The impact of prior entrepreneurial exposure on
perceptions of new venture feasibility and
Article from:Entrepreneurship: Theory and Practice | Article date:September 22, 1993
Entrepreneurship studies typically identify pre-existing entrepreneurs and ventures. Few follow the process
from the idea stage to the ultimate decision to initiate the venture (Long & McMullan, 1984). Research on
intentions clearlydemonstrates that intentions are the single best predictor of planned behaviors. (Starting a
business is hardly an example of simple stimulus-response behavior.) Entrepreneurship is clearly a process
where intentionality is central (Bird, 1988; Katz & Gartner, 1988). Yet, relatively few studies explicitly test
theory-driven process models in entrepreneurship (Carsrud, Gaglio, & Olm, 1987; Gartner, 1988; Low &
MacMillan, 1988) or in strategic management (Stubbart, 1989). Intentions models offer a coherent,
parsimonious and robust framework for pursuing a better understanding of entrepreneurial processes. One
intentions-based model is Shapero's model of the entrepreneurial event (1975, 1982), which hypothesizes that
the intent to start a business derives from perceptions of both desirability and feasibility and from a propensity
to act upon opportunities.
It is encouraging to note the increased interest in studying entrepreneurial intentions. Intentions provide critical
insights into behavior processes (Ajzen, 1987). Shapero's intentions-based model of the decision to initiate a
new venture (1975) has never been explicitly tested (nor has any other formal model of entrepreneurial
intentions). The Shapero process model thus merits testing. Empirical support for the model would
demonstrate the value of the intentions approach, providing a useful framework for researchers, teachers, and
THE IMPORTANCE OF INTENTIONS
In general, intentions toward a purposive behavior are absolutely critical to our understanding the antecedents,
correlates and consequences of that behavior (Ajzen, 1987; Ajzen & Fishbein, 1980). In particular,
entrepreneurial intentions are crucial to understanding the overall process of entrepreneurship because
intentions establish key initial characteristics for new organizations (Bird, 1988; Katz & Gartner, 1988; Krueger
& Carsrud, 1993).
Models based on Ajzen and Fishbein's theoretical framework continue to dominate social psychological
research into intentions. Intentions represent the degree of commitment toward some future target behavior.
Intentions robustly predict and explain that behavior. In turn, attitudes toward a behavior will affect intentions.
Exogenous factors influence intentions and behavior through these attitudes. In their theory of reasoned action,
attitudes toward a behavior consist of two components: an attitude based on expectancies and an attitude
based on social norms (Ajzen & Fishbein, 1980). Ajzen's theory of planned behavior takes these attitudes
representing the attractiveness of a behavior and adds another attitude, perceived behavioral control. This
represents perceptions that the behavior is within the decision maker's control, a necessary precondition for
the behavior to be personally feasible (Ajzen, 1987). "Intentions" here refers to the specific target behavior of
starting a business. This goal behavior is, by definition, planned. Usually, though, the plan or vision of how to
achieve the goal and the specific details of the goal are formulated after identifying the intended goal, Since
starting a business constitutes a complex, distal behavior, intentions (end) and the plan (means) will likely co-
Perhaps the most important consideration is that this class of parsimonious models exhibits a remarkable
robustness in practice. These models are particularly valuable for planned, purposive behaviors (e.g., starting a
business), even more so if the model uses measures of comparable specificity, preferably of moderate
specificity (Sheppard, Hartwick, & Warshaw, 1988). In practice, these models capture the motives underlying
intentional behavior far better than models using more retrospective methods (which are subject to
contaminants such as hindsight bias). Intentional processes such as entrepreneurship are best understood by
tracing subjects through the process regardless of their attitudes, intentions, or behavior. In the case of
entrepreneurial career choice, this suggests identifying samples of subjects currently facing actual major
MODELS OF NEW VENTURE INTENTION: THE SHAPERO MODEL
Shapero's model of new-venture initiation posits that the decision to initiate a new venture requires two things.
First, founders should perceive that starting a new venture is "credible" (i.e., they have intentions toward
entrepreneurship). Starting a new venture must be a believable opportunity. Second, new-ventureinitiation
requires some kind of precipitating (or "displacing") event. In turn, credibility requires at least a threshold level
of perceptions of both feasibility and desirability plus some propensity to act upon the opportunity. (As noted
above, perceived desirability roughly corresponds to the Ajzen-Fishbein model's attitudinal antecedent of
intentions; perceived feasibility corresponds to perceived behavioral control. The Ajzen-Fishbein framework
contains no explicit volitional measure akin to propensity to act.) This study focuses on the first component,
credibility (i.e., intentions). Shapero argued that attitudes toward entrepreneurship (perceived feasibility and
perceived desirability) should partly derive from prior exposure to entrepreneurial activity (Shapero, 1975, 1982)
and affect intentions (and thus behavior) through changing attitudes. Figure 1 graphically depicts Shapero's
model as applied to intentions. Each arrow represents a testable hypothesis.
Shapero (1984) also suggests that the process of forming intentions may prove more complex than shown in
Figure 1. Propensity to act is likely to also have indirect influences on relationships in the model, thus we
should test for moderating effects by propensity to act. Shapero also suggests that intentions may depend on
only a threshold level of feasibility and desirability perceptions, thus we may also want to attempt identification
of threshold effects.
This study begins by testing Shapero's model by examining the direct effect of feasibility and desirability
perceptions and propensity to act on entrepreneurial intentions. The second phase of the research tests the
influence of prior exposure to entrepreneurial activity on attitudes and intentions. The third phase assesses
some of the possible complexities of Shapero's model, including indirect effects of propensity to act. First, let
us examine key conceptual and measurement issues of the model.
Following the intentions literature, we define entrepreneurial intention as the commitment to starting a new
business. (We define entrepreneurship as starting a new business.)
The intentions literature suggests that subjects may have different conceptions of what they mean by
"intentions" (e.g., "intent to start a business" versus "likelihood of starting a business"). Attitudes and
intentions should clearly refer to the same target behavior (Ajzen, 1987).
The "buddingness" of potential entrepreneurs has not been defined or measured consistently. Buddingness is
typically a self-report measure of whether subjects intend to start a business, rarely specifying more about
what kind of business or how soon they intend to do so. Prior research on buddingness often used college
entrepreneurship majors as a convenient unobtrusive alternative. Even where entrepreneurship majors do not
exist, student samples reveal vocational preferences at a time subjects face important career decisions. This
approach explicitly includes subjects with a wide variety of both intentions and attitudes toward
entrepreneurship (Scherer, Adams, Carley, & Wiebe, 1989). Measurement Issues
Shapero proposed to measure entrepreneurial intentions with a dichotomous question about subjects' intent to
start a business. Most other intentions research uses Likert scales, often with multiple measures. However,
Shapero's notion of threshold effects is more easily tested with a dichotomous dependent measure. Also, as
noted above, it is more important that measures of intention and attitudes should be at the same level of
specificity (Sheppard, Hartwick, & Warshaw, 1988). Moreover, subjects should be facing realistic decision
scenarios; that is, they should actually be facing major career decisions (Ajzen & Fishbein, 1980).
In the future, we should ideally assess intentions by kind of venture (e.g., by growth-orientation, craftsperson or
opportunist, promoter or trustee) or by industry (e.g., manufacturing or retail). We should also assess how
soon subjects expect to launch a venture (e.g., "in the near future"). However, pretesting suggests that even if
such specifics may not yet have coalesced in subjects' minds, global intentions toward starting a venture
should still be quite informative (e.g., Krueger & Carsrud, 1993).
PERCEIVED DESIRABILITY AND PERCEIVED FEASIBILITY
Shapero defined perceived desirability as the degree to which one finds the prospect of starting a business to
be attractive; in essence, it reflects one's affect toward entrepreneurship. Perceived feasibility is the degree to
which one believes that she or he is personally capable of starting a business.
The intentions literature argues persuasively that attitudes influence behavior via intentions. The Ajzen-Fishbein
class of models includes constructs related to perceived desirability and feasibility. Perceptions of feasibility
appear closely related to perceptions of self-efficacy (Ajzen, 1987). Intentions and perceptions of feasibility and
desirability may not be related linearly. Attitudes may only need to be above some threshold level to elicit
intentions, much as managers tend to categorize strategic issues as either opportunities or as threats.
Shapero proposed a testable eight-item inventory of questions aimed at different aspects of perceived
desirability (e.g., "Would you love doing it?") and feasibility (e.g., "Do you know enough to start a business?").
The Ajzen-Fishbein model also suggests that multiple measures are useful.
Hypothesis 1: Perceived desirability is positively associated with entrepreneurial intentions.
Hypothesis 2: Perceived feasibility is positively associated with entrepreneurial intentions.
PROPENSITY TO ACT
Shapero conceptualized "propensity to act" as the disposition to act upon one's decisions. That is, propensity
to act reflects the volitional component of intentions ("Will I actually do it?").
Shapero's conceptualization, the propensity to act on an opportunity should depend on one's control
perceptions, specifically, the desire to gain control through taking action. Its effects are likely to be both direct
and indirect and thus its role in this model could be complex. Shapero's model could treat propensity to act as
a moderator rather than a direct influence. One is unlikely to have serious intentions toward a behavior without
perceiving a likelihood of taking action, thus arguing for propensity to act having a direct effect. However,
propensity to act might also be interpreted as moderating other relationships in the model. Recent evidence
argues that for intentions to predict behavior, intentions must be reasonably well formed. It is hard toenvision
well-formed intentions without a significant propensity to act (Bagozzi & Yi, 1989). Moderating effects are also
supported by current thinking both in management research (e.g., upper echelon theory) and in
entrepreneurship research (Gartner, 1988). Upper echelon theory argues that top management characteristics
influence organizations' strategy, but do so indirectly (e.g., Hambrick & Mason, 1984).
Propensity to act may influence the relative impact of experiences upon attitudes and of attitudes on
intentions. If propensity to act is very low, attitudes may be less predictive of intention and action. If propensity
to act is high, then taking action should be more likely seen as desirable and feasible and experiences may
have a greater impact on attitudes. This argues thatpropensity to act might be better viewed as a moderating
influence than a direct antecedent.
This entails identifying a measure that is consistently associated with initiating and persisting at achievement-
oriented behavior under uncertainty. Shapero suggested locus of control as a possible proxy, absent better
measures. Locus of control often fails to differentiate entrepreneurs from managers (Brockhaus & Horwitz,
1986). Burger's "desirability of control" scale has consistently explained and predicted commitment to
achievement-oriented behavior and done so in a variety of settings. Also, it consistently exhibits sound
psychometric properties, showing validity and consistent reliability (e.g., Burger, 1985).
Hypothesis 3: Propensity to act is positively associated with entrepreneurial intentions.
PRIOR EXPOSURE TO ENTREPRENEURSHIP
Ajzen-type models assume that prior experiences will influence intentions indirectly through attitude (including
social norms) and perceived controllability. Similarly, the Shapero model assumes that prior exposure to
entrepreneurial activity operates indirectly through feasibility and desirability perceptions. Prior management
research suggests that top managers share both breadth and quality of business experiences (McCall,
Lombardo, & McCauley, 1988). Thus, breadth of exposure should be a better predictor of attitudes toward
starting a new venture than any single type of experience. Moreover, good experiences should carry a more
positive influence on attitudes than would bad experiences.
Considerable evidence shows that entrepreneurs tend to have had role models of some kind (Brockhaus &
Horwitz, 1986). However, the existence of role models is not necessarily associated with entrepreneurial
intentions (Carsrud, Gaglio, & Olm, 1987). Many entrepreneurs have entrepreneurial parents, but entrepreneurs'
children do not disproportionately become entrepreneurs themselves (Brockhaus & Horwitz, 1986). Scott and
Twomey (1988) argue that entrepreneurial career preferences derive from multiple role models. Early exposure
to a family business appears to influence attitudes and intentions toward entrepreneurship (Krueger, 1993).
This suggests that relevant measures of prior exposure to entrepreneurship should address both quantity (how
much) of exposure and the perceived quality of that exposure (positive or negative). One's breadth of exposure
to different ventures should provide a better measure than using any one specific type of experience.
Quantity (breadth) of experience: A review of the literature suggests four likely sources of exposure: one's
family business, a business started by another relative or friend, working in someone else's small business,
and starting one's own business. All but three subjects had at least one such exposure. Quality (positiveness)
of experiences: If attitudes do indeed depend on experiences, then the impact of those experiences lies in how
subjects perceive them. An objectively bad experience (e.g., bankruptcy) from which the subject learned might
be rated as positive. Scherer et al. (1989) showed that role models did influence entrepreneurial career
preferences by influencing the intervening variable of perceived self-efficacy. Thus, subjects should rate the
quality of exposure as whether it was a positive or negative experience.
Hypothesis 4a: Breadth of experience is associated with perceived feasibility.
Hypothesis 4b: Breadth of experience is associated with perceived desirability.
Hypothesis 5a: Positiveness of experience is associated with perceived feasibility.
Hypothesis 5b: Positiveness of experience is associated with perceived desirability.
Phase I: Correlational analysis and t-tests tested the hypotheses regarding proposed direct antecedents of
entrepreneurial intentions: perceived feasibility, perceived desirability, and propensity to act (tests H1, H2, H3).
We also performed confirmatory analyses of Shapero's measures.
Phase II: Exploratory path analysis tested whether proposed measures of prior entrepreneurial experiences are
significant antecedents of entrepreneurial attitudes (tests H4ab, H5ab).
Phase III: Purely exploratory analyses tested the above relationships for possible moderator and threshold
DATA AND MEASURES
The sample consisted of 126 upper-division university business students (75 male, 51 female) currently
approaching career decisions. As noted above, such a sample provides subjects who have yet to face
important career choices. Career aspirations among adolescents are significantly predictive of eventual career
choice (Trice, 1991). This sample also provides a relatively homogeneous sample in terms of age and
education, yet offers a sufficient range of experiences, intentions and attitudes toward entrepreneurship, and
dispositions. This permits us to examine the entrepreneurial process prior to actual entrepreneurial activity.
Moreover, in non-entrepreneur samples, students appear reasonably comparable to non-students on
entrepreneurial attitudes and in strategic decision making (Bateman & Zeithaml, 1989). However, future
replications will compare actual managers and entrepreneurs and compare different types of entrepreneurs.
This sample should also be available for longitudinal follow-up. Given a sample size of 126, a one-tailed test
with alpha = .10, and an expected effect size of at least .2 (after adjusting for reliabilities), we anticipate less
than 20% chance of a Type II error.
Intention and Credibility Measures
Source: Pretested measures from unpublished questionnaire used in Shapero (1982, 1984).
Entrepreneurial intentions: "Do you think you'll ever start a business?" (yes or no item). |Also included was
Cooper, Woo, & Dunkelberg's (1990) measure of entrepreneurial optimism to validate the intentions measure.(*)
Credibility measures: Feasibility and desirability were assessed via eight 7-point Likert scales. Subjects were
asked, "If you actually started your own business, how would you feel?"
1) "How hard do you think it would be?" (very hard--very easy)
2) "How certain of success are you?" (very certain of success--very certain of failing)
3) "How overworked would you be?" (very overworked--not overworked at all)
4) "Do you know enough to start a business?" (know everything--know nothing)
5) "How sure of yourself?" (very sure of myself--very unsure of myself)
1) "I would love doing it" (I'd love doing it--I'd hate doing it)
2) "How tense would you be?" (very tense--not tense at all)
3) "How enthusiastic would you be?" (very enthused--very unenthusiastic)
Source: Pretested exposure items from Shapero; positiveness items developed for this study.
Breadth of entrepreneurial experience: Subjects were asked whether they had been exposed to each of four
possible types of entrepreneurial experience. Breadth of Experience is the sum of these four yes-no questions
(coded 1 for yes, 0 for no).
1) "Did your parents ever start a business?" |91 Ss reported Y~
2) "Did anyone else they knew start a business?" |97 Ss reported Y~
3) "Did they ever work for a small or new company?" |102 Ss reported Y~
4) "Did they themselves ever start a business?" |25 Ss reported Yes~
Positiveness of entrepreneurial experience: After subjects answered each of the four above items, they rated
the experience as positive or negative. For each item, positive responses were coded + 1 while negative
responses were coded as - 1. (No exposure was coded as 0.) Positiveness of experience equals the sum of
these items. Measurement of positiveness for subjects who have no exposure is problematic; however, there
were only three such subjects.
Propensity to Act Measure
"Desirability of control" (Burger, 1985) measures preference for having control over life events. It has
consistently predicted persistence at achievement-related behaviors. (Reliability: Cronbach's alpha was .78.
Burger reports alpha of .80 and a test-retest reliability of .81.)
Phase I Results: Intentions versus Credibility Perceptions
Initial examination of the data suggests that there are indeed strong relationships among the variables. Table 1
presents the results of t-tests comparing the means of the antecedent variables for intending and non-intending
subjects. Note the magnitude of most differences. The correlations in Table 2 suggest that both perception
measures and propensity to act correlated significantly with entrepreneurial intentions, supporting Hypotheses
1, 2, and 3. Collectively, they appear to explain a significant fraction of the TABULAR DATA OMITTED variance
despite the possible limitations of a dichotomous intentions measure (and the need to use biserial
correlations). Perceived feasibility and perceived desirability were not correlated. However, some of the
correlations may prove spurious, thus we need to test them further using path analysis (see Phase II).
Correlation Matrix: Entrepreneurial Intentions and Antecedents
(2) (3) (4) (5) (6)
(1) Entrepreneurial Intentions(*) .32(a) .23(b) .47(a) .37(a) .21(c)
(2) Perceived Feasibility -.03 .19(c) .17 .32(a)
(3) Perceived Desirability .25(b) .19(b) .24(b)
(4) Propensity of Act .25(b) .11
(5) Breadth of Experience .23(b)
(6) Positiveness of Experience
a Significant at .001
b Significant at .01
c Significant at .05
* Note: correlations with intentions are biserial correlations.
We also need to examine the reliability and validity of the measures employed. Entrepreneurial Intentions had
a biserial correlation coefficient with Entrepreneurial Optimism of + .33. This coefficient is significant at .001,
suggesting the likely validity of the intentions measure. Factor analyzing the feasibility and desirability
measures yielded two clear factors onto which the eight items loaded cleanly. The perceived desirability scale
had a Cronbach's alpha of .77; the perceived feasibility scale had a Cronbach's alpha of .57.
Phase II Results: Exposure and Credibility Perceptions
In this model, testing Hypotheses 4 and 5 implies examining the influence of the experience variables on both
the attitude measures and on intentions directly. This implies a method that will identify correlations which are
spuriously present (or absent). Path analysis is a useful way to do this. Path analysis isvaluable for identifying
a parsimonious model where one has at least an implicit causal ordering and most variables are correlated, as
in Table 2 (Asher, 1976).Path analysis entails regressing each model variable on all "prior" variables to control
for spurious correlations. The final model is specified by redoing theregressions after pruning all non-significant
paths. The standardized regression (beta) coefficients comprise the path weights (and show the relative impact
of predictors). Path analysis thus permits us to refine our models by this pruning process. The result in Figure
2 is a model that differs slightly from that hypothesized in Figure 1.
For instance, Table 2 shows that intentions and breadth of experience are significantly correlated. However,
path analysis shows that we can explain this covariance by intervening variables. The direct path between
intentions and breadth of experience is now non-significant, after controlling for causally prior variables.
Similarly, note that the direct path between propensity to act and intentions is larger than the biserial
correlation coefficient might suggest. Shapero presumed that perceptions of desirability and feasibility would be
correlated, but the path analysis verified that his measures were independent.
Factor Analysis: Feasibility and Desirability Measures
Factor Loadings (decimal point omitted)
(FEAS1) 52 -02
(FEAS2) -32 04
(FEAS3) 43 -20
(DESIR1) 21 -76
(DESIR2) 34 53
(DESIR3) 01 41
(DESIR4) -13 -56
(DESIR5) 02 81
Principal factor analysis, Harris-Kaiser oblique rotation.
Only boldface factors are significantly correlated.
Although intentions were measured dichotomously (as per Shapero's formulation),
this dichotomy is arbitrary, given that intentions are typically considered as a continuum. Thus, the path
analysis need not use logistic regression (Neter, Wasserman, & Kutner, 1983).(1) Instead, for consistency
across the path analysis, least squares regression was employed. (Logistic regressions yielded substantively
similiar results.) The regression results showed that intentions were significantly associated with perceived
feasibility, perceived desirability, and propensity to act. However, intentions are no longer associated with
either exposure variable. Note that the three hypothesized measures account for over half of the variance (54%)
Table 4 presents both the initial and final regression for each dependent variable (e.g., the second regression
on intentions prunes the non-significant paths and provides the standardized coefficients).
The analysis continues with regressions on the other critical variables. Breadth of Experience is still
associated significantly with perceived feasibility but no longer with perceived desirability. (This supports
Hypothesis 4a, but not Hypothesis 4b.) Positiveness of Experience is still significantly associated with
perceived desirability but not with perceived feasibility. (This supports Hypothesis 5b, but not Hypothesis 5a.)
Figure 2, which includes all significant path coefficients (p |is less than~ .01), argues for two conclusions. First,
Shapero's model is significantly supported (Hypotheses 1, 2 and 3). Second, we can explicitly add measures
of prior entrepreneurial exposure (supporting Hypotheses 4b and 5a).
TABULAR DATA OMITTED
Phase III Results: Possible Moderator and Threshold Effects
As noted above, Shapero's conception of the model included the possibility of moderator and threshold effects.
We now briefly explore these possibilities.
Possible moderator effects: Moderator effects are usually illustrated by (1) making a median split on propensity
to act, then (2) re-analyzing critical relationships in the model for the two levels of the hypothesized moderator.
However, this approach has a serious limitation. Using a median split more than halves the statistical power
which significantly increases the likelihood of Type II errors. That is, the significance of paths may be
spuriously understated (e.g., Mossholder, Kemery, & Bedeian, 1990). To answer this, a moderated regression
tested the significance of proposed moderator effects, followed by a median split approach (but strictly for
The moderated regression included terms for interactions between the measure of propensity to act and the
attitude measures (desirability and feasibility). The overall model showed a significant increase in |R.sup.2~,
but neither interaction term was significant. Despite the ambiguous results, the improvement in |R.sup.2~
suggests that predictability does apparently increase for higher levels of propensity to act.
Figure 3 shows the re-estimation of the path model for high and low propensity to act. They include only those
paths which remain significant at p |is less than~ .10. (Given the lower statistical power, we increased alpha to
partly compensate.) Under low propensity to act, the regression on intentions had an adjusted |R.sup.2~ of
only .160 (p |is less than~ .01) while the adjusted |R.sup.2~ for high propensity to act was .322 (p |is less
than~ .001), more than double.
The differences between the models suggests that the indirect influence of propensity to act (i.e., on
predictability) is meaningful. For low propensity to act, perceived feasibility is no longer a significant antecedent
of intention. For high propensity to act, positiveness of experience retains a direct influence on intentions,
beyond its indirect impact through its effect on desirability. Recall that Shapero assumed that, given a
significant propensity to act, past experiences should have a greater impact on attitudes and intentions. For
instance, the exposure measures were correlated only for
high propensity. That is, subjects with high propensity to act appear more likely toperceive a given level of
entrepreneurial experience as positive. However, given that the sample size is now halved, the likelihood of
Type II errors (spurious non-significance) has substantially increased, arguing for further analysis. In sum,
propensity to act appears to play some indirect role in the formulation of entrepreneurial intentions (as well as a
significant direct influence). However, ambiguous results emphasize the need to further explore the model's
Threshold effects: Shapero (1984) suggests that for a new venture to be perceived as credible, both perceived
desirability and perceived feasibility need only cross some threshold. However, differentiating true threshold
phenomena is far from simple; a joint threshold effect requires additional analysis beyond the scope of this
study. Table 5 does show significantly higher entrepreneurial intentions where both perceived feasibility and
perceived desirability were above both their respective medians, significantly lower below both. Above both
thresholds, 86% expressed an intent to start a business (versus 58% overall). Below both, only 38% did so. If
subjects were above only one threshold, a below-average 54% reported intending. A chi-square test of this
threshold phenomenon was statistically significant at p |is less than~ .001. These results could be interpreted
as a joint threshold effect (i.e., bothattitudes need to be above some threshold). In any event, threshold effects
seem also worthy of further investigation.
CONCLUSIONS, DISCUSSION AND IMPLICATIONS
Confirmation of Shapero's Model
This exploratory study found significant support for Shapero's propositions that entrepreneurial intentions derive
largely from (1) perceptions of feasibility, (2) perceptions of desirability, and (3) a propensity to act which
derives from control beliefs. Path analysis demonstrated that the impact of prior entrepreneurial exposure on
intentions is indirect, operating through perceived feasibility. The positiveness of those experiences also
indirectly influences intentions through perceived desirability. Finally, we found suggestions that intentions
formation may be more complicated than the tested model (e.g.,propensity to act might be a significant,
complex moderator of key relationships in the model).
Percent of Subjects Intending, by Credibility Levels
Low 37.5% 51.5%
High 55.2% 85.9%
Implications of Shapero's Model
These findings argue that we should test richer models of entrepreneurial antecedents and use better-refined
measures. In particular, these results clearly endorse further research applying formal models of intentions
(e.g., testing the empirically robust Ajzen-Fishbein model). These results should hearten proponents of
process models in entrepreneurship and encourage further process-based research. As noted above,
intentional behavior typically involves identifying goals before identifying the means to achieve them. This helps
explain Brockhaus's findings that a majority of entrepreneurs decide to start a business prior to deciding what
type of business to start; it is characteristic of intentional behavior. Increased use of process models may help
us lay to rest additional "entrepreneurial folklore" (Brockhaus, 1987).
An intentions-based framework offers a mechanism to assess the relative impact of various hypothesized
exogenous influences (e.g., perceptions of resource availability) on intentions and, ultimately, behavior. Martin's
(1985) model proposes an interesting menu of potential antecedents. It is a critical distinction to find that
exogenous factors influence entrepreneurship indirectly, not directly, and do so only insofar as they affect
attitudes. We now have evidence to explain why Carsrud, Gaglio, and Olm (1987) found less impact by role
models on entrepreneurial intentions than Scherer et al. (1989) did. That is, role models must affect attitudes in
order to affect intentions.
Some argue that entrepreneurship models ought not include trait measures unless there is a clear theoretical
justification (Gartner, 1988). The measure used for propensity to act (desirability of control) is already linked,
conceptually andempirically, to a strong goal orientation (Burger, 1985). Its role in this model suggests that
this instrument may find other productive uses in entrepreneurship research.
Implications for Researchers
Again, these results confirm 20 years of social psychology: process models are clearly indicated for studying
intentional behavior. We also need not use samples of actual entrepreneurs (though we must still carefully
select our samples). Intentions models typically use multiple measures of critical variables, analyzing them via
structural covariance modeling (Ajzen, 1987).
The Shapero model offers refinements to the Ajzen-Fishbein approach. Such models lack an explicit volitional
dimension (such as Shapero's propensity to act). Also, in research examining the intention-behavior linkage,
Shapero's notion of the precipitating event may prove useful. Finally, Shapero's two components(perceived
feasibility and desirability) offer even more parsimony than does the Ajzen-Fishbein model. These results raise
the question of how experiences get translated into attitudes. For instance, increasing subjects' tacit
knowledge could increase perceptions of feasibility (see Reuber, Dyke, & Fischer, 1990). Research should now
examine whether (and how) intentions become reality. Given the often large time lags involved, the intention-
behavior link may be weak (Ajzen, 1987; Katz, 1989). Studying this as a process itself should be fruitful (Bird,
1988; Katz & Gartner, 1988). As noted below, we have an opportunity to extend these results to a higher level
of specificity (exposure to a high-growth business versus attitudes and intentions toward starting a high-growth
business). A within-subjects approach could prove illuminating.
Implications for Training Entrepreneurs
Training should focus on increasing perceptions of both desirability and feasibility. Teaching a relevant skill is
not enough. The trainee/student needs to perceive that actual use of the skill is feasible. Research on
perceived self-efficacy strongly supports this conclusion.
This model also suggests implications for selecting trainees based on existing attitudes and propensity to act.
Where we cannot observe these, we may be able to measure their prior exposure. Employers may find ways to
use these background experiences to identify entrepreneurial managers.
Implications for Practitioners
Both current and potential entrepreneurs may find that negative experiences have biased them against certain
strategies needed for their proposed businesses. Since this approach captures motives (rather than relying on
possibly biased recall measures), it should help prospective entrepreneurs better understand their motives and
critical assumptions about their businesses. Useful Future Extensions and Refinements
Several improvements immediately suggest themselves, particularly through increasing the specificity of the
measures employed. Improving the perceived feasibility measure's reliability of only .57 is certainly called for.
Subjects' intentions need clarification. It should be instructive to distinguish amongthose entering their own
family business (perhaps through inheritance), starting a new business, or other self-employment. Measuring
exposure could also be much more sophisticated. Reviews of entrepreneurial background invariably note that
specific industry experience plays a meaningful role (Brockhaus & Horwitz, 1986). Positive experiences in one
specific industry may not transfer to different industries. Even the most positive exposure to craftsperson firms
may lend little credibility to launching an opportunist-type firm. We can separately analyze the impact of
exposure on intentions by type of venture to which subjects had exposure (growth-orientation, craftsperson-
opportunist, industry sector). Does exposure to one type of venture affect perceptions and intentions of other
types? This would lend itself to analysis of within-subject differences. We can analyze entrepreneurial
intentions and attitudes by type of exposure (e.g., vicarious versus hands-on experience, employment versus
managerial involvement in an entrepreneurial business) and by duration of exposure.
We should also further test the complexities explored in Phase III. A non-dichotomous measure of intentions
might clarify ambiguous moderator effects. Shapero assumed the existence of threshold effects, though
management research rarely tests for such effects. Explicitly identifying the threshold effects would contribute
to a better understanding of managerial cognitions such as strategic issue diagnosis. As perceived self-efficacy
is already linked to opportunity and threat recognition (Krueger & Dickson, 1993), perhaps perceived
desirability and propensity to act might also help predict opportunity and threat identification.
In sum, the statistical support found for Shapero's model argues that research into the processes of
entrepreneurship (especially models based on intentions) will prove most fruitful.
* Entrepreneurial Optimism equals the second probability asked below subtracted from the first. "If you actually
started your own business, what is the probability that in 2 years your business will be successful? If someone
else like you started this business, what's the probability of success in 2 years?"
1. One reviewer correctly pointed out that the intentions measure is an arbitrary dichotomy, based on a
continuous underlying phenomenon. If so, this would permit use of least squares instead of logistic regression,
if the measures of intentions and attitudes had a bivariate normal distribution (Neter, Wasserman, & Kutner,
1983). In any event, results using logistic regression did not differ materially. This insight was appreciated.
Ajzen, I. (1987). Attitudes, traits and actions: Dispositional prediction of behavior in social psychology.
Advances in Experimental Social Psychology, 20, 1-63.
Ajzen, I., & Fishbein, M. (1980). Understanding attitudes and predicting social behavior. Englewood Cliffs, NJ:
Asher, H. (1976). Causal modeling. Beverly Hills: Sage.
Bagozzi, R., & Yi, Y. (1989). The degree of intention formation as a moderator of the attitude-behavior
relationship. Social Psychology Quarterly, 52, 266-279. Bateman, T., & Zeithaml, C. (1989). The psychological
context of strategic decisions: A model and convergent experimental findings. Strategic Management Journal,
Bird, B. (1988). Implementing entrepreneurial ideas: The case for intentions. Academy of Management Review,
13, 442-454. (1987).
Brockhaus, R. (1987). Entrepreneurial folklore. Journal of Small Business Management, 25(3), 1-6.
Brockhaus, R., & Horwitz, P. (1986). Psychology of the entrepreneur. In D. Sexton & R. Smilor (Eds.), The art
and science of entrepreneurship, pp. 25-48. Cambridge, MA: Ballinger.
Burger, J. (1985). Desire for control and achievement-related behaviors. Journal of Personality and Social
Psychology, 48, 1520-33.
Carsrud, A., Gaglio, C., & Olm, K. (1987). Entrepreneurs-mentors, networks and successful new venture
development. American Journal of Small Business, 12(2), 13-18.
Cooper, A., Woo, C., & Dunkelberg, W. (1988). Entrepreneurs' perceived chances for success. Journal of
Business Venturing, 3(2), 97-108.
Gartner, W. (1988). "Who is the entrepreneur?" is the wrong question. Entrepreneurship Theory and Practice,
Hambrick, D., & Mason, P. (1984). Upper echelons--the organization as a reflection of its top managers.
Academy of Management Review, 9(2), 193-206.
Katz, J. (1989). Intentions, hurdles and start-ups: An analysis of entrepreneurial follow-through. Paper
presented at the Frontiers of entrepreneurship research, Babson Park, MA.
Katz, J., & Gartner, W. (1988). Properties of emerging organizations. Academy of Management Review, 13,
Krueger, N. (1993). Growing up entrepreneurial? Some developmental consequences of early exposure to
entrepreneurship. Paper presented at the Academy of Management, 1993.
Krueger, N., & Carsrud, A. (1993). Entrepreneurial intentions: Applying the theory of planned behavior.
Entrepreneurship and Regional Development, in press. Krueger, N., & Dickson, P. (1993). Perceived sell-
efficacy and perceptions of opportunity and threat. Psychological Reports, 72, 1235-1240.
Long, W., & McMullan, W. (1984). Mapping the new venture opportunity identification process. Paper
presented at the Frontiers of entrepreneurship research, Babson Park, MA.
Low, M., & MacMillan, I. (1988). Entrepreneurship: Past research and future challenges. Journal of
Management, 14, 139-151.
Martin, M. J. C. (1985). Managing technological innovation and entrepreneurship. Reston, VA: Reston.
McCall, M., Lombardo, M., & McCauley, N. (1988). The lessons of experience: How executives develop on the
job. Lexington, MA: Lexington Books.
Mossholder, K., Kemery, E., & Bedeian, A. (1990). On using regression coefficients to interpret moderator
effects. Educational and Psychological Measurement, 50(2), 255-263.
Neter, J., Wasserman, W., & Kutner, M. (1983). Applied linear regression models. Homewood, IL: Irwin.
Reuber, A., Dyke, L., & Fischer, E. (1990). Experientially acquired knowledge and entrepreneurial firm
performance. Paper presented to the Academy of Management, San Francisco.
Scherer, R., Adams, J., Carley, S., & Wiebe, F. (1989). Role model performance effects on development of
entrepreneurial career preference. Entrepreneurship Theory and Practice, 13, 53-81.
Scott, M., & Twomey, D. (1988). Long-term supply of entrepreneurs: Students' career aspirations to
entrepreneurship. Journal of Small Business Management, 26, 5-13.
Shapero, A. (1975). The displaced, uncomfortable entrepreneur. Psychology Today, 9 (November), 83-88, 133.
-----. (1982). Social dimensions of entrepreneurship. In C. Kent et al. (Eds.), The Encyclopedia of
entrepreneurship, pp. 72-90. Englewood Cliffs, NJ: Prentice Hall.
-----. (1984). Why entrepreneurship? Working paper, Babson College.
Sheppard, B., Hartwick, J., & Warshaw, P. (1988). The theory of reasoned action: A meta-analysis of past
research with recommendations and future research. Journal of Consumer Research, 15, 325-344.
Stubbart, C. (1989). Managerial cognition: A missing link in strategic management research. Journal of
Management Studies, 26(4), 325-347.
Trice, A. (1991). Relationship among first aspirations, parental occupations and current occupations.
Psychological Reports, 68(1), 287-290.
Norris F. Krueger is a Management Consultant with Entrepreneurial Strategies, Bozeman, MT.
Get full access to HighBeam Research articles
COPYRIGHT 2007 Baylor University