Implicit motives and their assessment

Current Personality Psychology distinguishes between two major elements of personality that affect and shape mental processes and behavior, namely personality characteristics or traits and motivational dispositions or motives [17,18]. These two distinct aspects of personality–traits and motives–have led separate lives, each developing its own measurement instruments. Yet, both streams of research converge on the idea that behavior is energized and directed by two distinct types of personality systems–namely an implicit system that operates outside of conscious awareness and control, and an explicit system that functions at a conscious level [19,20]. Indeed, implicit and explicit motives differ in their impact on behavior. In general, implicit motives predict spontaneous, uncontrolled behavior, and effort-related task performance, whereas explicit motives predict behavior that is subject to conscious thought and deliberation, such as self-reflective appraisals, judgments and deliberate choices [19,21]. For example, Brunstein and Schmitt [22] showed that the implicit motive to achieve was independent from the explicit motive to achieve, and that each type of motive predicted different types of behavior. That is, the implicit achievement motive exclusively predicted how hard participants tried to improve their performance on a task (i.e., effort-related task performance), whereas the explicit achievement motive uniquely predicted how much participants liked the task (i.e., self-reflective appraisal).

On the one hand, given the conscious representation of the explicit system, explicit motives can be easily assessed via self-reports, by use of questionnaires. These measures of explicit motives are well-established in the Management research domain, including Entrepreneurship [23]. On the other hand, however, because implicit motives operate outside of a person’s awareness and people often lack direct introspective insight into their implicit system, sensitive indirect measurement instruments are needed to measure these hidden individual differences [24,25]. Indeed, over the course of time, several measurement procedures have been proposed for the assessment of implicit motives. However, for reasons we will discuss below, the implicit motives measurement instruments are rarely applied in Management research [2,3], including Entrepreneurship [26].

The assessment of implicit motives can be traced back to the 1930s with the development of the Thematic Apperception Test (TAT) [27]. The TAT was developed to explore the unconscious dynamics of personality. Murray believed that people share the same basic sets of 27 motives, and differ only in their priority ranking of these needs [27,28]. He further argued that motives cannot be observed directly, but instead must be inferred indirectly. For this purpose, he developed the TAT. The TAT is a projective test that presents research participants with a series of ambiguous pictures. The participant is asked to spontaneously develop a story for each picture. The assumption is that the participant projects her or his own needs into the story, so that the stories can be analyzed and interpreted to uncover each participant’s underlying needs. For example, Atkinson and McClelland [29] find that the frequency of food-related interpretations of TAT pictures relates positively to the time that had passed since participants had their last meal.

A criticism of this early work has been that the classification of motives is just a lengthy inventory of needs, lacking sufficient empirical or theoretical evidence [28]. Because of their relevance for understanding economical and societal life, implicit motives researchers focused on Murray’s power, affiliation, and achievement needs [9]. Even though other basic motivational needs, such as hunger or sex, may exist, one of the greatest achievements of the early implicit motives researchers is that they developed an empirically validated system of content analysis for motives imagery and transformed the TAT into a major tool for scientific personality psychology [9,30–32]. To distinguish the original TAT from the version used to assess implicit motives, researchers started to refer to the latter as the Picture Story Exercise (PSE) [33,34]. To avoid any misunderstanding, we use the term PSE to refer to the TAT version developed to assess implicit motives.

A typical PSE consists of four to six pictures that depict people in a variety of social settings. For each picture, participants write an imaginative story. The content of these imaginative stories can be coded according to motivational coding systems, empirically derived and refined over decades [35,36]. The summed scores yield the person’s overall implicit motives scores. Even though recent work confirms the good psychometric qualities of PSE measures and other content-coding and interpretivist techniques [33,34,37,38], many researchers remain highly skeptical toward these so-called ‘subjective’ techniques [39–41]. Next to the criticisms on the so-called subjective nature of these techniques, also the heavy administrative and time-consuming data preparation procedures hamper a wider application of these techniques [42]. To illustrate, to complete a default PSE, each participant has to write stories about four to eight pictures, and each picture presentation and story-writing episode is advised to last about four to five minutes [33]. Afterwards, stories have to be independently content-coded by two or more trained coders. An experienced coder needs between two and five minutes to score one PSE story. Thus, a typical PSE with 6 picture presentations and 60 participants requires up to 30 hours of data collection and 12–30 hours of data preparation, and this only to assess the participants’ implicit motivational dispositions [33].

The Operant Motive Test (OMT) which can be seen as a modified PSE technique [43] partially addresses the shortcomings of the PSE. In a typical OMT, participants review 4–15 pictures and invent a story (without having to write it down). Then they offer spontaneous associations, in response to the following questions: (1) “What is important for the person in this situation and what is the person doing?”; (2) “How does the person feel?”; (3) “Why does the person feel this way?”; and (4) “How does the story end?” Next to motives imagery, the OMT scores also reflect motives-relevant traits (i.e., an implementation strategy of a motive), which should increase their predictive validity and test-retest reliability for motives measures. Recent studies on the validity of the OMT confirm its good psychometric qualities [11,44–47]. Yet, compared to the PSE, the administration of the OMT is indeed faster and renders shorter answers.

However, the OMT still requires relatively much effort of research participants who need to write down multiple sentences. Moreover, akin to the PSE, the OMT may also provoke resistance and skepticism, leading to disengaged or defensive responding by employees, entrepreneurs, managers or other participants in Management research [23]. Consequently, when new, yet ‘objective’ indirect measures were developed in the 1990s, offering an alternative–cheaper, quicker, and shorter–method, many researchers abandoned the subjective content-coding and interpretivist techniques in favor of the new objective techniques [39]. Researchers in implicit motives adopted these techniques, too, and provided new theoretical insights in the mechanisms underlying motives, behaviors and outcomes.

By now, a large number of new implicit assessment techniques circulate in the literature. In Management research, rooted in industrial and organizational psychology, excellent reviews are offered by Bing, LeBreton, Davidson, Migetz, and James [39], and Uhlmann and colleagues [23]; in personality and social psychology, examples of insightful overviews are De Houwer and colleagues [48], and Gawronski and De Houwer [49]. Perhaps strikingly, both domains–industrial and organizational psychology, on the one hand, and personality and social psychology, on the other hand–are not really connected, as signaled by the lack of cross-references. For example, the most popular new implicit measure for the assessment of implicit motives that is developed by Organizational and Management Researchers is the Conditional Reasoning Test [39], whereas the most popular technique that stems from Psychological researchers is the Implicit Association Test (IAT) [50].Yet, the CRT technique is not frequently referred to in the personality and social psychology literature, whilst the IAT instrument is not widely applied in the industrial and organizational psychology literature.

In what follows, we will first discuss the IAT and CRT, along with two other promising implicit measures that can be used for the assessment of implicit motives: the Brief Implicit Association Test [12] and the Implicit PsyCap Questionnaire [2]. We then argue why Management research may benefit especially from the introduction of the BIAT.

Implicit motives and entrepreneurship

So, the aim of our Entrepreneurship study is twofold. The first objective is to introduce the BIAT for the measurement of implicit motives. This is important because the availability of an easy-to-develop, easy-to-use and valid–and hence ‘appropriate’–measure could stimulate implicit motives research (and research into implicit attitudes, biases and traits, for that matter), especially in Management research where quick and easy data collection is crucial. Our second aim is to test BIAT’s predictive validity in a substantive study by focusing on the well-established observation that implicit motives predict long-term behavior and performance, whereas explicit motives are related to self-reflective appraisals, judgments, deliberate choices and decisions [31]. The selection of the criterion variables, and the underlying theoretical logic, is introduced below, leading to three hypotheses.

Firstly, the way in which business outcomes are measured is crucial. As long as business outcomes are captured by means of self-reported ‘subjective’ measures, we expect that primarily explicit motives profiles will predict these business outcomes, as then the explicitly self-reported entrepreneurial measures can be expected to correlate with the explicitly self-reported motives. Yet, if business outcomes are measured by means of ‘real’ or ‘objective’ (e.g., performance) metrics, measured through another source than self-reports (e.g., entrepreneurial ventures’ annual reports), then implicit motives measures should outperform their explicit counterparts [19,24,78].

1. Hypothesis 1 (H1): (a) Explicit motives are related to self-reported ‘subjective’ business outcome measures, whilst implicit motives are not; and (b) Implicit motives are related to ‘objective’ business outcome measures, whilst explicit motives are not.

An interesting entrepreneurship construct that can be taken as the ‘subjective’ business outcome dependent variable to explore this line of logic is the self-reported measure of Entrepreneurial Self-Efficacy (ESE) [79]. ESE is a highly relevant variable in Entrepreneurship research. ESE refers to the belief that one is able to perform activities required to start a business. ESE is a key determinant of entrepreneurial intention [80], venture creation [81], and entrepreneurial success [82]. ESE is found to be positively related to self-reported measures of need for achievement [83]. Regarding ‘objective’ business outcome measures, as we will explain below, we decided to use proxies for business founding and financial profitability, both on the basis of firm-level annual report data.

Secondly, the extensive literature reviews of Johnson [15] and Uhlmann [23] reveal that researchers in Entrepreneurship mainly focused on one particular motive: the need for achievement. Yet Murray compiled a list of 27 motives [27,28]. Because of their societal and economic relevance, Murray’s successors and contemporary motive researchers narrowed this lengthy motive inventory and drove the attention to three major motives: the power motive, the achievement motive, and the affiliation motive [31]. The power motive stems from a person’s desire to influence, teach or encourage others. Power-motivated individuals obtain satisfaction from exerting social, physical or emotional impact on others or on the world at large, but they experience social defeats and impacts from others as aversive [36]. People with a high need for achievement typically get satisfaction from mastering challenging tasks on their own, but experience the failure to master such tasks individually as dissatisfying [25].

Finally, people who are motivated by affiliation instead prefer to spend time with others they like. They love to create, maintain, and restore social relationships. They enjoy being a part of a group and have a desire to feel loved and accepted. Signals of rejection or hostility are experienced as unpleasant [84]. Interestingly, many aspects of business and career success have been linked to implicit motives. For example, McClelland and Franz [85] found that achievement-motivated individuals are less successful in jobs that require social skills because they are keen to function in an autonomous manner [25]. Similarly, results of McClelland and Boyatzis [86] indicated that, contrary to achievement-motivated employees, power-motivated employees strive for high-level positions in organizations so that they can control the direction in which their company is moving.

Tying these insights together, we expect that different implicit motives will become important at different developmental stages of a business’ lifecycle. Specifically, because many, if not most, companies start as micro-businesses, employing no or only a few people and with the ideas and skills of the entrepreneur being the most valuable [87,88], we expect that the achievement motive will drive business performance during the first years of a business. However, as the micro-business grows and employs more people, social skills become increasingly important [89]. In addition, as the growing business employs more people, not only the need to control the behavior of others increases, but also the opportunities that provide satisfaction to power-motivated entrepreneurs increase. Hence, growing businesses become ideal habitats for entrepreneurs who score high on need for power. So, the degree to which the entrepreneur is motivated by power will determine further business performance, rather than the entrepreneur’s level of achievement motivation.

1. Hypothesis 2 (H2): (a) During the first years after the venture’s founding, implicit need for achievement is stronger related to business performance than implicit need for power; and (b) during later years after the venture’s founding, implicit need for power is stronger related to business performance than implicit need for achievement.

Thirdly, business outcomes can take many different forms and shapes. As implicit motives orient, select and energize behavior [31,90], we expect that entrepreneurs will strive for goals that are congruent with their implicit motives. Power-motivated people tend to derive pleasure from their ability to have a physical, mental or emotional impact on others [91]. One of the consequences is that power-motivated people tend to engage in activities that increase their social visibility [91] and that provide a feeling that one is in control over her/his own life [21]. Therefore, on the one hand, we hypothesize that power-motivated entrepreneurs have founded multiple businesses because this provides expanded opportunities to increase their social visibility. The activities, behaviors, and outcomes associated with the venture-creation process also reinforce the entrepreneur’s notion that s/he is the master of her/his own fate [92,93]. On the other hand, as affiliation-motivated people prefer to spend time with others they like and love to maintain and restore social relationships [84], we expect that affilation-motivated entrepreneurs will mainly put their effort in a fewer businesses, trying to optimize social ties within one company before establishing another enterprise. Also, striving for good inter-personal relationships with their co-workers and customers, and placing long-term goodwill over short-term gain are competencies that distinguish average entrepreneurs from enterpreneurs who managed to turn an average business into a successful venture [94]. As these competencies are also characteristics of affiliation-motivated people [21], we expect that affiliation-motivated entrepreneurs will prefer to develop and expand fewer businesses.

1. Hypothesis 3 (H3): (a) Power-motivated entrepreneurs have founded more businesses; and (b) Affiliation-motivated entrepreneurs have founded fewer businesses.

Design and participants

Our entrepreneurial sample consists of 108 Belgian entrepreneurs who participated in two on-line surveys and a series of workshops organized in Antwerp and Namur on entrepreneurial motives. Seventy-nine entrepreneurs are male (73%) and their mean age is 47.44 (SD = 8.65). From the first on-line survey, administered in 2012, we gathered self-reported measures on their objective performance (that is, the number of businesses the entrepreneur established) and a control variable (i.e., the age of the entrepreneur). Implicit motives and self-reported measures on the subjective performance of the entrepreneurs (that is, entrepreneurial self-efficacy or ESE) were assessed during a series of workshops in the summer of 2013. After a short introduction, the participants were led to a computer room where they were provided with three implicit motives BIATs [12] measuring the relative strength of (a) the power vis-à-vis the achievement motive, (b) the power vis-à-vis the affiliation motive, and (c) the achievement vis-à-vis the affiliation motive.

The order of the three BIATs was randomized, and the self-reported outcome measures were administered immediately after the assessment of the implicit motives. Survey data about the explicit motives profiles from the entrepreneurs were collected online in 2013, which is two to three months before the workshops, as part of the enrolling process. The information collected during the workshops was later matched with objective business outcome measures and the founding year of the business. We obtained this data from the Bel-first database containing Belgian annual report information. So, we have data from four distinct (and differently timed) sources: two online surveys, an assessment exercise (workshops), and an annual report database.

Ethical statement

In Belgium and at the faculties to which the authors were affiliated with at the time that the data were collected, there were no requirements to obtain formal approval from an institutional review board (IRB) for non-clinical research studies, except when researchers might believe an approval is necessary (which was not the case here). Yet, the project, including the design of this study, was reviewed and monitored by the Federal Science Policy Office of the Belgian Federal Government (Belspo) and a Belspo-appointed project-specific advisory board. Belspo supports the Code of Ethics for Scientific Research in Belgium. All participants gave online (online survey) or verbal consent (exercises during workshop). Verbal informed consent for the workshops was viewed as sufficient as participants were not exposed to risky events, and all data were analyzed and stored anonymously.

Participants were also well informed of the purpose of the study. They were assured that no harm would come to them in the process of exercises, and were told that the exercise involved either the completion of questionnaires (online survey) or sorting tasks (exercises during workshop). Participants were also well aware and agreed that the information collected would be matched with measures that we obtained from other databases (such as the Bel-first firm accounts database). The results of all tests were kept private and participants were informed that they had the right to quit at any time during the completion of the survey or exercises. Participants were instructed to indicate their agreement to continue with the study by clicking forward after reading the informed consents (online survey) or by entering the computer room to start the sorting exercises (workshop) (see S1 Appendix for a detailed explanation of the data-collection process).

Measures and procedures

Implicit motives: BIAT.

We relied on the procedures of Slabbinck and colleagues [52–54] and Sriram and Greenwald [12] for the construction of the implicit motives BIAT. That is, we applied the instructions of Sriram and Greenwald [12] for the construction of the BIAT procedure, and adopted the recommendations of Slabbinck and colleagues [52–54] for the selection of the stimuli. The BIAT procedure consisted of seven blocks that allowed us to measure implicit power, implicit affiliation, and implicit achievement. The first block of the BIAT is a practice block, and the remaining six blocks capture the critical responses that were used to construct the BIAT scores. Critical blocks were presented in a random order. Each implicit motive involves two critical blocks, and each critical block consists of 20 trials. Table 1 presents an overview of the composition of the BIAT procedure. The full BIAT, including detailed instructions, can be downloaded from [reference withheld].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Structure of the implicit motives BIAT and construction of the BIAT scores.

https://doi.org/10.1371/journal.pone.0198094.t001

While completing the critical blocks, participants were requested to focus simultaneously on two labels. One label represented the focal concept (e.g., “Affiliation”) and the other label the focal attribute (e.g., “Pleasurable”). The labels appeared one under the other on the top center of the computer screen. Stimuli that were representative for the focal concepts plus focal attributes were presented one at a time on the center of the computer screen. Participants were required to categorize them as quickly as possible with one response key (e.g., the “i” key). Stimuli that were not representative for the focal concepts and attributes were also presented on the center of the computer screen, and participants were required to categorize them as “not belonging to the focal concepts and attributes” using another key (e.g., the “e” key). The practice block consisted of 20 trials. The structure of the practice block was identical to that of the critical blocks, with the exception that only the label of the focal attribute was provided (i.e., “Pleasurable”) and that participants were only required to categorize stimuli of the focal and non-focal attributes.

We followed Slabbinck and colleagues [52–54] regarding the selection of stimuli. For the target concepts, we used (Dutch or French translations of) the labels “Affiliation”, “Power”, and “Success”; for the attribute category, we took “Pleasurable”. The selection the attribute categories fits well with McClelland’s conceptualization of implicit motives, who defines ‘a person with a strong motive as a person who has a strong affective response to an incentive’ [9] and who argues that motives are based on affective experiences [31]. In a similar vein, Slabbinck and colleagues [52] demonstrated that IATs with affective attribute categories outperformed implicit motives IATs with attributes categories referring to the self (‘me’ vs ‘not-me’). The results of Slabbinck and colleagues [52–54] further indicated that motives can best be represented by pictorial rather than verbal stimuli. According to Gschwender and colleagues [96], pictorial stimuli enclose more concept-relevant information than verbal stimuli, which makes pictorial stimuli better suited for the assessment of traits such as motives. In addition, it is assumed that implicit motives are based on nonverbal experiences that are acquired early in life, before the development of language skills [97]. Because of this, nonverbal/pictorial stimuli that relate to implicit motives may be particularly suitable for capturing implicit motives. This is also shown by Slabbinck and colleagues [52], who tested different IAT variants for the assessment of implicit motives. They demonstrated that IATs with pictorial stimuli were superior to IATs with verbal stimuli.

Accordingly, the stimuli representing the target categories featured pictures demonstrating affiliation settings (e.g., kids walking hand in hand on the beach), power contexts (e.g., business man standing up straight at a meeting table), or achievement situations (e.g., a student who graduates and shows his degree). We had four pictures for each target category. For the attribute categories, we employed (Dutch or French translations of) “nice,” “friendly,” “pleasant,” and “lovely” to designate attractive, whereas we used “creepy,” “nasty,” “annoying,” and “undesired” to represent not attractive. We obtained all research materials from Slabbinck and his colleagues, who had carefully and systematically pretested all materials [52–54].

BIAT scores (e.g., the BIAT score of the power relative to the affiliation motive) were derived from the comparison of participants’ performance in two blocks that presented the same focal attribute (e.g., “Attractive”) but with two different focal concepts (e.g., “Affiliation” and “Power”). In total, we calculated three BIAT scores: that is, a BIAT score for the power relative to the achievement motive (BIAT_pow-ach), a BIAT score for the power relative to the affiliation motive (BIAT_pow-aff), and a BIAT score for the achievement relative to the affiliation motive (BIAT_ach-aff). Thus, a high score on, for example, the BIAT_pow-ach indicates that the participant found it easier to pair power stimuli with pleasurable words and achievement stimuli with non-pleasurable words (see Table 1 –critical block 4) than power stimuli with non-pleasurable words and achievement stimuli with pleasurable words (see Table 1 –critical block 6).

Data from all critical blocks were used to compute BIAT scores, according to the scoring recommendations of Nosek and colleagues [55]. Extreme latencies below 400 milliseconds and above 2000 milliseconds were recoded to these boundaries, and the first two trials of each block were discarded. Individual BIAT scores were computed using the D-measure [98]. The D-measure was computed as the difference between mean latencies of the two critical BIAT blocks divided by the standard deviation of latencies in the those blocks [55]. Following standard procedures for the estimation of the internal consistency of latency measures [52–54,99], we estimated internal consistency of the BIATs by dividing each critical block into two sub-blocks of equal length. The first sub-block contained the even trials and the second the odd trials, and BIAT scores were calculated for each sub-block separately. The Spearman–Brown coefficients revealed a good split-half reliability for all three BIATs (BIAT_pow-ach: .79 ;BIAT_pow-aff: .77; BIAT_ach-aff: .78). Participants required, on average, 3.45 minutes to complete the three BIATs (SD = 1.55).

Control variables.

Since business outcomes are contingent upon the age of the enterprise, the age of the entrepreneur and the size of the enterprise, we included the entrepreneur’s age, the year of establishment of the participant’s principal business, and the company’s paid capital as covariates [87] (see S2 Appendix for the survey questions). We subtracted the year of establishment from the year of the motives measurement, 2013, to obtain the age of the enterprise. We apply a natural log transformation to variables that are highly skewed, clustering toward either side of the distribution, with a long tail. Descriptive statistics are given in Table 2, indicating that multicollinearity is not an issue.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Descriptives, α coefficients, and correlations of all dependent and independent variables.

https://doi.org/10.1371/journal.pone.0198094.t002

Two remarks are worth making. A first comment relates to the absolute vis-à-vis relative nature of our measures, given the debate in the literature regarding difference scores [Edwards, 2001]. What we do here is the standard way of treating PRF measures and comparing them with implicit motives scores [22,51,54,101]. Thus, transforming the absolute PRF measures into relative PRF measures would make our study less comparable with other implicit motives research. In addition, Schultheiss et al. [73] demonstrate that implicit and explicit measures are measures of distinct constructs, and that correlations between measures of both constructs are not affected by the methods with which both constructs are measured, even if the methodological correspondence between the explicit and implicit measures is maximized. Further, the independence of implicit and explicit motives, regardless of the way in which they are measured, has recently been confirmed in a meta-analysis [8]. Indeed, our correlations between implicit and explicit motives are not affected by the method that is used to calculate motive scores. By way of robustness check, we calculated relative PRF measures in a similar way as the BIAT scores were constructed (i.e., subtracting the average score of a PRF scale from the average score of another PRF scale, divided by the standard deviation of all the items of those scales). Inspection of the bivariate correlations (available upon request) shows that the usage of relative instead absolute PRF measures does not affect the magnitudes of the correlations between implicit and explicit motive measures.

A second comment involves the timing of our measurements, and what this implies for causality claims [102]. To cope with the postdictive nature of our dependent variable, we linked our entrepreneurial sample with the Bel-first database. By doing so, we obtained financial performance data of 85 companies for the years 2012 and 2013. We realize that this approach does not fully resolve the issue of the postdictive nature of our dependent variable (the motives of the entrepreneurs were assessed in the middle of 2013]. However, because (implicit) motives and personality traits seem to evolve only slowly in adulthood [103–106], we believe it is legitimate to conclude that our results are reliable and valid. Of course, we acknowledge that making causal claims would not be warranted. Thus, instead of talking about ‘determinants’, which imply causality, we refer to associations between motives and business performance.

Explicit motives predict declarative behavior

We estimated an ordinary least squares regression model with total entrepreneurial self-efficacy (total ESE) as a function of implicit motives and explicit motives. To assess the relative importance of implicit and explicit motive measures in predicting behavior, we performed of relative weights analysis [107,108]. Relative weights analysis transforms a set of predictors into a new set of orthogonal variables that are as highly correlated to the original set of variables. The dependent variable (i.e., total ESE) is then regressed on the new uncorrelated predictor variables. To compute estimates of relative importance, these coefficients are then squared and combined with the squared standardized regression coefficients obtained by regressing the original variables on the orthogonal variables. Finally, the raw relative weights are transformed into proportions by dividing them by the sum of all raw relative weights, yielding a proportional contribution of each original variable [108,109]. In addition to the relative weights, we used a bootstrapping procedure with 1,000 bootstrap samples to construct confidence intervals for the relative weights, following Johnson’s procedure [110]. The results are summarized in Table 3.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Implicit motives and total entrepreneurial self-efficacy (ESE).

https://doi.org/10.1371/journal.pone.0198094.t003

The model explains 19 per cent of the variation in total ESE (p < .01). The regression results show that none of the implicit motives are significantly associated with total ESE (all p’s > .05). Yet, the results further yield significantly positive parameter estimates for explicit need for achievement (b = 0.39; p < .05) and explicit need for dominance (b = 0.37; p < .05). Explicit need for affiliation (b = 0.07; p = .57) is not predictive for total ESE. The PRF measures obtained the highest relative weights, and these weights are in line with the parameter estimates. Even though confidence intervals of the relative weights of PRF and BIAT measures overlap, explicit motives explain more than 90 per cent of the variance in total ESE, compared to eight per cent for the implicit motive measures. These findings are in line with previous literature and offer support for H1.

In exploratory analyses, we estimated the above ESE models for all three sub-dimensions separately (full tables are available upon request). In line with the above, we find that none of the BIAT measures is significantly related to any of the ESE sub-dimensions and that the relative weights of the BIAT measure are always considerably lower than those of the PRF measures. Moreover, explicit need for affiliation is non-significant, as for total ESE, except for the “growth and expansion” sub-dimension. Explicit need for achievement co-predicts all three sub-dimensions, explicit need for dominance does so for “Growth and expansion” and “Problem solving”, and explicit need for affiliation for “Growth and expansion”. Again, as in the case of total ESE, this pattern of results is fully in line with H1.

Implicit motives predict objective business outcomes

First, in order to examine the relationship between implicit motives and objective financial profitability, we analyzed our data by means of repeated measures with linear mixed models. The log of gross profit was treated as the repeated dependent variable, and time as a repeated independent variable. Before examining the relationship between implicit motives and gross profit, we first estimated an unconditional model (with no predictors involved) and found significant variation in the time variable (Fig 1). This confirmed that a repeated-measures mixed model was the correct data analysis technique to use. We estimated seven hierarchical models to test whether the implicit achievement motive predicts objective financial profitability (log of gross profit) during the first years of a business, whereas the power motive is predictive for later objective financial profitability.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Implicit motives profiles and business performance (gross profit).

https://doi.org/10.1371/journal.pone.0198094.g001

In Model 1, we added time as a fixed variable. In Model 2, we entered the control variables (age of entrepreneur, log of age of enterprise, and log of capital). The measures of explicit power and achievement motives were entered in Model 3, and their interactions with the log of the age of the enterprise were included in Model 4. The three-way interaction between the explicit power measure, the explicit achievement measure, and the log of the age of the enterprise was added in Model 5. The measure of the implicit power versus achievement motive was entered in Model 6. The interaction between the measure of the implicit power versus achievement motive and the log of the age of the enterprise was included in Model 7. Finally, we also estimated a model that only included the control variables, together with the implicit power versus achievement motive, the log of the age of the entrepreneur, and the interaction between the implicit power motive and the log of the enterprise’s age (Model 8).

We grand mean-centered all of the independent variables to facilitate interpretation [111] and used − 2 log likelihood (− 2LL) and BIC to assess model fit. To assess the incremental fit of our models, and hence the predictive validity of our measures, we evaluated the significance of the change in − 2LL from one model to the next. To do so, we performed a series of chi-square tests with the degrees of freedom equal to the difference in the number of parameters for the pair of nested models. We also inspected the evolution of the BIC as this measure includes a penalty function that is based on the number of estimated parameters. Additionally, we performed a relative weights analysis on Model 7 to evaluate the relative importance of explicit and implicit motive measures in predicting business performance. The results of these analyses are presented in Fig 1.

The incremental fit statistics (change in -2LL) in Fig 1 show that neither the addition of explicit motive measures, nor their interaction with the age of the enterprise did improve model fit. This is in line with H1b. However, the -2LL statistic improved significantly when the interaction between the implicit motive measure of power versus achievement and the log of the age of the enterprise was entered to the model. Removing explicit power, explicit achievement and their interactions with the log of the age of the enterprise did not deteriorate model fit (Model 8 vs Model 7) and fitted the data marginally significantly better than a the model that only comprised the control variables (Model 8 vs Model 2: Change in -2LL: 4.80; p = .091). Inspection of the BIC values further reveals that inclusion of explicit motive measures increased the BIC values (Model 3–5) and that BIC only decreased when the interaction between the implicit motive measure and the log of the age of the enterprise was entered into the model (Model 7 vs Model 3–6). However, the BIC value of Model 8 was higher than the BIC value of the Model 2 (model with control variables only). The parameter estimates of Model 7 show that older enterprises are more profitable than younger enterprises. Yet, more importantly, this effect is qualified with a significant interaction with the implicit motive measure.

To analyze this interaction in depth, we performed a Johnson-Neyman analysis [112]. This technique allows us to estimates the conditional effect of the implicit power versus achievement motive for different age levels of the enterprise and identifies regions of significance. For ease of interpretation, we do this plot for the non-logged age of the enterprise. Robustness analyses revealed that the results presented above are similar if we use non-logged rather than logged age of the enterprise (available upon request). These regions indicate for which ages of the enterprise the entrepreneur’s implicit power versus achievement motive has a significantly positive, non-significant, or significantly negative impact on the enterprise’s gross profit. Fig 2 illustrates the relation between the age of the enterprise and the impact of the entrepreneur’s implicit motive profile on the enterprise’s performance. The figure shows that companies grow (marginally) faster in the first six years after establishment when their entrepreneur is implicitly motivated by achievement rather than by power (age of enterprise = 6.41 years: b = -1.95; p = .10). The gross profit of mature companies seems to not to be affected by the entrepreneur’s implicit power and achievement motives as it takes up to 26 years after establishment before gross profit starts benefitting from an entrepreneur with a dominant implicit power motive (age of enterprise = 26.5 years: b = 1.49; p = .10).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Johnson-Neyman plot of the region of significance for the effect of implicit power versus achievement on gross profit across the life span of an enterprise.

https://doi.org/10.1371/journal.pone.0198094.g002

Altogether, the fit statistics, the parameter estimates and the relative importance of the predictors indicate that it is not worth adding explicit motive measures to explain business performance, whereas most evidence shows that implicit motives do explain an important part of an enterprise’s business performance. This largely confirms our H2a and H2b.

Second, to test H3a and H3b that implicit power-motivated entrepreneurs have founded multiple businesses, whereas implicit affiliation-motivated entrepreneurs mainly put their effort in a single business, we analyzed the relationship between the log of the number of businesses each entrepreneur established in her or his life and her or his motives profiles, using an ordinary least squares regression model with heteroscedastic consistent standard error estimator (HC3). We conducted three regression models. In the first model, we regressed the logged number of businesses on the age of the entrepreneur, and the PRF dominance and PRF affiliation scores. In the second regression model, we entered the interaction between both PRF measures. Finally, in the third model, we entered the BIAT_pow-aff score and compared the incremental validity of this model by assessing the significance of the change in R² from the second to third model. We also performed a relative importance analysis on the parameter estimates of the last model.

Table 4 contains the regression coefficients, the fit statistics of the estimated regression models and the relative importance weights of the parameter estimates The results show that none of the explicit motive measures reached significance (all p’s > .05) and that explicit motives were relatively unimportant in predicting the log of the number of businesses each entrepreneur established, which again is in line with H1b. Consistent with our prediction, the significant parameter of the BIAT_pow-aff score and its high relative importance weight (68%) indicate that entrepreneurs who are implicitly motivated by power rather than by affiliation established more businesses than entrepreneurs who are implicitly motivated by affiliation rather than by power (b = .82; p < .05). In addition, the significant change in R² from Model 2 to Model 3 established the incremental validity of the BIAT_pow-aff in predicting this aspect of business performance.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Implicit motives and number of established businesses.

https://doi.org/10.1371/journal.pone.0198094.t004

Methodological implications

The BIAT has definitely a unique advantage that condones its use in Management research. That is, the BIAT substantially reduces the data administration and preparation time in comparison with other techniques such as OMT and PSE. In our study, it took 3.45 minutes to obtain reliable and valid measures of three implicit motives. If we had used three traditional IATs instead or if implicit motives had been assessed by means of the PSE or CRT, participants would have needed up to 20–30 minutes to complete these tests [12,33,52]. Data preparation of IAT and BIAT procedures is fully automated [12]; but in case, had we used the PSE, we would have spent up to 60 hours to prepare our data for analysis [33]. Moreover, we constructed new BIATs to measure the implicit needs for achievement, affiliation and power. These three BIATs all work well, and are ready to be used in other Management research, given their efficiency, reliability and validity.

BIAT’s ease of use and flexibility imply that new BIATs can be relatively easily developed and validated. Hence, the BIAT might help to overcome the lack of research into implicit constructs in areas where time-efficient data collection is crucial, of which studies in managerial and organizational settings are a key example. Indeed, we argue that the lack of Management research into implicit motives may well be due to the unavailability, to date, of an easy-to-administer and easy-to-develop measurement instrument, which is critical in the area of field work in the context of research involving difficult-to-access and busy target groups like employees, entrepreneurs and managers (different from the students routinely used in much psychology research, mostly in lab settings). However, the added value of the BIAT is not limited to its time efficiency. As we argued, the BIAT is a non-symbolic measure in contrast to most, if not all, other implicit measures that are currently used in Management research, being symbolic measures. As different types of measures may predict different types or facets of (business) behavior, and as method and theory inevitably interact [114], we believe that the use of the BIAT in Management research will bring novel insights in Management theory. We provide some ideas in the next section. We also argued that the BIAT is complementary to both explicit and other implicit measures from a theoretical perspective. Yet, future research needs to discover the unique contributions of the BIAT and other implicit measures in predicting business behavior. To do so, we encourage researchers to include different types of implicit measures in their research design, and to strive for a methodological fit between the selected research methods and the studied behavior. This type of research may yield more complete and holistic insights.

Practical implications and future research

Our results have also important implications for research in different (sub-)disciplines in the broad domain of Management, from Organizational Behavior and Strategy to Accounting and Marketing, and many more. We speculate that the introduction of implicit constructs (such as implicit attitudes, biases, motives and traits) in Management research (including Entrepreneurship) has the potential to generate many novel insights. Our study shows that the entrepreneur’s implicit motives are important determinants of her or his venture’s success. Failing to do so would imply that Entrepreneurship research into motives could wrongly be considered to be a dead-end, and certainly would produce biased and incomplete insights.

Interestingly, for instance, our lifecycle-dependent motives logic might offer an explanation for a well-known puzzle in the Entrepreneurship literature: gazelles are very rare, and the R² associated with gazelles studies is very low [115]. Gazelles are defined as entrepreneurial ventures that grow sustainably over a long period of time. H2 would predict that this is difficult to achieve because (a) different implicit motives are key during different stages of a venture’s lifecycle, and (b) such implicit motives evolve only slowly in adulthood [105,106]. H2 suggests that turning a successful start-up into a gazelle requires a switch in the dominance of implicit motives from achievement to power. An entrepreneur with both high implicit power and achievement motives may have the ability to shift focus as necessary without “changing” his or her personality. Yet, if the entrepeneur’s is only motived by achievement, such a shift is unlikely to happen, requiring other mechanisms to achieve the needed implicit motive shift. For example a replacement of an achievement-motivated founding entrepreneur by a power-motivated successor, or adding a power-motivated CEO to the venture’s founding team may be good strategies to turn promising start-ups into successful gazelles. Of course, also other variables such as creativity, work ethics, timing, and networking do also play an important role in entrepreneurial succes, and many of these variables may interact with or operate independently from the motives that serve as the focus of this paper. Regrettably, we cannot explore these potential explanation with our cross-sectional data. Future research may explore these lines of logic further.

As said, the applicability of the BIAT moves far beyond our illustrative example of implicit motives in Entrepreneurship research. Here, two further examples should suffice. First, implicit constructs fit squarely with the recent plea to develop a Behavioral Strategy literature. According to Powell et al. [116] (p. 1369), “Behavioral strategy applies cognitive and social psychology to strategic management theory and practice.” A clear example relates to upper echelon theory and top management team studies. Here, a key question revolves around the impact of deep-level characteristics of (teams of) top managers on strategic decision-making, of which implicit motives and traits are prime examples. Second, implicit constructs are important in the context of many Organizational Behavior issues. As Harms and Luthans [2] (p. 589) observe, “Implicit psychological constructs are effective predictors of behavioral outcomes but are rarely used in organizational settings because of real or imagined problems with measurement validity and administration.” They use the example of an implicit measure of psychological capital to illustrate that this current state of affairs can be turned around.

The study of implicit psychological constructs should not be limited to implicit achievement, power and affiliations motives, assessed by BIATs. As mentioned before, contemporary implicit motive research is rooted in Murray’s research on human motives. Murray believed that people share the same basic sets of 27 needs and differ only in their priority ranking of these needs [27,28]. Murray’s successors, and especially McClelland, Veroff, Heyns, and Atkinson [25,117,118], drove the focus on the achievement, affiliation, and power motives. Yet, this selection of motives was based on their personal interests, not on any theoretical ground or empirical evidence. Hence, even though the list of potential motives is, in all likelihood, limited to a small number of motivational systems [119], it may be worthwhile to go back to the roots of implicit motives research and study if other, still unexplored motives do exist from which Management research could benefit.

Additionally, future applications should also not be limited to BIAT studies. As we argued before, different types of implicit measures are most likely complementary and not interchangeable, tapping into different mechanisms and relating to different outcomes. Schoen and colleagues [59], for example, recently developed a CRT measure to assess a person’s implicit creative personality and found that it is a substantial predictor of creative performance. Yet, other implicit measures of creative personality, such as a BIAT, an OMT or a I-PSQ-type of questionnaire, could most likely add to the knowledge of creative personality and performance.