). The Afterward, the method is applied to a specific marketing ex-
overriding characteristic of historical method is that all evi- ample: an analysis of long-term market share stability.
dence is approached critically or skeptically. Any witness,
whether living or dead, may have made mistakes or intended Stage 1: Select a Topic and Collect Evidence
to mislead (Salvemini 1939). Therefore, application of the Historians can study a wide range of topics from the dis-
historical method acts as a defense against fanciful tale, tant as well as the more recent past, as long as sufficient
willful distortion, and even honest error (Breisach 1983). records of these events are available (Brooks 1969; Fogel
The modern approach to historical method was first ex- and Elton 1983). Fortunately, most marketing research top-
pounded in the late 1800s by Ernst Bernheim (Gottschalk ics focus on twentieth-century events for which data are
1969; Johnson 1934) and Langlois and Seignobos (1898). plentiful. In addition, recent records are more likely to in-
Langlois and Seignobos felt the need to develop an au- clude the quantitative measures desired by some researchers.
tonomous method, because they believed that instinctive After selecting a topic, historians prepare questions of in-
methods would become irrational methods. Therefore, his- terest rather than develop hypotheses from established the-
Historical Method 159

Table 2
STAGES OF THE HISTORICAL METHOD

Stage Elements of Stage


1. Select a topic and collect evidence Select research topic.
Determine if sufficient material is available on research topic.
Collect data: published materials, field interviews, archival materials, cultural artifacts.
Prepare questions of interest.

2. Critically evaluate the sources of the evidence Textual criticism: Verify that documents are original or best copy available.
Are witnesses reliably recorded?
Investigation of authorship: Who wrote the document, and where, when, and under what circum-
stances was it written?
Is the author able to report correct information?
Classify authenticity of sources based on temporal and geographic proximity to event being
recorded, purpose of written record, and expertness of author.

3. Critically evaluate the evidence Interpretive criticism: Evaluate and determine the author’s meaning.
Negative internal criticism: Evaluate veracity of author’s statements for mistaken or purposely
false information.
Is the author willing to report correct information, or are there personal interests or biases?
Evaluate independence of observations. Is there corroboration from an equally credible witness?

4. Analyze and interpret the evidence Organize events chronologically within relevant subjects.
Draw easily deduced inferences from the data.
Consider objectivity of people who created historical record and the completeness of their reports.
Look for generalizations, explanations, and implications in the evidence.
Suitable data may be analyzed quantitatively and econometrically.
Assess the relative importance of causal factors.

5. Present the evidence and conclusions Select evidence to present.


Forms of presentation include chronologies, descriptive statistics, model parameters, and narrative
discussion.
Narrative discussion of evidence and conclusions includes key events and the thesis or explana-
tion of those events.

ory (Goodman and Kruger 1988). They approach sources in With so much data available, a common mistake of re-
an attempt to find their answers (Shafer 1974; Todd 1972). searchers is to spend too much time collecting unnecessary
The general approach is that the “material must precede the data. Researchers must be vigilant in focusing on relevant
thesis” (Tuchman 1981, p. 9). After some evidence has been rather than merely interesting data; otherwise, they can be
evaluated, working hypotheses can be developed (Johnson overwhelmed (Eisenhardt 1989). The researcher’s questions
1934; Shafer 1974). However, these hypotheses should be in of interest and working hypotheses must be used to focus data
the interrogative rather than the declarative form to help the collection efforts. Researchers should constantly ask whether
researcher maintain a noncommittal position (Fogel and data being collected support or refute their propositions.
Elton 1983; Gottschalk 1969). When researchers collect evidence, notes are often consid-
Potential data sources are published materials, field inter- ered sufficient (Gottschalk 1969). Transcribing exact quota-
views, archival materials, and cultural artifacts (e.g., adver- tions is recommended in cases in which that information may
tisements, billboards, structures). Initial research for most contradict the accepted story. I have found that making copies
questions begins in the library, where books are available on of all relevant articles is most efficient. Important information
almost all topics. Electronic databases (e.g., Dow Jones and can be highlighted rather than transcribed, and the entire text
ABI/Inform) are extremely useful for more recent articles. is readily available for quotation and further analysis.
Older articles can be found through Business Periodicals Although historical research may end up confirming es-
Index, Readers’ Guide to Periodical Literature, New York tablished theory, its more useful application may be gener-
Times Index, and Wall Street Journal Index. A much more ex- ating new theory from new data. It has this potential because
tensive and Internet-accessible collection of articles is cur- this method can be used to study phenomena that are not
rently being prepared by Bell & Howell’s UMI division. Its suited to other methods. The process of discovering gener-
Digital Vault Initiative will soon offer access to 5.5 billion alizations about previously unknown phenomena may be the
pages of its collection, which ranges from fifteenth-century more common mode of scientific progress. Simon (1968, p.
literature to recent business periodicals, as well as more than 458) states that “histories of science written in terms of the
1.4 million dissertations. The availability of this collection processes that discover patterns in nature would seem closer
will dramatically reduce the time required to conduct histor- to the mark than histories that emphasize the search for data
ical research and expand the possibilities for obtaining new to test hypotheses created out of whole cloth.”
insights. If published materials are not sufficient, researchers
most likely will try to obtain data from company archives or Stage 2: Critically Evaluate the Sources of the Evidence
through personal interviews. Additional sources of written The primary purpose of this stage is to evaluate the au-
evidence include records from government libraries, court- thenticity of documents and to exclude all inauthentic evi-
houses, churches, and private papers. dence (Gottschalk 1969). Many historians refer to this stage
160 JOURNAL OF MARKETING RESEARCH, MAY 2000

as “external criticism.” During this stage, researchers seek After completing external criticism, researchers should
to determine who wrote each document and where, when, have established the testimony of their witnesses (as
and under what circumstances it was written (Shafer 1974). recorded in documents most often) and have an understand-
If complete documents are not available, incomplete texts ing of each witness’s capability to report that testimony. The
are established in their most accurate form. Langlois and next stage analyzes the particulars of that testimony.
Seignobos (1898, p. 113) considered external criticism of
utmost importance: “Sure methods of external criticism Stage 3: Critically Evaluate the Evidence
have raised history to the dignity of a science.” Historians The primary purpose of this stage is to evaluate the par-
sometimes criticize social scientists for not sufficiently ap- ticular contents of a document and extract credible testi-
plying the criteria of external criticism and accepting data mony from it (Gottschalk 1969). Many historians refer to
too easily (Marwick 1970). this stage as “internal criticism.” During this stage, re-
Langlois and Seignobos (1898) list three steps in external searchers usually start with a literal interpretation of the ev-
criticism. The first step is textual criticism, which calls for idence and then consider context to determine if the real
the examination of documents to see if they are original or meaning is different from the literal meaning (Shafer 1974).
the best copy available. It is preferable to obtain multiple Documents are evaluated for two types of errors, deliberate
copies, but if only a single copy exists, it is assumed to con- and unintentional (Shafer 1974).
tain errors. The next step is investigation of authorship, Again, Langlois and Seignobos (1898) provide a useful
which calls for the determination of who wrote the docu- summary of internal criticism. One step is interpretive criti-
ment, where it came from, and when it was written. Finally, cism, which seeks to determine what the author meant.
there is a classification of sources, in which verified docu- Special attention is required if authors are from outside the
ments are arranged according to several principles. researcher’s culture, because determination of intended
Documents are considered more authentic if they are written meaning must be based on an understanding of the authors’
with the following characteristics (Gottschalk 1969): culture and the period in which they lived. Semantic misun-
derstandings can be minimized by using a dictionary from
•close to the event being researched (or rely on records written the authors’ culture and time. Also, a witness’s testimony
close to the event), must be considered in its full context rather than in isola-
•for the sole purpose of making a record (e.g., legal documents),
•for confidential communication,
tion. Otherwise, a humorous or ironic statement may be
•for communicating with a small number of people (personal misunderstood.
correspondence), A second step is negative internal criticism, in which the
•for a personal record or memory aid (diary), veracity of the authors’ or witnesses’ statements is evalu-
•for making a public record (newspapers and magazines), and ated. What witnesses expressed may not be what they be-
•by experts with broad knowledge of the events of interest. lieved, because they might have lied; in addition, what they
believed might not have actually happened, because they
Newspaper or magazine reports are often rated high in au-
might be mistaken. Therefore, historians do not feel obli-
thenticity because the time between events and their report-
gated to accept all the testimony of even authentic witnesses.
ing is short. In addition, reporters generally have a vested in-
An important consideration in determining veracity is the
terest in reporting events accurately, and the public nature of
witnesses’ objectives in presenting testimony. Personal in-
the reports leaves them open to correction. The reputations
terests may distort a report for the witnesses’ benefit. An au-
of these sources must be considered too.
thor with a literary style may recount hearsay without the
Documents, especially older ones, must be evaluated for
proper disclaimer. Egocentrism may increase the witnesses’
forgery. Chemical tests can be used to validate the date of
role. A desire to please may cause omission of important de-
documents and artifacts. Anachronisms within a document,
tails or a less harsh presentation of events. Conditions that
as well as the means by which it was prepared, have re-
promote veracity include the witnesses being indifferent to
vealed many forgeries. Therefore, it is important to have
the subject or event, the witnesses being harmed by their
broad knowledge about the events, culture, and language of
own testimony, and the witnesses relating events that would
the period of interest. Today, the risk of accepting forgeries
be common knowledge among their audience.
may be much less, because multiple copies are usually avail-
The final step is evaluating the independence of observa-
able. Some modern documents that must be scrutinized for
tions. Testimony from multiple, independent witnesses is
authenticity are press releases, propaganda, personal apolo-
preferred. Because evaluating credibility is central to inter-
gies, and private correspondence. As documents are stored
nal criticism and important to external criticism, Gottschalk
and retrieved electronically, the issue of potential forgery
(1969) proposes four criteria for this critical task:
becomes more important again.
Another useful classification is whether documents are •Competence: Is the witness able to report correct information?
based on primary or secondary sources.3 Primary sources Important factors are nearness to the event and to the recording
are eyewitness accounts of an event. They also could be of the event (geographical and temporal) and familiarity with
based on audio or video recordings of an event. Secondary the subject (expertness).
sources are testimony from witnesses who were not present •Objectivity: Is the witness willing to report correct information?
at the event of interest. Although the testimony of secondary Important factors are assessing personal interests, biases, and
desire to please.
sources is not as authentic, it can provide important corrob-
•Reliability: Is the witness accurately reported? Does the re-
oration or add missing details that are consistent with the porter have a reputation for integrity? Does the document lack
testimony of primary sources. internal self-contradictions?
•Corroboration: Is there confirmatory evidence from equally
3Some historians place this discussion under internal criticism. credible witnesses?
Historical Method 161

Evidence is considered credible if it passes all four crite- rather than the existence of certain evidence or causes. Even
ria.4 In particular, the final criterion is vital for developing though historians sometimes hold competing theories and
confidence in the data. Conformance with other facts and explanations, this situation is shared with other sciences. In
witnesses is a critical test of evidence. Uncorroborated evi- terms of the data on which these theories depend, historians
dence should be reported accordingly. When secondary evi- have an advantage over some other scientists. The evidence
dence is evaluated, credibility is determined for the primary of historical events is generated independently of any effort
evidence on which it is based (Gottschalk 1969). Doubts to study them. Finally, historians may develop generaliza-
about particulars of an event usually arise from the lack of tions and try to interpret events in a way that is relevant to
testimony based on primary evidence. the present (Fogel and Elton 1983).
The goal of internal criticism is to verify the events on When suitable quantitative data exist, most historians
which the historical record will be based. In many cases, conduct descriptive analyses to summarize their evidence
these events are straightforward, well-documented, and and develop generalizations. Some historians develop and
rarely disputed among historians. These cases typically do test econometric models (Fogel and Elton 1983). Their in-
not involve value judgments or contradictory evidence and terpretations come from evaluating parameter estimates.
are logically acceptable. These conditions apply to pure de- This approach may be most familiar to marketing re-
scriptions of many events. Fortunately, in marketing re- searchers who analyze historical data.
search, these types of descriptions often provide the basis for Even though the historical method seeks to minimize bi-
generating and testing our theories, models, and frameworks. ases, the potential for biased analysis still exists. Previous
Stage 4: Analyze and Interpret the Evidence studies have found that researchers may leap to conclusions
based on limited data (Kahneman and Tversky 1973), be
Only evidence that has passed the stringent tests in Stages overly influenced by vividness (Nisbett and Ross 1980), or
2 and 3 is qualified for further analysis and interpretation weight more elite respondents disproportionately (Miles and
(Baron 1986). At this point, many historians organize their Huberman 1984). Several safeguards may help limit these
data chronologically within each subject area (e.g., person, occurrences. First, the many steps of the historical method
institution, country). This arrangement has two advantages. are designed to derive conclusions from the data rather than
Contradictory evidence can be viewed in the appropriate from preconceived notions. Second, the public nature of the
context, and potential explanations are more easily recog- data makes it easier to falsify misguided interpretations.
nized, because causes precede effects. Proper synthesis of
Third, some techniques for generating theory from case
evidence usually requires multiple readings of documents.
studies may help ensure thorough analysis of all data
Historians may use inferences at this stage to complete in-
(Eisenhardt 1989). These techniques call for selecting cate-
formation that is not specifically provided (Shafer 1974).
gories or dimensions to look for within-group similarities
However, these inferences are usually restricted to situations
and between-group differences. Dimensions may be sug-
in which they are much more precise than educated guesses.
gested from the current research or existing literature. Two-
An example of the type of inference used is when a historian
has information that Caesar was in Rome on a certain date or three-dimensional matrices may prove useful for insight-
and in Gaul on a later date. Even though no specific evi- ful classification of events. The technique of classifying ex-
dence of a trip between cities exists, the historian feels con- amples on the basis of selected dimensions may be particu-
fident in inferring that a trip did take place. Inferences be- larly useful for developing generalizations and new theory
yond what the data specify should be clearly delineated. from historical data.
After organizing their data, historians analyze and inter- Stage 5: Present the Evidence and Conclusions
pret them. This process is straightforward with simple de-
scriptions of common events. Often, there is little dispute When historians present their research, they must use
among historians in these cases (Shafer 1974). However, their judgment to make many decisions. A thesis developed
historians may disagree about the objectivity of people who from analysis and interpretation of evidence often guides
created the historical record and the completeness of their these decisions. They must select evidence to present it,
reports. arrange it, and decide how it should be emphasized or min-
The process of interpretation becomes complicated when imized. Although it is rarely possible to present all the evi-
historians try to provide a higher level of meaning for a col- dence that has been collected, historians include important
lection of events. Because most historians seek to under- evidence in relating their results and conclusions.
stand the meaning of events and not just report facts, multi- The form of presentation depends on the data that were
ple potential causes should be considered (Hughes 1964). collected to address the research objectives. Different types
These explanations are often the source of disagreement of evidence may be presented as chronologies, descriptive
among historians (Gottschalk 1969). Yet these disagree- statistics, or model parameters. Most commonly, historians
ments tend to focus on the relative degree of importance present some or all of their evidence, analysis, and conclu-
sions in narrative form. This type of presentation enables
historians to communicate a rich understanding of events,
4In evaluating the credibility of evidence, historians recognize the limi- especially when the evidence collected is primarily qualita-
tations of any testimony or evidence being completely objective (Breisach tive. It links events together in a way that is not possible
1983; Malin 1946). Some historians make use of this fact by trying to glean with simple chronologies.
information about authors and institutions. They do this by evaluating
which evidence was selected for collection and presentation. Articles on
The next section illustrates how the historical method can
this approach can be found in History and Theory (Martin 1997; Quillen be applied in marketing and whether this method can gener-
1998). ate new insights.
162 JOURNAL OF MARKETING RESEARCH, MAY 2000

ANALYSIS OF LONG-TERM MARKET Table 3


SHARE STABILITY MARKET SHARE RANK OF BRANDS: 1923 VERSUS 1983
(BASED ON ADVERTISING AGE 1983)
Stage 1: Select a Topic and Collect Evidence
The historical method is well suited to address strategic Brand 1923 Rank 1983 Rank
issues in marketing, in which relationships must be exam- Swift’s Premium bacon 1 1
ined over a long period. This section addresses a specific Kellogg’s Corn Flakes 1 3
strategic research question: How stable are the market shares Eastman Kodak cameras 1 1
of leading brands over prolonged periods? The next para- Del Monte canned fruit 1 1
Hershey’s chocolates 1 2
graphs provide a brief description of the current evidence. Crisco shortening 1 2
Early research by Bain (1956) finds evidence of strong Carnation canned milk 1 1
and stable product preference patterns due to product differ- Wrigley’s chewing gum 1 1
entiation advantages of established brands. Carpenter and Nabisco biscuits 1 1
Eveready flashlight batteries 1 1
Nakamoto (1989, p. 285) report that “leading brands outsell Gold Medal flour 1 1
their rivals for years and sometimes decades.” Kotler (1997, Life Savers mint candies 1 1
p. 352) reports that “19 out of 25 companies who were mar- Sherwin Williams paint 1 1
ket leaders in 1923 were still the market leaders in 1983, Hammermill paper 1 1
sixty years later.” On the basis of this evidence, maintaining Prince Albert pipe tobacco 1 1
Gillette razors 1 1
market leadership is surprisingly common, because more Singer sewing machines 1 1
than 75% of leaders maintained their leadership for at least Manhattan shirts 1 top 5
60 years. Coca-Cola soft drinks 1 1
More recently, the concept of stable or stationary market Campbell’s soup 1 1
Ivory soap 1 1
shares has been found to be an empirical generalization. Lipton tea 1 1
DeKimpe and Hanssens (1995) and Lal and Padmanabhan Goodyear tires 1 1
(1995) cite an impressive list of previous articles that sup- Palmolive toilet soap 1 2
port stable market shares (Bass 1974; Bass and Pilon 1980; Colgate toothpaste 1 2
Ehrenberg 1988, 1991). More important, on the basis of ex-
tensive empirical evidence, Lal and Padmanabhan (1995, p.
G104) conclude that “market shares of most brands are in about leading brands refers to a study from Advertising Age
equilibrium.” Similarly, on the basis of a database of more (1983; see Table 3). These findings appear to be the basis of
than 400 prior analyses, DeKimpe and Hanssens (1995, p. Kotler’s (1997) report.
G118) conclude that “even when looking at prolonged peri- Because the Advertising Age study serves as the basis for
ods of time, market shares tend to be in a long-run equilib- much of the conclusions about long-term leadership, it is
rium position.” critical that the principles of the historical method be ap-
At this point, the answer to the research question seems plied to this document. This analysis raises several con-
clear. Long-term leadership and stable market shares seem cerns. From the standpoint of the criteria of external criti-
to be well-documented empirical generalizations. Rather cism, there is no author listed for this article. The source of
than end the investigation here, I complete the remaining the article’s findings is the president of a small communica-
stages of the historical method. tions design and marketing research firm. His expertness
cannot be determined. Also, it is not clear why the study was
Stage 2: Critically Evaluate the Sources of the Evidence undertaken. Furthermore, the study is far removed from the
Multiple copies of original documents for all the refer- time of the 1923 data on which its results depend. On the ba-
ences mentioned previously are available. Therefore, it was sis of the criteria of internal criticism, the objectivity of this
easy to establish authorship and date of publication and ver- study is questionable. The study claims to show that long-
ify that these findings are the result of scholarly inquiry. term success is due to well-managed promotions and con-
Furthermore, these authors are recognized experts in the temporary graphic presentations. Although no evidence on
field, with a broad knowledge of marketing phenomena. The these factors is presented, it seems clear that increased use
only concern from this stage of analysis is that the authors of these factors would benefit the source’s firm. Another
are not reporting contemporaneous events. Rather, they are concern is that the Advertising Age results depend on a book
relying on data or reports written many years before their published in 1923.
own reporting. The original 1923 book reveals a startling finding about
the commonly referenced data that “19 out of 25” market
Stage 3: Critically Evaluate the Evidence leaders maintained their leadership for at least 60 years.
Here, I focus on the findings of Carpenter and Nakamoto Although this finding of long-term leadership has been
(1989) and Kotler (1997), because the original data underly- widely reported in marketing textbooks and journals and in
ing the results in DeKimpe and Hanssens’s (1995) and Lal mass-market publications, it is based on a biased sample of
and Padmanabhan’s (1995) work are not available to me. categories. The original 1923 study was not done on 25 cat-
The criteria of this stage seem reasonably well satisfied egories, but rather on 100 categories (Hotchkiss and
except for corroboration. Kotler (1997) attributes his report Franken 1923). The sample of 25 categories was selectively
about “19 out of 25 companies” to Carpenter and Nakamoto chosen to demonstrate long-term leadership. Therefore, the
(1989), even though this information does not appear in Advertising Age study is dramatically flawed, and the re-
their article. Carpenter and Nakamoto’s (1989) statement ports of long-term leadership are overstated. But what is the
Historical Method 163

proper estimate of long-term leadership? And what is the an- pends on the average rate of leadership across all categories
swer to the original research question, How stable are the rather than the results from a single one. Therefore, even mi-
market shares of leading brands over prolonged periods? nor errors in the sources on which these data are based still
leave a high degree of confidence in the final results.
New Analysis of Research Question Table 4 presents a sample of the 100 categories and the
Historians consider multiple sources of evidence to ad- leading brands in 1923 and 1997.5 These 15 categories are
dress their research questions. To illustrate this approach, grouped according to a classification from the 1923 study.
I collect and analyze two types of evidence. In this sec- The numbers in parentheses beside the 1923 leaders are the
tion, a comparison of leading brands in 1923 and 1997 re- numbers of respondents who mentioned those brands.
sults in the presentation of descriptive statistics on long- Results. This extensive database provides several ways to
term leadership. In the following section, analysis of address the research question about long-term market share
qualitative data results in the narrative history of the stability. We can assess first the frequency of maintaining
causes of success and failure for brands in two product the leading market share position. Table 5 compares mar-
categories. keting’s current knowledge (which is based on a biased sam-
Data collection and evaluation. By comparing the leading ple) with the new findings that are based on the full sample
brands in 1923 with the leading brands today in all 100 cat- of categories.
egories, the actual percentage of former leaders that have The difference in the findings is striking. Leading brands
maintained leadership can be determined. In addition, the maintain their leadership at a rate less than one-third of that
market share stability of leading brands over this period can currently believed. Table 6 presents a broader set of findings
be considered. I now briefly describe the data from the 1923 based on these data. Because there may be differences be-
study and the data collected in 1997. tween durables and nondurables (durables last for many
The 1923 study was conducted by two faculty members uses, whereas nondurables do not), Tables 7 and 8 present
from New York University. They collected data in 1920 and findings for these two classes of goods.
1921 from 512 men and 512 women at a representative sam- Summary of major findings. Among the many findings
ple of U.S. colleges. Most of the colleges were in the presented in Tables 6–8, the following seem most important:
Northeast and Mid-Atlantic states, and colleges from the
•More of the leading brands in 1923 failed than remained
South, Midwest, and West were included too. Subjects were leaders.
given a list of 100 categories and asked to write the brand or •More of the top three brands in 1923 failed than remained
manufacturer they first thought of for each category. The among the top five brands.
analysis of these data considers all brands that were men- •Market shares over this prolonged period are not stable; regres-
tioned by at least 50 people. The only exceptions are 4 cat- sions of rank-order market share versus time show a significant
egories in which the leading brand is included even though decrease in market position over time (separate regressions
it was mentioned by fewer than 50 people. This threshold were run for each starting position: Number 1, Number 2,
ensures that nearly all the brands considered were reason- Number 3, and Number 4).
ably strong and more likely to be sold beyond a regional •The long-term success or failure of brands is proportional to the
strength or weakness of their starting positions; however, even
market.
the most successful brands in 1923 (Number 1 brand, Group 1)
The data collected in 1997 are the leading brands in each have a rate of long-term leadership (42%) much lower than that
category based on market share. Many data sources were currently believed for all leading brands.
used to compile this list, including Gale’s Market Share •These 100 categories may be biased in favor of the null hy-
Reporter, Simmon’s Study of Media and Markets, trade pub- pothesis of long-term leadership and market share stability, be-
lications, and multiple sources referenced in Business cause there are more nondurables when innovation is less im-
Periodicals Index, Readers’ Guide to Periodical Literature, portant and branding is more so.
and Lexis-Nexis. Reports of market share are primarily •For durable goods, the rate of maintaining leadership is lower
based on 1996 sales. After investigating these 100 cate- than the overall average, and the rate of failure is higher than
the overall average.
gories, 5 no longer seem relevant for today’s consumers.
•The subgroup of clothing has no brands that maintained leader-
Independent public reports of market share were available ship and 67% of former leaders that failed.
for 85% of the remaining categories. In the other categories, •For nondurable goods, the rate of maintaining leadership is
market shares are estimated on the basis of (1) company re- higher than the overall average, and the rate of failure is lower.
ports and (2) visits to multiple stores in New York City and •The subgroup of food and beverages has 39% of brands that
Los Angeles. A potential limitation in comparing these data maintained leadership and only 21% of former leaders that
is that the 1923 brands are not ranked on market share but failed.
rather on mind share. However, market share and mind
share should be highly correlated, so the rank order of Why Do Some Leaders Maintain Leadership Whereas
brands on these two measures should be similar within cat- Others Do Not?
egories. More important, the 1923 data have been the refer- The previous findings describe the rate at which market
ence point for assessing market share stability in the mar- leaders maintain leadership. In addition, the historical
keting literature for many years, so it seems appropriate to method can be used to develop explanations for these out-
prepare results on the basis of this sample. comes. To address this question partially, I selected two cat-
When the criteria of external criticism and internal criti- egories for analysis: one nondurable representative of cate-
cism are applied to the 1997 data, minor concerns about au-
thenticity, credibility, and corroboration of the data sources
still exist. However, the answer to the research question de- 5The complete list of categories and brands is available from the author.
164 JOURNAL OF MARKETING RESEARCH, MAY 2000

gories in which leaders are more likely to maintain leader- major search for contemporaneous articles describing events
ship and one durable representative of categories in which in each category. These articles helped minimize biases that
they are less likely to do so. Data collection consisted of a might come from a retrospective recollection of events.

Table 4
SAMPLE OF LEADING BRANDS IN 1923 AND 1997

Product Category 1923 Leaders 1997 Leaders


Single dominant brand in 1923 (well-known categories)a
Cleanser Old Dutch (744) Comet
Soft Scrub
Ajax
Chewing gum Wrigley’s (664) Wrigley’s
Adams (97) Bubble Yum
Bubblicious
Motorcycles Indian (564) Harley-Davidson
Harley-Davidson (156) Honda
Kawasaki
Single dominant brand in 1923 (less well-known categories)b
Five-cent mint candies Life-Savers (436) Breath-Savers
Tic Tac
Certs
Peanut butter Beech-Nut (435) Jif
Heinz (140) Skippy
Peter Pan
Razors Gillette (396) Gillette
Gem (87) Bic
Ever Ready (50) Schick
Single leading brand in 1923c
Soft drinks Coca-Cola (353) Coca-Cola
Cliquot Club (85) Pepsi
Bevo (75) Dr. Pepper/Cadbury
Hires (51)
Coffee Arbuckle’s Yuban (224) Folgers
White House (100) Maxwell House
Hotel Astor (56) Hills Bros.
George Washington (55)
Laundry soap Fels Naptha (192) Tide
Octagon (93) Cheer
Kirkman (83) Wisk
Ivory (82)
Babbitts (51)
Crystal White (51)
Brands sharing leadership in 1923
Typewriters Underwood (394) Smith Corona
Remington (265) Brother
Oliver (100) Lexmark
Corona (53)
Cigarettes Camel (256) Marlboro
Fatima (156) Winston
Pall Mall (90) Newport
Murad (72)
Lucky Strike (64)
Hosiery Holeproof (180) L’eggs
Onyx (156) Hanes
Phoenix (66) No Nonsense
Luxite (60)
Leading brands in 1923 that do not have pronounced leadership
Shoes Douglas (146) Nike
Walkover (120) Reebok
Hanan (52)
Candy Huyler’s (144) Hershey
Loft (94) M&M’s/Mars
Page & Shaw (91) Nestlé
Whitman (89)
Jelly or jam Heinz (62) Smucker’s
Welch’s
Kraft
aThe leading brand was mentioned by a majority or near majority of subjects.
bThe leading brand was mentioned by a majority of subjects who responded to the question for that category.
cThe leading brand was mentioned by more than twice as many subjects as was the next brand in the category.
Historical Method 165

More than 100 articles were collected, primarily from major First, over the years, Wrigley has maintained a strong
magazines and newspapers.6 These articles were written for commitment to its brands. In 1927, it was an early sponsor
the public record at a time when many potential readers of national radio. By some reports, Wrigley was the largest
were fully aware of the facts being presented. Furthermore, single-product national advertiser in the 1930s. Perhaps
major events were corroborated with independent, similarly Wrigley’s greatest commitment to its brands was shown in
credible reports. Explanations developed from this evidence World War II. Scarce supplies of ingredients meant that
are presented in narrative form, because this approach can Wrigley could not produce its normal quality gum. So it
link events in a way that is not possible with a simple withdrew its traditional brands and introduced another
chronology. brand, which they advertised as not up to Wrigley standards.
Category in which leader has maintained leadership: At the same time, Wrigley continued to advertise its tradi-
chewing gum. Wrigley has dominated chewing gum for the tional brands with the slogan “Remember this wrapper.”
past nine decades. By 1923, Wrigley had sold the leading After the war, sales soon surpassed their previous level. As
brand for more than ten years and was one of the strongest far back as 1964, Wrigley had well-formulated brand per-
brands in the 1923 study of market leaders. Wrigley’s early
success is primarily attributed to building strong brands
through extensive advertising. Its main competitor, Table 5
American Chicle, was formed by combining several gum LONG-TERM SUCCESS RATE OF 1923 MARKET LEADERS
companies into the chewing gum trust. This company
tended to maximize profits by minimizing its marketing ex- Current Knowledge New Findings
Later Market (based on biased (based on complete
penses, whereas Wrigley invested in building its brands. Share Rank 1983 sample) 1997 sample)
Since 1923, Wrigley’s success is primarily based on three
factors: maintaining and building strong brands, focusing on Number 1 76% 23%
Number 2 16% 8%
a single product, and being in a category that has not Number 3 4% 9%
changed much. Top 5 4% 8%
Top 10 0% 7%
Below 10 0% 16%
Failed 0% 28%
6These references are available from the author.

Table 6
COMPARISON OF STARTING AND ENDING MARKET SHARE POSITIONS

Ending (1997)

Starting (1923) Sample Size Number 1 Number 2 Number 3 Top 5 Top 10 > 10 Failed
Number 1 brand 97a 23% 8% 9% 8% 7% 16% 28%
Group 1 19 42% 5% 11% 5% 5% 5% 26%
Group 2 11 18% 18% 18% 9% 9% 0% 27%
Group 3 25 24% 8% 12% 12% 8% 20% 16%
Group 4 15 13% 7% 7% 20% 13% 20% 20%
Group 5 27 15% 7% 4% 0% 4% 26% 44%
Number 2 brand 70 11% 9% 3% 4% 9% 26% 39%
Number 3 brand 43 5% 7% 2% 5% 9% 14% 58%
Number 4 brand 26 4% 4% 4% 4% 8% 42% 35%
Number 5 brand 12 0% 0% 25% 0% 17% 42% 17%
Number 6 brand 5 0% 0% 0% 0% 20% 20% 60%
Number 7 brand 1 0% 0% 0% 0% 0% 0% 100%
aThe authors of the 1923 study determined that three categories had names mentioned that were not actual brands. The remaining 97 categories include 45
durables, 51 nondurables, and one service.

Table 7
COMPARISON OF STARTING AND ENDING MARKET SHARE POSITIONS FOR DURABLE GOODS

Ending (1997)

Starting (1923) Sample Size Number 1 Number 2 Number 3 Top 5 Top 10 > 10 Failed
Number 1 brand 45 16% 4% 13% 7% 2% 16% 42%
Number 2 brand 28 11% 7% 0% 0% 11% 25% 46%
Number 3 brand 17 6% 6% 0% 6% 6% 6% 71%
Number 4 brand 8 13% 0% 13% 13% 13% 38% 13%
Number 5 brand 4 0% 0% 50% 0% 25% 0% 25%
Number 6 brand 2 0% 0% 0% 0% 0% 0% 100%
Number 7 brand 1 0% 0% 0% 0% 0% 0% 100%
166 JOURNAL OF MARKETING RESEARCH, MAY 2000

Table 8
COMPARISON OF STARTING AND ENDING MARKET SHARE POSITIONS FOR NONDURABLE GOODS

Ending (1997)

Starting (1923) Sample Size Number 1 Number 2 Number 3 Top 5 Top 10 > 10 Failed
Number 1 brand 51 29% 10% 6% 10% 12% 18% 16%
Number 2 brand 41 10% 10% 5% 7% 7% 27% 34%
Number 3 brand 25 4% 8% 4% 0% 12% 20% 52%
Number 4 brand 17 0% 6% 0% 0% 6% 47% 41%
Number 5 brand 7 0% 0% 14% 0% 0% 71% 14%
Number 6 brand 2 0% 0% 0% 0% 0% 50% 50%
Number 7 brand 0 — — — — — — —

sonalities for its major products. More recently, Wrigley has as early as 1930. Their combined 90% market share and co-
continued to manage its brand equity carefully. When sug- ordinated pricing enhanced short-term profitability. One
arless gum and bubble gum became popular in the 1970s, year after the indictment, Underwood and the other three
Wrigley did not automatically extend its brand. Several manufacturers agreed to a consent decree that prohibited
years after the sugarless gum market had taken off, Wrigley their monopolistic practices.
introduced sugarless gum, but only under the same brand During World War II, many companies developed tech-
they had used for the subpar gum during World War II. A nology that would be useful after the war; however,
stronger commitment to sugarless gum did not emerge until Underwood did not. Its profitability from 1945 to 1955 was
Extra in 1984. In bubble gum, Wrigley used a subsidiary to due to pent-up, postwar demand for typewriters; a growing
introduce Hubba Bubba and still limits the association of economy; and sales from the Korean War effort. During this
Wrigley with bubble gum, because Wrigley targets a more period, Underwood continued to pay high dividends rather
adult market. In 1998, Wrigley explored chewing gum with than invest more in product development and manufactur-
caffeine, but only through one of its subsidiaries rather than ing. Whereas other companies were investing heavily in
with the valuable Wrigley brand. computer technology, Underwood only acquired tiny
The second factor in Wrigley’s long-term dominance is its Electronic Computer Corporation of Brooklyn in 1952.
unrelenting focus on chewing gum. Even though the Finally, in 1956, several factors exposed Underwood’s
Wrigley family owned the Chicago Cubs baseball team for weaknesses. First, lower-priced foreign competitors’ share
many years, this was considered a sideline to its real busi- of manual typewriters began a five-year increase from 15%
ness. For more than 100 years, Wrigley has been run by to 40%. Second, outdated manufacturing facilities became
three generations of Wrigleys. Collectively, the family still too costly. Third, electric typewriters became more compet-
owns more than 40% of the company. In the 1960s, the firm itive, eventually surpassing sales of manual typewriters in
chose not to expand into other businesses and considered di- the early 1960s. These conditions led to increasingly large
versification a dirty word. The firm rejected joining a con- losses and failed efforts at new strategic directions.
glomerate, American Home Products, even though Warner- Underwood spent $12 million to diversify into computers
Lambert and Squibb had acquired their major competitors. but gave up after 18 months, lacking the necessary techni-
As a single-product company, Wrigley has continued to cal expertise. In 1957, the firm sought mergers with IT&T
grow and prosper, both domestically and internationally. and National Cash Register but was already too weak to be
Over the past ten years, dollar sales have grown nearly 10% attractive to these firms. After a few more unprofitable
per year in what is considered a mature category. years, Underwood was acquired by Olivetti. Although
The third factor contributing to Wrigley’s long-term domi- Olivetti’s U.S. subsidiary returned to profitability from the
nance is that chewing gum has changed very little over the mid-1960s until 1970, it lost money throughout the 1970s
years. In some ways, the market has simplified. In the 1920s, and early 1980s. The Underwood name was phased out, and
at least 25 flavors of chewing gum were available. Today, only today, Olivetti is only a small part of a relatively unimpor-
mint-flavored gums dominate sales. These minimal changes tant category.
have made later entry difficult. In fact, Beech-Nut (now part Underwood’s lack of innovation stands in stark contrast
of Nabisco) was the last major chewing gum company to en- with IBM. Although electric typewriters were well known in
ter, way back in 1911. Although chewing gum has changed the 1920s, there was not a meaningful effort among the ma-
relatively little, Wrigley maintains a serious research and de- jor typewriter manufacturers to introduce them. In contrast,
velopment effort to improve its products and packaging. IBM bought Electromatic Typewriter in 1933 and intro-
Category in which leader has not maintained leadership: duced the first successful electric typewriter in 1935. In
typewriters. Underwood was the leading typewriter brand in 1941, IBM introduced proportional letter spacing. In 1961,
1923. Over the next three decades, Underwood was prof- the firm introduced the Selectric typewriter with no moving
itable; however, its success was based on a harvesting strat- carriage and the “golf-ball” typing element. In 1964, it in-
egy. During this period, Underwood’s core mistake was its troduced an automatic typewriter that stored information on
unwillingness to innovate. Rather than try to outdo its com- magnetic tape. These advances and others enabled IBM to
petitors, Underwood sought to collude with them. dominate electric typewriters through the late 1970s.
According to a 1939 federal antitrust indictment, the four Afterwards, advances in electronics enabled removal of
largest typewriter manufacturers met to coordinate activities many mechanical components. This change lowered barriers
Historical Method 167

to entry, made the category less profitable, and contributed diffusion models have included a measure of advertising,
to IBM’s eventual withdrawal. none has included a measure of unpaid media. One archival
More important than succeeding in typewriters, IBM measure of unpaid media is the number of articles on a new
leveraged its strong position to become the dominant firm in product in the Readers’ Guide to Periodical Literature, an
computers. During World War II, IBM developed the Mark index of more than 100 periodicals. The results of estimat-
I computing calculator. The firm sold its first computer to ing a diffusion model with this measure of unpaid media
the government research facility at Los Alamos, N. Mex., in and advertising are presented in Table 9. When both mass
1953 and soon afterwards captured market leadership. media variables are included, the unpaid media variable is
IBM’s leadership continues today in important segments of significant in one of three categories, whereas advertising is
the computer industry. Similarly, Remington-Rand moved not significant in any. When the variables are considered
successfully from typewriters to computers by acquiring the separately, the unpaid media variable is significant in two
Eckert-Mauchly Computer Company in 1950. Eckert and categories, whereas advertising is significant in one. These
Mauchly were responsible for developing one of the world’s results suggest that unpaid media may be at least as impor-
first computers during World War II. This acquisition en- tant as advertising in measuring mass media influence.
abled Remington-Rand to sell the first commercial com- Although a complete historical analysis of the role of unpaid
puter, a Univac, to the Census Bureau in 1951. Today, the media in new product adoption is beyond the scope of this
corporate descendants of Remington-Rand continue to sell article, this example illustrates the ability of archival
computers and computer services as Unisys Corporation. sources to provide more complete measures of constructs in
Underwood’s demise demonstrates the importance of quantitative models.
continuous innovation. At least until the mid-1930s, A third strength of historical method is that it provides an
Underwood was well positioned to be the dominant firm in approach for dealing with complex phenomena that cannot
office automation. Its revenues were approximately equal to be studied adequately in a laboratory or with readily avail-
IBM’s in 1937, but by 1957, IBM’s revenues were about 15 able data. Although case study research methods have been
times those of Underwood. Even though Underwood should used to address similar phenomena, the four stages of case
not be expected to have duplicated the success of IBM, his- research (drift, design, prediction, and disconfirmation;
tory suggests it did not even really try. Bonoma 1985) seem imprecise in comparison with the his-
DISCUSSION torical method.
The strengths of the historical method must be viewed
Laboring the difference between science and the hu-
manities has long been a fashion, and has become a against its limitations, some of which apply to other social
bore. The method of problem solving, the method of science methods. First, because the historical method evalu-
conjecture and refutation, is practiced by both. It is ates phenomena in their complex real-world environment,
practiced in reconstructing a damaged text as in con- developing and confirming theory precisely can be difficult.
structing a theory of radioactivity. (Popper 1972, p. 185) However, even when a single study does not generate the-
ory, it can contribute to new theory. For example,
As the previous analysis of market share stability illus-
Chandler’s (1962) historical study of organizational struc-
trates, the historical method is capable of producing new
ture and economic efficiency contributed to the develop-
knowledge about marketing phenomena. However, the
ment of transaction cost analysis theory (Williamson 1981).
method does have weaknesses as well as strengths relative
to more conventional methods. After discussing these A second limitation of historical method is that it is more
strengths and weaknesses, the concluding paragraphs sum- capable of descriptions than of explanations or predictions.
marize the key points of this research. However, historians attempt to determine causal factors and
As noted previously, historical method has several evaluate their relative importance when working with qual-
strengths that make it appropriate for studying marketing itative data. With quantitative data, their approach to devel-
phenomena. First, the method is helpful in collecting and oping explanations and predictions is similar to that of mar-
evaluating data for longitudinal analysis. This type of analy- keting researchers.
sis is useful for understanding strategic issues and could A third limitation of historical method is that it is subject
help restore marketing’s influence in the strategy literature to interpretive biases in the selection and evaluation of data
(Day 1992). One alternative for collecting data on strategic (Calder and Tybout 1987). Some interpretive researchers in
issues is survey research. Although survey research is useful marketing assert that it is meaningless or impossible to dis-
for collecting current information, it can produce biased in- tinguish causes from effects (Hirschman 1986; Hudson and
formation when past events are recalled. These biases are re- Ozanne 1988). In contrast, most historians make concerted
duced when contemporaneous accounts are used, as is called efforts to determine the immediate and underlying causes of
for in the historical method. Furthermore, the emphasis of events. Furthermore, competing explanations of historical
the historical method on authenticating, establishing the events may be more easily evaluated, because the original
credibility of, and corroborating evidence remains a key data are more commonly available than in many marketing
strength relative to data collection in much survey research. studies.
A second strength of the method is that it can be used to A final limitation of the historical method, which prevents
corroborate findings from existing data (e.g., PIMS) or ob- its use by most researchers, is that it is very time consuming.
tain new or more precise measures of constructs in quanti- Instead of using established databases, researchers must
tative models. For example, potential sources of mass media construct their own databases or at least verify existing data-
influence in diffusion models are advertising and unpaid bases. Unlike surveys and experiments, researchers do not
media reports (Horsky and Simon 1983). Although some construct the scenario under which their data are recorded.
168 JOURNAL OF MARKETING RESEARCH, MAY 2000

Table 9
PARAMETER ESTIMATES FROM DIFFUSION MODEL

Room Air Conditioner Color Television Clothes Dryer


Model with Model with Model with
Unpaid Model with Unpaid Model with Unpaid Model with
Variable Media and Unpaid Model with Media and Unpaid Model with Media and Unpaid Model with
(parameter) Advertising Media Advertising Advertising Media Advertising Advertising Media Advertising
Unpaid .00022** .00025*** NS .00011** NS NS
media (b) (.00009) (.00006) (.00005)
Advertising (c) NS .00062** NS NS NS NS
(.00030)
External
influence (a) NS NS .00649*** NS .00255** .00512*** .01063** .01360*** .01063***
(.00195) (.00113) (.00127) (.00438) (.00365) (.00264)
Internal
influence (q) .4034*** .4196*** .3567*** .6704*** .6704*** .6354*** .3007*** .3267*** .3006***
(.0459) (.0380) (.512) (.1533) (.0403) (.1442) (.0522) (.0386) (.0480)
Market
potential (m) 16.7*** 16.6*** 18.1*** 38.2*** 38.2*** 40.2*** 17.2*** 16.5*** 17.2***
(1.4) (1.3) (1.7) (8.3) (1.8) (8.6) (1.7) (1.3) (1.6)
*p < .10.
** p < .05.
*** p < .01.
NS = not significant.
Notes: Sales and advertising figures are from Bass, Krishnan and Jain’s (1994) study. The following model is estimated with nonlinear least squares (for
details, see Srinivasan and Mason 1986):
Sales(t) = m[F(t) – F(t – 1)] + u,
where

(1 − e − (p + q) t )
F(t) = ,
q
(1 + p e −( p+ q )t )

and
p = coefficient of external influence (mass media or innovation),
q = coefficient of internal influence (word of mouth or imitation),
m = market potential, and
u = error term.
The coefficient of external influence is reparameterized as
a + b × Unpaid Mediat + c × Advertisingt.
Parameters for color television were also estimated with a second source of unpaid media, the Business Periodicals Index. Color television is the only cat-
egory in which this reference was published during the years of interest. In this estimation, unpaid media considered alone was statistically significant, whereas
advertising, considered alone or with unpaid media, was not. These results provide some corroboration of the findings in Table 9.

They must depend on records of events that were made in- bility and the conditions in which it is more and less com-
dependently of their efforts to study them. mon. Additional areas for further research with the histori-
In conclusion, this article contributes to the marketing lit- cal method are presented in Table 10. I hope this table will
erature in several unique ways. First, it is the only article to encourage marketing researchers to consider historical
discuss the objectives of historical research as viewed by analyses in their research.
historians. Second, it presents a complete description of the Historians believe that a reasonable outcome of histori-
historical method that can be used as a reference by market- cally generated knowledge is being able to manage better in
ing researchers who apply this method. Third, it demon- similar situations (Tuchman 1981). Although historical
strates the usefulness of this method by presenting new evi- method is not suited for all research questions, it is capable
dence on long-term market share stability. The findings
of producing empirical generalizations in areas in which
from this analysis show that the rate of long-term leadership
is much lower than currently believed. In addition, these other methods cannot. It is particularly well suited to ad-
findings raise doubts about the currently accepted empirical dress longitudinal issues in which recollections of past
generalization that market shares are stable over prolonged events may be biased or wrong. Also, it is useful for uncov-
periods. Although market shares may be stable over shorter ering data that may not be available from more conventional
periods, they are not stable in these categories over the pe- data sources. At a minimum, as a substantially different
riod considered. This finding opens up several avenues for method, it can enhance the value of findings based on all
further research into the real duration of market share sta- other research methods.
Historical Method 169

Table 10
POTENTIAL APPLICATIONS OF HISTORICAL METHOD IN MARKETING RESEARCH

Previous studies that described limitations or proposed research on the application of the historical method
Market entry Data on survivors and failures necessary to establish how a firm’s competencies and resources Kerin, Varadarajan, and
are converted into sustainable advantages. Peterson (1992)
Longitudinal data and “thick” empirical studies needed to determine whether corporate-level Gruca and Sudharshan
strategies and resources enhance entry deterrence. (1995)
Examine the relationship between order of entry and short-term and long-term survival rates. Kalyanaram, Robinson,
and Urban (1995)

New products Study of brand extensions could be strengthened by including failures. Smith and Park (1992)
Insights from new product development process are limited by using cross-sectional data and Olson, Walker, and
could be enhanced with longitudinal data. Ruekert (1995)
Need a greater understanding of how organizations develop breakthrough products or Wind and Mahajan
discontinuous innovations. (1997)
A broader set of variables is needed to explain variance in product development cycle time. Griffin (1997)
Evaluating the determinants of new product success in a longitudinal framework would Song and Parry (1997b)
provide additional insight.
Evaluate cross-functional integration and internal commitment at various points during the Song and Parry (1997a)
new product development cycle.

Business Further research on organizational relationships should include longitudinal methods to Weiss and Anderson
relationships provide additional insights. (1992)
Develop an understanding of the forces leading to long-term customer relationships, strategic Webster (1992)
partnerships with vendors, alliances for the codevelopment of technologies, and the issues
involved in creating, managing, and dissolving these partnerships over time.
Limits of cross-sectional design could be overcome with longitudinal study of organizational Weiss and Heide (1993)
search process.
Longitudinal analysis could examine ethics across different types of exchange relationships. Gundlach and Murphy
(1993)
History of relationships could provide insight into the governance of marketing channels. Heide (1994)
Detailed case studies over substantial time periods could enhance understanding of dyadic Anderson, Hakansson,
business relationships. and Johanson (1994)
Limitation of cross-sectional data to study an undoubtedly dynamic process of relationship Fein and Anderson
development could be overcome with historical data. (1997)

Other Insights into decisions not to extend a brand name or take other marketing actions. Smith and Park (1992)
Future studies of standardization versus adaptation in international marketing should adopt a Szymanski, Bharadwaj,
longitudinal approach. and Varadarajan (1993)
The largest payoff for understanding the capabilities of market-driven organizations will come Day (1994)
from thoughtful clinical investigations of best practices and failed transformation efforts. Germain, Droge, and
Greater confidence in model of just-in-time selling could be achieved with longitudinal data. Daugherty (1994)
Competition cannot be understood without a longitudinal view of competitors and their Prahalad (1995)
competencies.

Additional areas of research in which the historical method could be applied


Market entry Survival durations of brands under various conditions.
and exit Order of entry–market share relationship after controlling for survival bias across all entrants.
Appropriate marketing-mix investment at time of entry.

New products Role of patents in successful new product introductions.


Operationalizing innovation in diffusion models by measuring the number of media reports on
a new category.
Role of firm size in new product development success.
Role of incumbency in new product development success.
Process by which consumers learn about, adopt, and use new products.
Elements of continuity versus discontinuity in new product development from the firm’s
perspective.
The determinants of sales takeoff for new products.
Incorporation of product features in sales growth models.
Causes of new product failure.
Organizational decision process to withdraw products from the market.

Product life Profitability or stock market returns over stages of PLC.


cycle (PLC) Pricing strategies over the PLC, for example, penetration and skimming.
Entry and failure rates over the PLC.

Other Survival duration of business relationships, alliances, and so forth.


Conditions under which preannouncements are more and less common.
How industry structures evolve over time.
Impact of new media on marketing; lessons from advent of radio and television for Internet.
170 JOURNAL OF MARKETING RESEARCH, MAY 2000

REFERENCES ——— and Robin Wensley (1983), “Marketing Theory with a


Strategic Orientation,” Journal of Marketing, 47 (Fall), 79–89.
Aaker, David A. and George S. Day (1986), “The Perils of High-
DeKimpe, Marnik G. and Dominique M. Hanssens (1995),
Growth Markets,” Strategic Management Journal, 7
“Empirical Generalizations About Market Evolution and
(September/October), 409–21.
Stationarity,” Marketing Science, 14 (3, Part 2 of 2),
Advertising Age (1983), “Study: Majority of 25 Leaders in 1923
G109–G121.
Still on Top—‘Old Standbys’ Hold Their Own,” (September Ehrenberg, Andrew S.C. (1988), Repeat Buying: Facts, Theory and
19), 32. Data, 2d ed. New York: Oxford University Press.
Agrawal, Jagdish and Wagner A. Kamakura (1995), “The ——— (1991), “New Brands and the Existing Market,” Journal of
Economic Worth of Celebrity Endorsers: An Event Study the Market Research Society, 33 (4), 285–99.
Analysis,” Journal of Marketing, 59 (July), 56–62. Eisenhardt, Kathleen M. (1989), “Building Theories from Case
Anderson, James C., Hakan Hakansson, and Jan Johanson (1994), Study Research,” Academy of Management Review, 14 (4),
“Dyadic Business Relationships Within a Business Network 532–50.
Context,” Journal of Marketing, 58 (October), 1–15. Elton, G.R. (1967), The Practice of History. New York: Thomas Y.
Armstrong, J. Scott and Fred Collopy (1996), “Competitor Crowell.
Orientation: Effects of Objectives and Information on Fein, Adam J. and Erin Anderson (1997), “Patterns of Credible
Managerial Decisions and Profitability,” Journal of Marketing Commitments: Territory and Brand Selectivity in Industrial
Research, 33 (May), 188–99. Distribution Channels,” Journal of Marketing, 61 (April),
Bain, Joe S. (1956), Barriers to New Competition. Cambridge, 19–34.
MA: Harvard University Press. Fogel, Robert William and G.R. Elton (1983), Which Road to the
Baron, Salo W. (1986), The Contemporary Relevance of History. Past? Two Views of History. New Haven, CT: Yale University
New York: Columbia University Press. Press.
Bass, Frank M. (1974), “The Theory of Stochastic Preference and Fullerton, Ronald A. (1987), “The Poverty of Ahistorical Analysis:
Brand Switching,” Journal of Marketing Research, 11 Present Weaknesses and Future Cure in U.S. Marketing
(February), 1–20. Thought,” in Philosophical and Radical Thought in Marketing,
———, Trichy V. Krishnan, and Dipak C. Jain (1994), “Why the A. Fuat Firat, Nikhilesh Dholakia, and Richard P. Bagozzi, eds.
Bass Model Fits Without Decision Variables,” Marketing Lexington, MA: Lexington Books, 97–116.
Science, 13 (Summer), 203–23. Germain, Richard, Cornelia Droge, and Patricia J. Daugherty
——— and T.L. Pilon (1980), “A Stochastic Brand Choice (1994), “The Effect of Just-in-Time Selling on Organizational
Framework for Econometric Modeling of Time Series Market Structure: An Empirical Investigation,” Journal of Marketing
Share Behavior,” Journal of Marketing Research, 17 Research, 31 (November), 471–83.
(November), 486–97. Ghemawat, Pankaj (1991), “Market Incumbency and
Bayus, Barry L. (1992), “The Dynamic Pricing of Next Generation Technological Inertia,” Marketing Science, 10 (2), 161–71.
Consumer Durables,” Marketing Science, 11 (3), 251–65. Golder, Peter N. and Gerard J. Tellis (1993), “Pioneer Advantage:
———, Sanjay Jain, and Ambar G. Rao (1997), “Too Little, Too Marketing Logic or Marketing Legend?” Journal of Marketing
Early: Introduction Timing and New Product Performance in the Research, 30 (May), 158–70.
Personal Digital Assistant Industry,” Journal of Marketing ——— and ——— (1997), “Will It Ever Fly? Modeling the
Research, 34 (February), 50–63. Takeoff of Really New Consumer Durables,” Marketing
Bonoma, Thomas V. (1985), “Case Research in Marketing: Science, 16 (3), 256–270.
Opportunities, Problems, and a Process,” Journal of Marketing Goodman, Robert S. and Evonne Jonas Kruger (1988), “Data
Research, 22 (May), 199–208. Dredging or Legitimate Research Method? Historiography and
Breisach, Ernst (1983), Historiography. Chicago: University of Its Potential for Management Research,” Academy of
Chicago Press. Management Review, 13 (2), 315–25.
Brooks, Philip C. (1969), Research in Archives. Chicago: Gottschalk, Louis (1969), Understanding History: A Primer of
University of Chicago Press. Historical Method. New York: Alfred A. Knopf.
Calantone, Roger J., Jeffrey B. Schmidt, and X. Michael Song Green, Donna H., Donald W. Barclay, and Adrian B. Ryans (1995),
(1996), “Controllable Factors of New Product Success: A Cross- “Entry Strategy and Long-Term Performance: Conceptualiza-
National Comparison,” Marketing Science, 15 (4), 341–58. tion and Empirical Examination,” Journal of Marketing, 59
Calder, Bobby J. and Alice M. Tybout (1987), “What Consumer (October), 1–16.
Research Is ...,” Journal of Consumer Research, 14 (June), Griffin, Abbie (1997), “The Effect of Project and Process
136–40. Characteristics on Product Development Cycle Time,” Journal
Carpenter, Gregory S. and Kent Nakamoto (1989), “Consumer of Marketing Research, 34 (February), 24–35.
Preference Formation and Pioneering Advantage,” Journal of Gruca, Thomas S. and D. Sudharshan (1995), “A Framework for
Marketing Research, 26 (August), 285–98. Entry Deterrence Strategy: The Competitive Environment,
Chandler, Alfred D., Jr. (1962), Strategy and Structure: Chapters in Choices, and Consequences,” Journal of Marketing, 59 (July),
the History of the American Industrial Enterprise. Cambridge, 44–55.
MA: MIT Press. Gundlach, Gregory T. and Patrick E. Murphy (1993), “Ethical and
Cochrane, Charles Norris (1929), Thucydides and the Science of Legal Foundations of Relational Marketing Exchanges,”
History. London: Oxford University Press. Journal of Marketing, 57 (October), 35–46.
Cusumano, Michael A., Yiorgos Mylonadis, and Richard S. Handlin, Oscar (1979), Truth in History. Cambridge, MA: Belknap
Rosenbloom (1992), “Strategic Maneuvering and Mass-Market Press.
Dynamics: The Triumph of VHS over Beta,” Business History Heide, Jan B. (1994), “Interorganizational Governance in
Review, 66 (Spring), 51–94. Marketing Channels,” Journal of Marketing, 58 (January),
Day, George S. (1992), “Marketing’s Contribution to the Strategy 71–85.
Dialogue,” Journal of the Academy of Marketing Science, 47 Hexter, J.H. (1971), The History Primer. New York: Basic Books.
(Fall), 43–55. Hirschman, Elizabeth C. (1986), “Humanistic Inquiry in
——— (1994), “The Capabilities of Market-Driven Marketing Research: Philosophy, Method, and Criteria,”
Organizations,” Journal of Marketing, 58 (October), 37–52. Journal of Marketing Research, 23 (August), 237–49.
Historical Method 171

Horsky, Dan and Leonard S. Simon (1983), “Advertising and the Nisbett, R. and L. Ross (1980), Human Inference: Strategies and
Diffusion of New Products,” Marketing Science, 2 (Winter), Shortcomings of Social Judgment. Englewood Cliffs, NJ:
1–17. Prentice Hall.
Hotchkiss, George Burton and Richard B. Franken (1923), The Olmstead, Alan L. and Paul W. Rhode (1995), “Beyond the
Leadership of Advertised Brands. Garden City, NY: Doubleday, Threshold: An Analysis of the Characteristics and Behavior of
Page & Company. Early Reaper Adopters,” Journal of Economic History, 55
Hudson, Laurel Anderson and Julie L. Ozanne (1988), “Alternative (March), 27–57.
Ways of Seeking Knowledge in Consumer Research,” Journal Olson, Eric M., Orville C. Walker Jr., and Robert W. Ruekert
of Consumer Research, 14 (March), 508–21. (1995), “Organizing for Effective New Product Development:
Hughes, H. Stuart (1964), History as Art and Science. New York: The Moderating Role of Product Innovativeness,” Journal of
Harper and Row Publishers. Marketing, 59 (January), 48–62.
Iggers, Georg G. (1979), “Federal Republic of Germany,” in Parker, Philip M. (1995), “‘Sweet Lemons’: Illusory Quality, Self-
International Handbook of Historical Studies: Contemporary Deceivers, Advertising, and Price,” Journal of Marketing
Research and Theory, Georg G. Iggers and Harold T. Parker, Research, 32 (August), 291–307.
eds. Westport, CT: Greenwood Press, 217–32. Pollay, Richard W. (1985), “The Subsidizing Sizzle: A Descriptive
Johnson, Allen (1934), The Historian and Historical Evidence. History of Print Advertising, 1900–1980,” Journal of Marketing,
New York: Charles Scribner’s Sons. 49 (Summer), 24–37.
Jones, D.G. Brian and David D. Monieson (1990), “Early Popper, Karl R. (1968), “Remarks on the Problems of Demarcation
Development of the Philosophy of Marketing Thought,” Journal and of Rationality,” in Problems in the Philosophy of Science,
of Marketing, 54 (January), 102–13. Imre Lakatos and Alan Musgrave, eds. Amsterdam: North-
Kahneman, Daniel and Amos Tversky (1973), “On the Psychology Holland Publishing Co., 88–102.
of Prediction,” Psychological Review, 80 (2), 237–51. ——— (1972), Objective Knowledge: An Evolutionary Approach.
Kalyanaram, Gurumurthy, William T. Robinson, and Glen L. Oxford: The Clarendon Press.
Urban (1995), “Order of Market Entry: Established Empirical Prahalad, C.K. (1995), “Weak Signals Versus Strong Paradigms,”
Generalizations, Emerging Empirical Generalizations, and Journal of Marketing Research, 32 (August), iii–viii.
Future Research,” Marketing Science, 14 (3), G212–G221. Quillen, Carol E. (1998), “Crossing the Line: Limits and Desire in
Keep, William W., Stanley C. Hollander, and Roger Dickinson Historical Interpretation,” History and Theory, 37 (February),
(1998), “Forces Impinging on Long-Term Business-to-Business 40–68.
Relationships in the United States: An Historical Perspective,” Reddy, Srinivas K., Susan Holak, and Subodh Bhat (1994), “To
Journal of Marketing, 62 (April), 31–45. Extend or Not to Extend: Success Determinants of Line
Kerin, Roger A., P. Rajan Varadarajan, and Robert A. Peterson Extensions,” Journal of Marketing Research, 31 (May), 243–62.
(1992), “First-Mover Advantage: A Synthesis, Conceptual Salvemini, Gaetano (1939), Historian and Scientist. Cambridge,
Framework, and Research Propositions,” Journal of Marketing, MA: Harvard University Press.
56 (October), 33–52. Savitt, Ronald (1980), “Historical Research in Marketing,” Journal
Kotler, Philip (1997), Marketing Management: Analysis, Planning, of Marketing, 44 (Fall), 52–58.
Implementation, and Control, 9th ed. Upper Saddle River, NJ:
Shafer, Robert Jones (1974), A Guide to Historical Method.
Prentice Hall.
Homewood, IL: The Dorsey Press.
Lal, Rajiv and V. Padmanabhan (1995), “Competitive Response
Sherry, John F. (1991), “Postmodern Alternatives: The Interpretive
and Equilibria,” Marketing Science, 14 (3, Part 2 of 2),
Turn in Consumer Research,” in Handbook of Consumer
G101–G108.
Behavior, Thomas S. Robertson and Harold H. Kassarjian, eds.
Langlois, C.V. and C. Seignobos (1898), Introduction to the Study
Englewood Cliffs, NJ: Prentice Hall, 548–91.
of History. New York: Henry Holt and Co.
Lieberman, Marvin B. and David B. Montgomery (1988), “First- Simon, Herbert A. (1968), “On Judging the Plausibility of
Mover Advantages,” Strategic Management Journal, 9 Theories,” in Logic, Methodology and Philosophy of Science III,
(Summer), 41–58. B. Van Rootselaar and J.F. Staal, eds. Amsterdam: North-
Low, George S. and Ronald A. Fullerton (1994), “Brands, Brand Holland Publishing Company, 439–59.
Management, and the Brand Manager System: A Critical- Smith, Daniel C. and C. Whan Park (1992), “The Effects of Brand
Historical Evaluation,” Journal of Marketing Research, 31 Extensions on Market Share and Advertising Efficiency,”
(May), 173–90. Journal of Marketing Research, 29 (August), 296–313.
Malin, James C. (1946), Essays on Historiography. Ann Arbor, MI: Smith, Ruth Ann and David S. Lux (1993), “Historical Method in
Edwards Brothers. Consumer Research: Developing Causal Explanations of
Martin, Raymond (1997), “The Essential Difference Between Change,” Journal of Consumer Research, 19 (March), 595–610.
History and Science,” History and Theory, 36 (February), 1–14. Song, X. Michael and Mark E. Parry (1997a), “A Cross-National
Marwick, Arthur (1970), The Nature of History. London: Comparative Study of New Product Development Processes:
MacMillan. Japan and the United States,” Journal of Marketing, 61 (April),
McCullagh, C. Behan (1984), Justifying Historical Descriptions. 1–18.
New York: Cambridge University Press. ——— and ——— (1997b), “The Determinants of Japanese New
Menon, Ajay and Anil Menon (1997), “Enviropreneurial Marketing Product Success,” Journal of Marketing Research, 34
Strategy: The Emergence of Corporate Environmentalism as (February), 64–76.
Market Strategy,” Journal of Marketing, 61 (January), 51–67. Srinivasan, V. and Charlotte H. Mason (1986), “Nonlinear Least
Messinger, Paul R. and Chakravarthi Narasimhan (1995), “Has Squares Estimation of New Product Diffusion Models,”
Power Shifted in the Grocery Channel?” Marketing Science, 14 Marketing Science, 5 (Spring), 169–78.
(2), 189–223. Szymanski, David M., Sundar G. Bharadwaj, and P. Rajan
Miles, Matthew B. and A. Michael Huberman (1984), Qualitative Varadarajan (1993), “Standardization Versus Adaptation of
Data Analysis: A Sourcebook of New Methods. Newbury Park, International Marketing Strategy: An Empirical Investigation,”
CA: Sage Publications. Journal of Marketing, 57 (October), 1–17.
Nevett, Terence (1991), “Historical Investigation and the Practice Tedlow, Richard S. (1996), New and Improved: The Story of Mass
of Marketing,” Journal of Marketing, 55 (July), 13–23. Marketing in America. Boston: Harvard Business School Press.
172 JOURNAL OF MARKETING RESEARCH, MAY 2000

Tellis, Gerard J. and Peter N. Golder (1996), “First to Market, First Williamson, Oliver E. (1981), “The Economics of Organization:
to Fail? Real Causes of Enduring Market Leadership,” Sloan The Transaction Cost Approach,” American Journal of
Management Review, 37 (Winter), 65–75. Sociology, 87 (3), 548–77.
Todd, William (1972), History as Applied Science. Detroit: Wayne Wind, Jerry and Vijay Mahajan (1997), “Issue and Opportunities in
State University Press. New Product Development: An Introduction to the Special
Tuchman, Barbara (1981), Practicing History. New York: Alfred Issue,” Journal of Marketing Research, 34 (February), 1–12.
A. Knopf. Workman, John P. (1993), “Marketing’s Limited Role in New
Webster, Frederick E., Jr. (1992), “The Changing Role of Marketing Product Development in One Computer Systems Firm,” Journal
in the Corporation,” Journal of Marketing, 56 (October), 1–17. of Marketing Research, 30 (November), 405–21.
Weiss, Allen M. and Erin Anderson (1992), “Converting from Zaltman, Gerald (1997), “Rethinking Market Research: Putting
Independent to Employee Salesforces: The Role of Perceived People Back In,” Journal of Marketing Research, 34
Switching Costs,” Journal of Marketing Research, 29 (November), 424–37.
(February), 101–15.
——— and Jan B. Heide (1993), “The Nature of Organizational
Search in High Technology Markets,” Journal of Marketing
Research, 30 (May), 220–33.

You might also like