Reply on RC2

L20: ‘Local authorities and managers use different indicators’: which ones? Please name some of them for clarity.>> Your comment has been integrated in the abstract, it now reads: ”Local authorities and managers use different indicators, usually modelled using statistical data such as % of vulnerable groups, minorities, per capita income, to quantify rescue and urbanism plans. However, existing risk indicators are often unidimensional and have certain limitations.”


Specific comments
In what concerns the first contact with the study objectives, the reader expects to learn a lot about risk awareness sessions, as the title implies. However, the article does not present sufficient information about these sessions, except for their impact on the willingness of the surveyed people to prepare against natural hazards' risks. This is a bit disappointing, since it would be useful for the analysis and the assessment of outputs to learn eg what kind of sessions were launched in the 2 study areas, under which framework, what was their purpose, when did they take place, were they open to the public? etc…Moreover, following these concerns, is the Title correct when it refers to lessons learned ABOUT awareness sessions? >> Indeed, a study focused on the content of risk awareness sessions, their outputs and impact on the participants would be a useful and significant contribution to improved risk management. That would also fit better to the initial version of the paper's title. However, in this study, the focus was only on the impact of risk awareness sessions that respondents experienced personally, as a mitigation tool to prepare against natural hazards. We analysed the impact of 'risk awareness actions' in general, without a priori knowledge or focus on specific types of actions that have taken place in the study area.

Introduction
In Introduction, the conceptual framework of the study, in which the statistical models are going to be based on, needs to be presented in a concise way. >> In the new version the procedure is now concisely presented in Introduction: 'The fieldwork procedure was both quantitative and qualitative, organized around questionnaires and interviews. In quantitative part of the study, the used regression model was binary logistic which allows dividing variables into more blocks, tracking the model's progress, with dependent variable of investing money. Secondly, the use of interviews in the North Moroccan study served as an additional source of information to be crossed with the responses from the questionnaires.' Also, the Introduction was reorganised, removing the sub-sections and integrating them into the introduction as such, in a more concise way.
L.39: 'that risk': which one? This sounds too general, even though the article is definitive. >> True indeed, no article needed since the discourse is on risk as a general concept. 'They might not share a common understanding of risk.' L54. The phrase 'in order to explore what could contribute to increased precautionary behaviour in a form of future financial readiness to protect of the local, general population' is not well understood. What future financial readiness means should be clarified at this point. >> Modified and simplified: "In this paper, North Morocco and West Sardinia, two case studies from the Mediterranean basin, are compared, in order to explore what could contribute to increased risk awareness and precautionary behaviour of the local, general population." Sub section 1.2 would rather be a Methodology sub section. Usually the study area is described in Methodology.>> Well noted. This section is now moved to the Methodology section.

Methodology
In Methodology, there should be a sub-section about statistical analyses implemented, description of the measures/concepts included in the models, etc. In general, statistical methods must be presented in a separate paragraph, following the appropriate scientific reporting: what kind of methods, and for which purpose each one will be applied; the level of significance for accepting correlation/regression results; measures should be clearly presented, how they were constructed, based on which survey questions, the items used to construct them, if multi-item assessment was applied, such as factor analyses and cronbach's alpha for scaling reliability etc. It must be clearly described which is the dependent variables, which are the independent ones (when regression is applied), which are the control variables (the dummy ones as well). >> Well noted, in the new version an additional paragraph was added under the section 2.3 Quantitative part of the study: Survey content, variables and analysis, as well as an updated Table 1 present the precise questions used for each variable. 'The first set of variables used in the regression model consists of demographic ones: gender (value 1 for women), age (value 1 for those respondents older than the mean age of 36 years old), education (value 1 for those with bachelor's degree or higher), and zone of living (Morocco: Atlantic, Martil, Rif, Sardinia: urban and rural, five dummies for five zones). The second block includes variables related to risk awareness that are considered to describe better future willingness to invest money, as an indicator of precautionary behaviour. Risk awareness was described using questions related to the level of information about natural hazards, i.e. how much the respondents considered themselves informed about natural hazards in their region (each hazard listed and evaluated on the Likert 1-5 scale), rating their level of information about natural hazards (drought, flood, landslide, coastal storm, sea-level rise, heatwave, earthquake, coastal erosion and wildfire), and about climate change, and whether they participated or not in the awareness sessions or environmental education campaigns organized by the municipalities or associations), (e.g., Domingues et al., 2018;Ivčević et al., 2020a). The first study was the North Moroccan when the variables were explored and constructed. Items were submitted to a factor analysis (Kaiser-Meyer-Olkin Measure of Sampling Adequacy = .897; Bartlett's test of sphericity: χ² = 1679, df = 55, p < .001), and the reliability was tested. One factor retained was 'risk informed' (α = .890), and the other 'risk sessions' (α = .381), used in the following analysis. Besides, the dichotomous variable of personal experience of a natural hazard (yes/no) was entered as dummy variable, whose importance in risk perception was showed in similar studies (e.g., Papagiannaki et al., 2019b). Moreover, respondent's belief in the existence of climate change was assessed (yes/no). Furthermore, the respondents were asked if they have previously taken any precautionary measures (yes/no). The final block of variables, related to trust, environmental identity and place attachment, was to test the supplementary variables about the impact on future precautionary behaviour. The place attachment was measured using items such as 'I think this municipality offers a good living environment' and represents a relationship of the resident with her municipality (Bonaiuto et al., 2016). These items were averaged to a single variable with a satisfactory internal consistency (α = .827). The environmental identity was measured using items such as 'I think of myself as someone engaged in environmental issues' and represents environmental concerns of residents (Bertoldo and Castro, 2016), averaged into a single variable with a good consistency (α = .846). The social trust, regarding science and institutions involved in risk management, is of importance especially when respondents cannot manage risks on their own (Achterberg et al., 2017). The items were submitted to a factor analysis (Kaiser-Meyer-Olkin Measure of Sampling Adequacy = .768; Bartlett's test of sphericity: χ2 = 1230, df = 21, p < .001) which split them in two dimensions: trust in science (α = .887) and trust in state (α = .804). When the variables were constructed, the Sardinian data was added later when it was collected. Later, all those items were measured using a 5-point Likert scale, and then dichotomized: participants with responses above mean were considered to have a high value, those under mean to have low value, and are as such used in a binary logistic model, schematically presented in Fig. 4. For more details on variables chosen refer to publications by Ivčević et al. (2020aIvčević et al. ( , 2020b.' The questionnaire could be added as supplementary material instead of just being available upon request. This would help the reader to better understand the conceptual framework and the variables. >> We agree. The questionnaires will now be available as supplementary material. For example, the authors mention: 'Risk awareness was described with two indicators: one related to the level of information about natural hazards (each hazard listed and evaluated on the Likert 1-5 scale), other related to risk sessions (whether the respondents participated or not in the awareness sessions or environmental education campaigns organized by the municipalities or associations)', but the reader cannot understand exactly the questions posed, for which specific hazards, what kind of information was asked to be rated? An appendix table e.g. showing the questions, the expected/constructed variables would be helpful. Furthermore, in what concerns the sessions, the only questions were about participation or were there any other questions about the content of these sessions, when they took place, and who organized them? >> Thank you, we agree with your comment. In the new version the Table 1 was enlarged showing the exact questions posed to get the constructed variables used in the model. In addition, regarding the specification on the content of risk sessions, organized when and by whom, this was not the scope of this study and the text is now added in methodology 'It is to note that the authors of this study did not organise any risk awareness sessions in the studied region, only the previous experience of respondents regarding risk sessions was registered, as well as the information on occurrence, on who organised it and on which topic.' And additionally, under the section 4.3. Future perspectives to clarify: 'Finally, organizing a series of risk awareness sessions in the local community, opened and destined to all stakeholders, would certainly represent an instructive research activity that would contribute to clarifying the existing learning processes between the local stakeholders.'

A graphical presentation of the model would be attractive and helpful to realize the explanatory variables studied.>>
Thank you for this suggestion, a graphical presentation is now included in the manuscript (Figure 4).

Methodology presented in 2.3 subsection is not a well-understood procedure. Concepts such as indicators, risk memory, risk experience, need to be developed here, in the methodology. A table presenting the basic questions, and the interviewed groups would be helpful. As well as what method is applied to assess the answers. Also, I did not understand whether the awareness sessions asked to be rated by the surveyed population were the same discussed during the interviews. >>
Thank you for this important suggestion. The Table is added and the section is rewritten: 2.4 Qualitative part of the study: Interviews and secondary literature In North Morocco, the interviews were based on an interview guide, which consisted of five main parts. The general topic was the influence of the representation that associates, administrators and scientists have on the relationship between citizens and risk; and on the influence of memory and the territorial context in the development of management tools, particularly preventive communication. Firstly, the life in the region and the relationship and attachment the respondent has with the place of living was discussed. Then, the knowledge and history on risk phenomenon in the region were questioned, followed by the personal experience and risk memory. There the interlocutors were asked about the different types of knowledge (scientific and vernacular) that the institutional practices base their risk awareness activities on when talking about risks to the general population. Thirdly, the information, prevention and risk management in the region, as well as the importance of risk awareness sessions in risk management strategies were debated, followed by the consideration of the region within the context of national and international risk policies. Finally, the meaning and use of indicators in risk management were argued. There were twenty-five interviews conducted, and the interviewees were grouped into three: a) the elected or appointed administrators (eight), b) the scientists and technicians that work in the risk research or management (eight), and c) the members of civil society, i.e. associations (nine). All twenty-five narratives were recorded and transcribed and basic discourse analysis was carried out using the software Iramuteq (Ivčević, 2020c). The main parts of the interview guide and the groups interviewed are presented in Table 2. As far as the Sardinian case study is concerned, the secondary literature on risk management at the regional (Sardinia) and national (Italy) level was obtained as public documents free from the administrative web pages. These sources served to contrast the population's responses with the official directions in managing natural risks. Since this literature was easily accessible, the qualitative study in form of interviews with the local authorities was not considered to be essential for this research, conversely to the Moroccan case study.

The use of the word indicator for the parameters/variables/measures that are correlated or regressed is not an appropriate one. Indicators of what? Please consider rewording. Moreover, if a reference on indicators of awareness or preparedness is made, it would rather be part of the discussion, which excuses a less formal writing.>>
Reworded in variables: ' Table 2 presents the correlations of five different areas (binary variable for each area) of the study with main variables.' Table 1 does not explain properly the model variables. Which is the dependent, which are the explanatory variables, the control variables…I suggest results and tables to be reviewed by someone with good statistical knowledge. >> Table 1 was now enlarged with additional information on variables used in the models, and questions posed in the questionnaire on which the variables were constructed. We agree with your comment. A more adapted statistic would be mean comparison (oneway ANOVA): what are the means of the measures in the different areas, and how significant is their difference, and between which groups. Even more precise comparison would be Pearson's chi-square test, adapted for dichotomous variables, that does not rely on the assumption of having continuous normally distributed data. However, based on the additional input in the new version of the paper, we find this correlations part not even relevant for the main discourse. We will fully remove it from the new version of the paper.
Please keep consistency in terminology. Willingness to invest, readiness to invest, invest money, financial precautionary behavior, are all describing the same measure? If yes, please consider using one of them. >> Thank you for this remark. Readiness to invest is now changed to willingness to invest, which is the same form of precautionary behaviour of investing money. 4.3. Again, dynamic indicators are discussed in a informal way. At last, which exactly are the suggested ones? How were they selected and how exactly they relate to this study? E.g. the 'participatory approach' is an issue aroused from the survey or the interviews? Many of the proposed indicators seem to be rather a philological discussion than revealed through this study. E.g. the content of sessions is shown as significant for the level and effectiveness of the obtained risk awareness; however, we have little knowledge about the sessions analysed in the frame of this study, as I mentioned in previous comments, thus we cannot assess the past sessions. >> Thank you for your remark regarding this section. This truly is theoretical reasoning discussed informally. The section is, as such, rewritten, in order to underline that there were no past studies underlined, but the importance of raising risk awareness was discussed in general with the interlocutors. The section is rewritten, starting and finishing with the clarification. "Another risk management recommendation of this study regards dynamic indicators in risk management, where risk awareness sessions particularly contribute. Although the content of past risk awareness sessions was not analysed within this study, risk awareness sessions seem to be considered as useful in preparedness and risk management strategy by the participants of these research activities. As such, they could be imagined as effective risk management tool and their true utility should be tested to confirm the effectiveness of risk awareness sessions in adopting precautionary behaviour."[…] "Finally, organizing a series of risk awareness sessions in the local community, opened and destined to all stakeholders, would certainly represent an instructive research activity that would contribute to clarifying the existing learning processes between the local stakeholders and improving risk management. " Please also note the supplement to this comment: https://nhess.copernicus.org/preprints/nhess-2021-159/nhess-2021-159-AC2-supplement .pdf