find research objectives/ hypotheses

Scope:

Objectives:

from the paragraph below,

paragraph:

Early studies of entrepreneurial career choice mainly focused on issues such as personality variables, demographics, personal history, and social contexts in their explanations of individuals' choices and preferences with respect to their entrepreneurial status (Gibb Dyer, 1994). As these explanations are all distal (Rauch and Frese, 2000), they merely apply to broad classes of behaviour, and their explanatory power turned out to be low, which makes it difficult to formulate guidelines for intervention. In the 1990s, researchers started to use social psychological models involving more proximal variables. The explanation of entrepreneurial intentions (EI) is an area of research where a sizeable body of comparable studies has

emerged. So far research has focused on the prediction of EI rather than on its realisation. Meta-analyses show that intentions are strong predictors of actual behaviour in other applied settings (Sutton, 1998; Armitage and Conner, 2001). To date, two models dominate the literature.

Explaining entrepreneurial intentions

The first is Ajzen's (1988, 1991) theory of planned behaviour (TPB), which explains

intentions by means of attitudes, perceived behavioural control (PBC), and subjective

norms. The second model is proposed by Shapero and Sokol (1982), and explains EI on541the basis of perceived desirability, perceived feasibility and the propensity to act.

Although Kruegeret al.(2000) regard these models as competing, they overlap to a

large degree. Shapero's perceived desirability and perceived feasibility correspond to

Ajzen's attitudes and perceived behavioural control, respectively (Krueger, 1993;

Kolvereid, 1996b; Autioet al., 1997). So in both models intentions are explained by

willingness and capability. Both models have consistently received empirical support.

In a direct comparison of the two models, Kruegeret al.(2000) conclude that both

models provide satisfactory predictions (Shapero model adjustedr2140:41 (p,0:00)

and TPB model adjustedr2140:35 (p,0:00)). Effects for the PBC/feasibility

component tend to be stronger than for the attitude/desirability component (Autioet al.,

1997; Davidsson, 1995; Kolvereid, 1996b; Kruegeret al., 2000; Tkachev and Kolvereid,

1999).

In our research design we use the theory of planned behaviour. This means that we do not use the Shapero model, and we disregard additional variables outside of the TPB that might explain EI. Our reasons for not using the Shapero model are related to its specification. Shapero and Sokol (1982, p. 83) conceptualise desirability in terms of social norms, while propensity to act is operationalised in terms of control measures (Krueger, 1993). Both specifications are confusing. The theoretical specification of the TPB is more detailed and consistent, and a great deal of research has been devoted to testing, advancing and criticising the TPB in a wide variety of fields (Fayolleet al., 2006; Shooket al., 2003).

With regard to the additional variables, additional variables such as gender, work experience, parental role models, and personality traits do in fact enhance our understanding of EI. However, we assume that the effect of these variables is mediated by the influence of the components of the TPB on EI. For example, DePillis and Reardon (2007) show that differing cultural perceptions of entrepreneurship in Ireland and the USA affect the level of EI, which would be mediated in the "subjective norms" component of the TPB. Drennanet al.(2005) demonstrate that parental business experience as well as a difficult childhood have a positive impact on both perceived desirability and perceived feasibility of starting one's own business. Whether the effects of these variables on EI are indeed only indirect is not the focus of this paper. By solely concentrating on the components that make up the TPB, we facilitate comparisons with previous research.

The attitude/desirability component has been researched with more specificity than the PBC/feasibility component. For example, attitude to autonomy was found to be related to EI by Douglas and Shepherd (2002) and by Kolvereid (1996b), although Davidsson (1995) did not find such an effect. Douglas and Shepherd (2002) further found attitudes towards risk to be related to EI, while workload and income attitudes were not. Davidsson (1995) reports that attitudes toward achievement and towards change explain EI. A range of other studies explaining EI by means of attitudes only

CDI 13,6

542

report outcomes for attitude on a generic level (see Table I). This goes for nearly all studies investigating the PCB/feasibility component (see Table I), with the exception of Kristiansen and Indarti (2004). They find EI to be explained by self-efficacy as well as by instrumental readiness, a contextual variable reflecting the individual's situation with regard to access to capital, information and networks. It is very useful to establish results on the level of the variables that make up the attitude and PBC components. Detailed knowledge of this kind is necessary for the design of interventions that may influence EI.

In addition, these variables should be selected with care. Attitudes, PBC, and subjective norms are theorised to be determined by two elements:

1. (1)beliefs about outcomes; and
2. (2)evaluations of these outcomes.

Pre-selection of relevant beliefs should be done carefully, as beliefs can be expected to vary among different populations. Younger people, for example, might have a different perception of the desirability or undesirability of factors associated with self-employment than older people. In order to capture the beliefs relevant to our population sample, we therefore included pilot studies in our research. Applying this procedure to our research project resulted in factors associated with students' perceived attractiveness or unattractiveness of self-employment as well as factors related to the perceived feasibility of starting or running one's own business. Still, we also included variables found to be important in the literature if they were hardly mentioned by the students. No previous studies (see Table I) selected variables based on the samples they studied, although Kolvereid (1996a) empirically derived career

Table I.

Designfeaturesemployed in this paper

Notes:SCV, split components into variables; ATT, attitudes component; PBC, perceived behavioural control; VSDP, variable selection derived from population; CAVO, check for association of the variables with the outcome (organisational versus self-employment); URDV, report on a range of dependent variables; RRSC, replicate results in comparable samples;aused theory of planned behaviour;, design includes feature;W, not so

reasons from a sample of MBA students which he later (Kolvereid, 1996b) applied to the study of EI among university students.

While Kolvereid (1996a) found no such conflicting perceptions, Brenneret al.(1991)

report that students sometimes favour organisational or self-employment for the same

reasons: both groups expected to earn a higher income, to work with people they

respect, and to have greater opportunities for development in their preferred

employment mode as opposed to their non-preferred one. This means that these543variables will only predict career status choice within the groups of preference, while

for the aggregate sample the effects will cancel out. This adds another feature to our design, not included in previous studies (see Table I): to correct, if necessary, for association beliefs concerning the attributes of organisational employment versus self-employment.

We now turn to the dependent variable. Intentions represent a person's motivation to make an effort to act upon a conscious plan or decision (Conner and Armitage, 1998). However, in the social psychological literature controversy has emerged about the measurement of intentions (Warshaw and Davis, 1985; Bagozzi, 1992; Bagozzi and Kimmel, 1995; Armitage and Conner, 2001). Depending on the formulation of the questions, these measures represent desires (do you want to start a business?), preferences (if you could choose between being self-employed and being employed by someone, what would you prefer?), plans (are you planning to start a business?), or behavioural expectancies (estimate the probability that you will start your own business in the next five years).

This is especially relevant when studying the EI of samples of undergraduate students. Some students are as yet undecided when it comes to their career preference, and are still exploring their options (Nabiet al., 2006; Schein, 1978, 1990). Others have goals that often change; in other words, they suffer from goal instability (Multonet al., 1995). In a review on career decidedness types, Gordon (1998) postulates seven different subtypes ranging from very decided to chronically indecisive. In addition, it has been argued that behavioural expectancies provide better predictions of behaviour than other measures of intention (Warshaw and Davis, 1985). This is because behavioural expectancies include considerations regarding the possible choice of other, competing behaviours (Armitage and Conner, 2001; Silvia, 2001). Non-committal measures, such as desires, take no account of facilitating or inhibiting factors.

Different operationalisations of EI result in differing explanations and predictions. Empirical evidence (Bagozzi and Kimmel, 1995; Armitage and Conner, 2001) shows measures of perceived behavioural control (PBC) to be more closely associated with commitment measures, such as behavioural expectations. In studies of EI, Phanet al.(2002) found attitude to be a stronger predictor of interest in self-employment than of likelihood to start. Most studies listed in Table I have ignored this issue, or investigated other types of dependent variables (such as planning to start in one's own field or a different one; Lee and Wong, 2004). In this study, we operationalise EI in several ways, and verify whether this results in different levels of importance of the explanatory variables. In the remainder of this paper we will use the label "entrepreneurial intention" (EI) as an overarching term.

Replication is the final feature of our design. The obvious rationale for testing the robustness of our findings is to ascertain the stability and reliability of the results. Deploying multiple datasets in a single study to test a theoretical model makes it

Explaining entrepreneurial intentions

CDI 13,6

544

possible to test for the stability of results (Davidsson, 2004). In previous studies on EI using replication designs, results were found to be stable across countries (Autioet al., 1997; Kristiansen and Indarti, 2004), regions (Davidsson, 1995), and samples of MBA and university students (Wilsonet al., 2007).

Pilot studies and hypothesis development

Pilot studies

Pilot studies were conducted among undergraduate students of business administration of the Free University of Amsterdam (n14200) and the University of Amsterdam (n14173). Rather than working with individual beliefs, it is customary in studies based on the TPB to focus on modal beliefs identified in pilot studies (Ajzen, 2002). Usually a frequency-of-elicitation procedure is used, according to which the beliefs most often mentioned are included in the major study. So outcomes, referents and inhibitory/facilitating factors are not assessed per individual, but they are pre-selected. Using modal beliefs has the advantage that the whole sample can be compared on the basis of similar variables.

For the attitude formation open questions were used to gather data on outcome beliefs. Two questions were asked:

1. (1)What aspects do you think are attractive about self-employment?
2. (2)What aspects do you think are unattractive about self-employment?

The answers were used to select the attitude variables in our study (see below). Open questions were also used to determine the students' control beliefs:

1. (1)What is needed to set up a business?
2. (2)What is needed to successfully run a business?

The answers were used to select the PBC variables in our study (see below). Subjective norms were not investigated in our pilot studies. We assumed that they are related to spouse, family, friends and "important others" (Kolvereid, 1996b).

Attitudinal variables in the model

The answers to the open questions in the pilot study were content-analysed. Four outcomes occurred with the highest frequency, i.e. autonomy and challenge (attractive aspects), and lack of financial security and workload (unattractive aspects). Based on these beliefs, we hypothesised the importance attached to autonomy and challenge to be positively related, and the importance attached to financial security and the avoidance of work load to be negatively related to the students' intentions of setting up a business. In addition, based on the literature, we decided to add the importance attached to the accumulation of income and wealth as an additional variable. It is likely that students mention these factors less often out of social-desirability considerations (materialism is not regarded as a favourable trait by the Dutch, although one would hardly guess that from their behaviour). Wealth in the context of self-employment refers to the increase in the value of the firm as well as in salary and benefits. When working for an organization the amount of wealth that one can accumulate is relatively fixed, whereas in self-employment the opportunities to acquire wealth are (at least theoretically) infinite.

H1.Students who attach more importance to autonomy, challenge, and wealth accumulation, and less importance to financial security and work load avoidance, are more likely to have intentions of starting a business.

Perceived behavioural control variables in the model

Explaining entrepreneurial intentions

The answers to the two questions regarding PBC beliefs were generally identical. In

the opinion of the students, starting as well as successfully running a business545primarily depend on perseverance and creativity. Based on the student beliefs, we

hypothesise perseverance and creativity to be positively related to EI. In addition,

based on the literature, we decided to add measures of entrepreneurial alertness and

self-efficacy. Entrepreneurial alertness (Kirzner, 1973) was added, since business

students might have overlooked that sensitivity to detecting business opportunities is

a precondition for entrepreneurship. Self-efficacy was added as it is a common

operationalisation of PBC. PBC was originally formulated as the perceived ease or

difficulty with which the behaviour involved is performed. According to this

formulation, PBC is compatible with Bandura's (1977, 1982) concept of perceived

self-efficacy (Ajzen, 1991). Previous work presents empirical support for the

relationship between entrepreneurial self-efficacy and EI (Kolvereid, 1996b, Krueger

et al., 2000; Wilsonet al., 2007).

H2.Students who rate themselves higher in terms of perseverance, creativity, entrepreneurial alertness, and entrepreneurial self-efficacy, are more likely to have intentions of starting a business.

Subjective norms

Previous EI research has proven subjective norms to be important (Krueger, 1993; Kolvereid, 1996b). One reason for this might be that generally students are still in the stage of finding out their career choice preferences. The opinions of parents, partners, friends and important others might be influential in this process.

H3.Students whose subjective norms towards self-employment are more positive are more likely to have intentions of setting up a business.