Preliminary Analysis

Because no complete scales had been employed in the research, it was important to examine the factor structure of the items actually used. Three separate principal components factor analyses, each with a varimax rotation, were conducted. The first examined the Kirton items alone, the second involved the EAO items alone, the third included both the SOC items and the risk aversion items. In each factor analysis, items were retained on a factor if their loading exceeded .40 on the primary factor but did not exceed .40 on any secondary factor. Results of these factor analyses are presented in Tables 2 to 4.



Factor Loadings for Creativity Items

  Efficiency Rules Originality
Eigenvalue: 4.50 1.86 1.14
% variance explained: 37.5 15.5 9.5
Chronbach alpha: .85 .79 .59
Thoroughness .85 .10 .13
Originial ideas .81 .02 .09
Methodical .70 .23 .16
Master details .64 .10 .17
Enjoy detailed work .69 .22 .30
Conform .22 .84 -.16
Agree with team .00 .75 .25
Never bend rules .04 .73 .29
Fit into "System" .37 .70 -.19
Rather Create .16 -.14 .75
Handle new ideas .22 .27 .65
.>Stimulating .>.47 .>.15 .>.55

With the exception of the item, "originality," all items loaded on the three factors originally identified and named by Kirton (1976). Consequently, our factor names are based on his description of the content of each factor. Even with an abbreviated version of the Kirton scale, we have managed to capture the essence of the distinctions between individuals whose principal motivation is to adapt to an ongoing business environment and individuals whose major objective is to strike out on their own. Together the three factors accounted for 62.5% of the variance, and the scales created by combining items within each factor showed acceptable Chronbach internal consistencies.


Factor Loadings for Entrepreneurial Attitude Orientation Scale Items

  Achievement Business Activities Innovation Behavior Time Management
Eigenvalue: 4.37 2.02 1.48 1.19
% variance explained: 27.3 12.6 9.2 7.4
Chronbach alpha: .67 .59 .56 .36
Do every job thorough (23) .75 .21 .24 -.02
Make most of resources (27) .75 .30 .13 .08
Make org. work better (9) .70 .28 .08 -.16
Never put off important (3) .67 -.18 .01 .05
Do meaningful work (48) .19 .78 -.03 -.01
Approach tasks unique (58) .17 .70 -.10 .23
Analyze before allocate (44) .02 .66 .24 -.20
Accepted ways (73) .34 .08 .76 -.22
Rarely question (74) -.06 -.04 .74 .29
Follow accepted (38) .29 .06 .59 -.27
Several ideas same time (39) .24 .12 -.05 .64
Delegate routine tasks (43) -.18 .24 .39 .62
Seldom follow instruct. (2) -.05 -.05 -.11 .56

In this Table, the original EAO Scale item numbers are shown in parentheses. For Achievement Behavior and Innovation Behavior, the items loaded here on the same scales as in the EAO.

The remaining items, however, clustered here into two additional factors not found in the EAO. On the basis of the content of their respective items, we called these factors "Business Activities" and "Time Management." The former included three items that have to do with the way in which the person approaches work. A desire to do meaningful work every day, a claim that one approaches work in a unique fashion, and an assertion that one analyzes the situation carefully before committing resources all represent a desirable level of conscientiousness in one’s business activities. By contrast, the claim that one can deal successfully with several ideas at a time, the idea that one should delegate routine business tasks to others, and the claim that one seldom follows instructions may represent either a recognition that successful time management requires delegation (hence the name for the scale) or an inflated self-importance. It remains to be seen which of these views is the more plausible.

Of the four scales created from these factors, only Achievement Behavior had a Chronbach approaching the recommended level of .70. The Business Activity scale and Innovation Behavior scales showed Chronbach internal consistencies above .50, whereas the Time Management scale showed an internal consistency too low to permit analysis. The first three factors together accounted for a total of 49.2% of the variance.


Factor Loadings for Control and Risk Items

  Social Control Personal Control Risk
Eigenvalue: 2.64 2.37 1.80
% variance explained: 15.5 13.9 10.6
Chronbach alpha: .68 .66 .40
Cannot control living cost .81 .04 .03
No influence on big business .73 -.05 .26
Econ. probs. beyond control .67 .30 -.01
Cannot guide conversation .65 -.26 -.05
Can make and keep friends -.01 .76 .13
Make own plans work -.07 .76 -.14
Get things by hard work .10 .64 -.09
Desire to rock-climb .11 -.22 .77
Lend car to reckless teen .09 -.16 .61
Never to old to start business .13 .33 .45

Although the control items had been selected from three separate subscales (personal control, interpersonal control, social control), in our data the 12 control items sorted themselves onto only two factors, one reflecting general social control and one reflecting personal efficacy. These two factors together accounted for 29.4% of the variance, and each scale’s Chronbach value was very near the .70 range. The third factor contained three of the risk items and accounted for an additional 10.6% of the variance, but did not have an acceptable level of internal consistency.

In summary, the factor analyses revealed three creativity factors, three usable factors derived from the EAO scale, and two locus of control factors. The internal consistencies of the Time Management factor from the EAO and the Risk factor were low enough that we elected not to examine them further at this time. Because we had conducted the factor analyses separately within domain (creativity, attitude, control) we next examined the degree to which the resulting scales might be related to one another. An index score was created for each of the eight scales. This index score was simply the sum of the items in the scale divided by the number of items. We then conducted Pearson correlations on the index scores, and the results are shown in Table 5.


Correlations Among Scales Derived From Factor Analysis

  (2) (3) (4) (5) (6) (7) (8)
Creativity Scales              
Efficiency (1) 35**. .49** .52** . 36** .29** .08 .34**
Rule following (2)   .21* .31** .14 .51** .11 .38**
Originality (3)     .38** .41** .10 -.07 .41**
EAO Scales              
Achievement behavior (4)       .33** .30** -.01 .63**
Business activities (5)         .14 -.21* .44**
Innovation behavior (6)           .24* .42**
Control and Risk Scales              
Social control (7)             .01
Personal control (8)              

Correlations marked by a single asterisk are significant at p < .05, those marked with two asterisks are significant at p < .01. Quite a few of the possible correlations are significant, but two aspects of this picture deserve detailed comment. First, despite having produced three factors that matched Kirton’s original pattern, the creativity scales were all significantly correlated with one another. By contrast, the two control scales were uncorrelated, and only one of the three EAO scales was correlated with the other two. Second, two scales, the Efficiency scale from the creativity items and the Personal Control scale from the locus of control items, were significantly correlated with all other measures except Social Control. Thus these two scales appear to be good proxies for a great deal of what is happening in the data. By contrast, Social Control appears to be tapping an aspect of behavior that is different from what is captured by the other measures. It will be interesting to see how these three scales in particular are associated with later success in new venture creation.

Demographic Variables

To test for demographic differences in response to the eight scales shown in Table 5, we conducted analyses of variance on the index scale scores. Each analysis was a mixed design, with two between-subjects factors (Sex and Race) and one repeated measure (the number of scales in each domain). For creativity, the design was thus a 2x2x3 (Sex x Race x Efficiency/Rule/Originality); for EAO it was also 2x2x3 (Sex x Race x Achieve/Activity/Innovation), but for Control it was only a 2x2x2 (Sex x Race x Social/Personal). Mean scores and standard deviations for all of these index scales are shown in Table 6.


Means And Standard Deviations For Index Scales By Domain

Participant Race: Minority White
Participant Sex: Female Male Female Male
Creativity Scales        
Efficiency 5.46 6.00 6.16 5.88
SD 1.94 .74 .57 .93
Rule following 4.78 4.78 4.99 4.77
SD 1.90 1.34 1.51 1.16
Originality 5.47 5.72 5.60 5.69
SD 1.77 .99 .83 1.02
EAO Sacles        
Achievement beh 5.91 6.04 6.06 5.42
SD 1.52 1.07 .87 1.26
Bus. activities 6.02 6.47 5.79 6.00
SD 1.45 .53 1.03 .70
Innovation beh 5.23 4.57 5.03 4.79
SD 1.19 1.37 1.38 1.16
Control Scales        
Social control 3.14 2.95 3.48 3.08
SD 1.54 1.70 1.19 1.28
Personal control 6.38 6.15 6.17 6.03
SD 1.46 .80 .81 94

The analysis of variance on the creativity scales revealed only one significant difference, and that difference occurred on the within-subjects factor. Specifically, scores for the rule-following scale were significantly lower than scores for either efficiency or originality, F (2, 198) = 28.28, p < .001.

The analysis of variance on the scales from the EAO also revealed a significant difference on the within-subjects factor. Here the scores for innovation behavior were significantly lower than scores for either achievement behavior or for business activities, F (2, 210) = 40.14, p < .001. (Note: The degrees of freedom shown differ across the analyses because of differential completion rates from one domain to another.) This analysis also revealed an interaction between sex and scale, F (2, 210) = 3.47, p < .05. The difference between innovation and the other two scales was greater among male respondents than among female respondents.

Finally, the analysis of variance on the control scores revealed only a significant within-subjects difference, F (1, 99) = 266.57, p < .001. Scores for all participant groups were substantially lower on expressed social control than on expressed personal control.

Attributional Coding

Coding of the open-ended responses to the question, "Why do you want to start this business" proceeded in three steps. The first step involves determining how many separate answers are actually given to this question. Ordinarily a period can be taken as the end of a sentence, but this does not absolve the coder from the responsibility to determine, to the best of his or her ability, whether a second sentence might be merely an amplification of a prior reason instead of an additional reason. When the responses are not delimited by periods, then the task becomes a bit more challenging. The 116 participants in the study provided us with a total of 231 separate reasons for why they wanted to enter business. Our two coders, working independently, disagreed on the number of separate statements in only 4 of these 231 instances.

The second step in the coding process is to determine whether a reason given is internal to the person or external to the person, and here there were disagreements in only 3 of the 231 instances. The third step is to decide whether a particular internal or external reason is "stable" or "variable." This is the most difficult of the interpretive tasks, but even here the disagreement between coders was under 10%, with disagreements on 18 of the 231 reasons.

It is important to notice that the high degree of agreement between coders represents only the reliability of the coding that was accomplished in this study. Whether the placements of answers into the four categories defined by internal/external versus stable/variable has acceptable validity has yet to be determined. We are in the process of comparing the coding in this study to the coding (done by other individuals) in the Gatewood, Shaver, & Gartner (1995) study to assess convergent validity. As this task has not yet been accomplished, we need to alert readers to the possibility that the attribution results reported here may be subject to change if recoding is needed.

With that caveat, we think it is appropriate at this time to present one attributional result, the percentage of times that the first reason given for entering business was "internal" as opposed to "external," and this proportion (with the base being number of individuals in each demographic category) is shown in Table 7.


Percentage Of "Internal" First Reasons Offered For Entering Business

Participant Race: Minority White
Participant Sex: Female Male Female Male

These percentages were compared statistically, but they were not significantly different x2 (1 df) = 2.47, n.s. Indeed, assuming for the moment that the coding is valid, it is interesting that even in the cell with the smallest percentage, two-thirds of the respondents gave an internal explanation instead of an external one. Unless this outcome is changed by our cross-study comparison of codings, it would give comfort to those who argue that personal reasons, rather than external market opportunities, are the primary drivers of decisions to create new ventures.

Top of Page
Previous Page Main Page Next Page
Return to 1996 Topical Index

(c) 1997 Babson College All Rights Reserved
Last Updated 4/25/97 by YuBei Teng

To sign-up for the Center for Entrepreneurial Studies' publication lists,
please register with the
Entrepreneurship WebTeam.