Академический Документы
Профессиональный Документы
Культура Документы
0022-2380
CHARLES C . SNOW
JAMES B . THOMAS
ABSTRACT
This article discusses the contributions that field methods have made to the
theory-development process in strategic management. Field studies drawn from
the literature are classified according to their research goal (description, explana-
tion, or prediction) and according to whether they built or tested theory. The
overall conclusion is that field research methods will continue to be used heavily
to develop strategy theory. However, certain conditions must be met to maximize
the contribution of field methods to strategy research. These conditions include a
balanced research agenda, muldfaceted research approaches, innovative data-
gathering techniques, and an applied futuristic orientation.
INTRODUCTION
Address for reprints: Cbarles C. Snow, Tbe Pennsylvania State University, Tbe Smeal College of Busi-
ness Administration, Dept. of Management and Organization, 411 Beam Business Administration
Building, University Park, PA 16802, USA.
© Basil BlackweU Ltd 1994. Published by Blackwell Publishers, 108 Cowley Road, Oxford OX4 lJF, UK
and 238 Main Street, Cambridge, MA 02142, USA.
458 CHARLES C. SNOW AND JAMES B. THOMAS
and outcomes - that is, they provide mechanisms for observing strategists and
organizations in their natural settings. In strategic management, field studies have
taken many forms, including single and comparative case studies, surveys, simu-
lations, and natural experiments. Sample sizes have ranged from the single case
to several hundred observations, and the unit of analysis has spanned the indivi-
dual decision-maker, the top management team, a business unit of a firm, and a
global corporation.
The purpose of this article is twofold. First, we discuss the contributions made
by field research methods to the development of theory in strategic management.
This is done by presenting a framework for classifying field studies by their (a)
purpose {description, explanation or prediction), and (b) focus (on theory building or
testing). Second, after identifying major trends in the literature, we discuss the
capacity of field methods to make further contributions to theory development.
The article is divided into three major sections. The first section classifies the
field methods commonly used in strategic management research and provides
examples of each method drawn from the literature. The contributions field
methods have made to theory building and testing are discussed in the second
section. The third section presents recommendations concerning the role of field
methods in future strategic management research.
Arcbival analysis
pose a similar classification problem. Had the questionnaire been sent to these
managers' firms to be completed, such a study could be classified straightfor-
wardly as a field study. Because the questionnaire was administered on a uni-
versity campus, however, it is not clear whether this study should be classified as
field research. Thus, it should be noted that the boundaries between types of
organizational research methods, as well as those between different types of field
techniques, are sometimes fuzzy. Although we encountered a number of classifi-
cationaJ difficulties during the review process, it is our belief that all of the studies
cited below are primarily if not entirely field studies.
Type 2: Interviewing
Asking questions of those who have information about a phenomenon that the researcher
has not been able to observe directly
Examples
Open-ended: Miles and Snow (1978)
Structured: EgelhoflF (1988)
Group: McDaniel et al. (1987)
Longitudinal: Johnson (1988)
Telephone: Javidan (1984)
Type 2: Interviewing
Interviewing involves asking questions of those who have information about a
phenomenon that the researcher has not been able to observe directly. Interviews
may require respondents, among other things, to (a) speak about themselves, (b)
inform on the attitudes and actions of others, (c) recall events that have occurred
in the past, and (d) speculate about future situations (Cannell and Kahn, 1968).
Responses are most commonly elicited from single respondents through open-
ended or structured interviews. For example, the studies by Miles and Snow
(1978) were based on open-ended interviews, whereas structured interviewing
provided the data in studies by Lenz and Engledow (1986), Egelhoff (1988) and
Walter and Barney (1990). However, there have been variations on the typical
interview format in strategic management research. These include (a) the use of
combined open-ended and structured interviews (Duhaime and Grant, 1984); (b)
group interviews (McDaniel et al., 1987); (c) unscheduled interviews over multiple
time periods (Johnson, 1988); and (d) telephone interviews (Javidan, 1984). The
study by Grinyer et al. (1986) is notable for its large-scale interviewing process
conducted in 48 randomly-selected firms.
Interviewing typically involves less interaction with the situation than direct or
participant observation, so objectivity may be easier to attain. However, inter-
viewing as a method relies heavily on the opinions, perspectives, and recollections
of respondents. Several useful interviewing guidelines for reducing inaccuracies in
historical accounts of strategic decision-making have been offered (e.g. Huber
and Power, 1985). In many field studies, interview data need to be combined
with observational (and other) data to arrive at a valid characterization of the
research problem (Eisenhardt, 1989).
dents. Questionnaires are 'written interviews', and they can be mailed to respon-
dents (e.g. Dess and Davis, 1984) or administered on site (e.g. Miller et al., 1982).
The questionnaire's prime advantage is its efficiency (speed, low cost, quantifica-
tion) in generating large amounts of data that can be subjected to statistical
analysis. In the strategic management literature, questionnaires have been used
with (a) single respondents in a firm (Balkin and Gomez-Mejia, 1990; Zajac and
Shortell, 1989); (b) multiple respondents in a firm (Norbum, 1986); (c) single
(Rosenberg, 1983) or multiple (Sousa de Vasconcellos e Sa and Hambrick, 1989)
expert panels; and (d) both the corporate and subsidiary offices of a firm
(Ghoshal and Nohria, 1989; Kriger, 1988). Questionnaires are also used to
collect data across several time periods (e.g. Gomez-Mejia, 1988).
The major limitation of questionnaires is their typically low response rate
(Gaedeke and Tootelian, 1976). Low response rates are problematic because they
reduce confidence about the extent to which survey results generalize to the
population from which the survey is drawn. To increase response rates, Dillman
(1978) recommends the 'total design method' consisting of (1) identifying and
coping with factors that can influence the number and quality of responses, and
(2) conducting the survey so that all plans are completely implemented. Dillman
reports that 48 questionnaire surveys using this method produced an average
response rate of 74 per cent, with none of the surveys obtaining less than a 50
per cent response.
Against the backdrop of the total design method, strategic management
surveys produce relatively low response rates. We compiled data on all strategy
studies using surveys from 1981 to 1991 in three major outlets for strategy
research: Strategic Management Journal, Academy of Management Journal, a n d Adminis-
trative Science Quarterly {n - 209). The average response rate for this group o
studies was 63 per cent. However, when weighting for the size of the sample, the
response rate was 55 per cent, suggesting that smaller surveys generate pro-
portionally more replies. Also, 33 per cent of the studies (« = 69) had response
rates lower than 50 per cent, indicating that a sizable proportion of studies may
have considerable problems with generalizability.
Taken together, the above results indicate that strategy surveys may be doing
a less than desirable job of obtaining responses compared to the broader social
science surveys referenced by Dillman. To explore this issue further, we also
looked at the response rates of different types of respondents. As might be
expected, surveying top managers produces the lowest rate of response (an
average of 52 per cent). Research using other managers as respondents achieved
an average rate of return of 61 per cent. Strategy studies using employees or
students generally obtained much higher response rates (72 per cent and 79 per
cent, respectively), but the validity of the strategic-level information provided
could be questioned.
Our examination of survey research in strategic management provides insight
into the dilemma of the response rate-validity tradeoff. In general, the following
features are associated with high response rates: (a) contacting respondents in
advance (e.g. Galbraith and Merrill, 1991); (b) administering surveys on site espe-
cially during work hours (e.g. Pearce, 1983); (c) convincing top managers to dis-
tribute surveys (e.g. Hambrick, 1982); (d) contacting non-respondents and/or
sending them another mailing (e.g. Robinson and Pearce, 1983); and (e) doing
© BasU BlackweU Ltd 1994
FIELD RESEARCH METHODS IN STRATEGIC MANAGEMENT 463
the survey in conjunction with interviews (e.g. Hambrick, 1981). On the other
hand, low response rates were generally obtained when surveys were mailed
directly to respondents without prior contact. This was especially true when
intended respondents were top managers.
researchers may need to augment field methods with other types such as labora-
tory experiments and/or computer simulations (Denzin, 1978).
Perhaps some of the best uses of multiple methods for within- and between-
method triangulation are the studies by Pettigrew (1973) and Pettigrew et al.
(1992). In these studies, Pettigrew and his colleagues made use of multiple field
methods to confront the problems of bias and validity. These methods included
participant and direct observation over time, interviews (using multiple inter-
viewers across levels of hierarchy), questionnaires, and unobtrusive measures (e.g.
personnel files). They also content-analysed internal documents (including firm
reports, memos, minutes of meetings and letters) and compared their personal
diaries of observations. Finally, in the organizational decision-making study
(1973), the research team also embarked on an intensive historical analysis
relying heavily on external data (e.g. newspaper advertisements) as a validity
check of their field findings.
Unfortunately, strategy research has normally not used between-method trian-
gulation in the form found in the Pettigrew studies (or in other forms). Venkatra-
man and Ramanujam's (1987) study of the measurement of business
performance, which used both perceptual measures obtained from managers
within organizations and objective measures compiled outside the organization, is
one of the few notable examples. Although both data-gathering techniques could
arguably be considered field methods, their variance on key dimensions (sub-
jective vs. objective, internal source vs. external source) allowed the investigators
to make a strong case for convergent validity.
In summary, this brief review of the major field research methods has high-
lighted their diversity, versatility, and limitations. It is now appropriate to assess
the contributions that field methods have made to strategic management theory
development.
One means of determining the contributions that field research methods have
made to theory development in strategic management is shown in figure 2. This
matrix is defined by (a) the stage of theory development {building or testing and (b)
the purpose of theory {description, explanation or prediction). Each cell of the matrix
includes a description of that specific stage of the theory-development process
and illustrative examples drawn from the literature.
Theory Building
Before a theory can be tested, it must be constructed. The process of theory gen-
eration typically includes steps such as the identification of relevant constructs,
development of hypotheses about relationships, and the proffering of explanations
for these relationships (Eisenhardt, 1989; Yin, 1984). Each of these steps addres-
ses a different research goal.
1 2 3
Key question is 'what'. Key questions are 'how' Key questions are 'who'.
Identify key constructs and 'why'. Establish 'where' and 'when'.
and variables. Studies relationships among Examine boundary
are usually based on constructs and provide conditions of a theory.
Theory observation (e.g. theoretical rationale for Result may be a middle-
building Mintzberg, 1973) and/ observed relationships. range theory (e.g.
or interviewing (e.g. Studies usually use Eisenhardt and Bourgeois,
Quinn, 1980). observation and/or 1988). Studies use
interviews (e.g. observation, questionnaire
Chandler, 1962; MUes surveys, and interviewing.
and Snow, 1978). Multiple methods are also
employed.
4 5 6
Focus is on developing Focus is on Focus is on testing
and validating measures documenting competing theories of the
of key constructs. relationships among same phenomenon
Studies usually use variables through through crucial
questionnaire, surveys hypothesis testing. experiments (e.g.
and/or interviews (e.g. Large samples are Venkatraman, 1990;
Shortell and Zajac, frequently used with Cosier and Aplin, 1980).
Theory 1990; Herbert and questionnaire surveys Because of the dearth of
testing Deresky, 1987). Several (e.g. Hitt and Ireland, this type of study, no
conceptual articles have 1985a,b; Snow and pattern in field method
also been important. Hrebiniak 1980) or field usage can be discerned.
simulations (e.g.
Thomas and McDaniel,
1990). Because causal
links are examined or
implied, researchers
must be wary of
common-method bias.
The success of descriptive theory building can be seen when other investigators
extend initial theorizing by adding more variables to a framework, by refining
variable definitions, and by exploring relationships among variables in larger-
scale studies. The work of Mintzberg et al. (1976), for example, has spawned a
substantial amount of conceptual and empirical research on strategic issue for-
mulation (e.g. Cowan, 1986; Dutton et al., 1983; Lyles, 1981; Thomas and
McDaniel, 1990).
Explanation. This aspect of theory building is an attempt to establish how and why
key variables are related (Whetten, 1989). The goals of the investigator in this
type of research are to: (a) explore the nature and degree of association among
major variables, (b) decide if additional variables are needed to provide a more
accurate description of the phenomenon, and (c) offer theoretical explanations of
observed relationships. With explanation as a research objective, the investigator
must address the issue of causality among variables. Although a particular study
may not be able to test for causality adequately, it is important to present a
theoretical rationale for any observed relationships.
An example of this type of strategic management research is Chandler's (1962)
explanation of the relationship between strategy and structure as key variables in
the determination of firm performance. Chandler first identified strategy and
structure as variables affecting firm performance. Next, through a detailed chron-
ological analysis, he showed how strategy and structure are related, including the
assertion that strategic change caused structural change. Finally, he suggested
that additional variables, such as leadership, also affected the basic strategy-
structure relationship.
Another example of explanatory research is Miles and Snow's (1978) investiga-
tion of the relationships among competitive strategy, organization structure, and
management processes. This study refined the fundamental variables described
by Chandler (1962) and indicated specific configurations of relationships among
the variables (so-called organization types). Most importantly, the study offered
theoretical reasons why such organization types exist. This rationale was buUt
largely on the constructs of enactment (Weick, 1969) and perceived environ-
mental uncertainty (Duncan, 1972; Lawrence and Lorsch, 1967).
Theory Testing
After a theory has been assembled, it can be tested. Similar to theory building,
theory testing moves through several distinct phases characterized by increasing
specification and rigour. However, because of statistical requirements, theory
testing usually requires studies of larger size than theory building. Ultimately,
theories that purport to explain the same phenomenon can be tested against
each other.
Prediction. It has been claimed that prediction is the ultimate research goal
(Braithwaite, 1955; Kaplan, 1964). Prediction without explanation, however,
leaves the researcher (and the strategist) with little understanding of the phenom-
enon in question. In situations where predictions are incorrect, the researcher has
little or no guidance in modifying the theory. Therefore, the predictive study that
is built on explanation represents the ultimate form of theory development. In its
ideal state, a predictive study presupposes validly defined and measured variables
as well as theoretically grounded explanations of key relationships.
Where competing theories exist, 'crucial experiments' (Platt, 1964), which pit
hypotheses against each other, can be employed to reject one or more of the
theoretical perspectives. Given the volume of theory development that has tran-
spired in strategic management, crucial experiments may now be necessary to
further advance knowledge. Unfortunately, these kinds of studies are rarely
found in the strategy literature. Indeed, it would be difficult to single out any
study that strictly meets all of the criteria of a predictive study. However, some
studies are clearly of this type and may approximate the ideal profile. One such
study is Venkatraman's (1990) examination of the effects on performance of
strategic coalignment. Three forms of coalignment suggested by previous
research were tested in a questionnaire survey of 201 strategic business units.
Another predictive study, in which three different strategic planning perspectives
were tested by presenting managers with case scenarios, was conducted by
Cosier and Aplin (1980).
Summary
It is clear that field research methods have played a major role in the develop-
ment of the strategic management literature. Their overall contribution can be
summarized in several observations. First, field methods have been used more
to generate theory than to test theory. Second, observation and interviewing
have been the main methods used to build theory (Cells 1 - 3 of figure 2), while
questionnaire surveys and field simulations/experiments have been used mostly
to test theory (Cells 4-6). Third, field methods, especially the questionnaire
survey, have been used to create instruments for measuring key variables such
as strategy and environment (Cell 4 of figure 2). Fourth, field methods have
been the most frequent means of conducting multi-level studies of strategic
decision-making within various team, firm, and environmental contexts. Last,
due to the organizational accessibility which they require, field research methods
have been the primary means of linking strategic management theory to
practice.
1 2 3
Theory
Building
5 6
Theory
Testing
Multifaceted Approaches
More sophisticated studies can be conducted by taking a multifaceted approach
to strategy research. Exemplary studies in this regard have exhibited some or all
of the following features: multiple levels, time periods, and methods. For
example, the studies by Eisenhardt and Bourgeois (1988), and Thomas and
McDaniel (1990), considered several levels of analysis simultaneously. These two
studies investigated various combinations of characteristics associated with
individual executives (e.g. the CEO), management teams (e.g. politics), organ-
izations (e.g. strategy), and environments (e.g. high-velocity). Multilevel studies
are needed to identify the boundary conditions of strategy theory (Cell 3 of figure
2).
The time frame of reference also has important implications for the choice and
application of research methods. For example, the studies by Fombrun (1990)
and Coser et al. (1984) combined observation, interviewing, and questionnaires
over an extended period, to examine the dynamics of organizational configura-
tions. Similarly, in their study of strategic decision-making. Gray et al. (1988)
used interviews with several hundred executives across three overlapping periods,
coupled with observations and company archival analysis, to analyse 150 strate-
gic decisions. These studies defy the conventional wisdom that field research
cannot obtain large samples because it is too dme-consuming. As a general rule,
strategy researchers should examine their phenomenon of interest for much
longer periods. Summer et al. (1990) noted that complex alignments of organiza-
tional and environmental factors often take years to achieve or change. There-
fore, more sequenced or programmatic empirical studies are needed (McGrath,
1964). Many of the newest strategies and organizational forms are developing in
industries that are undergoing rapid and substantial change. Formerly distinct
areas are converging - computers and telecommunications, microelectronics and
microbiology, medicine and microelectronics - and such convergence blurs tradi-
tional industry boundaries and relationships. Field studies can help to sort out the
reasons why competitors suddenly become customers or collaborators, thereby
laying the needed theory-building groundwork for later, more extensive theory-
testing research.
Finally, sophisticated multi-method approaches to the study of strategy can be
accomplished in field research. Mark and Shotland (1987) review three such
approaches, referring to them as the (1) triangulation model, (2) bracketing
model, and (3) complementary-purposes model. The triangulation model requires
the use of multiple methods to converge on a single 'answer'. For example, both
Pettigrew (1973) and Eisenhardt and Bourgeois (1988) combined several field
methods in an attempt to converge on a mid-range theory of politics. The brack-
eting model assumes that multiple methods will not converge on a single answer
but rather provide a range of estimates that are likely to include the right answer.
© BasU BlackweU Ltd 1994
nELD RESEARCH METHODS IN STRATEGIC MANAGEMENT 473
CONCLUSION
NOTE
*The original version of this article was presented at the conference on Theory Building
in Strategic Management, University of Illinois, 15-16 May 1990. We thank the con-
ference participants for their ideas and suggestions. Subsequently, the article benefited
from the comments of Rocki-Lee DeWitt, Dennis Gioia, Reuben McDaniel, Alan Meyer,
Raymond Miles, Howard Thomas, and Edward Zajac. We are especially grateful to
Shawn Clark and David Ketchen, who provided both research and editorial assistance,
and the two journal reviewers for their insights.
REFERENCES
ZAHRA, S. A . and PEARGE, J. A., Ill (1992). 'Priorities of CEOs and strategic management
professors for future academic rtsearch,' Joumal of Management Issues, 4, 2, 171-89.
ZAJAC, E . J. and SHORTELL, S. M . (1989). 'Changing generic strategies: Likelihood, direc-
tion, and performance implications'. Strategic Management Joumal, 10, 3, 413-30.