A Way Forward to Integrating Cascading Risk into Local Resilience Strategies in Central Asia
Description
This research developed a way forward to integrate cascading risk into local resilience strategies in Nur-Sultan, Bishkek, and Dushanbe, the respective capitals of Kazakhstan, Kyrgyzstan, and Tajikistan. To do so, it met three core objectives: 1) To examine stakeholder awareness and perceptions of cascading risk and interconnected failures. 2) To evaluate the extent to which cascading risk and interconnected failures are considered in existing strategies for resilience. 3) To understand local capacities and priorities for actions to integrate cascading risk into local resilience strategies. It was hypothesised that if the integration of cascading risk and interconnected failures in existing resilience strategies was insufficient, then the results could outline: 1) Inadequate awareness (consciousness) and understanding (knowledge) of cascading risk amongst key stakeholders. 2) Acknowledgement of the relevance of cascading risk to resilience but limited local capacities to develop and implement strategies. Research followed a mixed methods approach comprising a questionnaire and semi-structured interviews. This dataset includes only the questionnaire data, to preserve anonymity. Data collection was carried out in July 2021 and targeted mixed stakeholders working in fields relating to disaster risk reduction, resilience, infrastructure, urban development, and urban policy in Central Asia. The questionnaire was designed in SurveyMonkey, comprising 20 questions across six pages including an introduction which stated the target audience and completion time (5–10 minutes). Judgment and snowball sampling identified participants. The questionnaire was made available in English and Russian and completed online on a self-administered basis, accessible via two web-links distributed via email and LinkedIn. Question 1 asked what location participants answers best related to, with an option for general/regional if participant knowledge was not specific to one locality. The questionnaire received 115 responses, of which 55 were considered valid based on a completion rate of at least 50 percent. The high number of responses with low completion was considered a result of the questionnaire design and sampling method. Almost half of participants only answered Q1 (location). Subsequent questions were not visible until ‘next’ was clicked, and whilst the introduction stated the target audience, participants contacted through snowball sampling may have deemed their experience not relevant after reviewing questions in the following sections. Sheet 1 'Raw' is the raw data for the 115 responses collected. Sheet 2 'Accepted' includes the 55 responses that were accepted for this research, based on a completion rate of at least 50 percent. Sheet 3 'Accepted & Translated' includes the accepted responses, with the 15 Russian responses translated to English. Sheet 4 'Numerical' details the accepted responses, converted to numerical data where possible.
Files
Steps to reproduce
The questionnaire comprised four sections. Section 1 examined stakeholder awareness and understanding of cascading risks and interconnected failures, and their relevance to the locality, resilience, and climate change. Section 2 assessed the degree to which existing strategies consider cascading risks and interconnected failures, and to which they are coordinated at multiple levels. Section 3 explored stakeholder priorities for action and barriers to integration and was considered central to defining the roadmap. It asked participants to rank five actions and five barriers to integration, informed by the literature review and developed in collaboration with an expert in organisational resilience. Section 4 asked participants for their professional affiliation and experience, gender identity, and level of training. The questionnaire was provided in English and Russian. Questionnaires followed the same format, with slight differences in wording. Q16, which asked for participants’ affiliation, translated to Russian as ‘area of activity’, leading to minor differences in responses. Q19 (gender identity) of the Russian version did not provide an option to self-describe, owing to cultural specificities. Questions primarily followed a quantitative approach, utilising multiple choice, rating scale, and ranking answers, although each section included at least one open question to promote active engagement. Nine answers were anchored to a 0 to 3 Likert-based scale. Descriptive attributes included category labels from 0 (not at all) to 3 (extremely), and capacity/resilience levels, for example, 0 (no identification of CIs) to 3 (CIs and failure chains are identified) (Q7). The omission of a mid-value sought to mitigate moderacy response bias. Participants were given the option ‘don’t know’ if the question did not relate to their expertise. Questionnaire responses were exported to Microsoft Excel, cleaned, and verified for consistency and accuracy. This is what the attached dataset provides. This involved disregarding responses with a completion rate below 50 percent (the validity threshold) and comparing numerical values with actual text values. Next steps to replicate this research: Basic statistical analysis of quantitative answers was conducted in Microsoft Excel. This included calculating the mean (M) and standard deviation (SD) to measure central tendency and variance. Calculations were not skewed by ‘don’t know’ values, but these were reported where significant. To test for statistical difference in rating answers by location, analysis used the Kruskal-Wallis test in R Studio. The null hypothesis was that medians were approximately equal between groups. P-values were considered significant at the 0.05 level. Pairwise post-hoc analysis was performed using Dunn’s test for stochastic dominance; the familywise error rate controlled using Bonferroni adjustment. Qualitative data derived from open-ended questions were analysed thematically.