of the two variables are highly skewed, and therefore mean is not a relevant
description of the entire data, but since the median is 5 in both cases, a higher
mean for buying clothes is an indication to higher scores compared to buying food.
10.4.1 The Sustainable Food Consumption (SFC) Scale
The internal consistency of the initial sustainable food consumption (SFC) scale is
rather low, with a Cronbach’s alpha of 0.53 (descriptive statistics are presented in
Appendix Table 1). Some of the variables are highly skewed, for example, C1, C2,
or C9, making their inclusion in a factor analysis not only irrelevant but also a source
of estimation bias. C1 and C2 describe consumption habits and are highly concentrated toward the upper side of the scale, providing an indication that the participants
in the study eat meat and dairy products on a regular basis. By contrast, C9 is
concentrated toward the lowest values of the scale and signals that the majority of
students do not buy products right before their expiration date, even if, under normal
circumstances, we would consider this an efficient approach. Figure (10.1) presents
the correlations among the variables involved in SFC and clearly shows that many
items need to be dropped in an attempt to find a factorial structure of the scale.
The new scale consists of seven items (Table 10.1), with an acceptable level of
internal consistency. The Bartlett’s test of sphericity confirms that looking for
significant correlations within variables makes sense, while the decent value of the
Kaiser–Meyer–Olkin statistic (0.73) indicates that factor analysis is appropriate to
analyze our data. Dropping any of the remained items cannot further increase the
internal consistency of this reduced scale.
It is the intuition provided by the visual presentation of the scale (Fig. 10.1) that
suggests that the remaining scale is not only one-dimensional but also a parallel
analysis. Figure (10.2) shows that only one factor has an eigenvalue higher than
1, with the loadings of each variable in this factor presented in Table (10.3).
The poor consistency of the scale becomes evident once again in Table (10.2); the
items which remained in the reduced SFC cover a broad range of concerns, capturing
many aspects that normally would be assigned to different dimensions. One possible
explanation is the low level of preoccupation regarding the sustainability of food
consumption, making students behave in relation to the plain need for food, and not
in response to a specific and more profound motivation in choosing food.
Another possible explanation that goes hand in hand with the previous one is that
some of the answers are in line with socially desirable behaviors or clichés, for
example, keeping a healthy diet, buying fare trade products, or supporting local
farmers, and not with students’ awareness regarding the issue as such. A good
argument in line with our assumption is that SFC 7 and SFC 13, which would signal
a type of awareness regarding the sustainable dimension of food consumption, have
loadings below 0.4, a cutoff preferred in many research papers. Given our large
sample size, we accepted these items as relevant for the scale, but still their low
loadings raise a question mark regarding the algorithm (rationality, awareness versus
164
R. Ianole-Călin et al.
description of the entire data, but since the median is 5 in both cases, a higher
mean for buying clothes is an indication to higher scores compared to buying food.
10.4.1 The Sustainable Food Consumption (SFC) Scale
The internal consistency of the initial sustainable food consumption (SFC) scale is
rather low, with a Cronbach’s alpha of 0.53 (descriptive statistics are presented in
Appendix Table 1). Some of the variables are highly skewed, for example, C1, C2,
or C9, making their inclusion in a factor analysis not only irrelevant but also a source
of estimation bias. C1 and C2 describe consumption habits and are highly concentrated toward the upper side of the scale, providing an indication that the participants
in the study eat meat and dairy products on a regular basis. By contrast, C9 is
concentrated toward the lowest values of the scale and signals that the majority of
students do not buy products right before their expiration date, even if, under normal
circumstances, we would consider this an efficient approach. Figure (10.1) presents
the correlations among the variables involved in SFC and clearly shows that many
items need to be dropped in an attempt to find a factorial structure of the scale.
The new scale consists of seven items (Table 10.1), with an acceptable level of
internal consistency. The Bartlett’s test of sphericity confirms that looking for
significant correlations within variables makes sense, while the decent value of the
Kaiser–Meyer–Olkin statistic (0.73) indicates that factor analysis is appropriate to
analyze our data. Dropping any of the remained items cannot further increase the
internal consistency of this reduced scale.
It is the intuition provided by the visual presentation of the scale (Fig. 10.1) that
suggests that the remaining scale is not only one-dimensional but also a parallel
analysis. Figure (10.2) shows that only one factor has an eigenvalue higher than
1, with the loadings of each variable in this factor presented in Table (10.3).
The poor consistency of the scale becomes evident once again in Table (10.2); the
items which remained in the reduced SFC cover a broad range of concerns, capturing
many aspects that normally would be assigned to different dimensions. One possible
explanation is the low level of preoccupation regarding the sustainability of food
consumption, making students behave in relation to the plain need for food, and not
in response to a specific and more profound motivation in choosing food.
Another possible explanation that goes hand in hand with the previous one is that
some of the answers are in line with socially desirable behaviors or clichés, for
example, keeping a healthy diet, buying fare trade products, or supporting local
farmers, and not with students’ awareness regarding the issue as such. A good
argument in line with our assumption is that SFC 7 and SFC 13, which would signal
a type of awareness regarding the sustainable dimension of food consumption, have
loadings below 0.4, a cutoff preferred in many research papers. Given our large
sample size, we accepted these items as relevant for the scale, but still their low
loadings raise a question mark regarding the algorithm (rationality, awareness versus
164
R. Ianole-Călin et al.