Golder's historical method in research in marketing

Ann-Marie K. Thompson
AUT Business School, Auckland University of Technology, D93, Private Bag 92006, Auckland 1142, New Zealand
a b s t r a c t a r t i c l e i n f o
Article history:
Received 1 July 2009
Received in revised form 1 August 2009
Accepted 1 September 2009
Keywords:
Historical method
Market share stability
Surprising ndings
Seminal article
This article reviews a seminal article in the historical marketing literature Historical Method in Marketing
Research: With New Evidence on Long-Term Market Share Stability (Golder, 2000). The present study uses
citation analysis and commentary to consider the contributions of Golder (2000) to the marketing discipline.
This review explores why the article is seminal and how Golder's article inuences marketing historical
research. Golder's article is important for two main reasons. First for its ndings, which contradict commonly
held beliefs about long-term market share stability, though the relevant literature largely ignores these
ndings. The literature does recognize Golder's (2000) second contribution of the historical method: his
article offers clear guidelines to new and emerging research in historical marketing.
2009 Elsevier Inc. All rights reserved.
1. Introduction
According to Armstrong, 2003, academic literature in marketing
does not provide many important ndings. Criteria to judge if ndings
are important are whether or not they are replicable, valid, useful, and
surprising (Armstrong, 2003). Golder (2000) is one article that ts
these criteria. Golder's article presents a method for historical research
and uses the method to analyze the commonly held belief that market
share is stable over time. Golder analyses Kotler's (1997, p. 352)
statement, 19 out of 25 companies who were market leaders in 1923
were still market leaders in 1983. Using his historical method Golder
(2000) nds this statement to be incorrect.
Woodside (2009) states that multiple methods are necessary for
evaluating the importance of ndings. Possible importance metrics
include the number of citations in general, the average number of
citations per article in that year, and the number of citations in high
impact journals.
The present article evaluates the importance of Golder's (2000)
ndings based on Armstrong's (2003) criteria and citation analysis
(Woodside, 2009). This paper concludes that Golder's (2000) article is
important not only because of the surprising ndings, but as achieving
seminal article recognition for the historical method in marketing. The
present study shows that though Golder presents important ndings
that contradict a commonly held belief, the relevant literature largely
ignores these ndings.
2. Historical method The rst contribution
Golder's (2000) article presents steps to undertake the historical
method and illustrates these steps using the example of brand
leadership studies. He synthesised prior historical methods literature
and places the historical method for marketing within the landscape
of cross-disciplinary historical research. He seeks to describe how
historians view their own research method and its objectives. His
description of the historical method, which also includes the strengths
and weaknesses of the method, provides a thorough reference for
marketers wishing to undertake historical research. The steps to
undertake the historical method include the following actions:
1. select a topic and collect data,
2. critically evaluate the sources of evidence,
3. critically evaluate the evidence,
4. analyze and interpret the evidence, and
5. present the evidence and conclusions.
Historical studies in marketing include studies about marketing
history and studies about the history of marketing thought. Marketing
history studies look at the history of marketing activities and
practices (Jones and Keep, 2009, p152) such as the representation
of women in advertising over 100 years. History of marketing thought
studies look at the history of ideas about those phenomena (Jones
and Keep, 2009, p152), for example, a study may look at inuences on
the development of marketing between 1900 and 1930. Golder's
(2000) article offers a method of undertaking historical research in
both of these areas.
Historical research offers two approaches. One method follows the
German school of development and is more hermeneutic/construc-
tionist in its assumptions (Firat, 1987; Jones and Monieson, 1990).
Journal of Business Research 63 (2010) 12691272
Tel.: +64 9 9219999x5826; fax: +64 9 9219990.
E-mail address: ann-marie.thompson@aut.ac.nz.
0148-2963/$ see front matter 2009 Elsevier Inc. All rights reserved.
doi:10.1016/j.jbusres.2009.09.005
Contents lists available at ScienceDirect
Journal of Business Research
This rst method focuses on change or continuity over time, and the
specic contexts, geographies, events, and times where this change or
continuity took place (Hollander et al., 2005; Savitt, 1980;). The
method specically seeks to investigate the causes and processes of
change, by looking at the deeper contextual happenings surrounding
the events (Dray, 1957; Hollander et al., 2005; Smith and Lux, 1993).
The second approach follows the French schoolAnnalesof
development, and is more positivistic in its assumptions than the
rst method (Firat, 1987). A more positivist denition of historical
research looks at verifying collected data and using the data to
interpret and present causal evidence (Savitt, 1980). This version of
the historical approach aims to collect information that is falsiable
(Calder and Tybout, 1987; Fogel and Elton, 1983) and able to generate
causal scientic knowledge if collected using the historical method
(Cochrane, 1929; Gottschalk, 1969; Iggers, 1979). Golder (2000)
presents a historical method in keeping with this second philosophy.
When comparing Golder (2000) to another positivist historical
method (Savitt, 1980), the key differences are that Golder (2000)
provides a more detailed account of the historical method which is
more rmly grounded in the literature. This expansion and extension
of the historical method by Golder (2000) also gives more specic
detail on verication and validity methods when assessing data,
legitimizing the method further. The signicance of Golder's (2000)
article is in the detailed presentation of historical research as
scientic. By synthesizing past historical method literature, Golder
paves the way for the use of historical methods in academic journals
that are more supportive of positivist methods. By laying out the
specic steps for the historical method, secondary data's evaluation,
validation steps, and verication, Golder gives legitimacy to the
method within a positivistic framework. Golder's (2000) article
facilitates historical research by providing a thorough resource to
new and emerging researchers in marketing history.
3. Contradictory ndings: The second contribution
In the second part of the article Golder (2000) demonstrates the
usefulness of his method by re-examining the empirical generaliza-
tion that market share is stable over prolonged periods. He rst
examines Kotler's (1997) statement that 19out of 25 companies studied
were consistent market share leaders between 1923 and 1983. He nds
that this statement is based on Carpenter and Nakamoto (1989) who
source the nding from a study in Advertising Age (1983). By critically
evaluating the sources of evidence as well as the evidence itself, Golder
(2000) nds that the original Advertising Age (1983) article is based on
a book, andonreviewing the book, nds that the original study included
100companies, rather than25. The 25wereselectedbytheauthor of the
Advertising Age (1983) article to support their long-termbrand leader-
ship viewpoint.
Consequently, Golder (2000) undertakes his own analysis of the
original 100 companies by comparing the original companies fromthe
original study with market share data from 1997. He nds that more
of the brand leaders from 1923 failed by 1997 than reported, and that
market share of the market leaders from 1923 was unstable over that
time. This nding is contrary to the Advertising Age (1983) article and
against the commonly held belief that successful brands have long-
term market share leadership as well as the belief that market share
leadership is stable over the long-term (Golder, 2000). These ndings
increase understanding of long-term market share by then providing
a historical narrative of the organizations. In his analysis Golder
identies factors that do contribute to market leadership and brand
failure.
4. Criteria of importance
According to Armstrong (2003) criteria for important ndings are
that they are replications/replicated, valid, useful and surprising.
Replication implies consistent ndings when studies are redone with
different samples; replications are one of the most crucial measures of
importance (Leone and Schultz, 1980; Rosenthal and Rosnow, 1984).
Replications are especially important because they often do not nd
the same results as the original study (Hubbard and Vetter, 1996).
However replication reports seldomappear in print (Kerr et al., 1977).
Golder (2000) is an important study because his study is a replication,
as well as being an example of a replication study that does not support
previous ndings.
The second criterion is that the ndings are valid. While Armstrong
(2003) states that research seldom assesses validity, Golder (2000) is
an example of where validation is not only conducted, but includes a
thorough description of the research steps taken.
Golder (2000) also provides a historical narrative which describes
the context and variables that were present in each of the companies.
This historical narrative allows for assessing the validity of the
ndings as the narrative gives alternative sources and measures of the
success of the rms and brands in the study.
Usefulness is the third criterion needed for ndings to be important
(Armstrong, 2003). Findings are useful if they are actionable by
practitioners, consumers, or policy makers directly (Armstrong, 2003).
The historical narrative by Golder (2000), which includes an analysis of
why some market leaders maintained their position and others did not,
gives detailed information on the ndings and the contextual elements
of success and failure; Golder provides important insights for organiza-
tions. These achievements make the ndings important and useful.
Lastly, Armstrong (2003) suggests that ndings also need to be
surprising to be important. However Armstrong also states that
surprising ndings often contradict generally accepted ndings and so
are disliked and hard to get published (Armstrong and Hubbard,
1991). The ndings of Golder (2000) are surprising. The ndings go
against the commonly held statistics of market share leadership over
time and also contradict beliefs in the stability of market share.
So, according to the criteria of importance set out by Armstrong
(2003), Golder's (2000) ndings are extremely important as his study
is a replication that is valid, useful, and surprising. Turning now to
citation analysis, the present article assesses the impact of these two
important contributions.
5. Citation analysis
Woodside (2009) suggests that judging the importance of ndings
using a single measure is inadequate. Turning to citation analysis,
Golder (2000) is also an important article based on the number of
citations, though the expectation might be more citations for such a
surprising nding. Delving further, the use of Golder (2000) as a
citation also shows the importance of the article (see Table 2). The
citation analysis here uses Google Scholar as the most comprehensive
database and only includes citations in peer reviewed academic
journal articles. The analysis here does not include self or book
citations. Examining Table 1, the earliest citations of Golder (2000)
were in 2000 (Chandy and Tellis) and 2001 (Brown et al., 2001). This
diffusion rate is quick in general, but shows Golder's (2000)
importance in comparison with the other articles in the same issue
who were mostly not cited until 2002.
Table 1
Earliest and most recent citations of Golder (2000).
Earliest citations Most recent citations
Chandy and Tellis (2000) 114 Tellis et al. (2009) 6
Brown et al. (2001) 8 Sood et al. (2009) 3
Montaguti et al. (2002) 8 Bianchi(2009) 0
Mizik and Jacobson (2003) 136 Bronnenberg et al. (2009) 0
Tellis et al. (2009) 120 Ewing et al. (2009) 0
1270 A.K. Thompson / Journal of Business Research 63 (2010) 12691272
Lookingat the evolutionof the use of citations (see Table 2) toGolder
(2000) increases understanding of the impact of the article through
both its contributions. Chandy and Tellis (2000) used Golder's (2000)
historical method in their article, while Brown et al. (2001) includes
Golder (2000) in the literature review stating that his article discusses
historical methods and includes a longitudinal analysis of market share.
Montaguti et al. (2002) discuss the limitations and strengths of the
historical method in their limitations/future research section using
Golder (2000). Tellis et al. (2003) provideanother use of Golder's (2000)
article for justication of the use of historical data.
Mizik and Jacobson (2003) focus on Golder's (2000) surprising
ndings. Mizik and Jacobson (2003) recognize that Golder invalidates
the ndings in the Advertising Age article though they do not dwell on
this point. Mizik and Jacobson also discuss the factors that Golder
(2000) nds that do contribute to long-term market share stability.
The most recent citations also describe the use of Golder's (2000)
historical method, specically the use of his method in data collection
(Bianchi, 2009; Sood et al., 2009; Tellis et al., 2009). Ewing et al. (2009)
cite Golder (2000) and discuss his market share ndings more in-depth
andalsouse his historical narrative tolookat factors that cause brands to
fail. Bronnenberg et al. (2009) make a passing comment which refers
readers to Golder (2000) for a discussion of survivor bias.
Moving on to the most cited articles that cite Golder (2000), a similar
picture forms of the impact that Golder's (2000) historical method has,
and the lack of impact that his ndings have had. Nine of the ten most
cited articles use Golder's (2000) historical method in some way. Some
examine Golder's method for data collection (Chandy and Tellis, 2000),
or use the method to justify their data sources (Tellis et al., 2003) or use
historical methods in general (Echambadi et al., 2006; Shocker et al.,
2004). SoodandTellis (2005) cite Golder (2000) multiple times usinghis
method and the strengths and weaknesses of the method. Hollander
et al. (2005) discuss Golder (2000) as anexampleof a positivist approach
to historical research.
Only one article discusses the ndings fromGolder (2000) Mizik
and Jacobson (2003). Mizik and Jacobson state that Golder (2000)
does not agree with the length of time that market share stability lasts
and still state the statistic from the Advertising Age article. So the
appearance is that Golder (2000) has more of an impact for his
historical method than for discrediting the ndings on market share
stability. The average citations across the other articles that were in
the same issue as Golder (2000) support this conclusion; see Fig. 1.
While the number of citations by year for Golder (2000) is steady, one
would expect the level of citations to be higher than the average of
other articles in the issue, given its importance.
When comparing the level of citation of Golder's (2000) article
with the rest of the articles in the same issue, a possible problem with
averaging out the citations of the other articles is the one article in the
issue that is very highly cited (Jap and Ganesan, 2000). To compare
citations then, Fig. 2 takes the average citations per year per published
article in the Journal of Marketing Research and compares this average
with the number of citations of Golder (2000) per year (self-citations
are excluded). Similar to the other analyses of citations, no clear
nding trend occurs indicating the importance of the article based on
the level of citations.
Possible reasons for the lack of citations come from Armstrong
(2003) who states that surprising or contradictory ndings are not
often accepted, and are extremely hard to get published in journals
(Armstrong and Hubbard, 1991). This view receives support when
considering the average citations per year for the discredited statistic.
Using citation analysis on Kotler (1997) is difcult as Kotler (1997) is a
textbook. So turning insteadto citations of the original study that relates
to market leadership (Carpenter and Nakamoto, 1989), the level of
citations toCarpenter andNakamoto(1989) per year does not dropafter
Table 2
Most cited articles citing Golder (2000).
Articles Number of citations Usage
Mizik and Jacobson (2003) 136 Discusses the ndings of the article.
Tellis et al. (2003) 120 Used to justify using publicly available data sources.
Chandy and Tellis (2000) 114 Use Golder's historical approach to data collection.
Srinivasan et al. (2004) 84 Used Golder's historical method.
Sood and Tellis (2005) 53 As one article that uses the historical method. Strengths of historical method. Procedure used.
Bianchi and Arnold (2004) 27 Used methods outlined in article to evaluate the sources of evidence and the evidence itself.
Shocker et al. (2004) 24 Justication for using historical methods for their research topic.
Srinivasan et al. (2006) 19 Used Golder's historical method.
Hollander et al. (2005) 17 Golder's historical method is presented as a positivist method.
Echambadi et al. (2006) 16 As an advocate for the use and examination of longitudinal data to obtain causal explanation.
Fig. 1. Average citations for articles in the same issue as Golder (2000) with standard
errors. Fig. 2. Average citations per year of all articles in the Journal of Marketing Research.
1271 A.K. Thompson / Journal of Business Research 63 (2010) 12691272
Golder (2000). Infact, the number of citations per year for Carpenter and
Nakamoto (1989) actually increases (See Fig. 3) as the interest in rst
mover and pioneer advantage increases. On assessing the usage of
Carpenter and Nakamoto (1989) in citations, most, if not all, of the
citations do not quote the discredited statistic but instead list the early
pioneer advantages that Carpenter and Nakamoto (1989) identify.
The impact of Golder (2000) becomes clear, however when con-
sidering the journals where its citations appear. The article receives
citations in six articles in the Journal of Marketing, two in both Marketing
Science and the Journal of Business Research, and once each in the Journal
of Marketing Research and the Journal of Management Studies. Each of
these journals are high impact journals being among the top 15 journals
among journals having the highest-weighted productivity (Woodside,
2009). This nding shows that Golder (2000) receives citations in many
high impact journals adding to its importance (Woodside, 2009).
6. Conclusions
The present article discusses the two major contributions of Golder
(2000). The rst is his detailed description of the historical method that
places the method in a more positivist position, giving the method high
legitimacy. Golder's details on the historical method sythesizes the
historical method literature facilitating more historical research in
marketing. The second contribution is the important and surprising
nding that a well-known nding for market share leadership (Kotler,
1997; Carpenter and Nakamoto, 1989) is invalid. Not withstanding
these two contributions, the level of citations of Golder (2000) is much
lower than would be expected of such an article. This conclusion
suggests that the ndings of Golder's (2000) historical analysis are
largely ignored and instead, the key contribution that literature focuses
on is Golder's presentation on the historical method.
References
Advertising Age. Study: majority of 25 leaders in 1923 still on top Old standbys hold
their own; 1983. September 19:32.
Armstrong JS. Discovery and communication of important marketing ndings:
evidence and proposals. J Bus Res 2003;56:6984.
Armstrong JS, Hubbard R. Does the need for agreement among reviewers inhibit the
publication of controversial ndings? Behav Brain Sci 1991;14:1367.
Bianchi CC. Retail internationalisation from emerging markets: case study evidence
from Chile. Int Mark Rev 2009;26:22143.
Bianchi CC, Arnold SJ. An institutional perspective on retail internationalization success:
home depot in Chile. Int Rev Retail Distrib Consum Res 2004;14:14969.
Bronnenberg BJ, Dhar SK, Dube JPH. Brand history, geography, and the persistence of
brand shares. J Polit Econ 2009:117.
Brown S, Hirschman EC, Maclaran P. Always historicize!: researching marketing history
in a post-historical epoch. Marketing Theory 2001;1:4989.
Calder BJ, Tybout AM. What consumer research is.... J Consum Res 1987;14:13640.
Carpenter GS, Nakamoto K. Consumer preference formation and pioneering advantage.
J Mark Res 1989;26:28598.
Chandy RK, Tellis GJ. The incumbent's curse? Incumbency, size, and radical product
innovation. J Mark 2000;64:1-17.
Cochrane CN. Thucydides and the science of history. London: Oxford University Press;
1929.
Dray W. Laws and explanation in history. Oxford: Oxford University Press; 1957.
Echambadi R, Campbell B, Agarwal R. Encouraging best practice in quantitative
management research: an incomplete list of opportunities. J Manage Stud 2006;43:
180120.
Ewing MT, Jevons CP, J Khalil EL. Brand death: a developmental model of senescence.
Journal of Business Research 2009;62:3328.
Fogel RW, Elton GR. Which road to the past? Two views of history. New Haven, CT: Yale
University Press; 1983.
Fuat Firat A. Historiography, scientic method, and exceptional historical events. Adv
Cons Res 1987;14:435.
Golder P. Historical method in marketing research with new evidence on long-term
market share stability. J Mark Res 2000;37:15672.
Gottschalk L. Understanding history: a primer of historical method. New York: Alfred A.
Knopf; 1969.
Hollander S, Rassuli K, Jones B, Dix L. Periodization in marketing history. J Macromark
2005;25:32.
Hubbard R, Vetter DE. An empirical comparison of published replication research in
accounting, economics, nance, management, andmarketing. J Bus Res 1996;35:15364.
Iggers GG. Federal Republic of Germany. In: Iggers GG, Parker HT, editors. International
handbook of historical studies: contemporary research and theory. Westport, CT:
Greenwood Press; 1979. p. 21732.
Jap SD, Ganesan S. Control mechanisms and the relationship life cycle: implications for
safeguarding specic investments and developing commitment. J Mark Res 2000;37:
22745.
Jones DGB, Monieson DD. Historical research in marketing retrospect and prospect.
J Acad Mark Sci 1990;18:26978.
Jones DGB, Keep W. Hollander's doctoral seminar in the history of marketing thought.
J Hist Res Mark 2009;1:15164.
Kerr S, Tolliver J, Petree D. Manuscript characteristics which inuence acceptance for
management and social science journals. Acad Manage J 1977;20:13241.
Kotler P. Marketing management: analysis, planning, implementation, and control. 9th
ed. Upper Saddle River, NJ: Prentice Hall; 1997.
Leone RP, Schultz RL. A study of marketing generalizations. J Mark 1980;44:108.
Mizik N, Jacobson R. Trading off between value creation and value appropriation: the
nancial implications of shifts in strategic emphasis. J Mark 2003;67:6376.
Montaguti E, Kuester S, Robertson TS. Entry strategy for radical product innovations: a
conceptual model and propositional inventory. Int J Res Mark 2002;19:2142.
Rosenthal R, Rosnow RL. Essentials of behavioural research: methods and data analysis.
New York: McGraw-Hill; 1984.
Savitt R. Historical research in marketing. J Mark 1980;44:52.
Shocker AD, Bayus BL, Kim N. Product complements and substitutes in the real world:
the relevance of other products. J Mark 2004;68:2840.
Smith R, Lux D. Historical method in consumer research: developing causal
explanations of change. J Consum Res 1993;19:595610.
Sood A, Tellis GJ. Technological evolution and radical innovation. J Mark 2005;69:15268.
Sood A, James GM, Tellis GJ. Functional regression: a new model for predicting market
penetration of new products. Mark Sci 2009;28:3651.
Srinivasan R, Lilien GL, Rangaswamy A. First in, First out? The effects of network
externalities on pioneer survival. J Mark 2004;68:4158.
Srinivasan R, Lilien GL, Rangaswamy A. The emergence of dominant designs. J Mark
2006;70:1-17.
Tellis GJ, Stremersch S, Yin E. The international takeoff of new products: the role of
economics, culture, and country innovativeness. Mark Sci 2003;22:188208.
Tellis GJ, Yin E, Niraj R. Does quality win? Network effects versus quality in high tech
markets. J Mark Res 2009;46:13549.
Woodside AG. Journal and author impact metrics: an editorial. J Bus Res 2009;62:14.
Fig. 3. Number of citations per year for Carpenter and Nakamoto (1989).
1272 A.K. Thompson / Journal of Business Research 63 (2010) 12691272