top of page


All too often, even high performing teams, when experiencing high levels of stress, isolation, and overt personal alignment - often with an over-bearing leader or leadership group - can ignore critical information and lack objectivity.  The information blindness and bias caused can often have catastrophic impacts as decision-making becomes increasingly introspective, polarised and isolated.

Space shuttle Challenger disaster. 76 seconds into flight, reddish-brown cloud envelops th

1.1       Groupthink



Analysis of the “teamship” literature highlighted the presence of antecedents and symptoms of groupthink in the England Rugby squad of 2003 where the notion of “teamship” was identified.  Considering the lack of research in “teamship”, it was considered that examination of the groupthink literature may reveal specific antecedents that could help in understanding the concept.

Seminal Contribution

Groupthink theory examines decision-making failures in highly cohesive groups (Janis, 1972, 1983).  Janis (1983) defined a set of twenty-four variables that relate to the construct: eight antecedents, eight symptoms of defective decision-making, and eight symptoms of Groupthink (Figure 3.3.1).  The main proposition of Janis’ framework is that when a group is moderately or highly-cohesive, the presence of specific other antecedent conditions increases the chances of the development of groupthink symptoms (Janis, 1983).



Figure 3.3.1       Groupthink antecedents and symptoms (Modified from source: Janis 1983)


Temporal Developments

Groupthink has attracted scholarly attention over the last fifty years, with heightening interest in the concept mirroring the instances of scandals in both governmental and commercial settings.  My research identified four specific literature reviews on groupthink (Park, 1990; Neck & Moorhead, 1995; Esser, 1998; Rose, 2011).  Appendix F provides a summary of meta-analyses of these studies. 

Park (1990) identified sixteen empirical studies, which show only partial testing of Janis’ framework, resulting in only partial support of the theory.  Eight of these were case-studies, seven experimental and one content-analysis.  Park found that despite the intentions of the researchers to examine all eight of Janis’ antecedents, methodological design of the various studies resulted in only four of them being satisfactorily assessed.  Of these, ‘group cohesiveness’ – emphasised by Janis above all other antecedents – was concluded to have less impact than other antecedents.  Park observed that this may be a direct result of the challenges of producing or interpreting cohesiveness from experiment or case study analysis respectively (Flowers, 1977; Courtright, 1978).  He also notes that the effect of leadership style was examined in five of the studies, but the conclusions in support of Janis’ hypothesis were split (Fodor and Smith, 1982; Leana, 1985; Moorhead and Montanari, 1986) suggesting more empirical research is required.  However, it is also notable that none of the empirical studies attempted to define group cohesiveness, nor determine how it could be measured.  There is a lack of examination of the factors that affect its presence in groups, nor of an understanding or exploration of whether it is a group process or emergent state, or whether it is constructed at an individual or group level.  Janis’ assertion of the high importance associated with group cohesiveness in groupthink is dismissed by Park in his analysis of these previous studies.


Neck and Moorhead (1995) proposed enhancements to Janis’ original framework, based on their own literature analysis.  They reviewed fifteen research papers focused on groupthink, differentiating between case study and experimental methodologies.  They note “…the scarcity of research examining its [groupthink’s] propositions is startling…” (Neck and Moorhead, 1995:538).  Their analysis identified that two additional antecedent conditions should be added to Janis’ framework, the high consequential implications of the outcome of the group’s activities, and significant time constraints on the group.  The antecedents of groupthink - other than cohesiveness - are identified empirically from the source studies, as well as symptoms of both groupthink and defective decision-making.  However, cohesiveness can only be assumed in the historical analyses, and only task cohesion can be reliably catalysed in laboratory conditions.  Their analysis did not find support in either case or empirical studies for group cohesiveness as an antecedent to groupthink, but they nevertheless include it as the primary antecedent in their revised conceptual framework of groupthink.   This challenges the trustworthiness of the conclusions regarding the importance of group cohesiveness in this meta-analysis.

Esser’s (1998) review of theoretical and empirical research also separates the literature into two categories (case and empirical studies).  A total of twenty-eight studies were analysed, identifying seventeen as case, and eleven as experimental.  His analysis of literature highlights the independent and dependent variables of each study, the results associated, along with comprehensive narrative interpretation of his findings.  Esser’s conclusions highlight the prevalence of analysis and re-analysis of the same original five case studies from Janis (1972), and the limitations of laboratory studies to reproduce all the conditions of the original twenty-four variables of groupthink.  Additionally, he notes that there has been no adoption of a consistent experimental paradigm.  The qualitative papers analysed were all historical, and - as pointed out by Esser - lack the ability to reflect on the private feelings or beliefs of the participants.  Esser posits that the inability to capture primary data on such subjective issues as how and why people are behaving in certain ways is a consistent weakness in all groupthink empirical studies; “…most of the symptoms of groupthink cannot be assessed easily by an outside observer.  Rather, most groupthink symptoms represent (private) feelings or beliefs held by the group members or behaviours performed in private.  Therefore, the most appropriate way to measure these symptoms is to ask the group members.  Only two studies have used questionnaires to assess the full set of eight symptoms of groupthink...” (Esser, 1998:136–137).  His call for more original empirical research echoes those from Park (1990), and Neck and Moorhead (1995).

Extending from his 1990 literature review, Park (2000) looked to address the limitations of previous experimental studies into groupthink with an ambitious experiment including 256 participants, split into sixty-four groups.  He posed two key questions: (a) Is Janis’ theoretical model correct; and (b) What is the contribution of each symptom?  His conclusions contradicted those of previous experimental research in that he did find clear evidence to support the presence of group cohesiveness when groupthink occurs.  

Groupthink is also explored across a range of settings, investigating how it applies to autonomous work groups (Manz and Sims, 1982) and self-managing teams (Moorhead, Neck and West, 1998; Flippen, 1999).  Schafer & Crichlow’s multiple case comparative analysis (Schafer and Crichlow, 1996) looked to answer the specific research question “Do certain antecedent conditions give rise to defective decision-making during times of crisis?” (p.417).  They reviewed the nineteen case study analyses used in the original development of groupthink theory and explored the presence of Janis’ antecedent conditions.  Their conclusions support Janis’ framework, but they have omitted - without explanation - group cohesiveness as a key antecedent.  This oversight undermines an otherwise comprehensive and well-executed research design.

Rose  (2011) provides the most comprehensive review of groupthink literature.  He examined over sixty scholarly articles since 1972.  His analysis provides clearer categorisation of the research, identifying not only the split between case studies, experimental studies, theoretical contributions and literature reviews, but also breaking down the results by whether the research paper was looking at isolated decisions, or multiple decisions, and whether the studies were attempting to examine all the groupthink variables, or specific subsets.  He debates the argument for and against support for Janis’ theory, as well as its application and potential application in other environments.  Rose also incorporates some of the suggested proposals for the constructive development of groupthink from the likes of Hart (1998) and Mohamed and Weibe (1996), who posited that “…the nature of the theory is still unclear.  This ambiguity represents a major barrier to theory testing” (p.417).  He concludes with two clear recommendations; (a) that groupthink theory should be redefined based on research to date, and (b) it needs to be clear whether groupthink is “…a process model or a risk mitigation approach.” (p.51).  A comparative summary of these four meta-analyses is provided in Appendix F.

Current Thinking

More recent research examines the presence of groupthink in a range of other settings, from failures in the airline, banking, and energy industries (Hermann and Rammel, 2010; R. R. Sims and Sauser, 2013) to the 1996 disaster on Everest (Burnette, Pollack and Forsyth, 2011), to its impact in educational settings (Brady, 2015).  Breitsohl, Wilcox-Jones and Harris (2015) examined the behaviours of online communities in regard to customer conformity, employing a survey method to 343 participants, concluding that group cohesiveness was present along with groupthink.  This is an interesting finding considering the lack of interpersonal relationships, shared purpose or group identity.  

The empirical studies provide contradicting conclusions, which may be explained by challenges of (a) replicating specific antecedents in experimental design, (b) lack of conclusive evidence of the presence of certain antecedents in quantitative analysis of case studies, and (c) limitations of survey techniques in assessing fewer tangible antecedents, such as group cohesion and leadership style.  

However, analysis of recent qualitative case study analysis offers a more holistic perspective, and additional insight (Burnette, Pollack and Forsyth, 2011; Brunkhorst, 2020; Lee, 2020).  Burnette, Pollack and Forsyth (2011) case study analysis of the 1996 Mount Everest climbing disaster asked the question whether groupthink was in part responsible for the events that occurred.  Their findings were positive, conclusively observing the presence of antecedents and symptoms of groupthink.  In addition, they determined the presence and type of group cohesiveness in this case, recognising both task and interpersonal cohesion.  The impact of leadership style and behaviour on group relationships, cohesion and efficacy was also noted.  Their summary highlights the presence of two specific antecedents which support Park’s (2000) quantitative conclusions: directive leadership and group cohesiveness.  This case study incorporated multiple data types and sources, and as a result overcame some of the issues stated regarding other research designs which seek to examine groupthink as a phenomenon.  Caya's (2015) examination of juvenile gangs in Turkey finds overwhelmingly in support of the presence of affective group cohesiveness in relation to groupthink-oriented decisions and behaviours.  However, Lee’s (2020) analysis of the presence of groupthink in the 1989 Tiananmen Square massacre concluded the absence of group cohesiveness and offered an alternate model for groupthink that specifically excludes group cohesiveness as an antecedent (Figure 3.3.2).  Whilst the methodological approach and rigour of execution of this study is compelling, employing an adapted version of Qualitative Comparative Analysis (QCA) (Ragin, 2014), its findings regarding group cohesiveness are problematic.  As with many other studies, the definition of group cohesiveness as a uni-dimensional or multi-dimensional construct is not considered, nor the functional or structural facets incorporated.  This in turn catalyses the question of “what is being measured in regard to the determination of group cohesiveness that may reliably conclude on its presence or otherwise?”


Figure 3.3.2       Revised Groupthink Model (Source: Lee 2020:456)


Summary and Discussion of Groupthink literature

The conceptual link between “teamship” and groupthink – both phenomena relating to group performance and outcomes – led to the detailed examination of groupthink literature.  The review of literature highlighted significant problems in the understanding of group cohesiveness as a construct, and of its importance in phenomenon such as groupthink.  Case studies examined in the review bring to light important considerations in regard to defective decision-making in groups that led to disastrous outcomes.  Whilst scholars are in agreement with the measurement and recognition of many of Janis’ suggested antecedents, there is a consistent problem with the identification and measurement of group cohesiveness (Esser and Lindoerfer, 1989; Esser, 1998; Park, 2000; Callaway and Esser, 2006; Rose, 2011; Lee, 2020).  The disaster on Everest highlights extreme levels of many elements of groupthink (Burnette, Pollack and Forsyth, 2011), and the presence of cohesion in the group was essential for many group processes to function (teamwork and various types of interdependence).  However, it is difficult to ascertain from these case studies what caused the defective decision-making associated with groupthink.  The QCA analysis of the Tiananmen Square massacre (Lee, 2020) provides an empirical analysis of qualitative data that supports an amended version of Janis’ (1972, 1982) groupthink model, but - as with much of the theoretical, case and empirical research since 1972 - struggles to provide consistency or reliability in regard to the conclusions about the role of group cohesion in groups which make disastrously poor decisions.  

What we can conclude from all of these studies is that the role and importance of group cohesiveness on the performance of groups and teams is poorly understood in these research fields; on the one hand we have a collection of scholars who dismiss the construct as an irrelevant antecedent in the understanding of decision-making and outcomes of groups (Flowers, 1977; Courtright, 1978; Leana, 1985; Moorhead and Montanari, 1986; Callaway and Esser, 2006; Lee, 2020),and on the other that it is a key antecedent (Janis, 1983; Park, 2000; Burnette, Pollack and Forsyth, 2011).

The differing views of the importance of group cohesiveness in occurrences of groupthink is problematic (Park, 2000; Burnette, Pollack and Forsyth, 2011; Caya, 2015; Lee, 2020).  The challenge for scholars and researchers to reproduce conditions that might allow examination of the presence and importance of group cohesiveness is important.  It is not unreasonable to conclude that experimental or survey-based quantitative research designs are not conducive with creating conditions that may propagate genuine affective cohesion between group members and are – at best – only able to artificially create simulated and exaggerated instrumental task cohesion (Moorhead and Montanari, 1986; Neck and Moorhead, 1995; Rose, 2011).  

bottom of page