parameter and sub-dimension when aggregated to the higher grouping level (Sect.
6.3.1).
We also calculate the scores for each age group as the averages by each resilience
element (Sect. 6.3.2). Subsequently, the average of each element is used to obtain the
resilience score for each sub-dimension. Finally, each sub-dimension is averaged to
obtain the resilience dimension score and an overall resilience score that averages all
three dimensions for each community. Once again, the average of averages is used to
provide the same weight for each of the estimated resilience level.
Lastly, we assess interrater agreement (IRA) for each resilience sub-dimension
score to identify the level of consensus for each group. The IRA explores how each
rater (i.e. FGD participants in this case) assign similar values for each item (Gisev
et al. 2013). Therefore the IRA is a valuable analytical tool that allows for the
quantification of levels of consensus and dissent with regards to a particular resilience element (Burke and Dunlap 2002; O’Neill 2017). To estimate the consensus
and dissent for each resilience element, we use the critical values
5 for the 5-point
scale (Smith-Crowe et al. 2013). The IRA is limited to the resilience element since
higher levels are calculated and not directly derived from the FGD participants.
6.3 Results
6.3.1 Perceived Community Resilience by Study Community
Table 4 contains the overall mean resilience scores across each of the ten communities. Generally, our findings suggest that the perceived community resilience to
floods and droughts is “very low” to “low” across all study communities. The mean
resilience scores in drought-prone communities (1.99) are slightly higher compared
to flood-prone communities (1.96). When stratified by community, Yoggu (droughtprone) and Baluefili (flood-prone) have the highest overall resilience scores, with
both communities recording a mean value of 2.37. On the otherhand, Daboshe
(drought-prone) and Zowayili (flood-prone) have the lowest overall resilience scores
with mean values of 1.64 and 1.69, respectively. However, there is no considerable
variation in the resilience scores across the ten communities. In terms of resilience
dimensions, the perception of ecological resilience is higher (drought-prone ¼ 2.10;
flood-prone ¼ 2.07) compared to socio-economic resilience (drought-prone ¼ 2.07;
flood-prone ¼ 1.950) and engineering resilience (drought-prone ¼ 1.83; floodprone ¼ 1.86) (Table 4). These results strongly point to the fact that respondents
rate their community resilience to floods and droughts as rather weak. One reason
5 Critical values are assigned depending on the distribution skewness, with slight skew at 0.69,
moderate skew at 0.49, and heavy skew at 0.42. Level of skewness are measured with absolute
values ranging between 0–0.5, as slight skew; 0.5–1 as moderate skew; and > 1 as heavy skew
(Bulmer 1979).
204
Y. A. Boafo et al.
6.3.1).
We also calculate the scores for each age group as the averages by each resilience
element (Sect. 6.3.2). Subsequently, the average of each element is used to obtain the
resilience score for each sub-dimension. Finally, each sub-dimension is averaged to
obtain the resilience dimension score and an overall resilience score that averages all
three dimensions for each community. Once again, the average of averages is used to
provide the same weight for each of the estimated resilience level.
Lastly, we assess interrater agreement (IRA) for each resilience sub-dimension
score to identify the level of consensus for each group. The IRA explores how each
rater (i.e. FGD participants in this case) assign similar values for each item (Gisev
et al. 2013). Therefore the IRA is a valuable analytical tool that allows for the
quantification of levels of consensus and dissent with regards to a particular resilience element (Burke and Dunlap 2002; O’Neill 2017). To estimate the consensus
and dissent for each resilience element, we use the critical values
5 for the 5-point
scale (Smith-Crowe et al. 2013). The IRA is limited to the resilience element since
higher levels are calculated and not directly derived from the FGD participants.
6.3 Results
6.3.1 Perceived Community Resilience by Study Community
Table 4 contains the overall mean resilience scores across each of the ten communities. Generally, our findings suggest that the perceived community resilience to
floods and droughts is “very low” to “low” across all study communities. The mean
resilience scores in drought-prone communities (1.99) are slightly higher compared
to flood-prone communities (1.96). When stratified by community, Yoggu (droughtprone) and Baluefili (flood-prone) have the highest overall resilience scores, with
both communities recording a mean value of 2.37. On the otherhand, Daboshe
(drought-prone) and Zowayili (flood-prone) have the lowest overall resilience scores
with mean values of 1.64 and 1.69, respectively. However, there is no considerable
variation in the resilience scores across the ten communities. In terms of resilience
dimensions, the perception of ecological resilience is higher (drought-prone ¼ 2.10;
flood-prone ¼ 2.07) compared to socio-economic resilience (drought-prone ¼ 2.07;
flood-prone ¼ 1.950) and engineering resilience (drought-prone ¼ 1.83; floodprone ¼ 1.86) (Table 4). These results strongly point to the fact that respondents
rate their community resilience to floods and droughts as rather weak. One reason
5 Critical values are assigned depending on the distribution skewness, with slight skew at 0.69,
moderate skew at 0.49, and heavy skew at 0.42. Level of skewness are measured with absolute
values ranging between 0–0.5, as slight skew; 0.5–1 as moderate skew; and > 1 as heavy skew
(Bulmer 1979).
204
Y. A. Boafo et al.