THE EFFECT OF LEARNING FROM PERFORMANCE FEEDBACK ON ORGANIZATIONAL BEHAVIOR
S.S. SCHUURMAN University of Amsterdam
MSc. in Business Administration – Strategy Track Student number: 6176488
Supervisor: B. Lima 18/03/2015
Abstract: The topic of learning from performance feedback has fascinated the
organizational sciences since the behavior theory of the firm (Cyert & March, 1963). The behavioral theory of the firm predicts that organizations respond to performance shortfalls by an increase in problemistic search, change and risk-taking. Meanwhile threat-rigidity theory predicts a decrease of risk-taking and more rigid behavior when organizations face performance shortfalls. Despite the large number of studies examining the relationship between performance relative to aspirations and organizational behavior, little effort has been made to systematically investigate and explain these conflicting predictions. I conducted a meta-analysis of the relationship between performance relative to aspirations and organizational behavior to (a) determine the magnitude of the relationship; (b) test if the intrinsic level of downside risk associated with the organizational behavior, the type of performance and type of aspirations moderate this relationship; (c) test context-, and methods-related moderators of the relationship; and (d) suggest future directions for the performance feedback literature on the basis of the findings. The show meta-analytic results show that the relationship between performance relative to aspirations and organizational behavior is significant and negative (r= –.082). Moreover, the findings show significant findings consistent with the predicted moderators.
Key words: performance feedback, threat-rigidity, aspirations, downside risk, meta-analysis
STATEMENT OF ORIGINALITY
This document is written by Student Sannah Sophie Schuurman who declares to take full
responsibility for the contents of this document. I declare that the text and the work
presented in this document is original and that no sources other than those mentioned in
the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of
“Learning is experience, everything else is just information”
- Albert Einstein
There is much interest in the manner how experience can improve organizational performance and how feedback regarding organizational performance affects the likelihood of different actions and organizational behavior (Cyert & March, 1963; March & Shapira, 1987, 1992; Greve, 1998; Chen & Miller, 1994; Ocasio, 1995; Baum & Dahlin, 2007). Performance feedback draws back on the behavioral theory of the firm (BTOF) (Cyert & March, 1963) that argues that organizations learn from their experiences and are guided within their strategic behavior by the discrepancy between their set target levels and performance (Greve, 2003). Setting aspirations is central in order to achieve sufficient strategic decision making and has been a long-standing component of scientific management and strategic planning (Shinkle, 2011). According to the performance feedback theory, performance below the aspiration level will affect behavior of organizations by triggering problemistic search, organizations will engage in more strategic action and increase their risk taking. Performance above the aspiration level, on the other hand, will affect organizational behavior by a decrease in risk taking and less strategic change. Performance feedback can in that way be seen as a regulator of organizational change and risk taking. Learning from performance feedback plays a crucial role in the decision making processes of organizations and the actions that they take. The behaviors and actions that are affected by performance feedback are uncertain and consequential strategic choices are therefore of high importance for managers because their careers depend on it (Greve, 2003).
Research of performance feedback has focused on how feedback on firm performance relative to aspirations affects organizational behavior such as risk taking, R&D spending, change in capital structure, large scale organizational change, alliances, innovation, diversification, capital investment, divestment of business units, safety initiatives, bank lending practices, products quality, acquisitions and organizational misconduct (Bromiley & Harris, 2014). Despite burgeoning research, empirical findings are not always consistent and there has been an ongoing debate about the effect of learning from performance feedback on organizational behavior (March&Shapira, 1987, 1992; Ocasio, 1995; Sitkin and Pablo, 1992). This debate includes two different perspectives on how organizations respond to performance feedback: performance feedback theory and threat-rigidity theory. Performance feedback predicts that a decline in performance relative to aspirations will effect organizations behavior by engaging in more change and risk taking (Cyert & March, 1963; Kahneman & Tversky, 1979), whereas threat-rigidity theory suggest that when organizations face declining performance they perceive this as danger and threats what will lead to less risk taking behavior and more rigidity and inertia (Staw, Sandelands & Dutton, 1981). There is supportive evidence in the form of studies showing that on the one hand, performance below the aspiration level leads to increased search, change, and risk taking, as performance feedback argues (Fiegenbaum, 1990; Bromiley, 1991; Lant, Milliken & Batra, 1992; Wiseman & Bromiley, 1996; Greve, 1998, 2003; Chen & Miller, 2007). However, there are also studies showing that performance below the aspiration level leads to lower competitive aggressiveness and divesture of low performing units (Ferrier et al., 2002; Hayward & Shimizu, 2006).
Although the predictions of the BTOF have received empirical support, these conflicting findings give reason to investigate learning from performance feedback further and present a need for the performance feedback literature to be effectively summarized. Besides the conflicting findings, recent researchers have showed a change in results when considering moderating influences. For example, Vissa, Greve and Chen (2010) found that the form of the organization influences how organizations response to performance feedback conditionally on which type of search domain they use. Audia and Greve (2006) showed that performance below the aspiration level influence behavior by decreasing risk-taking in small firms, but found no significant result for risk-risk-taking in large firms.
These studies suggest that is necessary to investigate moderating factors to get a better insight into the link between learning from performance feedback and organizational behavior. This is an important concern because managers often operate in a complex and diverse context. This will lead to a variety of influences that may interact to shape their decision making behavior (Lim & Mccann, 2014).
To increase our understanding regarding the role that moderators play in the relationship of performance feedback and organizational behavior, I respond to the call of recent research (Iyer & Miller, 2008; Kacperczyk, Beckman & Moliterno, 2013) by focusing on the different levels of risk. Iyer & Miller (2008) suggest that when performance declines organizations may engage in local search rather than turn to mergers and acquisitions. In this meta-analysis, I specifically look at the level of risk that managers associated with different organizational actions, which I term intrinsic level of downside
risk. The different intrinsic levels of downside risk associated with organizational behavior
organizational behavior have limited downside risk, such as investment in R&D or local search, where as other actions have a higher probability of a substantial loss of investment, such as mergers and acquisitions.
In this meta-analysis it the following research question is investigated whether intrinsic levels of downside risk as a moderator might be able to resolve the debate between the two conflicting perspectives on learning from performance feedback. Besides the moderator of downside risk, I investigate if two other construct that are affected by the perceptions of managers (type of performance and type of aspiration) have a moderator effect on the relationship between performance relative to aspirations and organizational behavior. Furthermore, it is investigated if context- and methods-related moderators are of any influence on the relationship.
By performing a meta-analysis and searching for evidence for moderators this research will contribute to a better theoretical understanding of performance feedback and is important for scientific community and practitioners in five different ways. First, this research will summarize the effect of performance relative to aspiration levels is on organizational behavior. This brings not only insight in the decision making process of firms but it also presents us an overall starting point in the discussion between performance feedback and threat-rigidity theory. Second, and more importantly, this paper presents an alternative perspective on the theories by identifying an additional overlooked moderator. Third, due to the fact that there is a widespread use of the process in which organizations learn from their performance feedback it is important to investigate which moderators are of influence. This will help managers to make the learning process more conscious and to
be more aware on how their perceptions can influence the effect of learning from performance feedback.
Fourth, according to Bromiley and Harris (2014) there is, despite the big amount of research on the effect of performance feedback to explain organizational behavior, an insufficient level of theoretical and empirical understanding of organizational aspirations because of differences in measurement. To address this problem this research will take into account the differences in measurement while comparing empirical studies. Thereby the conclusions of the research of Bromiley & Harris (2014) will be used as a guideline. Fifth, there is no meta-analysis on performance feedback. By performing a meta-analysis the existing empirical literature will be summarized what gives a starting point for future research.
The present study is structured as follows: Firstly, based on existing literature, the theoretical perspectives on learning from performance feedback are briefly appointed. Secondly, possible moderators that influence the relationship between performance relative to aspirations and organizational behavior are hypothesized. Thirdly, the methods of data collection and analyzing are discussed. Furthermore, the results of the data analyses are presented are discussed. Next to this, the limitations of this study are presented. Finally, I will present the most important conclusion from this present study.
THEORETICAL PERSPECTIVES ON LEARNING FROM PERFORMANCE FEEDBACK
The relationship between performance relative to aspirations and outcome has been examined mainly by two contrary theories: (a) the performance feedback theory (Greve, 2003) and (b) threat rigidity theory (Staw et al, 1981). The different theoretical perspectives are summarized below.
Performance feedback theory
The idea that organizations learn from experience and make changes to practices, strategies and structures contingent on their performance plays a key role in organizational learning theory (Baum & Dahlin, 2007). It has led to an interest in the manner how experience can improve organizational performance and how feedback regarding organizational performance affects the likelihood of different actions and organizational behavior (Cyert & March, 1963; March & Shapira, 1987, 1992; Greve, 1998; March, 1988; Miller & Chen, 1994; Ocasio, 1995; Baum & Dahlin, 2007). Traditionally the organizational theory has been dominated by the concept of decision makers who are seen as bounded rational agents who are constrained by limits to their cognitive abilities (DiMaggio & Powell, 1983; March & Simon, 1993, 1958; Ocasio, 1997; Thompson, 1967, as cited in Jordan & Audia, 2012). The manner how organizations deal with these cognitive limitations during the decision making process is by learning from performance feedback (Audia, Locke, & Smith, 2000; Greve, 1998; Lant, Milliken, & Batra, 1992; Mezias, Chen, & Murphy, 2002; Miller & Chen, 1994, as cited in Jordan & Audia, 2012). The definition of learning used in this
research is that “an entity learns if, through its processing of information, the range of its potential behaviors is changed” (Huber, 1991, p. 89).
Organizational responses to performance feedback have been on the agenda of organizational researchers since the behavioral theory of the firm (Cyert & March, 1963) was created. The behavioral theory of the firm argues that organizations learn from their experience and are guided within their strategic behavior by the discrepancy between their set aspiration levels and achieved performance (Cyert & March, 1963). Goals, targets or also called aspiration levels, play a central role in how firms interpret their experience and, thus, how they respond to their experience (Lant & Shapira, 2008). An organizational aspiration level can be defined as ‘a reference point of the desired performance levels for a targeted organizational outcome by which performance is evaluated as a success or failure’ (March & Simon, 1958, as cited in Baum & Dahlin, 2007). The difference between the aspiration level set by the organization and its actual performance is called “attainment discrepancy” (Lant, 1992, as cited in Iyer & Miller, 2008). Aspiration levels arise from comparisons between two points of references that decision makers use to evaluate their own current performance: the historic performance of the organization itself, called historical aspiration level (Cyert & March 1963; Levinthal and March, 1981; Greve, 1998; Shinkle, 2011) and recent performance of the organization’s reference or peer group, referred to as social aspiration level (Festinger, 1954; Cyert & March, 1963; Pfeffer & Salancik, 1978; Greve, 1998). Consequently, organizations will gather performance measures and change their behavior if the performance differs from the set aspiration level. There are three different effects of performance relative to the aspiration level on organizational behavior (Greve, 2003, as cited in Jordan & Audia, 2012). First, if
performance falls below the aspiration level the search behavior of decision makers will be more oriented toward identifying alternatives to the current set of activities, referred to as “problemistic search”. Problemistic search can be defined as an effort to identify alternatives or actions to current activities that may resolve performance shortfalls and will lead to an outcome above the desired aspiration level (Iyer & Miller, 2008; Greve, 2008). For example, firms performing below their aspirations level often resort to acquisitions to accelerate growth and these acquisitions may contribute to sustained performance turnarounds (Slatter, 1984; Grinyer, Mayes & McKiernan, 1990, as cited in Iyer & Miller, 2008). Problemistic search will decrease when performance is above the aspiration level. Second, decision makers are more likely to engage in strategic change when performance is below the aspiration level than when performance exceeds the aspiration level (Greve, 2003; Jordan & Audia, 2012; Lant & Mezias, 1990; Lant & Shapria, 2008). Cyert and March (1963) stated that organizations rely on past operational procedures and only make changes if they fail to achieve the aspiration levels. When firms are performing above their aspiration level they have little incentive to change. For example, expenses in R&D, innovations and capital investments reduced more as performance improved above aspirations (Greve, 2003). Third, the tendency of decision makers to choose solutions with greater risk is increased when performance is below the aspiration level and is decreased when performance is above the aspiration level (Greve, 2003). Researches have argued that failure to meet an aspiration level motivates decision makers to accept the risks inherent in changing their organization (Bromiley, 1991; Fiegenbaum & Thomas, 1988; Greve, 2003; Lant et al. 1992, as cited in Greve, 2008).
To summarize, according to the theory of performance feedback, performance relative to aspiration levels has an effect on organizational behavior through the acceptance of risk, problemisitic search and implementing changes.
Threat-rigidity theory
The three different effects that Greve (2003, as cited in Jordan & Audia, 2012) describes are all about changes in organizational behavior as a reaction to performance relative to aspirations. According to a later article by Greve (2011) the assumption that organizations change in reaction to performance feedback relative to their aspiration levels has met two challenges. Firstly, organizations do not change as much and readily as assumed. Organizations do not necessarily change whenever performance feedback theory suggest they should. For example, decision makers become committed to losing courses of action (Staw et al., 1981) or see performance feedback in a way that allows them to stay inert (Hirschman, 1970; Milliken & Lant, 1991, as cited in Greve, 1998). Secondly, instead of changing when performance is perceived below aspirations organizations are rather rigid in their behavior. An explanation for this second challenge is that decision makers face competing pressures when performance is below aspirations (March & Shapira, 1992; Sitkin and Pablo, 1992, as cited in Greve, 2011). The performance below aspiration level can lead to a preference for risky alternatives, but the perception of a potential crisis can also trigger the cognitive and organizational changes to act less risky. The threat rigidity theory argues that performance below the aspiration level will lead to more centralized decision making, restricted information processing and organizational rigidity, lower competitive aggressiveness and divestment of low performing units (Staw et al., 1981;
Ferrier et al., 2002; Hayward & Shimizu, 2006, as cited in Greve, 2011). This view suggests, contrary to performance feedback, it is less likely organizations engage in risk-taking behaviors when they are focused on survival (Staw et al., 1981 in Iyer & miller, 2008). Organizations and managers see, according to threat-rigidity theory, low performance as a threat and thereby avoid new strategies, emphasis on cost reductions, reduce risk taking, conserve resources and limiting new strategic initiatives ((March & Shapira, 1987, 1992; Schendel, Patton, & Riggs, 1976; Starbuck & Hedberg, 1977, as cited in Chen & Miller, 2007; D’Aveni, 1989, as cited in Iyer & Miller, 2008). To summarize, according to the threat rigidity theory organizational behavior becomes less varied or flexible when facing threats (Staw et al., 1981).
One integrated approach
For both views there are findings that support the different assumptions and falsify the assumptions made by the contrary view. There has been some work to integrate these contrary views. In a study of Audia and Greve (2006) these theories were combined and firm size was used as the independent variable to define which of these theories dominates in which situation. They found that smaller firms were more likely to show rigidity to threats in comparison with large firms. Greve (2011) found that increased risk taking is a response to performance below aspirations of large firms, while increased rigidity is a response of small firms. As Greve (2011) argues, large firms do not become more rigid as a result of low performance, but they are already more rigid to begin with. That means that low performance gives small firms the same rigidity levels as large firms regardless of the
performance level. Considering the lack of consensus in literature about the effect of past performance on organizational behavior, I set up two competing hypothesis:
Hypothesis 1a. The effect of performance relative to aspirations on organizational
outcome is negative and significant
Hypothesis 1b. The effect of performance relative to aspirations on organizational
outcome is positive and significant
MODERATORS OF LEARNING FROM PERFORMANCE FEEDBACK
The process between the aspiration levels and the actual strategic behavior is influenced by moderators (Shinkle, 2011; Jordan & Audia, 2012). As mentioned above recent research has found interesting results when considering moderator effects (Audia & Greve, 2006; Greve, 2011). As Lim & Mccann (2014) argue by investigating moderator effects we can get more insights in the process of learning from performance feedback and see whether the two conflicting arguments of performance feedback and threat-rigidity theory can be integrated. In this next section I will discuss the moderators of interest in this study.
Intrinsic level of downside risk
Fluctuations around an outcome are generally used as a measure for risk, where risk can be defined as the ‘variation in the distribution of possible outcome, their likelihood and their subjective values’ (March & Shapira, 1987, p. 1404). When the variance gets higher the
outcome is less predictable and thereby risk entails a loss or a gain. Consequently projects are seen as highly risky when there is a wide spread of outcomes around the expected value in both the positive and negative direction. However, the view of managers and executives does not match with this classic view on risk. Executives are only considering negative outcome when thinking of risk, also called downside risk (March & Shapira, 1987). Downside risk can be defined as ‘the likelihood that a decision will result in a potential negative outcome’. The focus is on the magnitude rather than on the probability of the negative outcome. Different kinds of strategic actions have different intrinsic levels of risk. There are actions that have limited downside risk and may be reversible. An example of such action is investing in long term R&D, which can produce some basic learning and thus improve the firms overall R&D capabilities apart from the success or failure of any particular product. Action that do have downside risk and in that sense are more risky for example commercialization alliances, acquisitions, and divestitures because they can be very costly, and typically involve a higher probability of a substantial loss of investment (Markovitch, Steckel & Yeung, 2005). In this way organizational outcome can be divided into actions with low and high downside risk. Markovitch et al (2005) use two criteria to determine whether an actions has a high or low intrinsic level of downside risk, reversibility and the extent of associated downside. Actions have a high intrinsic level of downside risk if the action is a potential loss of investment or has a substantial negative impact on profits. Actions that do not possess these characteristic have a low intrinsic level of downside risk. Because managers play an important role in decision-making processes it is interesting to investigate if the intrinsic level of downside risk associated with different solutions can give us more insight in how organizations respond to performance relative to
aspirations. The performance feedback theory argues that if performance relative to aspirations declines the organization will behave more risky (this means a negative relationship between performance relative to aspirations and risk taking behavior). Contrary to this view threat-rigidity theory argues that low performance will trigger behavior that is more rigid and risk averse (this means a positive relationship between performance relative to aspirations and risk taking behavior). Both arguments make sense. With making the distinction between low and high level of downside risk associated with organizational behavior we can try to bridge this gap. Organizations will engage more in strategic actions that are associated with low intrinsic level of downside risk when performance relative to aspirations is declining, because it may even be reversible. In this way managers can try to improve performance without having to be scared of a substantial loss. The effect of performance feedback will be strong for actions that do have a low level of downside risk associated with it. On the other hand organizations will engage less in strategic actions that are associated with high intrinsic levels of downside risk when performance is declining, because these action have a higher probability of a substantial loss and managers are not willing to take that risk because their careers depend on it. The effect of performance feedback will be less on organizational behavior that entails higher levels of downside risk. This leads to the following hypothesis:
Hypothesis 2. The effect of performance feedback is stronger when outcome is
classified as a strategic action with a low intrinsic level of downside risk than
outcomes that are classified as an strategic action with a high intrinsic level of
Type of performance
As argued above managerial perspective does not always confirm with the theoretical reasoning. Therefore it is important that next to the effect of differences of downside risk levels we look at other dimensions that can be influenced by managerial perspectives. Organizational performance is an important, if not the most important, construct in strategic management research (Combs, Crook and Shook, 2005). Within organizational performance there a distinction can be made between three types of performances. Accounting returns, growth and stock market measures (Combs et al., 2005). Combs et al. (2005) found that managers see accounting returns prominently as a measure of overall performance. In the process of learning by performance feedback organizations are deciding whether or what action they should take depending on the performance relative to their own past performance and the performance of others. Due to the findings of Combs et al. (2005) we can expect that in the decision-making process managers will pay much more attention to accounting returns than to the other measures of organizational performance. Because accounting return is seen as overall performance, it can be expected that managers want to take actions when accounting return relative to aspiration declines. Eventually, their careers depend on how they deal with performance shortfalls. This urge to take action is expected to be lower with growth or stock market measures because they are measured on a higher level and thereby further away from the attention of managers. Another element is that growth and stock market measures may not be so obvious to interpret as accounting returns for managers. If they do not know how to interpret the performance feedback they can also not make decision on what kind of actions to take. Therefore I hypothesize the following:
Hypothesis 3. The effect of performance relative to aspiration on organizational
outcome is stronger for performance measured by accounting returns than
performance measured by growth of stock market measures.
Type of aspiration
Next to different kinds of outcome and performance another process can be of influence for the decision-making process of managers. As performance feedback argues, performance is measured/analyzed by a set target/benchmark. Whether historical or social benchmarks generate different influences on organizational behavior is unclear (Kim, Finkelstein & Haleblian, 2014). As Kim et al (2014) argue, because the two forms of aspiration levels are derived from distinct sources of performance feedback and are filtered through different cognitive and organizational processes, they may engender different interpretations, which in turn may induce different organizational responses. Supporting this claim Harris and Bromiley (2007) suggest that the effect of performance feedback on organizational outcome may vary depending on the kind of reference point. Historical reference points are derived from past performance and are related to the managerial capabilities and resources. The information that comes from this source tells something about how well the firm is on track compared to their past performance. Social reference points are derived from a reference group and give information about how well the organization could perform. While this is a useful benchmark, it can be hard to interpret because learning form the experience of others, is inherently more difficult than learning form one’s own (Baum & Ingram, 2002). To get a good interpretation managers not only need to have information
about the performance of the reference group but also how others achieved this performance. Unfortunately, managers do not always have access to the information required to accurately interpret other firms’ performance as such information is often private knowledge available only to the insiders of the firms (Kim & Minder, 2007; Menon & Pfeffer, 2003 in Kim et al., 2014). Therefore I hypothesize the following:
Hypothesis 4. Performance feedback relative to historical aspirations has a
stronger effect on organizational behavior than performance feedback relative to
social aspirations.
MEHTODS- AND CONTEXT-RELATED MODERATORS
Next to these theoretical-related moderators also context-related moderators are explored to see if they are of any influence on the relationship between performance relative to aspirations and organizational behavior. The context is also taken into consideration by making a difference between the type of firms investigated (manufacturing, service or high-tech industries). Next to a context-related moderator also methods-related moderators are explored. Specifically, three research design determinants (unit of analysis, observation plan and survey dummy) and two sample factors (organization type, sample location) are examined for possible moderator effect on the performance relative to aspiration and organizational behavior relationship.
METHODOLOGY Literature search
A literature search was conducted to identify published and unpublished studies that investigated the relationship between performance relative to aspiration and organizational behavior. The literature search included five different search techniques to identify prior empirical research. First, most studies were found by computerized indices (DeCoster, 2009). This means that articles were found by entering key words or a combination of them into different databases such as performance feedback, performance relative to aspirations,
threat-rigdity, aspirations, slack, problemistic search, organizational decline and reference
point. Second, the literature search included a descendent search (DeCoster, 2009). This
means that the articles that were published at early dates and that were important were used to locate later articles that cite them in their references. According to DeCoster (2009) this is a useful complement to the standard search of computerized indices. Third, an ancestor search, were the references of articles that were included in the analysis were checked to see if they contain any relevant studies. Fourth, Greve (2003,2008,2011) computed a list of interesting articles about performance feedback on his website (henrichgreve.com), this list was checked for missing articles. Finally, in an effort to identify relevant unpublished studies, I searched conference proceedings for the annual meeting of the Academy of Management in the years 2012 and 2013. The result of this search was a full candidate list (DeCoster, 2009). The search yielded 215 articles and dissertations.
Inclusion Rules and Sample
Of all these studies at least the title, abstract and methodology were examined. In these steps the studies were excluded that did not meet the criteria. First, only empirical papers that reported on correlations between performance relative to an aspiration and organizational behavior of any type were included. Theoretical and review papers were excluded. Second, studies that tested the relationship with a spline, measuring the relationship separate for performance above and below the aspiration level, were excluded. This decision was made because it was not possible to combine articles with and without a spline and there were more articles without a spline. Third, studies had to report sample size and a correlation coefficient between the investigated variables of interest. Fourth, the studies had to report how they measured performance relative to aspiration (historical, social, weighted average or switching model). Fifth, studies were excluded if they did not investigate action (for example, the effect of performance feedback on managers concerns). Based on these criteria, the final data set included 42 studies conducted between 1986 and 2014 yielding a total of 102 effect sizes.
Measures
Coding of each study was conducted by one coder using a structured coding sheet. The reliability of the coding was checked by a second coding of a subsample and comparing the results with the first coding session. This was done after sufficient time had passed so that the studies were not fresh in mind and without reference to the original coding. Next to this, the coding was checked with another researcher who is investigating performance feedback
of studies with a spline. Whenever there was a discrepancy in coding, the discrepancy was discussed until consensus was reached.
Level of downside risk. The studies were categorized into three types of downside
risk. Change, where risk is undetermined included correlation which an outcome that could not be coded with downside risk, but did have some kind of change. An example of this category is . Low intrinsic level of downside risk included outcomes as R&D investment or Search intensity. High intrinsic level of downside risk included outcomes as Acquisitions and Capital investment.
Type of performance. The type of performance was classified as operational
performance, organizational financial performance, organizational market performance and organizational growth performance. Operational performance was performance mostly within the firm including measures as product quality or innovation scales. Organizational financial performance included measures such as return on assets, return on sales and net income. Organizational market
performance was focused on measures as stock return an market to book value.
Organizational growth performance included measures such as market share and
growth scale (Combs et al., 2005)
Type of aspiration. We followed Bromiley and Harris (2014) when coding how
aspiration levels where measured. They make a distinction between a single measure of historical or social, a weighted average model and a switching model.
Context- Related Measure
Industry type. Service industries are characterized by frequent and close
interactions with customers. Manufacturing industries are characterized by fabrication, processing or preparation of products and are most of the time highly capital-intensive. High-technology industries are characterized by innovation and invention and compete in short-cycle and global product markets (Mikovich, 1987, as cited in Joshi & Roh, 2009).
Methods-Related Moderators
Following recommendations of previous meta-analysis study (Freund & Kasten, 2012; Judge et al., 2001, as cited in Park & Shaw, 2013), we coded several other aspects of the research design and publication-related factors for additional exploratory moderator analyses.
Sample location. Location of the sample was divided in US and NON US samples. Level of analysis. The studies were classified at the organizational or individual
level.
Observation plan. The studies were cross-sectional or longitudinal. Survey dummy. Most studies were obtained from archival sources (using
databases) others were obtained through survey.
Publication status. Only the papers from the annual meeting of AOM where coded unpublished, all the other articles were published.
Meta-Analytic Procedure
Meta-analysis researchers recommend using a random effects model that assumes that sampling error causes variability between effect sizes (Aguinis, Dalton, Bosco, Pierce, & Dalton, 2011; Erez, Bloom, & Wells, 1996 in Park & Shaw, 2013). However, Field (2005) argues that the random effect method requires at least 20 primary-level studies in order to obtain a properly confidence interval for the mean effect size assuming approximate normality of the super population of effect sizes (Field, 2005). The random effects model is used for the overall analysis and fixed effects model for the moderator analysis due to the small number of studies. I used Hedge and Olkin’s (1985) meta-analytic procedure to analyze the data. Zero-order correlations between performance relative to aspirations and organizational outcome were taken for each study. The meta-analysis was conducted in SPSS using a meta-analysis macro (Wilson, 2005).
Furthermore, to assess the adequacy of the mean effect size I tested for heterogeneity of validity, that is, the extent to which results varied across studies more than would be expected due to sampling error using the homogeneity analysis (Lipsey & Wilson, 2001). For this analysis I used the Q statistic, which indicates the level of variance across study results relative to sampling error variance (Hedges & Olkin, 1985). If the mean effect sizes are not found to be homogenous, it may be the fact that various variables moderate the effect size. A significant Q rejects the null hypothesis from of homogeneity and indicates that the variability among the effect sizes is greater than what is likely to have resulted from sampling error alone.
With performing a meta-analysis there are two complications (Lipsey & Wilson, 2001, p. 105). The first complication is that studies in a meta-analysis often generate more than one
effect size. While the meta-analyst may be interested in the full range of these multiple effect sizes, any two or more that come from the same study are statistically dependent. Including them in the same analysis would, therefore, violate the assumption of independent data point that is fundamental to most of the common forms of statistical analysis. However, Bijmolt and Pieters (2001) argue that reducing each study to a single value is generally unsatisfactory. Performing a meta-analytic procedure with only one value per study does not comply very well to recovering the true measurement of effects. They conclude that procedures representing each study by a single value should in general be avoided because it results in a serious loss of information. Although this recommendation is less obvious and contrary to leading meta-analyst as Hunter and Schmidt (2004) in this study the complete set of measurements was used on the recommendation of Bijmolt and Pieter (2001). The downsides of this decision will be discussed in the limitation chapter. The second complication is that the samples are different sizes (Lipsey & Wilson, 2001). Effect sizes are derived from sample statistics (correlations, means, standard deviations) and in that way depend on the underlying sample size. A large sample size contains less sampling error and the effect sizes based on it will be more precise and reliable estimations than on small sample sizes. So we need to treat them differently otherwise they will make the same contribution to the results. To ensure that correlations resulting from large sample sizes had greater weighting than correlations resulting for smaller samples, each correlation values are weighted by the sample size. I calculated weighted mean correlations by adopting the inverse variance weights and applying Fisher’s Z transformation procedures (Hedge & Olkin, 1985; Lipsey & Wilson, 2001). Next to correction of sample size I followed Hunter and Schmidt’s (2004) formula and corrected the correlations for
unreliability using the artifact distributions for subjective performance measures. For the empirical studies that used a survey I used the cronbachs alpha and for the other studies where the reliability of the measures was not reported, I followed other macro-level meta-analysis (Bommer, Johnson, Rich, Podsakoff, & MacKenzie, 1995; Dalton, Daily, Ellstrand, & Johnson, 1998; Dalton, Daily, Johnson, & Ellstrand, 1999, as cited in Park & Shaw, 2013) and used 0.8 for the reliability correction. Ninety-five percent confidence intervals and eighty percent credibility intervals were calculated around the sample-weighted correlation as a measure of accuracy of the effect size. The difference between confidence intervals and credibility interval is that confidence interval provide an estimate of the variability around the estimated average correlation whereas credibility intervals provide an estimate of the variability around the individual correlations in the population of studies. When a 95% confidence interval is excluding zero this indicates that the true score of the correlation is different than zero for 95% change. An 80% credibility interval that is excluding zero means that at least 80% of all the correlations reported are different than zero.
Outliers
The purpose of meta-analysis is to arrive at a reasonable summary of the quantitative findings of a body of research studies. This purpose is not usually served well by the inclusion of extreme effect sizes values that are notably discrepant for the preponderance of those found in the research of interest and, hence, unrepresentative of the results of that research and possibly even spurious. In addition, extreme effect sizes values have disproportionate influence on the values of the means, variance and other statistics used in
meta-analysis and may distort them in misleading ways. Therefor the distribution of effect sizes is examined, to determine if outliers are present (Hedges & Olkin, 1985). If so, Lipsey and Wilson (2001) advice to remove or adjust them to less extreme values before proceeding with the analysis. One approach is to recode the extreme values to more moderate ones. This procedure is called winsorizing. I looked for a break in the effect size distribution and coded the outliers back to the next largest cluster of effect sizes. From the sample size four cases got adjusted, all from the same study (Study ID 99). From the effect sizes six cases got adjusted (Study ID 5, 41, 108 and 43).
Publication bias
To assess the sensitivity to publication bias I conducted two types of analyses. Publication bias can occur when the studies used in the analysis differ from the full body of research on the topic. First, I used funnel plot to see if there were patterns of asymmetry that predict publication bias (Sterne, Becker & Egger, 2005, as cited in Morris et al., 2014). Second, to estimate what the average validity would be if the hypothetical ‘missing’ studies were included in the analysis I used a trim-and-fill analysis (Duval & Tweedie, 2000, as cited in Morris et al., 2014). The linear trimming estimator was used in this analysis.
RESULTS Overall Meta-Analysis
The top panel of Table 1 shows the results of the overall meta-analysis of the relationship between performance relative to aspirations and organizational behavior using all available independent correlations (kcorr= 102; N=934818). The average corrected correlation between performance relative to aspirations and organizational behavior across all studies was negative (ES = -0.078) and a 95% confidence level did not include zero (95% CI [-.10, -.06]). However, the homogeneity of effect sizes tests were significant across the analyses (QB = 5282,54, p < .01).
This justifies not only using the random effects model, but also indicates that moderators may be present for the relationship between performance relative to aspiration and organizational behavior. A few studies in the full analysis contained extremely large
Table 1: Meta-Analysis of the Relationship Between Performance relative to aspiration and Organizational Behavior: Overall Analysis
Sample characteristics K Kcorr N Weighted
Mean r SEr 95% CI 80% CV QB
T&F ∆K r
All studies 42 102 934818 -0,078 0,01 -0,10 -0,06 -0,09 -0,06 5282,54 0b 0,016 All studiesa 42 102 408238 -0,082 0,01 -0,11 -0,05 -0,10 -0,06 4113,76 0b 0,016
Note. Random effects model is used. k = number of studies; kcorr = total number of correlations; N = total sample size for all studies combined; Weighted Mean r = sample size weighted and corrected averaged observed correlation; SEr = standard error of Weigthed Mean r; 95% CI = 2.5%
lower and 97.5% upper limits of 95% confidence interval of ES; 80% CV = lower and upper bounds of the 80% credibility value for ES; QB =
homogeneity statistic Q; T&F: ∆K = number of effect sizes imputed by trim-and-fill analysis; T&F r = trim-and-fill estimate of average correlation.
a
Studies including the correction with winsorizing b Imputed on the left side of the distribution.
** p < 01.
samples or effect sizes. I used winsorizing to correct for such cases and results are shown in the second row of table 1, showing that the correlation magnitude increased slightly to – .082 (95% CI [-0,11, -0,05]). These findings clearly indicate the negative relationship between performance relative to aspiration and organizational behavior and thereby hypothesis 1a is supported and there is no support for the hypothesis 1b in the overall analysis.
Moderator-analysis
Table 2 shows the tests of the suggested moderators. All the categories that where coded as ‘mixed’ or ‘other’ are not included in this table, due to the small number of studies included in these categories. The top panel rows show the moderating effect of the level of downside risk (change, low risk and high risk). The between-group goodness-of-fit statistic QB shows that the correlations between performance relative to aspirations and organizational behavior were significantly different across levels of downside risk, QB = 298,99 p < 0,01. The results show that the size of the negative correlation between performance relative to aspiration organizational action that had a high intrinsic level of downside risk (ES = –.051, 95% CI [–.06, -0,05]) was smaller than the associated correlations with action that had a low intrinsic level of downside risk (ES = –.073, 95% CI [–.08, –.07] and actions that can be defined as change, where the downside risk level was not sure (ES= -.165, 95% CI [– .18, –.15]).
Note. Fixed effects model is used. Data is corrected by winsorizing. k = number of studies; kcorr = total number of
correlations; N = total sample size for all studies combined; Weighted Mean r = sample size weighted and corrected averaged observed correlation; SEr = standard error of Weigthed Mean r; 95% CI = 2.5% lower and 97.5% upper limits of
95% confidence interval of ES; 80% CV = lower and upper bounds of the 80% credibility value for ES; QB = homogeneity
statistic Q
** p < 01.
Analysis
k kcorr N ES SE 95% CI 80% CV QB
Level of downside risk
Change, risk not sure 8 19 42241 -0,165 0,01 -0,18 -0,15 -0,17 -0,16 298,99** Low risk 13 30 122822 -0,073 0,00 -0,08 -0,07 -0,08 -0,07
High risk 21 53 243175 -0,051 0,00 -0,06 -0,05 -0,05 -0,05
Type of performance 552,28**
Operational performance 6 17 8634 -0,069 0,0135 -0,10 -0,04 -0,09 -0,05 Org, Perf, - Financial 25 63 251438 -0,105 0,0025 -0,11 -0,10 -0,11 -0,10 Org, Perf, - Growth 7 12 9934 -0,064 0,0126 -0,09 -0,04 -0,08 -0,05
Type of aspiration level 133,10**
Historical 17 37 78787 -0,080 0,0045 -0,09 -0,07 -0,09 -0,07 Social 19 52 289346 -0,076 0,0023 -0,08 -0,07 -0,08 -0,07 Weighted average 6 13 40105 -0,001 0,0062 -0,01 0,01 -0,01 0,01
Context- & Methods-related moderators Type of Industry 644,25** Service 6 12 133415 -0,004 0,0034 -0,01 0,00 -0,01 0,00 Manufacturing 13 34 157383 -0,081 0,0031 -0,09 -0,08 -0,09 -0,08 High-Tech 9 26 10843 -0,116 0,0120 -0,14 -0,09 -0,13 -0,10 Sample Location US 25 61 343552 -0,078 0,00 -0,08 -0,07 -0,08 -0,08 102,80** NON-US 11 31 59699 -0,022 0,01 -0,03 -0,01 -0,03 -0,02 Type of organization 468,24** Public 6 10 7859 -0,001 0,0141 -0,03 0,03 -0,02 0,02 Private 25 69 312013 -0,054 0,0022 -0,06 -0,05 -0,06 -0,05 Level of analysis 0,53 Individual 4 11 5920 -0,058 0,0163 -0,09 -0,03 -0,08 -0,04 Organizational 38 91 402318 -0,070 0,002 -0,07 -0,07 -0,07 -0,07 Observationplan 480,17** Cross-sectional 18 45 145186 -0,012 0,0033 -0,02 -0,01 -0,02 -0,01 Longitudinal 24 57 263052 -0,101 0,0024 -0,11 -0,10 -0,10 -0,10 Survey dummy 14,85** Survey 5 16 6885 -0,012 0,02 -0,04 0,02 -0,03 0,01 Archival 37 86 401353 -0,071 0,00 -0,07 -0,07 -0,07 -0,07 Publication status 388,82** Unpublished 8 18 134356 -0,015 0,0034 -0,02 -0,01 -0,02 -0,01 Published 34 84 273882 -0,097 0,0024 -0,10 -0,09 -0,10 -0,09
Hypothesis 2 predicts that the effect would be stronger for action with low intrinsic level of downside risk than action with high level of downside risk, the results support this hypothesis. The second set of results in Table 2 shows the moderation results for the different performance measures. The variance of performance relative to aspiration and organizational behavior correlations was significantly different across performance measures (QB = 552,28, p<0,01). Specifically, the negative correlations between performance relative to aspirations and organizational behavior were large when performance was measured as financial organizational performance (ES = –.105, 95% CI [– .11, –.10]) The relationship was somewhat weaker but also significant and negative when performance was measured as operational performance (ES= -0,069, 95% CI [-.10, -.04]), and organizational growth performance (ES= -,064, 95% CI [–.09, –.04]). The correlations for organizational market performance came from only two studies, due to this small number of studies I removed this variable from the list. These results support hypothesis 3 that predicts that the relationship of performance relative to aspiration on organizational behavior will be stronger when performance is measured as organizational financial performance.
The third set of results shows the moderation effects of the different measures for the aspiration level: historical, social and weighted average model. The between-group goodness-of-fit statistic QB shows that the correlations between performance relative to aspirations and organizational behavior were significant when considering the measure of the aspiration level (QB = 133,10, p<0,01). The weighted average model did not show a significant result. The historical and social aspiration measures show similar results, but results show a slightly stronger negative correlation when performance was set against a
historical aspiration level ( ES=0.080, 95% CI [-.09, -.07]), than against a social aspiration level (ES=0.076, 95% CI [-.08, -.07]). Hypothesis 4 predicts a significant difference between the relationship of performance relative to historical aspirations on organizational behavior and the relationship of performance relative to social aspirations on organizational behavior, where performance relative to historical aspirations will have a stronger effect than performance relative to social aspiration has on organizational behavior. Results show that hypothesis 4 is supported.
Context- and Methods-Related Moderators
The bottom half of Table 2 shows the results for context- and methods-related moderators. The context-related moderating effect of industry was statistically significant (QB = 644,25, p<0,01). The relationship between performance relative to aspiration level and
organizational behavior was significantly different than zero in manufacturing (ES=-.081, 95% CI [-.09, –.08]), and high-tech (ES= -.116, 95% CI [-.14, -.09]) samples. The relationship was not significantly different than zero in service samples (ES= -.004, 95%CI [-.01, .00]). The between-group goodness-of-fit statistic QB shows that the correlations between performance relative to aspirations and organizational behavior were significant when considering the sample location (QB = 102.80, p<0,01). The correlations between performance relative to aspirations and organizational behavior were more negative in US (ES=-.078, 95% CI [-.08,-.07]) than NON-US (ES=-.022, 95% CI [-.03, -.01] samples. The moderator effect of the type of organization was significant (QB = 468,24, p<0,01). The correlations between performance relative to aspiration and organizational behavior were not significantly different than zero for samples that included private
organizations(ES=-.001, 95% CI [-.03, .03], but did significantly include zero for samples that included private organizations (ES=-0.054, 95% CI [-.06, –.05]. The moderation results for unit of analysis showed that correlation sizes were not significantly different across unit of analysis (QB = 0.53). The next panel shows that the observation plan is a significant moderator (QB = 480.17). The correlation between performance relative to aspiration and organizational behavior is significantly different than zero for cross sectional (ES=0.012, 95% CI [.02, -.01] and longitudinal samples (ES= -0.101, 95% CI [-.11, -.10]. The variance of performance relative to aspiration and organizational behavior correlations was significantly different across survey dummy(QB = 14,85, p<0,01). The correlation between performance relative to aspiration and organizational behavior is not significantly different than zero for survey (ES=-0.012, 95% CI [-.04, .02] but is significantly different than zero for archival samples (ES= -0.071, 95% CI [-.07, -.07]. The last result of table 2 includes the significant moderator of publication status (QB = 388,82, p<0,01). The results show that correlations between performance relative to aspiration and organizational behavior is are more negative and significantly different than zero for published (ES=-0.097, 95% CI [-.10, -.09] than for unpublished studies (ES= -0.015, 95% CI [-.02, -.01].
Publication analysis
The editorial practices of journal make it hard for studies with non-significant findings to be published (Morris et al, 2014?). Next to this, unpublished research is less accessible because researchers may be less motivated to share the unfavorable results (Morris et al, 20..). This can lead to a potential upward bias in the mean effect sizes because the effect sizes of studies that have lower validity are less likely to be included. There are several
methods that can be used to determine if publication bias is present. As presented above in the methods-related moderators the result of publication status shows a significant difference between published and unpublished studies (QB = 388,82, p<0,01). The findings show that the effect of performance relative to aspiration on organizational behavior is stronger for studies of published articles than studies of unpublished articles. However, the comparision of published to unpublished findings is a limited way to determine whether a publication bias exist, so next to this comparison two addition analysis have been performed. First, I used visual inspection of funnel plot asymmetry to detect publication bias (Sterne et al., 2005 in Morris et al., 2014). If the funnel plot is symmetrical one could expect that there is no publications bias. The funnel plot has a symmetrical pattern (see figure 1), suggesting that publication bias is not a concern with this data. Second, I used the trim-and-fill analysis (Duval & Tweedie,
2000 in Morris et al., 2014) to examining publication bias. Consistent with the funnel plot, the trim-and-fill analysis did not impute an effect size, meaning there is no concern of publication bias in this data.
DISCUSSION
Performance feedback is seen as an important part of the learning and decision-making process within the organization. In this study, I contribute to the research on performance feedback and practitioners by (a) meta-analyzing the relationship between performance
Figure 1. Funnel plot with pseudo 95% confidence limits
relative to aspiration level, (b) outlining and testing theoretically relevant moderators of the relationship and (c) testing context- and method-related moderators. I believe that researchers and practitioners can benefit by having more insight in the overall performance relative to aspiration-organizational behavior relationship. Furthermore, this will give a starting point for future research of other moderators. In this discussion, I review these meta-analytic results and discuss recommendations for future research.
Relationship between performance relative to aspiration level and Organizational Behavior
After correcting sampling errors, unreliability and outliers, the estimated meta-analytic correlation was -.082. One of the key findings from this meta-analytic review is that despite some variation across moderators, organizations change their behavior when performance relative to aspiration declines. Critics might argue that the effect of -.082 is modest and that learning from performance feedback only explains a small amount of the variance in organizational behavior. However, as other meta-analyst (Park & Shaw, 2013) argue, from a qualitative standpoint small effects can be considered relevant when the outcome variable has many legitimate predictors. In case of organizational behavior, there are countless influences on organizational, individual and macro levels. Therefore I expect that single predictors on organizational behavior provide modest explanation if you compare it to an outcome variable that has less influencing antecedents.
Moderators
The results of this meta-analytic review show that performance relative to aspirations and organizational behavior correlations are quite different in size comparing levels of downside risk. These findings show, as predicted, that actions with high level of downside risk have a weaker effect on organizational behavior than actions that have a low intrinsic level of downside risk. Although these results answer some question posed in the literature, several important unknowns remain. First, the effect that was measured for the variable
change, where risk is undetermined was much stronger than the two variable of downside
risk. An explanation could be that change does not always have to entail risk and in that way have no downside risk is associated with it. This is in line with Kacperczyck et al (2013) that argued there is a substantial difference between risk and change, and suggested that researcher must see them as two different constructs. Further research could take a more in-depth look at the difference between risk and change.
Second, the assumption that managers take a look at the downside risk rather than the variability of risk is not fully supported here, since the difference between these two constructs was not of interest in this research. Further research could test this assumption by making a distinction between the level of downside risk associated with the action and the variability of risk associated with the action. This is interesting to investigate, because most of the research on performance feedback uses the variability of risk as a measure instead of the level of downside risk.
Next to the level of downside risk, the type of performance has shown to be a significant moderator. The relationship between performance relative to aspiration and organizational behavior is much stronger when performance is measured as financial organizational
performance. Further research could focus on subgroup analysis of financial organizational performance to see whether there are significant differences between the different measures within financial organizational performance.
The results of the moderator analysis on the type of aspiration level showed significant results. Although it is a significant result, the result seems small regarding to the argument of the information that is available to managers. Therefore I performed a subgroup analysis to check whether there is more difference between historical and social aspirations levels when looking at the level of downside risk, the type of performance and type of industry. The findings are presented in appendix A.
The results of this subgroup analysis show that there is a substantial and significant difference between historical and social aspirations on the subgroup level. It shows that social aspirations have a stronger effect when considering low intrinsic downside risk action and historical aspirations have a stronger effect when considering action with change, where risk is unsure. When considering the different types of performance, social aspirations have a stronger effect than historical on operational and financial organizational performance. For market and growth measures it is the other way around. These findings suggest that there are significant differences between aspiration levels when looking at a subgroup level. However, due to the small number of studies this research could not investigate the relationship on subgroup level. Further research should consider doing this. Second, this present study included only studies with a continuous measure of performance relative to aspiration. Future research could investigate if there is a difference between studies that investigate the effect of performance feedback with or without a spline between above and below aspirations levels.
LIMITATIONS
As with all research, this study is not without limitations. The most obvious limitation of this study is the small number of samples available for inclusion. The current meta-analysis included only 42 studies, which made interpretation of results somewhat difficult.
A related issue is that the moderator analysis showed a significant difference between published and unpublished studies, but the funnel plot and trim-and-fill analysis showed different results. These contradictory findings suggest that we still need to consider publication bias within this data.
Another limitation is that of all measurements are included while not accounting for the dependency between measurements from the same study. Especially conclusions about study level moderators may be biased (Bijmolt & Pieters, 2001).
The reliability of the coding is another limitation. The research could be checked more often by a second coder to improve the reliability. Despite these limitations, I hope that this present study has offered clarity to the existing literature and research on learning from performance feedback and motivate researchers to investigate this process further.
CONCLUSION
This meta-analysis shows that performance relative to aspirations and organizational behavior are significantly and negatively related. The three moderators: intrinsic level of downside risk associated with strategic actions, the type of performance and the type of aspirations levels, have a moderator effect on the relation between performance relative to aspirations and organizational behavior. Furthermore are also context- and methods-related moderators of influence on this relationship. These results show that meta-analysis has an
important role, because single studies may not be able to investigate these moderators. I hope to encourage future researchers to examine this relationship to (a) distinguish measurement with or without a spline and (b) consider more organizational- and context related factors.
REFERENCES
(References marked with an asterisk indicate studies included in the meta-analysis)
*Alessandri, T. M., & Pattit, J. M. (2014). Drivers of R&D investment: The interaction of behavioral theory and managerial incentives. Journal of Business Research, 67(2), 151-158.
Audia P. G., & Greve H. R. (2006). Less likely to fail? Low performance, firm size, and f actory expansion in the shipbuilding industry. Management Science, 52, 83–94.
*Audia, P. G., & Brion, S. (2007). Reluctant to change: Self-enhancing responses to diverging performance measures. Organizational Behavior and Human Decision Processes, 102(2), 255-269.
*Audia, P. G., & Goncalo, J. A. (2007). Past success and creativity over time: A study of inventors in the hard disk drive industry. Management Science, 53(1), 1-15.
*Audia, P. G., Locke, E. A., & Smith, K. G. (2000). The paradox of success: An archival and a laboratory study of strategic persistence following radical environmental change. Academy of Management Journal, 43(5), 837-853.
Baum, J. A. C., & Dahlin, K. B. (2007). Aspiration performance and railroads' patterns of learning from train wrecks and crashes. Organization Science, 18(3), 368-385
Baum, J. A., & Ingram, P. (2002). Interorganizational learning and network organization: Toward a behavioral theory of the interfirm. The economics of choice, change, and organization: Essays in memory of Richard M. Cyert, 191-218.
*Ben-Oz, C., & Greve, H. R. (2012). Short-and long-term performance feedback and absorptive capacity. Journal of Management, 0149206312466148.
Bijmolt, T. H., & Pieters, R. G. (2001). Meta-analysis in marketing when studies contain multiple measurements. Marketing Letters, 12(2), 157-169.
*Boeker, W. (1997). Strategic change: The influence of managerial characteristics and organizational growth. Academy of Management Journal, 40(1), 152-170.
*Bolton, M. K. (1993). Organizational innovation and substandard performance: when is necessity the mother of innovation?. Organization science, 4(1), 57-75.
Bromiley, P. (1991). Testing a causal model of corporate risk taking and performance. Academy of Management journal, 34(1), 37-59.
Bromiley, P., & Harris, J. D. (2014). A comparison of alternative measures of organizational aspirations. Strategic Management Journal, 35(3), 338-357.
*Chattopadhyay, P., Glick, W. H., & Huber, G. P. (2001). Organizational actions in response to threats and opportunities. Academy of Management Journal, 44(5), 937-955.
