Espacios. Espacios. Vol. 31 (3) 2010. Pág. 36

Measuring Knowledge Exploitation and Exploration: An Empirical Application in a Technological Development Center in Brazil

Medición de la explotación y exploración del conocimiento: Una aplicación empírica en un Centro de Desarrollo Tecnológico de Brasil

Silvio Popadiuk, Marcos Antonio Franklin, Patricia Gonçalvez Vidal, Lilian Aparecida Pasquini Miguel y Vanderli Correia Prieto


Factor Analysis

The exploratory factor analysis started with the 46 original attributes in the scale, the first factor extraction, using the principal component model with varimax rotation, revealed the existence of 17 factors, and some factors had only one attribute. With the help of a matrix of correlation of these 46 attributes, we were able to discard the attributes that were not adding to the model. A new factor analysis allowed to the improvement of the model and the exclusion of other attributes that were not helping explain the model. With 25 attributes, we run the factor analysis again, and this analysis presented a KMS of 0.811, presenting a good fit of the sample, and the Bartlett´s test of sphericity was considered significant at the 0.0% level. Seven factors were extracted from the analysis of 25 attributes that explained 70.9% of the variance of the model. The eigenvalues and explained variance, as well as the factor loadings are presented in Table 4.

The RDI was structured in four departments with activities totally distinct, as well as a support department, we expected to find a large variation regarding the evaluation of the attributes shown in this survey. For this reason, we checked the dimension related to innovation, to see if there was significant difference in the groups. The option to examine the innovation dimension is explained by the fact that, in an R&D institute, the focus is on the search for innovation, either radical or incremental. To analyze the difference in groups, we used cluster analysis and identified two separated groups of employees in this institute. The first group was formed by 36 individuals. These individuals had the higher evaluation for the four attributes related to innovation, and, therefore, were classified as the Higher Innovators. The second group, Low Innovators, formed by 34 individuals, had the lower score for the innovation dimension.

Table 5 shows the statistics related to the factor analyses, the Cronbach´s alpha (its values are in parentheses), as well as the mean scores for each attribute evaluated, according to the dimension resulting from the factor analysis, as well as the cluster analysis of high and low innovators. The factor loadings are presented in the first column, associated with the mean scores for each attribute, considering the whole sample.

For each attribute, but three (competition, cost, and partnership), the Mann-Whitney U non-parametric test tended to be significant at the 1% level. This test compares two independent samples for distributions that do not meet the normality assumption. What this test revealed was that there were two groups of employees working at the RDI with total different view regarding the attributes investigated here. For the high innovator group the means tended to be close to the upper limit in the scale, i.e. positioning more favorably in agreeing with the attribute. For the dimensions knowledge, innovation and strategy, the tendency revealed that this group evaluated the RDI with an orientation tending more to the exploration of knowledge. However, the same group evaluated the dimensions efficiency, competition, cost and partnership with high values, i.e. the same group evaluated the RDI as tending to exploitation in these dimensions.

Table 5 presents the means associated to the seven dimensions, according to the two groups identified in this research. Table 5 also presents the significant correlations for the dimensions (significance level of 5% or lower). The top of the Table 5 presents the correlations for the High Innovator group, and lower part presents the correlations related to the Low Innovator group. Table 5 also presents the means for the dimensions for both groups: the Low Innovator (in the final column) and the High Innovator (in the line that reads means). The values for test of Mann-Whitney U for two independent samples are presented in the final line of Table 5. All the dimensions are significantly different, at 5% level of significance, but two of them: the dimensions of competition and of partnership. These two dimensions being not different makes sense if we think about the competition for the whole institute as well as the partnership with the major client (the multinational company) as well as with the major universities.

Table 4
Statistics for the Attributes associated to the dimensions

 

Loadings

Means

Level of innovation

Significance Mann -Whitney U

High

Low

Factor 1 – Knowledge dimension (0.875)

 

 

 

 

 

Volume of new ideas generation (Low – High)

0.619

3.50

4.28

2.68

0.000

Use o new sources of knowledge (Low – High)

0.780

4.03

4.39

3.65

0.013

Learning intensity (Low- High)

0.491

4.27

4.89

3.62

0.000

Form of capaciting the team (Sporadic – Continuous)

0.784

4.06

4.36

3.74

0.008

People development intensity (Low – High)

0.520

3.96

4.28

3.62

0.010

There is a constant knowledge sharing (Disagree – Agree)

0.453

3.59

4.14

3.00

0.000

Social interaction is part of organizational culture (Disagree – Agree)

0.420

3.61

4.00

3.18

0.006

Individual knowledge appreciation (Low – High)

0.673

3.77

4.22

3.27

0.002

Factor 2 – Efficiency dimension (0.780)

 

 

 

 

 

Degree of existing knowledge utilization (Low – High)

0.576

3.99

4.44

3.50

0.000

Importance level of efficiency  (Low - High)

0.794

4.39

4.78

3.97

0.008

Concerns about economy of scale (Low – High)

0.663

3.67

4.17

3.15

0.000

Level of knowledge exploitation (Minimum – Maximum)

0.331

3.62

4.16

3.03

0.000

Factor 3 -  Innovation dimension (0.814)

 

 

 

 

 

Focus on products and process totally new (Low – High)

0.572

 

4.53

2.82

0.000

Products improvement (Sporadic – Continuous)

0.747

 

4.75

3.18

0.000

Diversity of generating product or processes (Low – High)

0.583

 

4.14

3.14

0.000

Prototype development (Sporadic – Continuous)

0.774

 

4.22

2.53

0.000

Factor 4 – Strategy dimension (0.713)

 

 

 

 

 

Information technology orientation (Weak – Strong)

0.841

 

4.36

3.79

0.022

Strategic view focused on (Present – Future)

0.525

 

4.47

3.39

0.001

Time horizon for the organizational strategy (Short time – Long time)

0.528

 

4.11

3.12

0.000

Factor 5 – Competition dimension (0.766)

 

 

 

 

 

Emerging competitors with similar characteristics (Limited – Intense)

0.832

 

3.69

3.38

0.391

Competitor´s activities with similar characteristics (Reduced – Intense)

0.819

 

3.83

3.12

0.018

Factor 6 – Cost dimension (0.764)

 

 

 

 

 

Preoccupation with R&D cost (Low – High)

0.875

 

4.39

3.72

0.036

Cost focus (Low – High)

0.855

 

4.50

3.94

0.124

Factor 7 – Partnership dimension

 

 

 

 

 

Alliances forming (Situational – Durable)

0.585

 

3.64

3.32

0.318

The intensity level of the partnership contracts (Low – High)

0.787

 

3.64

3.03

0.025

Table 5
Spearman´s rho coefficients and summated scale means for dimensions

Variables

Means (High innovators)

Means (Low innovators)

Mann-Whitney U significance

1

2

3

4

5

6

7

1 - Knowledge

4.31

3.28

0.000

 

 

 

 

 

 

 

 

2 - Efficiency

4.32

3.47

0.000

0.506**

0.630***

 

 

 

 

 

 

 

3 - Innovation

4.41

2.92

0.000

0.214

0.462**

0.597***

0.363*

 

 

 

 

 

 

4 - Strategy

4.31

3.44

0.000

0.356*

0.557***

0,319

0.540***

0,137

0.423

 

 

 

 

 

5 - Competition

3.76

3.26

0.060

0,072

0,137

0,157

0,287

0,295

-0,120

 

 

 

 

6 – Cost

4.44

4.44

0.037

0,209

0,113

0,069

0,232

0,140

-0,53

-0,156

0,151

0,127

0,163

 

 

 

7  -Partnership

3.64

3.18

0.060

0.481**

0.403*

0.685***

0.401*

0.560***

0.191

0.284

0.415*

0.204

0.350*

0.049

-0.049

 

 

 

Regression with Optimal Scaling (CATREG)

As the previous results revealed that almost all attributes and the mean values for the dimensions are statistically significantly different between the two groups, two regressions for categorical data were performed, using the optimal scaling model from the SPSS, version 13. Regression with optimal scaling is also known by the acronym CATREG, for categorical regression with optimal scaling. Categorical variables serve to separate groups of cases, and the technique estimates separate sets of parameters for each group. The estimates coefficients reflect how changes in the predictor affect the response. Prediction of the response is possible for any combination of predictor values (Meulman and Heiser, 1999). The purpose of this procedure was to determine if the dimensions that explained the innovation dimension were the same for both groups. Since the focus of this research project is an R&D institute that serves a multinational company, we can argue that such type of institute focuses on developing innovation, and that the innovation depends on all the activities that the institute performs. Therefore, we can say that the innovation has the following relationship with the different dimensions: Innovation = F (knowledge, efficiency, strategy, competition, cost, partnership)

Table 6 presents the results for the two regression equations using the optimal scaling model (Meulman and Heiser, 1999). For the High Innovator group, the dimensions that are the most important to explain innovation are the knowledge (β=.503; p-value<0.01) and costs (β=-.465; p-value<0.001), this equation explain 64% of the variance (R2=.64, Adj. R2 =.399). The coefficient for knowledge is positive, indicating that more focus on knowledge will increase the innovation, while the coefficient for cost is negative, indicating an inverse relationship: more focus on cost, less innovation. For the Low Innovator group, the dimensions that are the most important to explain innovation are efficiency (β=.324; p-value<0.05), competition (β=.618; p-value<0.00), and cost (β=.317; p-value<0.05), this equation explain 74.5% of the variance (R²=.745, Adj. R² =.56).

Table 6
Categorical Regression with optimal scaling

 

High Innovator

Low Innovator

Knowledge

0.503***   (0.153)

-0.042       (0.125)

Efficiency

0.226         (0.142)

0.324**    (0.145)

Strategy

0.201         (0.163)

-0.244*     (0.132)

Competition

0.097         (0.138)

0.618***  (0.131)

Cost

-0.465***  (0.151)

0.317**    (0.124)

Partnership

-0.25*       (0.133)

0.012        (0.135)

N

36

34

0.64

0.745

Adjusted R²

0.399

0.56

F

2.661

4.04***

Note: standard error in parentheses
*** p < .01 **p < .05 *p < .10

These results show that there is a substantial difference between the two groups regarding the innovation dimension in the RDI. Moreover, the two regression results are very consistent to the theory. It seems natural that the group that is more innovative would be more focused on knowledge than in cost. In the other hand, the less innovative group could also be considering efficiency, competition and cost to be restricting innovation.

March (1991) and his followers (Cheng and Van de Ven, 1996; He and Wong, 2004; Popadiuk and Choo, 2006; Popadiuk, 2007) argue that exploration is related to new sources of knowledge, generation of new ideas, intense learning processes and organizational empowerment, sharing and valorization of individual and collective knowledge, as well as a high level of social interaction. This tendency towards valorization of knowledge should lead to search for totally new products and processes, product diversification, experimentation with new products and development of prototypes. Following these stream of thought, it seems that an organization that focus on exploration will also have a long-term strategic planning as well as a vision towards the future.

On the other hand, in an organization more oriented towards exploitation, it can be argued that the control mechanism should be more flexible (Burns and Stalker, 1961). The level of efficiency, the economies of scale, the use of explicit knowledge, the use of information technology, the costs of research and development should not be restrictive influence in the search for innovation. Regarding competition, during the exploration phase of an industry, the theory considers that there are few competitors with similar characteristics which makes it more difficult to develop alliances with partners. At this phase, the organization is still looking for specific partners to initiate its innovation generation process. Therefore, the alliances tend to be sparse and the alliance contracts can be established with less level of details regarding the technical and timing specifications.

One of the main purposes of this research project was to develop of a questionnaire that could assess the type of strategic orientation towards exploration and exploitation of knowledge (exploiter or explorer of knowledge). We started with 46 attributes, discarded 19 after the exploratory factor analysis. The final result of this factor analysis showed seven factors, which are interpreted as the dimensions commented previously.

From the results, the theoretical considerations about exploration and exploitation need to be bounded. What has been proposed by March (1991) and other researchers refers to an attempt to create a general model about the implications associated with the exploration and exploitation strategies and, probably, to specific business situations. The context of the discussion about the exploration and exploitation strategies needs to be well defined and explicit for one to argue that exploration involves more risks, more costs, emphasis on the development of activities to create organizational knowledge as well as to develop partnerships. Even in one organization, its departments may act in different manner regarding the exploration and exploitation strategies. Due to its specificities, some departments due to its activities need to be more explorer, as it should be the case for a P&D department, while other departments need to be more exploiter, for example the production department.

Also, it is necessary that the discussion uses the two perspectives to analyze the institute, the internal and external, considering the potential seven dimensions, as identified in this study. Moreover, this delimitation contribute to make explicit which situations can be characterized as present in an orientation towards explorer, exploiter, or, even, for a situation of equilibrium as proposed by March (1991).

Regarding the RDI, object of our study, the theoretical considerations about exploration and exploitation were not completely verified. For the High Innovator group, the theoretical expectation related to exploration regarding the dimensions (Table 3) was not confirmed for two dimensions: efficiency and costs. It was expected that in a situation of exploration, the preoccupation with efficiency and costs would be relatively lower than in a situation of exploitation. However, for both dimensions the contrary was true. This fact can be explained by the context that the RDI is in, where it needs to provide its only client with the lowest cost solution at the fastest time.

For the Low Innovator group, there were also some diverse results comparing to what was expected. For this case, it was expected that the efficiency dimension would have a significant influence, the competition dimension would be more intense, and the emphasis on cost would be higher, as well as the use of partnerships. However, the results showed that the Low Innovator group had inversed orientation, which translate in a divergent view from the theoretical considerations.

These divergent findings related to the current theory on exploration and exploitation should not be taken as conclusive, and they can also be explained by some possibilities, some of them related to the limitations of this research.

The first limitation refers to the conceptual model developed for this research. Since it is a first attempt to assess exploration and exploitation, in the proposed format, i.e., to verify the fit between theory and practice, the formulation of the attributes may have influenced the results, probably due to lack of definition for the attribute, or they may not be exactly what we were trying to measure. In this way, the respondents of this research may have had difficulties in understanding some attributes.

A second limitation refers to the sample used here. Only one institute of research was used in this assessment. Since the results presented here comes from this institute, which has a very specific range of activities, basically one client, the multinational company that it spin-out from; these results are not generalizable to other organizations.

Another limitation refers to the operationalization of the indicators to verify the model´s hypotheses. Since it was not possible to gather data from a larger number of organizations and, therefore, to classify them as High and Low Innovators, to analyze the tendencies of the dimensions developed in this research, we assumed that there was two sets of employees in the institute analyzed, according to their vision regarding innovation, assessed by the four attributes of these dimension proposed in this research. In an ideal situation, it would also be wise to have other attributes to measure innovation, for example, a measure of the number of innovations developed in the year.

Future studies could be developed focusing on improving the set of attributes related to exploration and exploitation, as well as the dimensions obtained in this experimental model. The samples should contemplate organizations in different sectors of activities, aiming to verify if the theoretical model can be generalized or if there are contingencies that need to be specified accordingly to the idiosyncrasies inherent to the specific segments of each organization.

From the academic perspective, the major contribution of this study is related to the experimental model developed here, which offers elements to start a process of identifying if the practices of exploration and exploitation are in line with what is expected in theory. From the practitioner perspective, the model allows to diagnose the elements of the management process as well as the process of organizational learning, the essence of exploration and exploitation, given that it offers an ample set of attributes related to the internal and external environment likely to be measure by the developed scale and, therefore, likely to be compared over time, as the organization promote actions to change the course in order to fulfill its strategic and operational objectives.

References

[anterior] [inicio]

Vol. 31 (3) 2010
[Índice]