Abstract
Competent research methods and data analysis are essential components for the progression of family business research. To identify and evaluate empirical trends, and make suggestions for future research, we examine 319 empirical articles published in Family Business Review since 1988. These studies are compared with 146 family business research articles published in top-tier journals not dedicated to family business research over the same timeframe. While we substantiate growth in rigor and sophistication, we address specific family business research challenges regarding construct validity, generalizability, causality, temporality, and multilevel issues. Suggestions are provided for future empirical research across six major topical areas.
Introduction
Family business has undergone significant changes over the past several decades that now point to its legitimacy as an independent academic field of study. Specifically, increases in terms of total number of studies published, the impact of these studies on the broader academic community, and the number of conferences and journals dedicated to family business demonstrate its growing prestige and acceptance as an established field (Sharma, Chrisman, & Gersick, 2012; Stewart & Miner, 2011; S. R. Wilson et al., 2014). Also, Family Business Review (FBR), the premier outlet for family business research, has repeatedly ranked in the top 20 journals in the field of business—as measured by impact factor—and ranked fourth of 110 business journals in 2014 (Sharma, 2015). Such indicators suggest that family business has been generally successful in addressing the three ingredients that Hambrick and Chen (2008) argue enhance an aspiring academic community’s chances of ascension—differentiation, mobilization, and legitimacy building.
While differentiation (i.e., the distinctiveness of the research domain) and mobilization (i.e., procurement of necessary structures, relationships, and resources for collective action) are important considerations in our efforts to further specify the boundaries and direction of research in the field, the focus of this review article is on legitimacy building, particularly the legitimacy gained through “conforming to the methodological or paradigmatic conventions of more well-established fields” (Hambrick & Chen, 2008, p. 38). More specifically, given the extensive growth of family business research in general (Sharma et al., 2012), this study’s purpose is to assess the state of empirical research to account for the past as well as guide future research efforts. Empirics—the practice of basing ideas, conclusions, or theories on testing, observation, or experience—provide a common language and legitimizing indicator for scholars across fields and, therefore, play an influential role in the development of knowledge and establishing relevance for a field (Cole, 1983; Pfeffer, 1993). As such, empirics, including methodological (i.e., the process of collecting data and information) and analytical (i.e., systematic examination of data or statistical approach) practices, play a key role in the forging of a field’s distinct position among established academic areas of interest (Harrison & Leitch, 1996; Sharma et al., 2012).
Family business scholars have previously asserted an improvement in methods and analytic techniques (e.g., Bird, Welsch, Astrachan, & Pistrui, 2002; Debicki, Matherne, Kellermanns, & Chrisman, 2009; Litz, Pearson, & Litchfield, 2012). However, the overall current state and historical trends of empirics in family business research lacks detail and clarity, particularly in terms of how research has advanced relative to more legitimated domains. In other words, uncertainty remains regarding how empirical improvements have been realized within the broader evolution of the literature, leaving a gap in what we currently know and ought to know about the state of empirical family business research. Hence, rooted in the argument that an explicit understanding of previous work is required for a field to progress (Dyer & Sanchez, 1998), this article thoroughly examines 465 published articles (319 published in FBR and 146 from prominent nonfamily business-specific journals), focusing on the research methods and analytical techniques used.
This review makes three key contributions. First, we provide a comprehensive review of the methodologies and analytics used in family business research by analyzing all empirical studies in FBR since its inception (i.e., 1988); this timeframe allows us to gain a comprehensive view of the field’s progression over an extended length of time. For while new family business–specific journals have recently emerged (e.g., Journal of Family Business Management, Journal of Family Business Strategy), they derive short-term benefit from the long-term legitimizing efforts of FBR. Second, we help clarify and legitimize progress by juxtaposing articles published in FBR to those family business studies published in other high-quality journals that are not dedicated to family business. With intent, we build on similar reviews of empirics in family business research (e.g., Bird et al., 2002; S. R. Wilson et al., 2014) to examine trends over multiple decades, but do so across a more comprehensive set of methods and analytical techniques and in a comparative fashion. Third, we identify distinct empirical challenges in family business research and provide suggestions to scholars regarding where future research can be focused in order to better advance the field. Specifically, we consider how family business research can be improved by directing more attention to four general problematic areas identified in our review: construct validity and reliability, generalizability, causality, temporality, and multilevel issues. Then, drawing on Dyer and Sanchez (1998), we extensively discuss empirical applications and employment relative to the six most prominent topical areas of research that appeared in our review. Since each of these areas of research (e.g., succession, governance) tends to focus on unique research questions and theoretical perspectives, we give specific attention to the empirics contained therein.
Overall, we argue that because the availability and applicability of empirical approaches can both stimulate and allow for the testing of new inquiries (Zahra & Sharma, 2004), it is imperative that scholars diligently consider empirics from the outset and throughout the course of any study. Furthermore, with the increasing complexity and proliferation of data and advanced statistical software, researchers can (and should) ask and test more sophisticated, empirically robust questions. However, while we focus explicitly on empirics in this review, we note the interconnectedness of theory and empirics, and the role of this relationship in increasing consensus on the boundaries and relevance of the field (e.g., Pfeffer, 1993). As the exclusive focus of this review, empirics serve to test and validate critical questions that provide a finer grained perspective of various family business phenomena; yet scholars must always consider the appropriateness of their empirical approaches from a theoretical and logical perspective. As there are many potentially valid approaches to empirically test family business research questions, scholars should be cognizant to account for both theory and context of the study, exercising good judgment to ensure the proper use of methods and analytical techniques (Bettis, Gambardella, Helfat, & Mitchell, 2014).
Methods and Analyses
Sample
Our study tracks trends in family business empirical research over time from 1988 through 2014. We begin our review in 1988 because it was the first year FBR was published. To identify our sample of empirical articles, we first screened each of the 855 articles published by FBR during this timeframe. Following Colquitt and Zapata-Phelan (2007) and Yu, Lumpkin, Sorenson, and Brigham (2012), we identified empirical studies as quantitative, qualitative, or mixed method data-based research designed to test research questions, models, hypotheses, or to develop propositions. All other articles were considered inappropriate and excluded from the sample, even if they contained some data. For instance, some studies consisted of simple interview transcripts (e.g., Bowman-Upton, 1993; Lank, 1991) and others described personal anecdotes from the perspective of various family business members (e.g., Correll, 1989); these types of articles were not included in our sample.
In total, 319 FBR empirical studies were deemed relevant for our review. Of these, 126 (39%) were published in the past 9 years of our sampling time frame (2006-2014), compared with 79 (25%) published in the first 9 years (1988-1996); this comes despite an overall trend downward in terms of total articles published. This change serves as a basic, yet telling, indicator of the growing prevalence of empirical research in family business. Figure 1 contains a breakdown of the frequency of FBR empirical articles over the sampling timeframe; more recently, the split between empirical and conceptual articles is relatively even.
Growth of empirical articles in Family Business Review (FBR), 1988-2014 (N = 855).
Given the increase of family business research appearing in both specialized (e.g., Bird et al., 2002; Chrisman, Chua, & Steier, 2003; Heck & Mishra, 2008; Rogoff & Heck, 2003) and mainstream journals (e.g., Chrisman, Chua, Kellermanns, Matherne, & Debicki, 2008; Debicki et al., 2009), as well as an emerging prominence and domain overlap among management scholars (Sharma, Hoy, Astrachan, & Koiranen, 2007), we next sought to benchmark FBR’s progress by examining an additional sample of empirical articles. Specifically, based on similar reviews of this kind (e.g., McLeod, Evert, & Payne, 2014; Short, Payne, & Ketchen, 2008), we gathered relevant articles from nine prestigious management-based journals that have published family business research between 1988 and 2014: Academy of Management Journal, Administrative Science Quarterly, Entrepreneurship Theory and Practice, Journal of Business Venturing, Journal of International Business Studies, Journal of Management, Journal of Management Studies, Organization Science, and Strategic Management Journal. Through comparisons with family business research appearing in these other prominent journals (also referred to as non-FBR articles throughout this review), we can assess if FBR differs in any significant way, in terms of empirics, from these more general outlets. Specifically, our findings can help address oft-repeated criticisms of family business studies directed at shortcomings in methodological rigor (Chrisman, Sharma, & Taggar, 2007) and statistical analysis (Debicki et al., 2009).
Our initial screen of the comparison sample journals returned 13,101 total articles. Next, we filtered these results by searching for articles with “family” in the abstract. Of these, we eliminated articles that did not address family businesses specifically. For example, articles that broadly studied the effects of the Family and Medical Leave Act on employee absenteeism were excluded (e.g., Johnson, Holley, Morgeson, LaBonar, & Stetzer, 2014); this vetting process left 407 potential articles. Additional screening and removal of nonempirical articles resulted in 146 articles for our final comparison sample. In total, we coded 465 articles, 319 from FBR and 146 from other high-quality journals that composed our comparative sample. 1 Relative to the articles sampled from FBR, this collection of studies showed a similar upward trend in terms of the number of empirical family business articles published, although the absolute numbers and resulting percentages are much lower (as would be expected). These results are displayed in Figure 2.
Growth of family business empirical articles in other journals, 1988-2014 (N = 13,101).
Coding Procedure
As an organizing structure for our review, we generally followed the coding process used by Hiller, DeChurch, Murase, and Doty (2011). Each of the 465 articles in our overall sample was read and rated by at least two authors; a third rater was commonly used to ensure reliability and to settle differences. Following the establishment of our common coding scheme, the raters individually coded a random sample of 35 articles and met to discuss discrepancies. These deliberations led to further refinement and specification of coding criteria. Throughout the coding process, raters regularly discussed coding protocol and results to maintain process consistency while resolving discrepancies. In sum, two authors jointly resolved 100 articles, which was more than 20% of the overall sample. With ample experience at this point of the coding and resolution process, the remaining sample was divided for completion. Any additional coding conflicts were adjudicated using the previously described process.
For the quantitative techniques, we initially used categories suggested by Shook, Ketchen, Cycyota, and Crockett (2003) and Dean, Shook, and Payne (2007). Over the course of our review, several techniques that are commonly used in multilevel studies (e.g., McKenny, Payne, Zachary, & Short, 2014) emerged and were added to the coding process. Coding schemes for qualitative techniques were developed concurrently with our initial review of several qualitative articles within our sample. After generating counts of prominent qualitative collection and analysis techniques (e.g., structured, semistructured interviews; observation, participant observation; inductive, deductive theory development), further specification and additions to our coding classifications were iteratively finalized following Pratt (2009) and Gephart (2004).
Consistent with protocol found in similar reviews (e.g., Shook et al., 2003; Stone-Romero, Weaver, & Glenar, 1995), we coded data analytic procedures specifically used to test hypotheses or the research question(s) in each article. Furthermore, the nature of our coding was exhaustive, screening each occurrence of a technique used in the process of hypothesis testing. For example, Barnett, Eddleston, and Kellermanns (2009) used factor analysis to assess concerns about common method bias, coefficient alpha to examine the internal consistency of the scale items administered, and hierarchical regression to test the study’s hypotheses. In this case, hierarchical regression was coded because it was the only technique used to directly test the hypotheses. Likewise, if a study tested multiple hypotheses, we coded each technique that was used to test at least one hypothesis. In other words, we did not restrict the number of methods or techniques to one for each article. Rather, we included the method/technique if it was used to test one or more hypotheses/questions, or explicitly used to develop propositions.
Finally, to classify articles according to topic, which we later use in the “Discussion” section, we used Dyer and Sanchez’s (1998) categorization scheme. However, during the course of our assessment we iteratively added the topic characteristics and attributes to the coding scheme. This additional categorization follows previous work by Bird et al. (2002), De Massis, Chua, and Chrisman (2008), and Sharma (2004) to account for characteristics and attributes at multiple levels (i.e., individual, family, firm) and subsumes the gender and ethnicity categorization that was part of Dyer and Sanchez’s (1998) original scheme. Specifically, studies classified in this way are those that focus on the distinguishing features or qualities of a person, group, or organization (Koiranen, 2002). One study, for example, that fell into this category was Fahed-Sreih and Djoundourian (2006), which examines firm age and number of employees in a sample of Lebanese family businesses.
The most prominent topics that appeared throughout our overall sample included management of the firm (40%), business performance and growth (39%), characteristics and attributes (32%), interpersonal family dynamics (29%), governance (28%), and succession (24%). As accomplished by Dyer and Sanchez (1998, p. 291), the categories were “not exclusive because articles often contain more than one topic”; this resulted in percentages totaling greater than 100%. Other categories included topics such as interpersonal business dynamics, wealth management, philanthropy, and estate issues.
Analyses
For each method or technique, we calculated percentage use indices (PUIs; Dean et al., 2007; Shook et al., 2003; Stone-Romero et al., 1995). PUIs were calculated by dividing the frequency of a method or technique’s use by the total number of sampled articles that appeared in a given period. For example, a PUI of .23 would indicate that 23% of the sampled studies during a specified time period relied on a particular method to test at least one hypothesis. Our coding protocol also included a range of research design dimensions such as level of analysis (e.g., individual, group, subunit, organization, interorganizational, industry, environment, or country/national), industry sample (e.g., single, multi, or nonindustry), geographic characteristics of samples (e.g., single or multiple country), and temporal scope (e.g., cross-sectional, longitudinal, or both). To best assess trends over time, we followed similar studies by computing correlations between annual PUIs and publication years for each coded empirical dimension (e.g., Dean et al., 2007; Shook et al., 2003; Stone-Romero et al., 1995). By correlating PUIs (vs. absolute counts) with year of publication, we account for changes in overall journal output over time as an alternative explanation of trends related to a given empirical method or technique. In search of a more segmented perspective of these year-over-year data, we also partitioned select study characteristics into three 9-year intervals (1988-1996, 1997-2005, and 2006-2014). These aggregations, integrated into the results and discussion, help further clarify and compare certain trends over time.
Results
Over the 27-year time span of our review, there has been a notable increase in the number of published empirical articles. For articles published in FBR, PUIs for the overall presence of empirical studies has increased significantly since the journal’s inception (r = .73; p < .001). This change suggests the increasing importance of empirical research to FBR, not simply in terms of relative numbers of articles published but the overall analytical sophistication in a given article. Furthermore, because paper submissions have increased substantially over time (Sharma et al., 2012), along with the number of outlets publishing family business research, the field is becoming increasingly competitive, which should indicate and give way to improvements in overall quality. Considering these general trends, we now report more specific results along two key areas: (1) research design and methods and (2) statistical techniques. Tables 1 and 2 summarize these results as well as provide basic comparative information between FBR and non-FBR journals.
Methodological Design Percentage Use Indices (PUIs) and Correlations With Year.
p < .10. *p < .05. **p < .01. ***p < .001.
Sample and Scope Percentage Use Indices (PUIs) and Correlations With Year.
p < .10. *p < .05. **p < .01. ***p < .001.
Research Design and Method
With respect to design and methods, we paid particular attention to article type, sampling, level of analysis, and sources of data. Of the 465 articles, there were 339 quantitative, 90 qualitative, and 36 mixed methods (incorporating a combination of quantitative and qualitative approaches) articles. As demonstrated in Table 1, FBR efforts indicate a substantial amount of growth taking place in the area of archival sources of data (r = .74; p < .001) and to a lesser extent, multiple sources of data (r = .31; p < .10). Of interest, while qualitative studies did not show a significant increase over time, they account for nearly a quarter (24%) of the empirical work published in FBR during our sample timeframe.
The large majority of FBR empirical studies (85%) examined just one level of analysis, as opposed to multiple levels simultaneously. However, we note that scholars have apparently heeded the call from Sharma (2004) to direct more attention to the organizational level of analysis; organizational-level studies have shown a dramatic increase relative to other levels of analysis in both FBR (r = .45; p < .01) and the comparative sample (r = .45; p < .05). Multilevel studies show a rise in utilization in the non-FBR sample (r = .58; p < .01), and to a lesser degree in the FBR sample (r = .38; p < .05). Furthermore, we found that cross-sectional designs dominate both samples (FBR: 74%; non-FBR: 55%), although the presence of longitudinal designs is on the rise for FBR (r = .55; p < .01). Indeed, our results show an increase in percentage use of longitudinal designs from 14% (1988-1996) to 34% (2006-2014) for FBR, whereas the non-FBR sample showed no significant change.
Table 2 shows that notable trends exist with regard to sampling as well. For FBR, we specifically see significant advancements in the use of international (i.e., non-U.S.; r = .67; p < .001) and multiple industry (r = .50; p < .01) samples. In terms of international studies, since 2006, FBR has published studies covering Europe (46), Asia (18), and Australia (9), with just seven studies sampling more than one nation (out of 98 empirical articles reporting geographic sample specifics during this time). Moreover, two of these seven multination studies used European samples. Across both samples, the overall growth of international samples has been specifically driven by increases in studies using European (FBR, r = .60; p < .001; non-FBR, r = .47; p < .05), Asian (FBR, r = .45; p < .05; non-FBR, r = .51; p < .01), African (FBR, r = .35; p < .10; non-FBR, r = .53; p < .01), and Australian (FBR, r = .42; p < .05; non-FBR, r = .32; p < .10) samples. By time period, in FBR we see the use of European samples especially popular, growing from 16% (1988-1996), to 35% (1997-2005), and 37% (2006-2014); this is comparable to the non-FBR sample showing growth from 13%, to 30%, to 34% in the same time periods.
Analytical Techniques
Given the importance and distinct presence of both qualitative and quantitative approaches in the study of family businesses, we discuss these separately. Table 3 provides a summary of findings for the quantitative techniques, while Table 4 provides information regarding qualitative techniques.
Quantitative Data Analytic Technique Percentage Use Indices (PUIs) and Correlations With Year.
Note. FBR = Family Business Review; ANOVA = analysis of variance; ANCOVA = analysis of covariance; MANOVA = multivariate analysis of variance; MANCOVA = multivariate analysis of covariance.
p < .10. *p < .05. **p < .01. ***p < .001.
Qualitative Data Analytic Technique Percentage Use Indices (PUIs) and Correlations With Year.
Note. FBR = Family Business Review.
p < .10. *p < .05. **p < .01. ***p < .001.
With respect to quantitative empirical studies in FBR, there was a distinct decrease in the use of simple descriptive statistics (e.g., means and standard deviations) to test hypotheses and research questions over time (r = −.65; p < .001). As expected, increases were observed in more advanced techniques over time, with hierarchical regression (r = .66; p < .001), panel data analysis (r = .51; p < .01), multinomial logistic regression (r = .48; p < .01), and structural equation modeling (SEM; r = .55; p < .01) showing the strongest upward trends in FBR. However, multiple regression analyses appear to be commonly used in the FBR and non-FBR samples, with overall PUIs at 13% and 40%, respectively. Panel data analysis had the largest increase for non-FBR articles (r = .80; p < .001), with multinomial logistic regression the only other technique showing a statistically significant increase (r = .39; p < .05). By and large, the trends noted in our sample of FBR articles mirror those of the non-FBR articles. However, exploratory and confirmatory factor analysis trends are both significant and positive (r = .38; p < .05; r = .46; p < .05) in FBR, which may reflect efforts to develop and validate new constructs.
Qualitative studies continue to represent a strong and important part of family business research, contributing to “rich, intriguing stories about family dynamics and the intersection of these dynamics with the operation of a business” (Reay, 2014, p. 100). Overall, the popularity of qualitative designs declined in FBR from 1988-1996 (25%) and 2006-2014 (19%); a more dramatic decrease was observed in the 1988-1996 (38%) to 2006-2014 (6%) periods for non-FBR journals. With the overall decline in the relative number of qualitative studies, we note a significant decline in the use of standardized interviews (r = −.48; p < .01), but an increase in the use of deductive application of qualitative methods (r = .38; p < .05) for studies in FBR. No statistically significant relationships were noted for non-FBR articles over time, a finding that may reflect more efforts among these journals to test existing theory, as opposed to develop new theory. However, the utilization of qualitative techniques appears relatively stable over time when accounting for newer or less well-established journals (Reay & Zhang, 2014).
Discussion
FBR—along with family business research in general—has made great progress over time in terms of notoriety and influence (Litz et al., 2012). What began as a conduit for the documentation and dissemination of ideas related to family businesses is now widely recognized as the “journal of choice” for family business scholars (e.g., Sharma, 2009, p. 8). As evidenced by our results, the mounting sophistication and complexity of the empirics used in family business research is an important element that contributes to the overall growth and legitimacy of the field (Sharma et al., 2012). Indeed, since empirics influence knowledge accumulation over time (e.g., Sackett & Larson, 1990; Schriesheim, Powers, Scandura, Gardiner, & Lankau, 1993), we expect that family business research will become increasingly more sophisticated and complex in the future, drawing ever closer to older and more established domains. So, in light of past efforts, we now discuss our results with an eye to the future. We do so in two ways. First, we discuss some key challenges, along with some solutions and exemplary articles, which seem to generally beleaguer the field; these are summarized in Table 5. Then, we more specifically discuss empirical efforts within key topical areas, while also making note of some opportunities for future research; Table 6 contains a summary of the more refined suggestions arranged according to the six most commonly researched topics in our review. Overall, we wish to advocate for more criticality and skillful judgment of existing literature, such that we might depart from some of the more normative perspectives that dictate family business research, and consider new research questions, especially given the availability of more advanced methodologies and analytical techniques.
Key Challenges and Suggestions for Family Business Empirical Research.
Selected Empirical Research Opportunities Involving Family Business.
Research Design and Methods
As noted in the “Results” section, the presence of more empirical studies has significantly increased over time. We find this to be encouraging because it suggests that the field of family business has advanced toward more construct development and theory testing. While theory building is necessary and desirable (e.g., Chua, Chrisman, & Steier, 2003), the field does appear to be building consensus around some key theories (e.g., agency theory, socioemotional wealth, stakeholder theory) and topics (e.g., succession, governance) that serve as a foundation for the field (Litz et al., 2012). While we did not find any significant change in qualitative studies over time in FBR, we do wish to note that qualitative studies represent a key method for researchers to not only answer important research questions but also to develop new questions (Reay & Zhang, 2014). Hence, we wish to encourage family business researchers to continue to conduct qualitative research studies, particularly those, such as ethnographies, that help us better understand the nuances of the family (e.g., kinship) and associated relationships within the business (Stewart, 2014).
In her editorial, Reay (2014) notes that qualitative articles can vary widely, and offers several suggestions for publishing high-quality qualitative research. Her suggestions include (1) ensuring access to sufficient, high-quality data, (2) setting up an appropriate research question to guide the article, (3) grounding the study in the relevant literature, (4) explaining the methods and showing your work, (5) telling an intriguing empirical story, (6) telling a convincing theoretical story, and (7) showing a clear contribution to the family business literature. We support these suggestions, and also note that these same considerations should be applied to quantitative research as well. Furthermore, we also encourage more mixed methods approaches as they help “provide the triangulation basis for convergence” (McGrath & Brindberg, 1984, p. 116). For example, Björnberg and Nicholson (2012) use strong qualitative methods, along with a subsequent quantitative analysis, to establish an intriguing story regarding the role of emotional ownership in the relationship between the next generation and the family firm.
Archival data, which include information obtained from documents such as 10K reports, analyst reports, directories, and memos, have increasingly been used in family business research. While archival sources are often inexpensive, provide oft-needed power, offer more generalizability, and allow for temporal application, these sources present key challenges as well. The most notable problems are associated with reliability, because of missing or inaccurate data, and internal validity, where available information does not exactly match up to the desired construct. For instance, family business scholars often lament the common and ongoing difficulty associated with identifying and measuring family businesses (Chua, Chrisman, & Sharma, 1999; Westhead & Cowling, 1998). Even with a clear definition to guide efforts, such as
a business governed and/or managed with the intention to shape and pursue the vision of the business held by a dominant coalition controlled by members of the same family or a small number of families in a manner that is potentially sustainable across generations of the family or families (Chua et al., 1999, p. 25)
there are numerous ways to operationalize the definition. Indeed, the real difficulty lies in measuring the various intangible attributes that are associated with a given definition—including intention, dominant coalition, and vision—and serve as important distinguishing factors that separate family businesses from nonfamily businesses.
There is a growing recognition that family businesses are heterogeneous (Naldi, Nordqvist, Sjöberg, & Wiklund, 2007) and that variance within family firms merits more refined examination (Astrachan & Shanker, 2003). Indeed, Chua, Chrisman, Steier, and Rau (2012) note that a greater focus on within-group differences of family businesses and away from simply comparing family businesses and nonfamily businesses is a logical and necessary step for the advancement for the field. The use of continuous variables instead of binomial variables will aid studies that focus on within-group differences of family businesses. For example, development of the Family Influence on Power, Experience and Culture Scale takes a strong step in the right direction (S. B. Klein, Astrachan, & Smyrnios, 2005). Furthermore, Holt, Rutherford, and Kuratko’s (2010) efforts to develop the Family Influence on Power, Experience and Culture Scale through replication and extension, including tests of convergent validity, exemplifies the type of scale development that is much needed in family business research. Therefore, following Pearson, Holt, and Carr’s (2014) arguments, we encourage more efforts to attend to issues of construct validity and reliability in family business research, but particularly in terms of how the family and its influences are accounted for across a highly heterogeneous group of firms.
The positive trend of archival data use is partially driven by the difficulties associated with gathering primary data about family businesses, which are often privately held and resistant to providing confidential information about the business (Neubauer & Lank, 1998). As a field, since we want to avoid overreliance on any one type of sample (e.g., large, publicly traded companies), it seems that more creative approaches to data gathering and methods are needed. As an example of innovative ways to assess information, we acknowledge McKenny, Short, Zachary, and Payne’s (2011) utilization of content analysis of organizational narratives to examine privately held family firms; specifically, they gathered and analyzed 77 “about us” website texts and 163 press releases from 93 companies to assess company goals. As another creative example, Herrero (2011, p. 893) combined individual-/family-, firm- (i.e., fishing vessel), and census-level data to examine the effects of agency costs on efficiency in a sample of “fishing firms” over a 9-year period. Overall, in an effort to reduce the possibility of inappropriate or inaccurate implications becoming generally accepted by the field, a variety of data sources and methodologies are ideal.
FBR studies using longitudinal designs have increased, which is encouraging since longitudinal approaches aid in the testing of causal relationships between constructs and variables (Litz et al., 2012). For instance, Craig and Moores (2006) used surveys, delivered at 10-year intervals, to investigate the relationship between shifting leadership and innovation in family firms; they noted that more protracted time periods between initial and follow-up surveys are required to examine changes in innovation. Despite increases in longitudinal approaches, cross-sectional designs remain an important and useful aspect in family business research. Brockhaus (2004) asserted that cross-sectional studies continue to dominate the field because of three key reasons. First, it is already difficult to get family businesses to participate in any study, let alone studies that require responses over multiple time periods. Second, family businesses often fail, making it difficult to obtain metrics that are not subject to severe survival bias. Third, the “publish or perish” environment encourages researchers to conduct studies that can be published quickly (i.e., cross-sectional) in lieu of studies that are more complex and take a longer period to conduct (i.e., longitudinal). Given these understandable concerns, we remain adamant that efforts converging on more longitudinal studies are necessary for the field to continue to progress toward a more thorough understanding of family business. Furthermore, as discussed in more detail below, we advocate for inquiries that account for time explicitly (Sharma, Salvato, & Reay, 2014).
One noteworthy finding involves the lack of FBR studies examining multiple levels of analysis in ways that partition the effects of each level. In FBR, while we observe a significant increase in these studies over time (r = .38; p < .05), we also observe this same trend in the comparative sample (r = .58; p < .01). Given the inherent nested nature of organizations—individuals within groups (i.e., the family), groups within organizations, and so forth (Aguinis, Dalton, Bosco, Pierce, & Dalton, 2011; Hitt, Beamish, Jackson, & Mathieu, 2007) —multilevel theory development and testing is a particularly fruitful area to explore (McKenny et al., 2014). Furthermore, scholars have called for more multilevel testing for some time (e.g., Chrisman et al., 2007; K. J. Klein, Tosi, & Cannella, 1999). We did note, however, some cross-level research (which may be considered a type of multilevel approach), where variables or constructs at one level are examined in relation to variables or constructs at a different level (Kozlowski & Klein, 2000). For instance, Huybrechts, Voordeckers, and Lybaert (2012) studied the influence of CEO tenure (individual level) on the family versus nonfamily levels of entrepreneurial risk taking (organizational level). While such research may contribute greatly to our understanding of family businesses, such studies must be approached carefully, both theoretically and methodologically. In particular, misspecification—misalignment between conceptualization and measurement—is a key problematic issue in mixed-level studies where measurement at one level is used to represent a construct conceptualized at a different level (Kozlowski & Klein, 2000).
In both FBR and non-FBR articles reviewed here, a positive sampling trend is the significant increase in studies using non-U.S.-only (i.e., international) samples; increased sampling diversity is essential to developing generalizability and establishing global applicability. For example, we advocate more studies like Micelotta and Raynard (2011), which examined corporate brand identity strategies using websites of 92 of the world’s oldest family firms covering 16 nations. However, the sampling trends described in our results suggest that scholars tend to sample from single countries, as opposed to multiple countries. In fact, we noted just 22 FBR studies (only 7% of the sampled articles) used samples from more than one nation. Additionally, research efforts need to extend to less developed economies. Currently, the large majority of research is based on highly developed and modernized economies (e.g., Australia, Germany, Italy, Spain, the United States). For an exception, see Fahed-Sreih and Djoundourian’s (2006) exploratory study of the determinants of longevity and success in Lebanese family businesses.
Whereas advancements are being made in terms of international samples, more attention to cross-border and internationally comparative issues would be preferred. Specifically, we argue that more critical consideration must be given to how studies contribute to our overall understanding of family business in general—apart from basic contextual differences. We are encouraged, though, by large-scale research efforts such as the STEP (Successful Transgenerational Entrepreneurship Practices) project, which is a collaboration of affiliated researchers to explore the entrepreneurial process in family businesses on a global basis. Likewise, Gupta and Levenburg (2010), through employment of nine cross-cultural dimensions of family business from the CASE (Culturally Sensitive Assessment Systems and Education) project, provided an exemplary study in their exploration of cross-cultural variations in family businesses. Through such efforts, we can not only better understand family businesses in general but also more fully address the contingencies associated with the various family business phenomena in question.
Analytical Techniques
Language ambiguity was a problematic issue with the reporting of statistical techniques; it was often difficult to determine the exact nature of the analysis procedure. As a basic example, several studies made explicit claims that regression was used to test hypotheses but on closer examination of the results, more specific types of regression were actually used (e.g., hierarchical, stepwise). Though this example is a seemingly minor concern, it exemplifies the overall lack of precision for many of the studies, particularly with regard to statistical procedures. Accordingly, as a simple suggestion for future research, more effort in terms of clarity and completeness is needed in the presentation of information. Indeed, as statistical sophistication and complexity increases, the fundamental need for clarity is all the more important because the ability to fully understand, and even replicate, empirics is necessary for the field to progress.
While FBR has made substantial strides in its application of more advanced statistical techniques, such as hierarchical regression, SEM, and panel analysis, there were important techniques that, while underrepresented, are important for discussion due to their potential utility. First, generalized method-of-moments (GMM) was used in several studies, primarily as a robustness check. GMM is a statistical modeling procedure that allows researchers to address endogeneity emerging from multiple sources, such as reverse causality and unobservable heterogeneity (Wintoki, Linck, & Netter, 2012). Specifically, GMM “uses internal instruments . . . that are based on lagged values of the explanatory variables that may present problems of endogeneity” (Sacristan-Navarro, Gomez-Anson, & Cabeza-Garcia, 2011, p. 82). A similar technique typically used to address endogeneity concerns, two-stage least squares, was used sparingly in FBR (e.g., Steijvers & Voordeckers, 2009), whereas two-stage least squares appeared in eight non-FBR studies (out of 146 articles). For example, Eddleston, Kellermanns, Floyd, Crittenden, and Crittenden (2013) used two-stage least squares because of the possibility of reverse causality between two types of planning (strategic and succession) and firm growth. Finally, Bayesian analysis offers the ability to account for family firm heterogeneity while using small and skewed samples (Block, Miller, & Wagner, 2014).
Another underrepresented, yet potentially useful, technique was meta-analysis. In our sample of 465 family business empirical articles, we only identified one meta-analytic study. Meta-analysis is useful in that it integrates conflicting statistical findings across studies “to reveal the simpler patterns of relationships that underlie research literatures, thus providing a basis for theory development” (Hunter & Schmidt, 2004, p. 17). Additionally, the prevalence of meta-analytic studies can be a signal that a field has achieved a greater degree of maturity and growing consensus with respect to key definitions, the operationalization of key constructs, and its nomological network (Cogliser & Brigham, 2004; Reichers & Schneider, 1990). The meta-analytic study, O’Boyle, Pollack, and Rutherford (2012), serves as a strong example of how meta-analysis might be used in family business research. 2 Given that family business research has seen a large number of studies on a common phenomenon, we believe that there are several opportunities to use meta-analysis techniques to sort out conflicting or ambiguous findings. Furthermore, we would expect the number of meta-analytic studies to increase as the family business field matures.
Earlier, we noted that studies conducted under multiple levels of analysis are lacking. From a statistical standpoint, software packages (e.g., SAS, STATA, R) with the ability to analyze multiple levels of analysis are more readily available and more easily used than ever before. Hence, we encourage scholars to consider what research questions might now be asked that up until recently could not be answered because appropriate techniques were not available. Indeed, random coefficient modeling, also referred to as mixed-effects modeling or hierarchical linear modeling, is a particularly useful technique because it allows researchers to understand how relationships might differ across groupings, but does not rely on the same assumptions as traditional linear approaches (Bliese, 2002). One exemplar of multilevel modeling applied to the family firm context is Dawson (2011), which analyzed private equity investment decisions, controlling for the nested nature of data decisions within private equity professionals, who in turn are nested in private equity firms.
Finally, while time-related variables such as long-term orientation (Brigham, Lumpkin, Payne, & Zachary, 2014) have been introduced to the literature, we identified few studies that explicitly accounted for time. As stated by Sharma et al. (2014, p. 10), such studies “are exceptions rather than the norm.” Hence, time, as a level of analysis, is a key component in numerous family business phenomena and should be included in more empirical analysis (e.g., Sharma et al., 2014). Indeed, the complex nature of time and its effects on organizations has led to the development of a variety of empirical approaches including hazard modeling (Bronnenberg & Mela, 2004), growth modeling (Bliese & Ployhart, 2002), and discontinuous change modeling (Singer & Willett, 2003). Taken together, time represents part of a new empirical frontier, which has great potential to inform the field of family business, enabling researchers to address novel and more complex research questions.
Empirical Issues and Opportunities by Topic
While we have discussed more general empirical trends and challenges associated with family business research, we recognize the variety of topics in family business research and how each area may differ in their employment and needs with regard to empirics (Sharma et al., 2012). As previously mentioned and summarized in Table 6, the six most prominent topics in order, were management of the firm, business performance and growth, characteristics and attributes, interpersonal family dynamics, governance, and succession. While many other topics exist, we limit our discussion to this group for parsimony purposes and because there was a substantial decrease in popularity to the next topic (family firms in international contexts). However, it is remarkable that these groupings closely resemble the clusters and numerical taxonomy developed by Yu et al. (2012), which was based on categorized outcome variables. Overall, our goal in this section is not to fully review the hundreds of studies associated with these topics, but to briefly point toward some future directions with regard to empirics. For a comprehensive examination of many of the most prominent topics in family business research, see Melin, Nordqvist, and Sharma’s (2014) edited book.
Management of the Firm
Articles classified under this topic categorization—following Kelly, Athanassiou, and Crittenden (2000) —include studies on strategic processes (i.e., the steps that firms use to assess or reassess organizational missions, goals, external environment, available resources, and commitment to their strategic vision), content (i.e., an organization’s strategic orientation and decisions made about explicit actions across organizational, competitive, and functional contexts), and implementation (i.e., how decisions are executed and evaluated). Accordingly, the management of the firm topic appeared 185 times (40%) in our sample—the most common of any topic. Management of the firm drew mostly from agency (22%) and family (12%) theories, followed by the resource-based view (7%), socioemotional wealth (SEW), and stewardship theories (6% each). Given SEW’s recent emergence, along with its ability to depict “the uniqueness of the family firms’ identity through the consideration of noneconomic factors” (Berrone, Cruz, & Gomez-Mejia, 2012, p. 269) we expect its overall use to increase significantly in the coming years. General linear models (e.g., analysis of variance [ANOVA], regression analysis) appeared commonly (44%) in these particular articles, followed by utilization of basic statistics (i.e., descriptive statistics and correlations; 24%) and test of mean differences (21%). Notably, the use of general linear models has increased from 23% in the 1988-1996 period, to 63% over the 2006-2014 period.
The large majority of the articles within this topical area examine management differences, in terms of content, process, and implementation, in a comparative way (i.e., testing for differences between family and nonfamily firms). For example, Daily and Dollinger (1992) found several structural, process, and strategic differences—assessed using Miles and Snow’s (1978) typology—between firms managed by family and nonfamily members. Likewise, in their study on strategic site location decisions, Kahn and Henderson (1992) found that the imprint of family systems led family firms to be more concerned with proximity to residence, whereas nonfamily firms seek locations that minimize facilities and employment cost.
While comparative studies regarding the management of family versus nonfamily businesses have remained consistently popular, they have also seen increased empirical rigor and complexity. For instance, though no difference was supported, Gudmundson, Hartman, and Tower (1999) used multifactor ANOVAs to test the relationship between organizational variables (i.e., family business status, market type, and product type) and strategic orientations of family and nonfamily firms. Additionally, after using cluster analysis to group features of the respondents and their businesses, Littunen (2003) used one-way ANOVA and logistic regression to compare strategic factors and style of management between 134 Finnish family and nonfamily firms. A more recent exemplar is found in Chrisman and Patel’s (2012) comparative study of R&D investment strategies, which examined panel data on 964 manufacturing family and nonfamily firms using fixed-effects Tobit regression.
Given the breadth and scope of this topic, we are generally encouraged by the extent of rigor being used in more recent studies. However, some empirical gaps remain. In particular, more work is needed to reveal the mechanisms through which the family influences the business such that valuable resources and capabilities are developed (e.g., Habbershon & Williams, 1999; Habbershon, Williams, & MacMillan, 2003). Indeed, because the traditional delineation between strategic content and process tends to blur in family firms (Salvato & Corbetta, 2013), this topic is necessarily intertwined and broad. Within this purview, however, strategic decision-making has received significant attention and serves as a good point of discussion with regard to empirical issues.
While various methodologies have been used to examine decision making, the most common in family business research are surveys (e.g., Romano, Tanewski, & Smyrnios, 2001). Although surveys have significantly advanced family business research, they (along with case studies) involve post hoc data collection, which often results in recall bias and revisionism (Golden, 1992). However, archival data are generally limited in their ability to capture decision-making processes and tie them to specific outcomes, and experiments often lack context specificity, which is important to family businesses in particular. Overall, it seems particularly difficult to partial out the decision processes from the content and the context of the decision for use in understanding family business decision making. As an exemplar of a study that overcomes many challenges, we note Koropp, Kellermanns, Grichnik, and Stanley’s (2014) study that uses a two-part, time-lagged survey design and SEM to analyze the impact of motivational elements (i.e., perception of family norms, attitudes, and perceptions of behavioral control) on the financial decision-making process in family firms.
The challenges associated with studying decision making in family businesses suggest multiple opportunities for future research. For instance, we identified no studies in our sample using causal mapping and just two articles using conjoint analysis. The exceptions include Dawson (2011) and Shepherd and Zacharakis’s (2000) investigation of decision making by potential family business leaders under different succession structures. Overall, conjoint studies, following more extensive use in the entrepreneurship literature (e.g., Drover, Wood, & Payne, 2014), should be given more consideration because of their ability to avoid the recall bias and post hoc rationalization commonly associated with surveys (Lohrke, Holloway, & Woolley, 2010; Shepherd & Zacharakis, 1999). Likewise, we suggest future research might leverage recent interest in conceptual frameworks developed within family psychology (von Schlippe & Schneewind, 2014) to capitalize on cognitive mapping’s ability to generate representations of various cognitive structures used by family and nonfamily members in strategizing activities (e.g., Salvato & Corbetta, 2013).
We also noted a lack of decision-making studies conducted in laboratory settings. While experiments in laboratory settings often receive criticism for not replicating the true thought processes of executives (Maule & Hodgkinson, 2002), there are many opportunities to tease out how family considerations influence business decisions. For example, researchers could use experiments to investigate the concept of SEW (e.g., Berrone et al., 2012; Zellweger, Kellermanns, Chrisman, & Chua, 2012) to frame decisions as a choice between alternatives that would either (1) diminish economic risk but lead to a loss of SEW or (2) protect the family’s SEW.
Finally, Sharma (2004) alluded that a true understanding of strategic processes in family firms must consider different levels of analysis. Indeed, “familiness” can manifest as a “source of strategic competence (distinctive) or encumbrance (constrictive)” in the management of family firms (Sharma, 2004, p. 21). Given this, and the innate multilevel nature of individual perspectives and group social processes through which family influence is imposed on organizational strategy decisions (e.g., McKenny et al., 2014), we think studies related to family business management could benefit from more robust use of appropriate techniques to test multilevel models. Throughout this review and despite research designs advanced by scholars that commonly measured variables of interest across multiple levels, multilevel effects were seldom addressed conceptually or empirically. Indeed, this trend has emerged in the face of increased numbers of techniques that allow for the application of multilevel principles and data (Bliese, Chan, & Ployhart, 2007; Payne, Moore, Griffis, & Autry, 2011). One notable exception that has used these more rigorous analysis tools is Dawson’s (2011) use of hierarchical linear modeling to analyze the nested nature of private equity decisions in family firms.
Business Performance and Growth
Generally viewed as how organizations perform from financial and nonfinancial perspectives, business performance and growth was the second most prevalent topic in our sample, appearing 180 times. Within this category, studies relied primarily on agency (27%), family (e.g., lifecycle and systems theories; 13%), and resource-based perspectives (11%). General linear modeling techniques dominated our sample as the most common method of data analysis (53%). Somewhat expected, our review indicates that early performance literature was characterized by a desire to establish if family firms do, in fact, perform differently than their nonfamily counterparts (e.g., Donckels & Frohlich, 1991; McCollom, 1988). More nuanced investigations of family firm performance and growth appeared as our review progressed. Generally, research on family firm performance has been mixed with some studies finding that family firms outperform (e.g., Anderson & Reeb, 2003; Villalonga & Amit, 2006), underperform (e.g., Claessens, Djankov, Fan, & Lang, 2002; Morck & Yeung, 2004), or do not differ from the performance of their nonfamily equivalents (Miller, Le Breton-Miller, Lester, & Cannella, 2007). Indeed, O’Boyle et al. (2012) advanced the field toward a potential resolution of the debate focused on family versus nonfamily financial performance. Their meta-analysis of 78 articles found that family involvement did not significantly affect firms’ financial performance; nonsignificance was also illustrated with potential moderating influences of multiple conceptual and methodologically based variables.
Given this general equivocality of the findings regarding financial performance, there are two key areas for future research with respect to empirics. First, there is the need for additional nonfinancial outcome measures that can help in better understanding how family firms differ from nonfamily firms and each other (O’Boyle et al., 2012). For instance, Dyer and Whetten (2006) demonstrate that family businesses, because of their image and reputation concerns, act more socially responsible than nonfamily businesses. However, as detailed in Table 6, along with the development of new measures comes the need to consider construct validity and reliability (e.g., Brigham et al., 2014). Future opportunities for the development of scales can be pursued in research in the areas of business sustainability (e.g., Olson et al., 2003), social responsibility (e.g., Marques, Presas, & Simon, 2014), organizational virtue (e.g., Payne, Brigham, Broberg, Moss, & Short, 2011), or familiness (e.g., Rutherford, Kuratko, & Holt, 2008).
As a second area for future research within the business performance and growth topic, there is a need to more explicitly consider the role of time (Sharma et al., 2014). Although this topic has seen a greater presence of longitudinal data (e.g., Barbera & Hasso, 2013) and the use of panel data (e.g., Moss, Payne, & Moore, 2014), there was a noted absence of more advanced longitudinal techniques related to repeated measures and growth modeling. Furthermore, we noted a lack of attention to potential causality issues. In some cases, it was unclear whether or not performance and growth was a cause or an effect of family firm structures and processes, bringing endogeneity concerns into play. Kowalewski, Talavera, and Stetsyuk’s (2010) work serves as an exemplary article by using GMM to address the ownership-performance endogeneity issue. In sum, because of issues related to causality and the inherently longitudinal nature of family business growth and performance, we see abundant future empirical opportunities involving time and temporality.
Characteristics and Attributes
We classified 150 articles under the topic of characteristics and attributes, which refers to the distinguishing features or qualities of a person, group, or organization (Koiranen, 2002). Nearly 35% of these studies relied on general linear models, with an additional 27% using mean difference tests and 23% drawing on basic statistical techniques. Although family theories (19%) and agency theory (10%) were prominent in this part of our sample, we did note that 41% of these particular studies contained no apparent theoretical foundation or dominant perspective. This coincides with the descriptive nature of many of the early attribute studies that often hypothesized about who might be considered the most desirable successor in a family business. In fact, within this topic, we note a steep decline of studies with no theoretical foundation, declining from 61% in the 1988-1996 timeframe to just 11% in the 2006-2014 period.
Heeding original calls for confirmatory attribute research made by Wortman (1994) and Handler (1992), Chrisman, Chua, and Sharma (1998) used exploratory testing, including t tests and ANOVAs, among 485 family business managers to rate the relative perceived importance of successor attributes. In a similar vein, Smyrnios, Tanewski, and Romano (1998) identified a set of broad traits using factor analysis and the development of a reliable and valid measure of family business. More recent characteristic and attribute work is still seeking to resolve imperfect measurements of various family business constructs. For example, using a transaction-cost perspective, Royer, Simons, Boyd, and Rafferty (2008) developed and tested a theoretical model—using exploratory factor analysis and confirmatory factor analysis —to investigate whether the merits of an internal or external successor depends, in part, on the characteristics of the potential successor. More recently, De Massis, Chirico, Kotlar, and Naldi (2014) used hierarchical multiple regression in finding a cubical relationship between firm age and proactiveness in family businesses. Qualitative work has yielded findings within this topic as well; Lambrecht’s (2005) analysis of 10 case studies uses entrepreneurial characteristics and education, in part, to build theoretical explanations about why generational succession succeeds in one family while failing in another. Additionally, Vera and Dean (2005), through structured qualitative interviews, provide insights on gender issues from their study on how daughters experience the succession process.
As a line of inquiry, family business research on characteristics and attributes is in need of a reawakening, both conceptually and empirically. We optimistically note that the research questions proposed by scholars became more advanced over time, typically involving the influence of individual-level attributes on firm-level outcomes; the proliferation of testing this basic relationship was recently augmented by calls for deeper studies of family systems variables in order to “better understand why, when, and how its characteristics (and) attributes are likely to influence the behaviors and performance of family firms” (Sharma et al., 2012, p. 10). Indeed, the innately nested nature of characteristics and attributes (i.e., individual characteristics and attributes are nested in families, which are nested in organizations) is deserving of more scholarly attention, particularly since families can also impose their influence at other levels of analysis such as interorganizational (e.g., networks) or environmental (McKenny et al., 2014). Echoing similar calls in other domains (e.g., Hitt et al., 2007; Kozlowski & Klein, 2000), we suggest a need for family business scholars to advance family-related characteristics and attributes by explicitly accounting for the nested nature of these characteristics at various levels and how each level affects family business outcomes differently.
Interpersonal Family Dynamics
Interpersonal family dynamics research appeared 133 times (29%) in our review. Studies classified under this topic examine the interactions and involvement among family members, such as conflict and disagreement, that occur in the course of running a family business (Sonfield & Lussier, 2004), Of these, nearly half were published in the last 9-year period of our sampling timeframe (2006-2014), which demonstrates a growing scholarly interest in this already strong topic area. Researchers used family (22%) and agency (16%) theories most often. While the SEW perspective seems particularly relevant to interpersonal family dynamics (Berrone et al., 2012), it was used in just six articles that addressed this topic, with four recently appearing in 2014.
Within interpersonal family dynamics studies, the most common statistical technique used was general linear models (35%). We found the use of confirmatory factor analysis, exploratory factor analysis, and reliability analysis techniques in 16 articles in this category, which seems particularly appropriate because processes associated with family businesses are “largely unseen patterns created, sustained, and modified by the family” (Dyer & Dyer, 2009, p. 218). However, some scholars still seem content with single respondent surveys and simply summing or averaging items to derive a single variable. As a result, studies unnecessarily introduce measurement error into the equations. An exemplar is Khanin, Turel, and Mahto (2012), which uses SEM, based on data from multiple respondents, to study family business job satisfaction and turnover intentions. In this article, the authors viewed embeddedness as interorganizational connectedness, measuring it with three single-item measures of value congruence. We encourage more of these types of research efforts that account for the latent, difficult to measure variables involved with the dynamics of family businesses.
Our review also points to a potential direction for future research by exposing limitations in use of interdependence techniques (e.g., network analysis, diffusion models, multidimensional scaling [MDS]). Although we found a practical and compelling example of the use of network analysis (Debicki et al., 2009), this study was not geared toward family business, but rather family business research. Hence, there appears to be the need for broader application of network techniques within the family dynamics literature. Also, the use of MDS was limited. A recent exemplar by Morris, Allen, Kuratko, and Brannon (2010) introduced family business creation as a “lived through” experience composed of interdependent events. Although the overall study partitioned perceptions of experiences across more traditional groupings (i.e., founders, nonfamily managers, and founders of nonfamily businesses), the use of MDS offered deep insights by visually approximating patterns of similar structures among test subjects’ affective experiences.
On its own, the Morris et al. (2010) study also provides an ample guide for future research, especially as scholars seek to investigate the extended influence of interpersonal family dynamics. Prevailing research has widely studied the outcomes of relationships between immediate family members, but these interpersonal dynamics may play a larger role than originally thought. Perhaps, as Sharma (2004) suggests, individual family members and the relationships within one generation may act to bridge a family business with the next generation. In the same work, she calls for more systematic study of family business dynasties and transgenerational sustainability; the overarching point is that family dynamics transcend traditionally studied relationships between father and son, or mother and daughter. As such, and as outlined in Table 6, we cannot think of a more important topic in which to use qualitative techniques leading to a deeper analysis of interdependence in and among family businesses. Since techniques that account for interdependence will ultimately lead to a greater understanding of family dynamics, they should also serve as a springboard to interesting opportunities for future quantitative research, perhaps even across levels and time, that hopefully result in more predictive, functional conclusions within the topic.
Governance
Although Mustakallio, Autio, and Zahra (2002, p. 219) once cautioned that “empirical research into the governance systems of family firms is scarce,” our review indicates substantial progress on the topic. In fact, governance was a key topic in 132 articles, with 77 of those (58%) appearing in the last 9-year period (2006-2014). For all governance articles, more than half used some version of general linear modeling and 27% reported the use of longitudinal data. And while several articles use mean differences (e.g., t tests) and descriptive statistics (21% and 14%, respectively), there is a substantial portion (13%) of studies that applied discrete event methods (e.g., logistic regression, discriminant analysis). Finally, methodological techniques that account for heterogeneity and grouping (e.g., cluster analysis, hierarchical linear modeling) were found in 5% of the articles that addressed governance issues.
Whereas most articles in our sample seemingly used appropriate methodologies, the variety of approaches appears narrow. Although regression can serve as an effective analysis technique, it seems that family business scholars should branch out from more basic designs to capture the complexities of ownership and control exclusively found in family businesses—particularly since governance structures in these settings must reconcile shared visions within the family while mitigating harmful conflicts (e.g., Carney, 2005). Indeed, one of the most unique aspects of family businesses is the multiple roles that family members often play within the firm (e.g., Tagiuri & Davis, 1996); hence, this distinctiveness could be leveraged with methods that force scholars to ask and answer more insightful governance questions. For example, while Lungeanu and Ward (2012) considered grant-making practices of family and nonfamily foundations, their analysis used several ANOVA statistical tests between these entities. However, because foundations and boards are typically made up of a blend of unique (and often related) individuals across various levels of analysis, the inherently nested nature of their influence leaves techniques such as ANOVA unable to adequately account for this variance (e.g., McKenny et al., 2014). As such, a random coefficients modeling technique may be a better option for future research of this kind.
Though overall research efforts in family business governance have made significant gains in recent years (Yu et al., 2012), we further emphasize the call by Mustakallio et al. (2002, p. 219) that governance research must remain “grounded on the unique characteristics of family firms.” Our review finds that consistently adhering to this principle is challenging for family business scholars. Indeed, although family businesses are a highly heterogeneous organizational form (e.g., Sharma et al., 2012), many studies attempt to generate broad governance findings based on survey responses from an individual executive (e.g., Corbetta & Montemerlo, 1999). Recognizing the inherent difficulties associated with gathering data on governance, we suggest that instead of reporting findings derived from the responses of a single executive within the firm, future research could take a mixed method approach. For example, Hatum and Pettigrew (2004) triangulated archival data with other sources derived from qualitative methods. As a result, their findings better account for heterogeneity among the dominant coalition, leading to more holistic predictions of how governance and ownership are linked to firm growth and, ultimately, a deeper understanding of organizational flexibility. In sum, to better capture the heterogeneity of ownership and control processes within family businesses (e.g., McKenny et al., 2014), future governance-related research could benefit from the implementation of similar triangulation approaches, whereby multiple perspectives are used to verify and clarify meaning (Jick, 1979).
Succession
Succession, typically drawing from family theory perspectives, was the sixth most popular topic in our sample, appearing 113 times. About 34% of these studies relied on regression-based methods of data analysis. While many articles used more traditional regressions with appropriate assumptions (i.e., normal distributions, homoscedastic error terms), some studies implemented more nuanced regression techniques. For example, logistic regression techniques were used in studies with succession-related measures serving both as the dependent variable (e.g., Blumentritt, 2006; Schröder, Schmitt-Rodermund, & Arnaud, 2011) and the independent variable (e.g., Brun de Pontet, Wrosch, & Gagne, 2007). Following regression, some early studies approached family business succession using basic techniques (21%) including descriptive statistics (e.g., Chau, 1991) or simple correlations (e.g., Fahed-Sreih & Djoundourian, 2006). Also, structural equations modeling (e.g., Marshall et al., 2006) and MDS were commonly used (e.g., Garcia-Alvarez, Lopez-Sintas, & Gonzalvo, 2002) within the succession literature.
Because succession is a process ascribed with numerous activities taking place over a protracted time period (Handler & Kram, 1988; Ward, 1987), temporal factors are an inherently important consideration within this topic. In short, succession takes time (Le Breton-Miller, Miller, & Steier, 2004; Sharma, Chrisman, & Chua, 2003). However, while 19% of the data used were gathered over time (e.g., Gagne, Wrosch, & Brun de Pontet, 2011), or combining surveys with multiyear archival data (Minichilli, Nordqvist, Corbetta, & Amore, 2014), only two used analytical techniques designed for examining longitudinal data (e.g., panel data analysis, repeated measures, growth modeling). The first (Wiklund, Nordqvist, Hellerstedt, & Bird, 2013) combined multiple databases to generate a 4-year panel of 12,125 firm-year observations, while the second study (Molly, Laveren, & Deloof, 2010) examined the effect of generational transitions on family business performance and capital structure decisions using an unbalanced panel of more than 2,000 firm-year observations. As Molly et al. (2010, p. 139) suggest, “cross-sectional analysis does not allow controlling for all time-invariant characteristics that might have an impact on the firm’s financial structure or performance.” Moreover, longitudinal designs can “more accurately identify the pure succession outcomes in a company” (Molly et al., 2010, p. 139).
Ultimately, while longitudinal techniques are needed in all aspects of family business research (see Tables 5 and 6), it remains especially salient to succession research—it is not, in fact, “simply a single step of handing the baton; it is a multi-staged process that exists over time” (Handler, 1994, p. 134). Thus, we add our voices to an earlier call by De Massis et al. (2008) and challenge researchers to seek out data sets that can be analyzed using longitudinal methods; this ensures succession research advances in a finer grained manner, effectively ruling out explanations that are clouded by the disadvantages inherent in cross-sectional methods. More important, an emphasis on longitudinal designs will strengthen claims of causality within the literature, a challenging proposition for family business research without relying on random experiments or violating key statistical assumptions (e.g., Antonakis, Bendahan, Jacquart, & Lalive, 2010). Last, we noticed a rise in the number of qualitative studies focused on succession throughout our review. This is encouraging whereby it allows scholars to gain a deeper understanding of the “temporal influences and triggers” that are inherent to the succession process (Murray, 2003, p. 31), but should be further examined on a larger scale.
Conclusion
Litz et al. (2012) recently surveyed family business researchers, gathering their perceptions about the family business field of study. Two important issues, relevant to this review, pervaded the advances, opportunities, and challenges discussed. First, respondents commonly mentioned not only the validity and reputation of the field, noting its increase, but also the necessity of continuing to improve legitimacy relative to adjacent fields. The second issue commonly mentioned, which is closely associated with the first, was methods or empirics; respondents perceived not only substantial progress in empirics but also noted the need for further improvement. Building on these concerns regarding the field of family business, the purpose of our review was to document FBR’s empirical progress since the journal’s inception in 1988, juxtaposed with family business research published in other prestigious business journals. Specifically from the discussion above, we note general challenges, and offer up some suggestions, regarding (1) reliability and validity, (2) generalizability, (3) causality, (4) temporality, and (5) multilevel considerations. On the whole, however, we document great improvements in empirics over time, which has brought about significant advancement of the field and legitimacy.
Footnotes
Acknowledgements
We appreciate the helpful comments and suggestions of Keith H. Brigham and four anonymous reviewers on earlier versions of this article. We are also grateful to the special issue editorial team, Jeremy Short, Tom Lumpkin, Allison Pearson, and Pramodita Sharma for their guidance throughout the process. John Martin extends special thanks to the Department of Management at the U.S. Air Force Academy, most specifically Steve Green, who availed significant department resources for completion of this project.
Authors’ Note
Opinions, conclusions and recommendations expressed or implied within are solely those of the authors and do not necessarily represent the views of the U.S. Air Force Academy, U.S. Air Force, the Department of Defense, or any other government agency.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
Notes
Author Biographies
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
