THE RESEARCH METHODOLOGY

 

The Survey and data processing

A total of 3,214 questionnaires were mailed to the owner-managers of manufacturing companies identified from Quebec’s Industrial Subcontracting databank (known by its French acronym STIQ). A total of 377 responses were received, for a response rate of 11.7%. Of these, 53 were rejected because they were incomplete. The survey questionnaire included five sections, concerned respectively with the technological characteristics of the firm, the state of its TECHSCAN practices, perception of the environment, firms characteristics, the manager’s profile and information networks. The data were processed in four steps: analysis of the overall characteristics of the target firms and their TECHSCAN activities using descriptive statistics, identification of the latent dimensions of TECHSCAN practices using factorial analysis, identification of scanning practices configurations using cluster analysis, and testing of the research model using the partial least square (PLS) method.

 

The Characteristics of the Final Sample

The final sample represented 10.1% of the total STIQ database enterprise population. Around half the firms (48.4%) were small (10 - 49 employees), 21.4% were medium-sized (50 - 199 employees) and 18.8% were very small (less than 10 employees). A variety of sectors were represented, with most of the firms belonging to seven sectors in particular: metal products (28.1%), plastic and rubber products (12%), transportation equipment (10.2%), machinery (10.2%), wood (7.5%), electrical and electronic products (7.7%) and furniture (6.8%).

In terms of technological characteristics, we found a relatively high dissemination level for computer assisted design equipment (44.1%), numerically controlled machine tools (35.2%), industrial computers (25%) and traditional computers (86.1%), for which the penetration rate was above average compared with all SMBs. On the other hand, robots (11.4%), electronic mail (14.2%), electronic data interchange (11.7%) and decision support tools (7.4%) were less common (Julien, 1995). Three-quarters of the firms (75%) said they carried out R&D - a much higher percentage than generally observed in the literature (Archibugi and Casaretto, 1989; Bernard and Torre, 1993). In most cases, R&D activities were informal, and involved an average of 4.2 employees on a regular basis. The penetration of ISO 9000 (25.9%) and Z299 (21.9%) quality standards was limited although not negligeable.

The level of education among managers was fairly high - 64.2% had at least a college education. More than 43% had a technical (college) or scientific (university) background. They were all fairly experienced, having worked on average for more than 19 years in the same sector, and having headed their present firms for at least 10 years.

 

DESCRIPTION OF RESULTS

 

The Overall Characteristics of Technological Scanning Activities

Analysis of the overall characteristics of TECHSCAN practices revealed a number of elements. First, interest in product, equipment and process innovations did not predominate over other types of interest (marketing, financial, etc.). The results, as might have been expected, highlighted the multidimensional nature of TECHSCAN in SMBs. The sources of information perceived as most important by the firms were also those that were used most often. The first ten were, in decreasing order of importance: customers, specialized reviews, production staff, executives, sales staff, suppliers, brochures and catalogues, industrial fairs and exhibitions, trade fairs and exibitions, and general business journals. Although informal methods (direct or telephone contact, ad-hoc meetings) predominated along the informational process, a non-negligeable number of firms also used formal methods (meetings or different types of formal study). Scanning activities were generally performed on an ad-hoc, informal basis under the leadership of the owner-manager, supported by a group of employees. Despite its informal nature, TECHSCAN nevertheless occupied an appreciable role in the strategic management of a fairly large proportion of firms (77.1%).

 

The Latent Dimensions of Technological Scanning Practices

Because of the number of variables used to measure TECHSCAN practices, principal component factorial analysis was used to identify the latent dimensions. The results suggested that 14 latent variables (or dimensions) could be used to analyze scanning practices: three related to objectives, three to information, six to information sources and the last two to organization of scanning activities (see Table 1). The Kaiser-Meyer-Olin index values (from 73% to 88%) confirmed that the factorial model matched the data (Norusis, 1991), and the Cronbach alpha values (from 73% to 91%) confirmed the internal consistency of the measures (Carmines and Zeller, 1979; DeVellis, 1991).

TABLE 1
The Latent Dimensions of Technological Scanning Practices

Variable Categories

Initial No. of Variables

No. of Latent Variables

Variance Explained

KMO

a Cronbach

Latent
Dimensions

Scanning objectives
Types of information
Information sources
Management practices

17
12
26
34

3
3
6
2

53.8%
65.8%
63.2%
59.8%

0.88
0.84
0.87
0.73

0.87
0.88
0.91
0.73

1: PERFOBJ
2: COMPOBJ
3: PRODOBJ
4: FINHRINF
5: INOVINF
6: MARKINF
7: KNOWSOU
8: INDUSOU
9: ASSISOU
10: INTERSOU
11: SALESOU
12: OPERSOU
13: DIVMETH
14:COMPMAN

 

The first three dimensions were objective-related: performance (PERFOBJ: improvement of service quality and brand image, satisfaction of demand, meeting of deadlines and product diversification), competitiveness (COMPOBJ: cost reduction and control, sustaining and improving of competitiveness) and production (PRODOBJ: increased production capacity, productivity and flexibility, and reduced manufacturing times). The next three dimensions were concerned with information needs for TECHSCAN purposes: financial and human resources (FINHRINF: cost and cost effectiveness of technological change, staff recruitment and training needs), innovation (INOVINF: product, process and equipment innovations) and marketing (MARKINF: identification of market potential, suppliers and competitors). The information sources used by these firms were grouped in six dimensions, identified respectively as sources of basic knowledge (KNOWSOU: universities, colleges, research centers, textbooks and consultants), industry-related sources (INDUSOU: fairs and exhibitions, business and specialized publications, advertising brochures and catalogues), support sources (ASSISOU: government organizations and publications, financial institutions), internal sources (INTERSOU: management, production staff and internal database), sales-related sources (SALESOU: sales staff, agents, representatives, customers), and operational sources (OPERSOU: order givers, subcontractors, suppliers and competitors). The last two dimensions were concerned with TECHSCAN management practices: the diversity of methods used in the informational process (DIVMETH: information gathering, analysis and dissemination methods) and management complexity (COMPMAN: frequency and level of formality of activities, staff involvement and inclusion of scanning in strategic management).

 

Configurations and Success Factors of Technological Scanning

Cluster analyses were performed to verify the existence of separate TECHSCAN practice configurations. The use of the jump heuristic (Norusis, 1991; Aldenderfer and Blashfield, 1984), combined with the results of the variance analysis, revealed an optimal solution comprising four scanning groups or configurations. Table 2 shows the means and standard deviations of the 14 dimensions by group, and the results of the variance analysis. It is worth noting that the dimensions obtained from the cluster analysis are standardized variables (mean= 0 and variance= 1).

TABLE 2
The Configurations of Technological Scanning Practices

Scanning
Dimensions

Group 1
(n=68)

Group 2
(n=71)

Group 3
(n= 104)

Group 4
(n=81)

F

 

Mean

Stand.
Dev.

Mean

Stand.
Dev.

Mean

Stand.
Dev.

Mean

Stand.
Dev.

Anova

Scanning objectives PERFOBJ
COMPOBJ
PRODOBJ

Types of information
FINHRINF
INOVINF
MARKINF

Sources of information
KNOWSOU
INDUSOU
ASSISOU
INTERSOU
SALESOU
OPERSOU

Management practices
DIVMETH
COMPMAN


-1.10
-0.18
-0.02


-0.54
-0.57
-0.65

-0.18
-0.34
-0.45
-0.53
-0.21
-0.41

-0.41
-0.86


1.15
1.17
0.85


1.22
1.06
1.14

0.96
1.08
0.82
1.01
0.98
0.94

0.64
0.96


0.53
-0.43
-0.98

-0.24
-0.30
0.11

-0.14
0.01
-0.10
-0.38
-0.39
0.23

-0.48
-0.10


0.71
0.95
0.90

1.05
1.08
0.93

0.85
1.04
1.03
1.08
1.07
1.04

0.61
0.68


0.31
0.04
0.60

0.22
0.48
0.32

0.05
0.04
0.53
0.19
0.14
-0.12

-0.36
0.55


0.58
0.90
0.64

0.74
0.70
0.80

1.07
0.95
0.90
0.80
0.97
0.92

0.57
0.86


0.05
0.48
0.11

0.38
0.13
0.02

0.22
0.22
-0.22
0.52
0.34
0.30

1.24
0.10


0.77
0.76
0.90

0.76
0.85
0.91

1.01
0.87
0.90
0.76
0.81
0.96

0.91
0.90


58.28***
12.86***
53.24***

15.76***
21.37***
15.51***

2.77*
4.25**
18.89***
22.67***
9.41***
8.91***

115.89***
37.53***

*: p =0.05 **: p =0.01 ***:p =0.001

The four TECHSCAN practices configurations differ significantly at a significance level of at least .01, except for one of the 14 dimensions which involve the use of basic knowledge sources (KNOWSOU). The means for each dimension, if compared between configurations, exhibit an upward trend. The four configurations clearly represent four distinct levels of TECHSCAN development that can be shown on a continuum measuring the state of scanning practice development. The 68 Group 1 firms clearly have the least developed practices. Their means are always negative and the lowest over 11 of the 14 dimensions. These firms consider all objectives and information to be relatively unimportant, make little use of information sources and have not organized their activities. At the other end of the continuum are the 81 Group 4 firms, whose TECHSCAN practices are clearly the most developed. The means of this group over all 14 scanning dimensions are generally positive and often the highest (in 8 of the 14 dimensions). The TECHSCAN configurations of the Group 2 and Group 3 firms fall between these two extremes. The state of TECHSCAN development in the 71 Group 2 firms is similar overall to the Group 1 firms. Their means are usually negative, but with higher values than the Group 1 firms in 8 of the 14 scanning dimensions. In other words, their scanning activities are more developed than those of their Group 1 counterparts. The TECHSCAN configuration of the 104 Group 3 firms resembles that of Group 4. Their means are generally positive, and are lower than the Group 4 means in 8 of the 14 dimensions, but higher than the Group 2 means in 12 of the 14 dimensions. TECHSCAN development between configurations follows an upward trend, implicitly suggesting increased awareness of the need to scan the environment, clarify information fields, intensify use of information sources and develop scanning organization. However, as the means show, TECHSCAN development does not follow a linear pattern.

The dimensions emphasized in TECHSCAN activities also differ between groups. This suggests that TECHSCAN development is reflected not only in the overall upward trend in the intensity of activities, but also in qualitative changes, expressed by the change of dominant dimensions (i.e. dimensions with the highest means) from one group to another. For example, firms in Group 1 focus on production-related objectives (PRODOBJ), basic knowledge (KNOWSOU) and sales-related information sources (SALESOU) and on scanning methods (DIVMETH), They place more or less equal value on the different types of information, make more use of different types of information. Group 2 firms, on the other hand, put more emphasis on performance-related objectives (PERFOBJ), marketing information (MARKINF), operational information sources (OPERSOU) and on the management of the scanning activities (COMPMAN). In the Group 3 firms, the focused dimensions revolve around production-related objectives (PRODOBJ), information related to innovations (INOVINF), information sources of support (ASSISOU) and management of scanning activities (COMPMAN). Finally, firms in Group 4 are concerned firstly with competitiveness objectives (COMPOBJ), informations related to finance and human resources (FINHRINF), internal sources (INTERSOU) and scanning methods (DIVMETH). Since this research was concerned with TECHSCAN, it is worth noting that only the Group 3 firms reported information related to innovations as being the most important. This is somewhat surprising. At the very least, it shows that TECHSCAN is not limited to technological aspects only. This results reveal that the definition which is generally given to the concept of TECHCAN in the literature has to be broadened in order to better reflect SMBs practices.

In addition to the two phenomena described above, TECHSCAN process development is also characterized by a number of recurrent dimensions. The recurrence phenomenon is usually accompanied by an increase in the relative weight of the dimensions concerned. For example, production-related objectives (PRODOBJ) are dominant for Group 1 firms (-0.02), marginal (i.e. having the lowest mean) for Group 2 firms (-0.98), and dominant once again for Group 3 firms (0.60). The same applies to eight other dimensions of TECHSCAN: PERFOBJ, FINHRINF, MARKINF, INTERSOU, ASSISOU, OPERSOU, DIVMETH, and COMPMAN. In the light of organizational change theories (Mintzberg and Westley, 1992), these data seem to indicate an upward spiral of TECHSCAN development. The additional tests we performed indicate that this developmental process could be best designed by a two-phase framework - digestive (or stability) and active (revolution).

In order to identify the TECHSCAN success factors, we performed anovatests to compare means for owner-managers profile, technological and organizational characteristics of the firms, information networks and characteristics of the environment (see Table 3). In all, wenty-one contingency factors were taken into account this analysis.

TABLE 3
The Firms Characteristics related to the Configurations

Characteristics

Group 1
(n=68)

Group 2
(n=71)

Group 3
(n= 104)

Group 4 (n=81)

F

analyzed

Mean

Stand.

Dev.

Mean

Stand.

Dev.

Mean

Stand.

Dev.

Mean

Stand.

Dev.

Anova

Managers:
DOMSPEC (1)
LEVEDUC (2)
MANEXP (3)
SECTEXP(3)

Firms:
TECHPRO (4)
TECHINF (4)
MODERN (4)
RDSCOPE (5)
RDSIZE (6)
STRATEGY (7)
EMPSIZE (8)
TURNOVER (9)
SIZEGROW(10)
TURNGROW (10)
EXPORT (11)

Networks:
PROFASS (12)
MEMBPA (12)
RECENTER (12)

Environment:
ENVTURB (13)
ENVUNCER (13)
INTCOMP (14)


3,27
3.03
10.36
19.90


1.33
1.88
3.22
1.08
2.47
1.93
60
9.59
2.75
2.83
14


0.43
0.52
0.46


2.89
2.89
27.13


1.33
1.00
7.23
11.15


1.54
1.75
2.78
0.97
2.94
0.75
18
29.95
1.41
1.44
23


0.77
0.74
0.73


0.91
0.77
36.70


3.27
2.93
10.28
19.87

1.15
1.66
2.81
1.03
3.34
2.20
56
10.24
2.78
3.18
13


0.33
0.55
0.38


2.85
2.98
33.58


1.30
0.89
6.56
10.52

1.34
1.29
2.35
0.98
6.15
0.78
133
29.62
1.29
1.41
23

0.68
0.85
0.73

0.72
0.71
39.16


3.18
2.90
11.75
19.90

1.34
1.92
3.26
1.30
4.95
2.40
60
6.34
3.05
3.10
14

0.49
0.58
0.35

3.00
3.15
22.80


1.28
0.95
8.94
10.91


1.43
1.55
2.69
0.89
7.43
0.80
114
14.99
1.44
1.42
25

0.86
0.84
0.57

0.78
0.80
33.99


3.72
3.42
9.28
17.43

1.75
2.56
4.32
1.68
5.24
2.32
167
36.07
3.01
3.16
22

0.86
1.00
0.56

2.97
3.13
27.63


1.16
0.85
8.36
9.40

1.39
1.49
2.36
1.13
6.58
0.81
400
11.55
1.49
1.42
30

1.00
1.14
0.68


0.81
0.83
37.23


2.64*
5.24**
n.s.
n.s.

n.s.
5.04**
4.90**
6.33***
3.21*
4.73**
4.59**
3.39*
n.s.
n.s.
n.s.

4.32**
4.10**
n.s.

n.s.
n.s.
n.s.

*: p =0.05 **: p =0.01 ***:p =0.001

(1): 1=General educ. to 5= Scientific education
(2): 1=Elementary to 5=University
(3): Number of years’ experience
(4): Number of technologies owned
(5): Number of R&D fields
(6): Number of employees assigned to R&D
(7): Miles and Snow’s (1978) typology
(8): Total workforce
(9): Turnover in $ millions
(10): 1=Drop to 5=growth above 30%
(11): As a percentage of turnover
(12): Number of prof. assocs. & research centers
(13): Miller et Dröge’s (1986) grid
(14): As a percentage of total competition.

The four groups differed with respect to 11 factors. In decreasing order of importance (considering the F value), these factors were: the scope of R&D activities (RDSCOPE), level of education of managers (LEVEDUC), the number of management technologies owned (TECHINF), the overall state of modernization (MODERN) which is measured by the total of management and production technologies owned, the strategy proactiveness (STRATEGY), the total workforce (EMPSIZE), the number of professional associations existing locally (PROFASS), the number of professional associations in which management is actively involved (MEMBPA), total turnover (TURNOVER), the number of employees assigned to R&D (RDSIZE), and the managers’ specialty field (DOMSPEC). The means of all these factors increase from one group to the next. TECHSCAN is therefore most developed in firms that have proactive strategies, larger workforces and turnovers, more modern technology, a greater diversity of R&D activities on which more people work, that are located near a larger number of professional associations and whose managers are better educated, have a technical or scientific background and are actively involved in professional associations.

Since the state of technological modernization, measured by the total number of new technologies owned, could be considered as an appropriate indicator of TECHSCAN success, we can deduce that the firms' R&D potential, the level of education of managers, firms' size (in terms of workforce and turnover) and the local presence of professional associations in which its managers are actively involved are all success factors for its scanning activities. The four groups did not significantfly differ along ten other factors including: professional (MANEXP) and activity sector experience (SECTEXP) of the managers, the firms' endowment in production technology (TECHPRO), employees (SIZEGROW) and turnover (TURNGROW) growth rate, the contribution of exports to turnover (EXPORT), the number of research centers (RECENTER) existing locally, perceived turbulence (ENVTURB) and uncertainty (ENVUNCER) in the environment and the intensity of international competition (INTCOMP).

 

Assessing the Measurement Model

The partial least square (PLS) method was chosen in preference to the better-known LISREL, since it is more appropriate for causal-predictive analysis. It is particularly useful in the initial phase of developing and verifying theories (Fornell and Bookstein, 1982). PLS is also more robust and does not require a large sample or normally distributed multivariate data (Fornell and Larcker, 1981). This was the case for our research. Figure 2 summarizes the results obtained.

The PLS structural equation model allows the research model to be tested and the properties of the underlying empirical model to be verified at the same time. Internal consistency of measures, i.e. their unidimensionality and their reliability, must be verified first. The observable indicators measuring a non-observable construct (or latent variable) must be unidimensional to be considered unique values. In our case, unidimensionality was satisfied since all measurement loadings were above 0.5, except for management technologies (TECHINF). Reliability can be verified by considering the value of the rho coefficient, defined as the ratio between the square of the sum of the loadings plus the sum of the errors due to construct variance. An rho of 0.7 shows that the variance of a given construct explains 70% of the variance of the corresponding measure. In Figure 2, this is the case for all the constructs in the research model. The second property to be verified is discriminant validity. It shows the extent to which each construct in the research model is unique and different from the others. PLS uses the correlations between each pair of latent variables as criteria. The shared variance between two variables must be less than the average extracted variance (AEV). In our case, this property was also satisfied, since the highest value of the shared variance between all the variable pairs was 0.36 (between objectives and information), and the lowest average extracted variance value was 0.54 (for sources).

 

Testing the Research Hypothesis

We began with the overall hypothesis that TECHSCAN practices are affected by the characteristics of the firm, the profile of the manager, the perception of the environment and the presence of and access to information networks.

FIGURE 2
Results of the PLS Analysis (N=324)

 

* p: = .05 ** p= .01 *** p=.001

Organizational characteristics of firms were separated from their technological characteristics. Of the six organizational factors, the firm’s strategy most affected TECHSCAN practices. The addition of size and growth factors (number of employees and turnover size) and the export factor in the analysis did not improve the overall results at this level. We used Miles and Snow’s (1978) generic strategy typology. These authors distinguished between defensive, reactor, analyser and prospector strategies. The four path coefficients were all positive (.09, .22, .17 and .14) and significant at least .01. They showed that a proactive strategic behaviour intensified TECHSCAN practices with respect to the importance given to the different types of objectives and informations, the use frequency with of information sources and the organization of activities. These results confirm those reported by Miles and Snow (1978) and Miller and Friesen (1982), but contradict those of Hambrick (1982) regarding strategy and those of Johnson and Kuehn (1987) regarding the effect of size. A positive causal relationship was also found between the strategic behaviour of the firms and their technological characteristics. So, a proactive strategic behaviour had a positive effect on new technology adoption. All the four factors measuring the firms' technological characteristics significantly affected scanning practices. These technological characteristics are best explained by the number of people regularly assigned to R&D activities. The more these firms adopted new technologies and enlarge the scope and the size of their R&D activities and had more R&D staff had a positive impact on their TECHSCAN practices. All path coefficients were significant and positive. These results confirm Julien’s (1995) theories and, implicitly, those of Cohen and Levinthal (1990) regarding the ability of firms to innovate.

The managers also had an impact on TECHSCAN practices in their firms. However, of the four initial factors, only level of education was found to be relevant. The addition of specialty field and professional experience did not improve the results. Contrary to the above results, the path coefficients were significant but negative, except in the case of organization of activities. These results therefore suggest that better educated managers give less importance to overall objectives and information and make less use of overall information sources. On the other hand, the better educated the managers, the more complex the organization of their scanning activities. A possible explanation of this would be that as the managers became more educated they tend to focus TECHSCAN activities of their firms on specific objectives, information needs and sources. This remains doubtful, however, because the same reasoning can be applied to experience factors that were found to be irrelevant. These results only partially confirm the theories suppoprted by Rothwell (1990), Radnor (1992) and Julien (1995).

The perception of the environment was found to be one of the most relevant factors explaining the state of TECHSCAN practices. The addition of international competition and activity sector did not improve the results. The factor that best explained the managers’ perception of their environment was perception of uncertainty. However, not all path coefficients were significant. This was the case for organization of activities. It was found that the intensity of uncertainty and perceived environmental turbulence had a positive impact on the importance given to the different objectives and types of information and the frequency with which information sources were used, but not on organization of activities. These results partially confirm our hypothesis, and support the results obtained by Daft et al. (1988) and Culnan (1983).

The presence of and access to information networks also had a positive impact on TECHSCAN practices. The factor that best explained the information network construct was the active involvement of managers in the network. However, not all path coefficients were significant. This was the case in particular for objectives. These results partly confirm our overall hypothesis and suggest that the existence of local professional associations and research centres, and management involvement in the associations, have an impact on information seeking, resource use and the ocmplexity of scanning activity management. They also confirm the results obtained by O’Reilly (1982) and Culnan (1983).

Generally speaking, the results confirmed the general research hypothesis. The firm’s strategy and their perception of the environment were the factors that had the greatest impact on TECHSCAN practices. The path coefficients showed that the factors had different impacts and could be arranged hierarchically. In decreasing order, these factors are: strategy, perception of the environment, technological and R-D characteristics, level of education of managers, and finally information networks.

Previous Page | Main Menu | Next Page

 

 

© 1997 Babson College All Rights Reserved
Last Updated 1/15/97 by Geoff Goldman & Dennis Valencia

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