Page 181 - Hojnik, Jana. 2017. In Persuit of Eco-innovation. Drivers and Consequences of Eco-innovation at Firm Level. Koper: University of Primorska Press
P. 181
Results 181
The number of expected factors was one, and the extracted factor was
one. In addition, the scree plot of the initial run indicated one factor as
an appropriate number. Further, one factor explains 50.254% of variance.
After consideration of each item’s communality index and its contribu-
tion, we removed one item called “The company is using eco-labeling”,
which had communality index below the threshold of 0.20 (0.194).
We then conducted exploratory factor analysis again to see how the
factor characteristics behave with six items to measure the construct of
product eco-innovation. In the second run, we noted in the correlation
matrix that one item –“The company is recovering end-of-life products
and recycling” – has low correlations with other items, ranging between
0.299 and 0.354. However, KMO was 0.846, which is still excellent for
sampling adequacy, and the Bartlett’s test of sphericity showed that the
correlation matrix has significant correlations (sig. = 0.000 for all items).
Moreover, the communalities after extraction were all above the thresh-
old of 0.20. The aforementioned item, “The company is recovering end-
of-life products and recycling”, had the lowest communality (0.283); how-
ever, this value did not imply that it should be removed. Moreover, the
percentage of variance explained has risen. With six items, we could ex-
plain 55.245% of variance.
For the aforementioned reasons, we decided to eliminate the item
“The company is recovering end-of-life products and recycling” and con-
ducted the exploratory factor analysis again. This time, KMO was 0.836,
still demonstrating an excellent sampling adequacy. Moreover, the Bart-
lett’s test of sphericity showed that the correlation matrix has significant
correlations (sig. = 0.000 for all items). After extraction, all the commu-
nalities were above the threshold of 0.20 (the lowest was 0.401), and we
retained all five items. The percentage of variance explained has risen for
approximately 5%. With five items, we are able to explain 60.687% of var-
iance. The reason we retained only four items in the final model to meas-
ure product eco-innovation is explained further in Section 7.3.4. When
conducting an exploratory factor analysis for all three dimensions (prod-
uct, process and organizational eco-innovation), one item (“The company
is using less or non-polluting/toxic materials (i.e., using environmentally
friendly material)”) loaded on both the product and process eco-innova-
tion factors. Moreover, while it loaded a bit higher on process eco-inno-
vation, it loaded on both with a low loading value. Therefore, we exclud-
ed this item to improve the results. In addition, Table 45 indicates the
KMO value and Bartlett’s test of sphericity for product eco-innovation,
including only four items. The lowest extracted communality was 0.355,
The number of expected factors was one, and the extracted factor was
one. In addition, the scree plot of the initial run indicated one factor as
an appropriate number. Further, one factor explains 50.254% of variance.
After consideration of each item’s communality index and its contribu-
tion, we removed one item called “The company is using eco-labeling”,
which had communality index below the threshold of 0.20 (0.194).
We then conducted exploratory factor analysis again to see how the
factor characteristics behave with six items to measure the construct of
product eco-innovation. In the second run, we noted in the correlation
matrix that one item –“The company is recovering end-of-life products
and recycling” – has low correlations with other items, ranging between
0.299 and 0.354. However, KMO was 0.846, which is still excellent for
sampling adequacy, and the Bartlett’s test of sphericity showed that the
correlation matrix has significant correlations (sig. = 0.000 for all items).
Moreover, the communalities after extraction were all above the thresh-
old of 0.20. The aforementioned item, “The company is recovering end-
of-life products and recycling”, had the lowest communality (0.283); how-
ever, this value did not imply that it should be removed. Moreover, the
percentage of variance explained has risen. With six items, we could ex-
plain 55.245% of variance.
For the aforementioned reasons, we decided to eliminate the item
“The company is recovering end-of-life products and recycling” and con-
ducted the exploratory factor analysis again. This time, KMO was 0.836,
still demonstrating an excellent sampling adequacy. Moreover, the Bart-
lett’s test of sphericity showed that the correlation matrix has significant
correlations (sig. = 0.000 for all items). After extraction, all the commu-
nalities were above the threshold of 0.20 (the lowest was 0.401), and we
retained all five items. The percentage of variance explained has risen for
approximately 5%. With five items, we are able to explain 60.687% of var-
iance. The reason we retained only four items in the final model to meas-
ure product eco-innovation is explained further in Section 7.3.4. When
conducting an exploratory factor analysis for all three dimensions (prod-
uct, process and organizational eco-innovation), one item (“The company
is using less or non-polluting/toxic materials (i.e., using environmentally
friendly material)”) loaded on both the product and process eco-innova-
tion factors. Moreover, while it loaded a bit higher on process eco-inno-
vation, it loaded on both with a low loading value. Therefore, we exclud-
ed this item to improve the results. In addition, Table 45 indicates the
KMO value and Bartlett’s test of sphericity for product eco-innovation,
including only four items. The lowest extracted communality was 0.355,
