Sie sind auf Seite 1von 4

ROT013

Promoting Experimentation for


Organizational Learning:

The Mixed Effects of


Inconsistency
Creating a consistent work environment containing ‘psychological safety’
can reduce employees’ fear of failure and promote experimentation.
by Amy Edmondson

Innovation is a rallying cry for most of today’s organizations. Yet encouraging innovation
and superb execution of routine work at the same time runs the risk of sending a mixed
message that can end up accomplishing neither.
For example, in 2000, Bank of America sought to become an industry leader in
innovation and established a program to promote experimentation in two-dozen real-life
‘laboratories’ – fully-operating retail bank branches in which employees were to experi-
ment with new product and service concepts, such as virtual tellers. Successful experiments
Amy Edmondson – determined on the basis of consumer satisfaction or revenue growth – would then be
recommended for a national rollout.
Bank senior management voiced strong support for innovation, and explicitly recog-
nized and communicated that experimentation with new ideas necessarily produced failures
along the way. Indeed, a failure rate of 30 per cent was targeted as indicative of ‘sufficient’
risk-taking and novelty. Initially, however, employee compensation continued to be based
on measures of routine performance (such as opening new customer accounts.)
The espoused goal of increasing innova- The role of failure in experimentation mistakes, errors, and failures against him or
tion thus was inconsistent with the reward Failures are unavoidable outcomes of her.Without psychological safety, individu-
system; individuals’ compensation could experimentation, because the outcome of als are likely to be concerned with the
suffer from time spent experimenting with any single experiment or trial is always interpersonal risks of failure and to be
new ideas or from failed experiments. uncertain in advance. For example, when reluctant to engage in experimentation.
Understandably, many employees were selecting one unknown key out of many to Differences across organizations in
reluctant to experiment much until man- unlock a door, one does not know in psychological safety have been shown to
agement made changes to align reward advance whether or not the key will work; affect the level of anxiety people feel when
systems with the organization’s new value of risk of failure is thus unavoidable. Such fail- confronting ambiguity and uncertainty.
experimentation. ures can be beneficial because they provide Psychological safety can be created by sup-
the experimenter with new knowledge portive structures such as information and
Defining ‘experimentation’ about the solution and thereby facilitate inno- reward systems, and by the words and
Experimentation is a trial-and-error process vation and performance in the long run. actions of high-level management. In
in which each trial generates new insights Research shows that when pharmaceu- particular, messages that indicate support-
on a problem. Learning by experimentation tical companies such as Eli Lilly launch new iveness, openness, and tolerance for error
is fundamental to solving problems for drugs, or automotive firms like BMW affect beliefs about the level of psychologi-
which outcomes are uncertain and where introduce new cars, the products are the cal safety throughout an organization.
critical sources of information are non-exis- result of as many failed experiments as suc-
tent or unavailable. Imagine trying to cessful ones.An innovation process is at least Normative vs. instrumental influences
unlock a door with a set of unfamiliar keys. partially based on ‘accumulated failure’ that Research shows that creativity is related to
Putting one key into the lock to see if it will has been carefully understood. Individuals organizational culture, reward systems,
turn is experimentation; even if the experi- who select tasks in which failures are likely supervisory encouragement, trust, and
ment fails, new knowledge is created that (rather than safe tasks in which they know resources. Feedback, help-seeking and
narrows the scope of subsequent trials. they can perform well) tend to persevere in issue-selling behaviours are all predicted
Each trial in experimentation gener- the midst of hardship and perform better in by supportive organizational norms, lead-
ates information about a solution that the the long run than others. ership openness, and trust. Proactive
experimenter could not know in advance. Despite its benefits, failure has costs, learning behaviours are related to sup-
Information learned in a previous trial can and is often avoided by both organizations portive organizational contexts (access to
be used to modify subsequent experimental and their members. Clearly, failures can resources, information, training, and sup-
designs, conditions, or even the nature of alienate customers, reduce business, and portive reward systems), leader coaching,
the desired solution. Tasks that are con- lead to dissatisfaction among employees. At and routines that encourage exchange of
ducive to effective experimentation are the extreme, failures can harm employees relevant information, reduce sensitivity to
those that allow multiple problem-solving or customers, financially undermine the feedback, decrease defensiveness, and
trials and present opportunities to use organization, and lead to the organization’s increase trust.
knowledge gained from earlier trials to demise. Yet, even when these costs of fail- Organizational variables that affect
enhance learning in subsequent trials. ure are greatly reduced, people are still innovation behaviour include both normative
Experimentation is critical to organiza- reluctant to experiment. One study and instrumental influences. Normative influ-
tional innovation. Important discoveries in showed that when new technologies dra- ences, such as organizational culture and
science – such as artificial vaccines – and matically reduced the economic costs, espoused values, influence employee beliefs
technology, such as the electric light bulb, time, and effort associated with experi- and behaviours by establishing norms and
resulted from constant trial-and-error mentation – such that incurring failures standards that define appropriate and inap-
experimentation through which inventors would not harm the organization’s budget, propriate forms of behaviour. Normative
systematically built up a knowledge base. deadlines, cost structure, employees or cus- values can be explicitly stated by leaders
Research shows that R&D teams spend tomers – individuals still seemed to avoid (e.g., through speeches, signs or memos) or
almost 80 per cent of their time on experi- experiments in which failures were likely. tacitly communicated in features of the
mentation, and that these experiments This avoidance can be explained by the organizational environment (e.g. organiza-
constitute an important source of technical interpersonal or social costs of failure. tional routines). There is evidence that
information. Experimentation advances Specifically, failures make one’s gaps in when normative values state that failures are
understanding of new concepts, promotes expertise and knowledge salient to others, expected and acceptable as part of learning,
new ways of thinking, and creates new while avoiding failure helps to maintain people are less hesitant to discuss mistakes
knowledge. More broadly, individuals who one’s image and professional standing and more willing to try novel tasks even at
constantly improvise, tinker, and experi- among colleagues. Interpersonal costs of the expense of incurring more failures.
ment are able to remain adaptive in failure are exaggerated when people lack The second category – instrumental
fast-paced industries where new ideas are ‘psychological safety’ – a belief that a group influences – pertains largely to formal
constantly in demand. or organization will not hold a person’s reward systems and incentives. Instrumental

Rotman Magazine Winter 2005 • 21


rewards influence the ‘instrumentalities’, or intended to identify and expose failures has Fourth, individuals under high and low
costs and benefits, of experimentation been shown to inhibit creativity and make evaluative pressures react to emotion dif-
behaviour. For example, when employees novel or unfamiliar tasks more difficult. ferently. Specifically, uncertainty often
do not have easy access to time, materials or Evaluative pressure makes failures espe- produces fear, which has been shown to
information to experiment with their ideas, cially salient – inhibiting admission of error produce two somewhat contradictory
experimentation can be too costly to be and help seeking – and is thus likely to responses: one is an ‘automatic’ or instinc-
practical. Rewards systems that punish fail- inhibit experimentation. tive response of behavioural inhibition (or
ures increase the costs of experimentation, There are observable and documented inaction) that requires no expenditure of
and may make individuals reluctant to differences between individuals under high mental or cognitive resources; the other is or
experiment. For instance, a study of airline and low evaluative pressure. First, when a more ‘controlled’ response of behavioural
employees showed that reward systems that facing ambiguity and uncertainty, people activation that requires one to exert some
punished individuals when problems arose under high evaluative pressure tend to be level of in-depth processing of the situation.
reduced employees’ willingness to adopt more aware of the potential for punishment Given that being under high evaluative
new routines. and thus are risk averse, leading to behav- pressure can tax one’s mental energy and
ioural inhibition and less experimentation. attention so that fewer cognitive resources
The effects of evaluative pressure In contrast, people under low evaluative are available, it is often the case that the fear
One critical factor in organizations is ‘eval- pressure facing uncertainty tend to be more that results from facing uncertainty produces
uative pressure’ – the degree to which aware of the potential to learn from the sit- the ‘automatic’ response of behavioural
individuals are being closely evaluated and uation and thus are willing to take risks, inhibition. In contrast, those under low
monitored on their performance. Being experience greater behavioural activation, evaluative pressure do not share the
closely evaluated inhibits experimental and engage in more experimentation. salience of failure and punishment, nor do
behaviour, because it reduces psychological Second, being evaluated creates the they experience the same drain on their
safety and the willingness to risk failure. psychological burden of being constantly mental energy and attention, and so they
Evaluative pressure is high when aware of and thinking about one’s perform- have more cognitive resources available with
bosses – or even peers – are seen as judging ance and of the impression one is making which to react to fear in a ‘controlled’ fash-
rather than enabling one’s performance. on the evaluator.This psychological burden ion, characterized by higher levels of
thinking, information processing, and action.

Being closely evaluated inhibits experi- Consistent vs. inconsistent


organizational conditions
mental behaviour, because it reduces How do combinations of organizational
variables affect innovation behaviour? The
psychological safety and the willingness combinational perspective assumes that the
combination of conditions employees face
to risk failure. may be as influential as the individual condi-
tions themselves. The Bank of America
Although most individuals in organizations can tax an individual’s mental energy and example illustrates what can happen when
are being evaluated to some extent, some attention, which in turn can prevent the normative values are changed to explicitly
employees face more evaluative pressure type of in-depth processing that is essential encourage experimentation and instrumen-
than others. Individuals under high evalua- for contemplative, strategic, and effective tal rewards discourage it. Inconsistency in
tive pressure receive intense scrutiny experimentation. organizational conditions may actually do
directed at rating their performance rather Third, evaluative pressure shapes the more harm than good, because it creates
than at providing helpful information or emotional experience of coping with uncer- uncertainty in which individuals do not
feedback. In contrast, those under low eval- tainty. For example, research has shown know which factor (e.g., normative values
uative pressure either receive helpful that, in response to uncertainty, individuals or instrumental rewards) will shape the
information and support from supervisors under high evaluative pressure are more organization’s response to their actions.
or other observers, or else simply face a likely to exhibit guilt, while individuals Consistency, or lack thereof, has impli-
lack of intense scrutiny. under low evaluative pressure are more cations for psychological safety, which is
Evaluative pressure is distinct from likely to exhibit anger. High arousal emo- greatest when organizational conditions such
coaching, in which close attention or mon- tions such as anger might produce more as instrumental rewards, normative values,
itoring is provided to facilitate rather than proactive behaviours directed towards and evaluative pressure are aligned, consis-
evaluate performance. Indeed, monitoring change and innovation – thus responding to tently encouraging experimentation. Under
in the context of supportive coaching can the uncertainty through search – as com- these conditions, the message that failure is
actually enable interpersonal risk taking, pared to the low arousal emotions of guilt, an acceptable element of the innovation
while close and constant evaluation which instead may inhibit proactivity. process is powerful and unambiguous. Thus,

22 • Rotman Magazine Winter 2005


when organizational conditions consistently Further, in consistently-discouraging thereby has a stronger effect on behaviour.
encourage experimentation, we expect conditions, people working closely together Inconsistency in organizational conditions –
more experimentation behaviour than when can experience a sense of solidarity based when some encourage and others discourage
organizational conditions consistently dis- on shared perceptions of negative work experimentation – may undermine experi-
courage experimentation. conditions, whereas inconsistent conditions mentation behaviour, with one factor
In contrast, inconsistency can reduce may lead to mistrust and suspicion that rendering the other ineffective. Inconsis-
psychological safety, and thus experimenta- undermine psychological safety. When an tency creates suspicion, mistrust, fear,
tion. First, inconsistent conditions make organizational context discourages experi- confusion, and risk aversion for individuals
the rules unpredictable and ambiguous.The mentation, the immediate interpersonal under high evaluative pressure.
uncertainty about whether one will be pun- context in a specific work group can still be Individuals under low evaluative pres-
ished creates a state of mild fear, which is characterized by psychological safety, sure are more likely to perceive
antithetical to feelings of psychological
safety. Second, facing the need to simulta-
neously serve contradictory aims itself may Inconsistency lowers psychological
create anxiety, lowering psychological
safety. Inconsistent messages place people safety, increases fear, and makes
in a bind because they communicate two
incompatible goals (e.g., “experiment with failure’s social costs more salient.
new ideas, but don’t fail”). Facing this, peo-
ple may experience emotions of fear or encouraging experimentation, albeit at a themselves as having control over external
anxiety that make taking action and not taking smaller scale. Although this suggests the environments, being able to influence oth-
action equally unpleasant alternatives. possibility that experimentation may be ers to see things their way, more able to
Third, inconsistency has been shown to cre- higher in the consistently ‘discouraging’ effect change, and as more likely to have
ate cognitive and emotional responses such conditions than under inconsistent condi- an internal locus of control. Thus, when
as suspicion, mistrust, and confusion, lead- tions, this goes beyond the core message of organizational conditions are inconsistent
ing to ‘threat rigidity’, a tendency towards this article. and the environment appears uncertain
risk aversion, behavioural inhibition, suppres- In a related vein of inquiry, a recent and unpredictable, individuals under
sion of activity, avoidance, lack of openness, study of HR practices showed that bundles lower evaluative pressure might be able to
and an inability to try novel behaviours. of interrelated and internally consistent HR draw upon these internal, psychological
Because inconsistency lowers psycho- practices create multiple, mutually-rein- resources in order to support experi-
logical safety, increases fear, and makes forcing conditions that support employee menting with different ways to effect
failure’s social costs salient, it in turn motivation and skill acquisition. In this way, change and exert control over this uncer-
inhibits experimentation behaviour. In this an organizational characteristic may have tain environment.
way, inconsistent organizational conditions beneficial effects in combination with one All of these explanations have in com-
can inhibit experimentation, compared to set of organizational attributes but a detri- mon the observation that, facing
consistent organizational conditions. While mental effect in combination with another unpredictability, individuals under high
this argument leads to the somewhat intu- set of organizational attributes. Taking this evaluative pressure are more likely to
itive idea that consistently-encouraging perspective means examining synergistic become inhibited, afraid, narrowly
organizational conditions would encourage and higher-order interactions rather than focused, and rigid, while individuals under
more experimentation behaviour than simply examining main effects or linear less evaluative pressure are more likely to
inconsistent conditions, it also suggests a relationships.This perspective also suggests become proactive, optimistic, thoughtful
less intuitive scenario. Specifically, it is that changing single organizational attrib- and risk seeking.The proposition that indi-
possible that individuals will engage in more utes in a piecemeal or incremental fashion viduals experience different responses to
experimentation behaviour when organiza- may be detrimental to outcomes. combinations of organizational conditions
tional conditions consistently discourage depending on their levels of evaluative
experimentation than when some condi- Conclusion pressure has far reaching implications for
tions encourage experimentation and The finding that inconsistency leads to less behaviour in organizations.
others do not. In the consistently-discourag- experimentation among those under high
ing situation, individuals are clear about the evaluative pressure is consistent with the Harvard Business School Professor Amy Edmondson is the
rules and constraints they encounter, and idea that multiple organizational conditions 2003 recipient of the prestigious Cummings Scholar
therefore may experience more psychologi- should be aligned in the same direction to Award, given annually by the Academy of Management to
recognize outstanding achievement by a researcher in
cal safety than they would when facing the support desired behaviours.When individu- mid-career. This article is based on a paper she wrote
with University of Michigan School of Business Professor
uncertainty created by inconsistent condi- als are exposed to consistent messages, each Fiona Lee and Harvard colleague Stefan Thomke that
tions. If so, they may experiment more. is more likely to be seen as credible and appeared in the June 2004 edition of Organization Science.

Rotman Magazine Winter 2005 • 23

Das könnte Ihnen auch gefallen