Page 168 - Hojnik, Jana. 2017. In Persuit of Eco-innovation. Drivers and Consequences of Eco-innovation at Firm Level. Koper: University of Primorska Press
P. 168
In Pursuit of Eco-innovation
Statistical information of the construct command-and-control in-
strument, pertaining to reliability (reliability coefficients) and conver-
gence (goodness-of-fit model indexes) based on the overall sample (N =
223), is indicated in Figure 8. The construct command-and-control in-
strument showed good reliability (Cronbach’s alpha = 0,946). Also the
goodness-of-fit indexes are shown in Figure 8 (NFI = 0.877; NNFI =
0.636; CFI = 0.879; SRMR = 0.043; RMSEA = 0.52), where we can see
that all the goodness-of-fit indexes showed slightly worse fit, except for
SRMR and RMSEA.
Second, the appropriateness of factor analysis was determined by ex-
amining the correlation matrix of the economic incentive instrument
items. The existence of sufficient correlations (the Bartlett’s test of sphe-
ricity) and the Kaiser-Meyer-Olkin measure of sampling adequacy higher
168 than 0.50 are more critical issues. The Bartlett’s test of sphericity showed
that the correlation matrix has significant correlations (p < 0.05). Fur-
thermore, the Kaiser-Meyer-Olkin measure of sampling adequacy was
examined and indicated similar results; specifically, the KMO value was
0.860, which indicates an excellent sample adequacy. The number of ex-
pected 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, explaining 60.265% of variance. Furthermore, the communal-
ity index showed good communalities (above the threshold of 0.20), ex-
cept for the item “The government provides environmental taxes on en-
ergy, transport, pollution/resources,” which had a communality index of
0.182. We deleted the item that had low communalities after extraction
– below the threshold of 0.20 and conducted the exploratory factor anal-
ysis once again.
This time, the KMO value was a bit lower (0.848), while the Bartlett’s
test of sphericity showed that the correlation matrix has significant cor-
relations (sig. = 0.000 for all items). Moreover, the communality index
showed good communalities (all items after extraction had communali-
ties above the threshold of 0.20; the lowest communality was 0.448) and
one factor was extracted, explaining 67.169% of variance. However, we
decided to remove the other three items that had high correlations with
each other in the correlation matrix (“The government provides green
public procurement”, “The government provides an opportunity to un-
dertake environmental tenders/calls” and “The government provides an
opportunity to undertake environmental projects”).
The third time we conducted an exploratory factor analysis, the val-
ue of KMO was 0.660, and the Bartlett’s test of sphericity showed that
Statistical information of the construct command-and-control in-
strument, pertaining to reliability (reliability coefficients) and conver-
gence (goodness-of-fit model indexes) based on the overall sample (N =
223), is indicated in Figure 8. The construct command-and-control in-
strument showed good reliability (Cronbach’s alpha = 0,946). Also the
goodness-of-fit indexes are shown in Figure 8 (NFI = 0.877; NNFI =
0.636; CFI = 0.879; SRMR = 0.043; RMSEA = 0.52), where we can see
that all the goodness-of-fit indexes showed slightly worse fit, except for
SRMR and RMSEA.
Second, the appropriateness of factor analysis was determined by ex-
amining the correlation matrix of the economic incentive instrument
items. The existence of sufficient correlations (the Bartlett’s test of sphe-
ricity) and the Kaiser-Meyer-Olkin measure of sampling adequacy higher
168 than 0.50 are more critical issues. The Bartlett’s test of sphericity showed
that the correlation matrix has significant correlations (p < 0.05). Fur-
thermore, the Kaiser-Meyer-Olkin measure of sampling adequacy was
examined and indicated similar results; specifically, the KMO value was
0.860, which indicates an excellent sample adequacy. The number of ex-
pected 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, explaining 60.265% of variance. Furthermore, the communal-
ity index showed good communalities (above the threshold of 0.20), ex-
cept for the item “The government provides environmental taxes on en-
ergy, transport, pollution/resources,” which had a communality index of
0.182. We deleted the item that had low communalities after extraction
– below the threshold of 0.20 and conducted the exploratory factor anal-
ysis once again.
This time, the KMO value was a bit lower (0.848), while the Bartlett’s
test of sphericity showed that the correlation matrix has significant cor-
relations (sig. = 0.000 for all items). Moreover, the communality index
showed good communalities (all items after extraction had communali-
ties above the threshold of 0.20; the lowest communality was 0.448) and
one factor was extracted, explaining 67.169% of variance. However, we
decided to remove the other three items that had high correlations with
each other in the correlation matrix (“The government provides green
public procurement”, “The government provides an opportunity to un-
dertake environmental tenders/calls” and “The government provides an
opportunity to undertake environmental projects”).
The third time we conducted an exploratory factor analysis, the val-
ue of KMO was 0.660, and the Bartlett’s test of sphericity showed that
