SECTION V СЕКЦИЯ V
4. Data Analysis and Results
4.1. The Indicators of the New Product in the Stages of the Innovation Process
The three tables, Table 6, Table 7, and Table 8, reflect, respectively, the number and percentage of enterprises that have improved the objectives of the new concept, the new prototype and the new final product.
Table 6. The number and percentage of firms in improving the design goals.
Number of improvements New Design Improvements
Total improvement
% of the improvement
% of the non- improvement
Easy to adapt to ingredients 3 60% 40%
Design time 3 60% 40%
Cost design 3 60% 40%
Table 7. The number and percentage of firms in improving the prototype development goals.
Number of improvements New Prototype Improvements
Total improvement
% of the improvement
% of the non- improvement
Quality of product components 4 80% 20%
Taste 5 100% 0%
Texture 4 80% 20%
Smell 3 60% 40%
Aesthetic image 4 80% 20%
Development time 2 40% 60%
Cost of development 1 20% 80%
Table 8. The number and percentage of companies improving production targets.
Number of improvements New Final Product Improvements
Total improvement
% of the improvement
% of the non- improvement
The number of products rejected 3 60% 40%
Number of non-compliant
products 5 100% 0%
Production time 5 100% 0%
Production cost 4 80% 20%
4.2. The Confirmatory Analysis
After interviewing the companies in each category, it turned out that five companies were all in disagreement with certain variables. Therefore, we chose to eliminate them because their variance was zero.
Eliminating these variables enabled us to keep the variables that give us an acceptable Cronbach alpha of and which prove that can significantly affect the improvement of the new product.
Thus, we identified through the following 2 Built:
・ The first built including: engineers in research and development, internal spending on research and development, external expenditure on research and development, spending on information technology, spending on equipment and materials, and spending on technology point ;
・ The second building including: Workforce;
We noted that the “patents” indicator does not exist either as an input to the innovation process or as a result of performance. Hence, hypothesis 3b cannot be verified in our present study.
Therefore, each element of the construct of the dependent and independent variables will divide into three sub-elements each with values 1 and 0.
This recoding is done with the aim of transforming the qualitative variables into quantitative variables that will be ready for the correlation test that we will apply later on.
For example, for the design cost of the new prototype which is measured on a Likert scale ranging from 1: low to 3: high; It will take the following values:
・ For design cost _1: 1 if the design cost improvement is low, 0 otherwise;
・ For design cost _2: 1 if the design cost improvement is average, 0 otherwise;
・ For design cost _3: 1 if the design cost improvement is high, 0 otherwise.
We conducted thereafter, purification of the scale of all constructs we have; those entries and those outputs. We ended up with the following results.
According to the confirmatory analysis, we assume that built our variables are one-dimensional (see Table 9). This condition is mandatory as part of a reflexive modeling but not in a formative modeling.
The results of the confirmatory analysis have restored to the resources of innovation the following two constructs:
The performance of the new prototype was divided into:
- Internal R & D expenditure;
- Workforce.
At the level of improvements made in each of the three phases of the innovation process, we had:
- The improvements made to the new design: easy adaptation to ingredients, improved design time, and improved cost;
- The improvements made to the new prototype: the improvement of the quality of the new prototype, its taste, its odor, its development time and its aesthetic image;
- The improvements made in the new final product: improving the number of discards, improving the number of non-compliant products, and improving the cost of production.”
Convergent validity of the measurement model scales is assessed by first examining the level and significance of the contribution factor (factor loadings) generated by the PLS algorithm (which are interpreted in the same way that ACP). The usual rules used in the factor analysis for structural equation models, contributions should be high (>0.5) and significant [23].
The convergent construct validity can also be evaluated by showing that the items measuring a built are more highly correlated with the construct with the other built the model [24].
Here, we find that the absolute contributions of variables 8 built are all above 0.5 (see Table 9).
Table 9. The results of the confirmatory analysis of the constructed inputs and performance indicators.
Factor Items Own
values
Factorial Contributions
(Cross loadings)
Reliability index:
Rho Dillon- Goldstein
The average variance extracted
(AVE) R&D Input Internal Spending in
R&D 1.0000
Human resources Input
Workforce 1.0000
New design improvements
Easy adaptation to
ingredients (T-1)_2 0.4658 0.8290 0.8858 0.7077
Design time
improvement (T-1)_3 0.1138 0.7513
Cost design
improvement (T-1)_1 0.0604 0.9335
New prototype improvements
Prototype quality
improvement (T-1)_3 0.8892 0.8773 0.9479 0.7518 Prototype taste
improvement (T-1)_2 0.1656 0.9403 Prototype smell
improvement (T-1)_2 0.0652 0.9403 Prototyping time
improvement (T-1)_2 0.0000 0.5805 Prototype aesthetic
improvement (T-1)_2 0.0000 0.9403
New product improvements
Improvement of rejected products number (T-1)_3
0.4000 0.8868 0.8824 0.7291 Improvement of non-
compliant products number (T-1)_3
0.1600 0.8868 Production cost
improvement (T-1)_2 0.0000 0.7839
From the table above, the average variance extracted (Average Variance Extracted” or AVE) also called average community is greater than the variance shared between the built and the other constructs of the model (the squared correlation between two built) [25]. This shows that constructs are discriminated against (see Table 9).
4.3. The Correlation Test between Innovation Resources and the Process Input Indicators
Since we recoded our variables (explanatory and explained) so that they become quantitative, we chose to apply a Pearson correlation test to detect the nature of the relationship between these two variables.
We remind our assumptions:
Hypothesis 1: The inputs in Team leaders are significantly and positively related to the improvement of the new concept and prototype;
Hypothesis 2: The inputs in R&D are significantly and positively related to the improvement of the new concept and prototype;
Hypothesis 3a: The importance given to material investment is positively associated with the improvement of prototype and final product.
We note that a weak improvement in design cost is positively associated but not significantly to the workforce (r = 0.77; p = 0.11).
Nevertheless, this improvement is positively and significantly associated to the spending on IT (r = 0.87; p = 0.05). We conclude that the assumptions
#1 and #2 are enabled for input in personnel and spending on information technology (see Table 10).
Table 10. The relationship between expenditure and human resources inputs and improvement of the new design improvements.
New design improvements Innovation resources
Easy adaptation to ingredients
(T-1)_2
Design time improvement
(T-1)_3
Cost design improvement
(T-1)_1 Internal Spending in R&D −0.5857 −0.2689 −0.3506
What most affects the moderate improvement, are the following inputs:
external spending on research and development (r = 0.87; p = 0.05), spending on equipment (r = 0.87; p = 0.04), spending on advanced technologies (r = 0.87; p = 0.05), and spending on information technology (r
= 0.91; p = 0.02). The average improvement of taste is associated with these inputs positively and significantly (see Table 11).
Table 11. The relationship between expenditure and human resources inputs and objectives of the new prototype improvements.
New prototype improvements
Innovation resources
Prototype quality improve-
ment (T-1)_3
Prototype taste improve-
ment (T-1)_2
Prototype smell improve-
ment (T-1)_2
Prototyping time improve-
ment (T-1)_2
Prototype aesthetic improve-
ment (T-1)_2 Internal Spending in R&D −0.3506 0.4624 0.4624 −0.1605 0.4624
Workforce 0.7799 0.1950 0.1950 0.7164 0.1950
We conclude that the improvement of taste at the new prototype is influenced largely by the first built of research and development factors and the third built of the information technology factor.
From these results, we can confirm the two hypotheses #1 and #2 for inputs in R & D and inputs of equipment and materials included spending on information technology.
We notice that a small improvement in time is associated positively but not significantly enough to the number of engineers in research and development (r = 0.76; p = 0.13), external spending on research and development (r = 0.61; p = 0.27), and spending on advanced technologies (r
= 0.62; p = 0.26) (see Table 12).
Table 12. The relationship between expenditure and human resources inputs and objectives of the new product improvements.
New product improvements Innovation resources
Improvement of rejected products
number (T-1)_3
Improvement of non-compliant products number
(T-1)_3
Production cost improvement
(T-1)_2
Workforce 0.2388 0.2388 0.8969
So, we come back to validate both hypothesis # 1 and # 2 for inputs in R & D personnel and expenditures in R&D.
Here we have the workforce (r = 0.77; p = 0.11), and spending on information technology (r = 0.87; p = 0.05) which play a role in improving low and moderately the non-compliance of the new final product.
After we have external expenditure on research and development (r = 0.87; p = 0.05) spending on equipment (r = 0.87; p = 0.04) spending on advanced technologies (r = 0.87; p = 0.05) and spending on information technology (r = 0.91; p = 0.02), which affect strongly and significantly enough the low and medium improved time of production.
So, we come back to validate the hypothesis 3a for entries in equipment and materials.
Here we have the engineers in research and development (r = 0.80; p
= 0.10) and labor (r = 0,071; p = 0.17) associated positively and significantly enough to the average improvement cost of production.
We also return to validate assumptions #1 and #2 for inputs in R&D personnel and workforce.