Page 135 - Hojnik, Jana. 2017. In Persuit of Eco-innovation. Drivers and Consequences of Eco-innovation at Firm Level. Koper: University of Primorska Press
P. 135
Methodology 135
questionnaire, the lack of an established relationship with the companies
and the use of email to gather the results. Many of the emails sent were
not delivered to the recipients, and not all of the companies to which the
questionnaire was sent were dealing with eco-innovations.
The 223 completed questionnaires were further analyzed for missing
data. We followed Hair et al. (2006), who suggest that an observed unit
(in our case, one questionnaire) missing less than 10% of values can be
retained for further analysis and that a separate variable (measurement
item) missing less than 15% of values can be retained as well. Therefore,
the extent and the pattern of missing data were checked.
We first checked for the extent of missing data concerning variables.
The overall amount of missing data was small, totaling 3.947% of miss-
ing values. In more detail, only two variable measurement items in the
questionnaire demonstrated missing data (measurement items had 0.4%
to 1.3% missing values) and therefore no variable measurement items
were removed from the analysis. The percentage of missing data is a bit
higher for firm performance, for which the data were required separate-
ly from the collected questionnaires for all the included companies (sec-
ondary data). With regard to firm performance (profitability indicator
ratios and company growth), two measurement items (growth of net sales
through two business years and ROS) had 0.4% of missing values, one
measurement item (ROE) had 1.3% of missing values and one measure-
ment item had 9.9% of missing values (growth of number of employees
over two business years). We did not remove these items from the analy-
sis. The higher proportion of missing data for these items may be due to
the fact that some companies do not report certain data because they do
not pertain to them (e.g., growth of employees through the last two busi-
ness years does not apply to a sole proprietorship).
The pattern of missing data was also examined. Missing data must al-
ways be addressed if the missing data are in a nonrandom pattern or more
than 10 percent of the data are missing (Hair et al. 2006). Missing data
can be considered random if the pattern of missing data for a variable
does not depend on any other variable in the data set or on the values of
the variable itself (Hair et al. 2006). We checked for a pattern among the
cases (questionnaires/companies) and found that there are only four cases
(companies) with missing values, and the overall amount of missing data
was 1.794%. When we add the firm performance variables (profitability
indicator ratios and company growth), there are 25 cases with a total of
11.21% of missing data. As we have explained previously, companies that
are of the legal form of a sole proprietorship do not report some data, such
questionnaire, the lack of an established relationship with the companies
and the use of email to gather the results. Many of the emails sent were
not delivered to the recipients, and not all of the companies to which the
questionnaire was sent were dealing with eco-innovations.
The 223 completed questionnaires were further analyzed for missing
data. We followed Hair et al. (2006), who suggest that an observed unit
(in our case, one questionnaire) missing less than 10% of values can be
retained for further analysis and that a separate variable (measurement
item) missing less than 15% of values can be retained as well. Therefore,
the extent and the pattern of missing data were checked.
We first checked for the extent of missing data concerning variables.
The overall amount of missing data was small, totaling 3.947% of miss-
ing values. In more detail, only two variable measurement items in the
questionnaire demonstrated missing data (measurement items had 0.4%
to 1.3% missing values) and therefore no variable measurement items
were removed from the analysis. The percentage of missing data is a bit
higher for firm performance, for which the data were required separate-
ly from the collected questionnaires for all the included companies (sec-
ondary data). With regard to firm performance (profitability indicator
ratios and company growth), two measurement items (growth of net sales
through two business years and ROS) had 0.4% of missing values, one
measurement item (ROE) had 1.3% of missing values and one measure-
ment item had 9.9% of missing values (growth of number of employees
over two business years). We did not remove these items from the analy-
sis. The higher proportion of missing data for these items may be due to
the fact that some companies do not report certain data because they do
not pertain to them (e.g., growth of employees through the last two busi-
ness years does not apply to a sole proprietorship).
The pattern of missing data was also examined. Missing data must al-
ways be addressed if the missing data are in a nonrandom pattern or more
than 10 percent of the data are missing (Hair et al. 2006). Missing data
can be considered random if the pattern of missing data for a variable
does not depend on any other variable in the data set or on the values of
the variable itself (Hair et al. 2006). We checked for a pattern among the
cases (questionnaires/companies) and found that there are only four cases
(companies) with missing values, and the overall amount of missing data
was 1.794%. When we add the firm performance variables (profitability
indicator ratios and company growth), there are 25 cases with a total of
11.21% of missing data. As we have explained previously, companies that
are of the legal form of a sole proprietorship do not report some data, such
