Social Identity and Risk Perception Explain Participation in the Swiss Youth Climate Strikes

Round 1

Reviewer 1 Report

Dear editor and authors,

I would like to congratulate the authors on their manuscript; it is highly relevant, both societally and scientifically, and well-founded on theory and existing evidence. Also, the article is highly structured and presented in a pleasant writing style. All in all, a pleasure to review; thank you for offering me this opportunity.

In my opinion, the manuscript would benefit from a sharper and more concise presentation of existing theory and evidence in the Introduction section. I feel that some concepts can be defined more elaborately, while others warrant less attention, as they do not seem to be explicitly addressed in the research at hand. 

Please view the following comments as suggestions to improve the presentation of your work.

1. P.2, line 82: Collective action is introduced as an overarching concept. However, the term itself is not defined. This is important, as it is a broad concept that my refer to several, distinct phenomena; e.g., behaviors towards public goods, versus acting in or as a group. In addition, the exact nature of the collective action needs to be clarified, as throughout the manuscript, protest/protesters and stike/strikers are used interchangeably, and without further elaboration or apparent connection to the concept of collective action.
2. Related, in the Introduction, many concepts are introduced but without explicitly mentioning the theories or models they derive from, and I find that to be an omission, as concepts can be part of multiple theories or models, each with more or less different definitions and operationalizations. This sort of culminates on p. 3, line 133, when the authors write 'Some theoretical frameworks (…)'; such vague language can and should be avoided. Please be explicit regarding what (and how many) theories you refer to.
3. Line 89: Again related to my first comment, the authors now use the term 'collective participation (framework)'; is this term interchangeable with collective action? I would suggest consistent use of concepts to increase clarity.
4. Throughout the introduction, and for the first time on p. 3, line 105, the authors discuss the relevance of intergenerational differences, (in)equity etc., sort of culminating in the section on p. 4, starting on line 145, where it says 'This raises the question about whether and how young people differ from adults with respect to climate action (…)' I believe a lot of this discussion is redundant, as this question is not addressed in the present research. I also believe this line of reasoning is not necessary to support the relevance of the research; ultimately weakening the underlying argumentation. 
5. In addition, the age issue remains problematic also for other reasons. On p. 5, line 200, the authors state that 'the current evidence base from the public perception research supports the notion that age plays an important role in various populations with respect to their evaluations of climate change risks and support for mitigating strategies.' However, in earlier sections, notably the paragraph starting on p. 4, line 152, studies from Corner (2015) and Mead (2012) are discussed that point to little or no age differences regarding risk perceptions. Thus, the conclusion by the authors is not sufficiently supported. Again, I feel that this line of reasoning is not necessary to support the relevance of the research. For one, because the climate strikes are evidently done by youth.
6. Reading along, issues with introducing theoretical concepts remain. On p. 4, line 190, (disengagement from) guilt is introduced; however, this concept is not operationalized in the study. I would suggest to presenting a concise theoretical framework, in which all concepts that were actually measured are presented, based on a short description of the theory and its empirical evidence. Concepts that are not measured should not be introduced, ideally.
7. The summarizing section starting at line 194 can be strengthened. The authors state that 'it is still unclear to what degree youth strikers share a social identity that would allow us to adopt established theories from the collective action literature.' However, this is not the main research aim; the actual research aim is to examine the extent to which social identity can explain participation in climate strikes. And again the notion that age plays a role is mentioned, but this is something different from te relevance of examining the views of a particular age group. Please consider formulating a more appropriate research aim, and related research questions, that are actually addressed in the research.
8. In the same section, it is stated that 'we adopt the common core constructs (…)'; however, it remains unclear what these constructs are, and why these are considered to be core constructs. All in all, I suggest presenting a clear and much more integrated theoretical framework, based on a more concise discussion of only the relevant (i.e. measured) constructs.
9. Chapter 2, Materials and methods, is generally very clear. Minor points of improvement: section 2.1 also states the study design, and this could be identified in the header. Second, the term 'survey questionnaire' is more specific and thus appropriate then the term 'survey'. Finally, please identify the statistical software that was used for the analysis. As a side note, I found the statistical analyses well done and clearly explained.
10. P. 7, line 262: Efficacy belief; this should be plural (i.e. beliefs).
11. p. 7, line 271: Social identification. This again is a theoretical points, as it seems that the authors use 'identification' and 'identity' interchangeably. Please choose one of those concepts and provide a clear definition in the Introduction.
12. Then I move on to the Results: Generally clearly presented. Points for improvement: Please omit the repetition of the statistical analyses (e.g., lines 322-325.
13. A short presentation on the demographic factors (as control variables) can also be interesting. Especially age and education, which appear to remain significant predictors, even in the full model?
14. When presenting statements such as 'a person’s level of worry about climate change best predicted 327 participation.', please provide the regression coefficient for that predictor, including its p-value. Consider doing this throughout the Results section.
15. It would also be very informative to discuss the explanatory power of each model, and how the explained variance (r2) increases with every subsequent model (i.e. addition of predictors). In addition, if the full model indeed explains 60% in striking behavior, this is quite high. I would note that explicitly and also discuss this number in the light of comparable studies. 
16. Discussion, p. 11, line 375: "Our findings would suggest 374 that, for climate change, this is not necessarily so." Please provide an alternative explanation for this finding. One explanation is suggested by the authors themselves, in line 402: goals that are in the interests of many could perhaps replace a 'perpetrator out-group'....?
17. Page 12, line 436: 'We used participation in protests as a dependent variable that is 436 ‘predicted’ by other constructs.' This can and should be explicitly linked to the cross-sectional study design.
18. Do the authors believe social desirability played a role?

Author Response

(Please note that we uploaded a more reader friendly version of our responses as a PDF file).

Dear reviewers

We thank you all very much for taking the time to read our manuscript so carefully and for your thoughtful and constructive feedback. We know that for many people pressure from work and private lives have increased during the current COVID-19 pandemic and that it is difficult to meet all our obligations. Against this backdrop, we especially appreciate the effort you put into your reviews.

A general comment that is probably best addressed here is that we needed to update the Pseudo R2s. During the summer, an error has been detected and fixed in the R package DescTools (see https://github.com/AndriSignorell/DescTools/issues/19). This resulted in somewhat smaller explained variance in the models.

 

 

 

Reviewer 1

Dear editor and authors,

I would like to congratulate the authors on their manuscript; it is highly relevant, both societally and scientifically, and well-founded on theory and existing evidence. Also, the article is highly structured and presented in a pleasant writing style. All in all, a pleasure to review; thank you for offering me this opportunity.

Thank you very much for the favourable assessment of our manuscript and your thoughtful and constructive suggestions.

In my opinion, the manuscript would benefit from a sharper and more concise presentation of existing theory and evidence in the Introduction section. I feel that some concepts can be defined more elaborately, while others warrant less attention, as they do not seem to be explicitly addressed in the research at hand.

Thanks for pointing this out and for the more detailed suggestions below.

Please view the following comments as suggestions to improve the presentation of your work.

1. P.2, line 82: Collective action is introduced as an overarching concept. However, the term itself is not defined. This is important, as it is a broad concept that may refer to several, distinct phenomena; e.g., behaviors towards public goods, versus acting in or as a group. In addition, the exact nature of the collective action needs to be clarified, as throughout the manuscript, protest/protesters and strike/strikers are used interchangeably, and without further elaboration or apparent connection to the concept of collective action.

We agree that the concept definitions in the previous version of our manuscript were perhaps a little unclear. We have now introduced more precise concept definitions and explicitly linked them to existing research and theory. For instance, we adopted landmann and Rohmann’s (2020) conceptualization of collective action as ‘action taken together to achieve a common goal’ (line 72–73). Furthermore, we add example actions (signing a petition, participating in a strike) to indicate the relative breadth of our concept definition and clarify that strike participation is one possible form of collective action. We believe this definition suits our research well because it doesn’t require the existence of existing groups and is sufficiently open to accommodate different types of actions.

We agree that the synonymous treatment of protesters and strikers leads to conceptual unclarity. We have therefore changed all instances of protest / protestors to strike / strikers when we speak about the specific events that this research focuses on.

 

1. Related, in the Introduction, many concepts are introduced but without explicitly mentioning the theories or models they derive from, and I find that to be an omission, as concepts can be part of multiple theories or models, each with more or less different definitions and operationalizations. This sort of culminates on p. 3, line 133, when the authors write 'Some theoretical frameworks (…)'; such vague language can and should be avoided. Please be explicit regarding what (and how many) theories you refer to.

Point well taken. We have now made the links to underlying theories more explicit whenever we introduce a new concept. For instance, we now specifically refer to van Zomeren et al.’s (2008) social identity model of collective action (SIMCA) and explain explicitly in what way we extended it.

1. Line 89: Again related to my first comment, the authors now use the term 'collective participation (framework)'; is this term interchangeable with collective action? I would suggest consistent use of concepts to increase clarity.

Thanks for pointing this out. We have added a more explicit definition of how we conceptualize ‘collective action’ (lines 72-73). Moreover, we now use the terms more consistently. For instance, when we talk broadly about research looking at different forms action, we use the term ‘collective action’. When we speak about the specific events on which the present research focuses, we refer to them as ‘strikes’ and the people who participate in them as ‘strikers’.

1. Throughout the introduction, and for the first time on p. 3, line 105, the authors discuss the relevance of intergenerational differences, (in)equity etc., sort of culminating in the section on p. 4, starting on line 145, where it says 'This raises the question about whether and how young people differ from adults with respect to climate action (…)' I believe a lot of this discussion is redundant, as this question is not addressed in the present research. I also believe this line of reasoning is not necessary to support the relevance of the research; ultimately weakening the underlying argumentation.

We agree that the passage on intergenerational differences is not necessary and have therefore deleted it (e.g., what was previously the last paragraph in section 1.1).

However, we strongly believe that some background about how young people think about climate change and steps to reduce it is essential to situate the present research, and that without this a paper that is about youth strikes on climate would be lacking important wider context. We added a passage that more explicitly raises the question of whether the processes identified in the collective action research (SIMCA) and risk research can be observed among younger people (lines 141–146).

1. In addition, the age issue remains problematic also for other reasons. On p. 5, line 200, the authors state that 'the current evidence base from the public perception research supports the notion that age plays an important role in various populations with respect to their evaluations of climate change risks and support for mitigating strategies.' However, in earlier sections, notably the paragraph starting on p. 4, line 152, studies from Corner (2015) and Mead (2012) are discussed that point to little or no age differences regarding risk perceptions. Thus, the conclusion by the authors is not sufficiently supported. Again, I feel that this line of reasoning is not necessary to support the relevance of the research. For one, because the climate strikes are evidently done by youth.

Thank you, we have amended this section as we agree that it was potentially unhelpful. The mentioned references from the risk perception literature (including Corner 2015 and Mead 2012) demonstrate the general importance of demographics in explaining risk perception of specific groups/situations. In the following paragraph 1.3. we have then outlined the more recent evidence (derived from opinion polls) that the currently younger generation is more concerned about climate change than the older population. Therefore, we think that it is crucial to keep the paragraph 1.3. to explain why it is so important to gather evidence from that particular group of young climate strikers to investigate their climate change risk perception and motivation to strike. Overall, while there is mixed research evidence for the relevance of age, we still think it is important to retain some discussion of these matters.

1. Reading along, issues with introducing theoretical concepts remain. On p. 4, line 190, (disengagement from) guilt is introduced; however, this concept is not operationalized in the study. I would suggest to presenting a concise theoretical framework, in which all concepts that were actually measured are presented, based on a short description of the theory and its empirical evidence. Concepts that are not measured should not be introduced, ideally.

We agree with this observation and accordingly deleted the passages about guilt.

1. The summarizing section starting at line 194 can be strengthened. The authors state that 'it is still unclear to what degree youth strikers share a social identity that would allow us to adopt established theories from the collective action literature.' However, this is not the main research aim; the actual research aim is to examine the extent to which social identity can explain participation in climate strikes. And again the notion that age plays a role is mentioned, but this is something different from the relevance of examining the views of a particular age group. Please consider formulating a more appropriate research aim, and related research questions, that are actually addressed in the research.

Thanks for pointing out this confusion. We revised this section and have stated the research goal more precisely:

“The present research aims to close this gap in the literature by testing how well the constructs of an extended version of the social identity model of collective action, which we henceforth refer to as ESIMCA and those from the risk perception framework can explain participation in the youth climate strikes. This enables us to consider factors that derive from disparate but potentially relevant fields across the social science of climate action.”

1. In the same section, it is stated that 'we adopt the common core constructs (…)'; however, it remains unclear what these constructs are, and why these are considered to be core constructs. All in all, I suggest presenting a clear and much more integrated theoretical framework, based on a more concise discussion of only the relevant (i.e. measured) constructs.

Point well taken. We now more explicitly introduce one specific model in more detail (SIMCA), explain how the model used is different from it, and avoid vague wordings such as those mentioned above. The relevant section has been thoroughly revised.

1. Chapter 2, Materials and methods, is generally very clear. Minor points of improvement: section 2.1 also states the study design, and this could be identified in the header. Second, the term 'survey questionnaire' is more specific and thus appropriate then the term 'survey'. Finally, please identify the statistical software that was used for the analysis. As a side note, I found the statistical analyses well done and clearly explained.

Points well taken. We have changed these passages accordingly.

1. P. 7, line 262: Efficacy belief; this should be plural (i.e. beliefs).

Thanks for pointing this out (changed accordingly).

1. p. 7, line 271: Social identification. This again is a theoretical point, as it seems that the authors use 'identification' and 'identity' interchangeably. Please choose one of those concepts and provide a clear definition in the Introduction.

We have added van Zomeren et al.s (2008) definition of ‘social identity’ (‘people’s subjective sense of identification with a group’) to the introduction section and briefly paraphrased it in the method section.

1. Then I move on to the Results: Generally clearly presented. Points for improvement: Please omit the repetition of the statistical analyses (e.g., lines 322-325).

We deleted the redundant paragraph.

1. A short presentation on the demographic factors (as control variables) can also be interesting. Especially age and education, which appear to remain significant predictors, even in the full model?

We have added a paragraph about the sociodemographics at the beginning of the result section.

1. When presenting statements such as 'a person’s level of worry about climate change best predicted participation.', please provide the regression coefficient for that predictor, including its p-value. Consider doing this throughout the Results section.

We added this information to the respective passages.

1. It would also be very informative to discuss the explanatory power of each model, and how the explained variance (r2) increases with every subsequent model (i.e. addition of predictors). In addition, if the full model indeed explains 60% in striking behavior, this is quite high. I would note that explicitly and also discuss this number in the light of comparable studies.

We now explicitly mention R² and how it changes in the result section. Note that the figures changed slightly because of an improved algorithm in the used R package (see general comment at the beginning of this document). Moreover, we have compared the R² of the full model to related research in the discussion section (lines 379–382).

1. Discussion, p. 11, line 375: "Our findings would suggest that, for climate change, this is not necessarily so." Please provide an alternative explanation for this finding. One explanation is suggested by the authors themselves, in line 402: goals that are in the interests of many could perhaps replace a 'perpetrator out-group'....?

Thank you for this suggestion. We added the following sentence to this paragraph:

‘Intead, and somewhat speculatively, it could be argued that in this context collective action is mainly motivated by the shared goal to protect the climate and the wellbeing of future beings.’

1. Page 12, line 436: 'We used participation in protests as a dependent variable that is ‘predicted’ by other constructs.' This can and should be explicitly linked to the cross-sectional study design.

We agree––and had already pointed out in our previous version––that this research is cross-sectional and that caution is warranted when making causal claims or interpretations. In response to comments from several reviewers, we have changed the language throughout the paper and now avoid using the misleading verb ‘predict’. In response to your comment, we now also explicitly mention the study design in the passage indicated above (lines 456–458 ).

 

1. Do the authors believe social desirability played a role?

We don’t believe that social desirability is likely to have affected our findings, not least as the main outcome variable (participation or not in the strikes) reflects self-report of actual participation in these activities. It is possible that some measures (e.g. concern about climate change) could have higher means due to social desirability (that is, people feel it is socially important to ‘seem’ concerned) but we don’t see this as an issue for the analysis undertaken.

Author Response File: Author Response.pdf

Reviewer 2 Report

This article combines the literature on risk perception and social identity related processes to examine what factors predict participation in climate strikes among youth (14-25) in Switzerland. Both risk perception (e.g., worry) and identification with protesters as well as beliefs about the efficacy of youth strikes stand out as important determinants. These results are generally what one would expect based on theory but useful to document as not much research has looked at this, especially not among youth so I think the paper is a welcome contribution.

Some revisions that I think are in order;

 

Literature

1) Generally good and comprehensive but the section on risk perception could use some clarity. For example, many of the variables mentioned can be organized under Van Der Linden's (2015, 2017) Climate Change Risk Perception Model (CCRPM) which combines  cognitive, experiential, and socio-cultural factors and is arguably the most comprehensive risk perception model on this topic (also see Xie et al's replication of van der Linden, already cited). Leiserowitz (2006) is also missing as a seminal risk perception paper, particularly around the role of affect.

 

2) The criticism that past survey work is atheoretical and not specifically designed to assess climate activism needs to be toned down a bit. For example, the Yale Program on Climate Change Communication has done very specific research on climate activism n the US that could be referenced:

https://climatecommunication.yale.edu/publications/climate-activism-beliefs-attitudes-and-behaviors-november-2019/

 

Method

3) The Cronbach's alphas need to be moved into the text in the relevant measures section as usual (I don't see why it needs to be in the supplement).

 

Results

 

4) Comparing predictors for relative importance on betas only is not great (see van der Linden, 2015; Xie et al, 2020 on variance decomposition). This is probably not feasible when using logistic regression due to the pseudo R2. Accordingly, I suggest the authors just report ODDS RATIOS. This is common practice as a measure of effect-size and easier to understand and interpret. I would not object to providing both but odds ratios should be included. Maybe the beta model could be moved to the supplement.

 

5) Hierarchical regression is sensitive to order. Some rationale needs to be provided for the proposed block order (see van der Linden, 2017, oxford encyclopedia entry on risk for arguments about order, e.g., why psychology research should start with socio-demographics, which may be helpful here to bolster the authors' case).

 

6) I find the whole factor analysis and multicollinearity discussion a bit questionable for deciding on the predictors. It's good that the authors are providing the data (appreciated) but I suggest the authors include an alternative hierarchical regression model in the supplement that is not based on the factor analysis groupings for transparency. 

 

7) I would appreciate a few sentences on the striker vs non-striker profiles based on Table 1. Little point in including this table if you're not going to elucidate some of the differences for the reader.

Author Response

(Please note that we uploaded a more reader friendly version of our responses as a PDF file).

Dear reviewers

We thank you all very much for taking the time to read our manuscript so carefully and for your thoughtful and constructive feedback. We know that for many people pressure from work and private lives have increased during the current COVID-19 pandemic and that it is difficult to meet all our obligations. Against this backdrop, we especially appreciate the effort you put into your reviews.

A general comment that is probably best addressed here is that we needed to update the Pseudo R2s. During the summer, an error has been detected and fixed in the R package DescTools (see https://github.com/AndriSignorell/DescTools/issues/19). This resulted in somewhat smaller explained variance in the models.

Reviewer 2

This article combines the literature on risk perception and social identity related processes to examine what factors predict participation in climate strikes among youth (14-25) in Switzerland. Both risk perception (e.g., worry) and identification with protesters as well as beliefs about the efficacy of youth strikes stand out as important determinants. These results are generally what one would expect based on theory but useful to document as not much research has looked at this, especially not among youth so I think the paper is a welcome contribution.

Some revisions that I think are in order;

 Literature

1) Generally good and comprehensive but the section on risk perception could use some clarity. For example, many of the variables mentioned can be organized under Van Der Linden's (2015, 2017) Climate Change Risk Perception Model (CCRPM) which combines  cognitive, experiential, and socio-cultural factors and is arguably the most comprehensive risk perception model on this topic (also see Xie et al's replication of van der Linden, already cited). Leiserowitz (2006) is also missing as a seminal risk perception paper, particularly around the role of affect.

Thank you for the suggestions. We agree that these references strengthen our argument (especially van der Linden 2015) and we have now added these references to paragraph 1.2.

2) The criticism that past survey work is atheoretical and not specifically designed to assess climate activism needs to be toned down a bit. For example, the Yale Program on Climate Change Communication has done very specific research on climate activism n the US that could be referenced:

https://climatecommunication.yale.edu/publications/climate-activism-beliefs-attitudes-and-behaviors-november-2019/

We fully agree, and on reflection the phrasing used was not that helpful. We have changed this section as follows and added the reference to the new Yale survey on activism:

‘Second, while some survey research on climate activism is based on academic research and established theories, the limited amount of research carried out on the youth climate strikes has not yet been able to investigate established theories, over and above consideration of more ad hoc assumptions and intuitions. The lack of research on the theoretical foundations of this new wave of protest constrains the ability to understand why young people do or do not participate in climate strikes.’ (lines 61–66).

 Method

3) The Cronbach's alphas need to be moved into the text in the relevant measures section as usual (I don't see why it needs to be in the supplement).

Done.

Results

4) Comparing predictors for relative importance on betas only is not great (see van der Linden, 2015; Xie et al, 2020 on variance decomposition). This is probably not feasible when using logistic regression due to the pseudo R2. Accordingly, I suggest the authors just report ODDS RATIOS. This is common practice as a measure of effect-size and easier to understand and interpret. I would not object to providing both but odds ratios should be included. Maybe the beta model could be moved to the supplement.

Thank you for this interesting methodical comment. We have tried to use Pratt’s (1987) method described in van der Linden’s (2015) paper but indeed seems not to work, probably because we are here dealing with Pseudo R2, as you rightly point out.

We realize that (unstandardized) odds ratios are more common in logistic regression and that many readers would find their addition to the paper useful. We have therefore added them to the supplementary materials. The reason that we want to keep the standardized coefficients––which are in fact still odds ratios––in the main text is that some variables are very differently scaled (e.g., age vs. worry) and cannot be meaningfully compared without a standardized coefficient.

5) Hierarchical regression is sensitive to order. Some rationale needs to be provided for the proposed block order (see van der Linden, 2017, oxford encyclopedia entry on risk for arguments about order, e.g., why psychology research should start with socio-demographics, which may be helpful here to bolster the authors' case).

Thanks for pointing this out. Your comment made us realize that the description of our analyses and the results wasn’t sufficiently clear. We did, in fact, not use hierarchical models in the sense that blocks of predictors were successively added. Instead, we constructed a series of separate logistic regression equations with four different blocks (socio-demographics, risk perception, social identity, full model) to predict participation in strikes. We revised the description of our analysis to make this clearer (lines 298–310). Among other things, we replaced the misleading term “block” with “model”.

Because we didn’t use hierarchical regressions, we think the discussion in van der Linden 2017 doesn’t need to be included in our manuscript.

6) I find the whole factor analysis and multicollinearity discussion a bit questionable for deciding on the predictors. It's good that the authors are providing the data (appreciated) but I suggest the authors include an alternative hierarchical regression model in the supplement that is not based on the factor analysis groupings for transparency.

We have added such a table to the supplementary materials.

7) I would appreciate a few sentences on the striker vs non-striker profiles based on Table 1. Little point in including this table if you're not going to elucidate some of the differences for the reader.

Thank you for this suggestion. We now discuss the differences between the strikers and non-strikers in section 2.1 and at the beginning of the result section.

Author Response File: Author Response.pdf

Reviewer 3 Report

This paper explores the psychological motivations for youth climate strikers. Youth-led climate strikes around the world are an extremely important development - and this study is one of the first that I am aware to study participant motivation with large-n empirical survey data.

The paper has several strengths. In particular, the paper offers an unusually crisp and clear theory section that appropriately situates the study within relevant academic literatures. It dissects the psychological bases for collective action with precision, and avoids some of the imprecise and often generic concepts that make other studies  difficult to follow. Moreover, it also presents an interesting review of age and generational differences in motivations (section 1.3); this is a literature I was not familiar but the authors introduce it with a lot of clarity.

Empirically, the factor analysis was also competently done and I have no concerns in how its psychological variables were constructed.

However, I also have two intersecting concerns about the papers' empirics and, jointly, they leave me with a much more mixed assessment of the paper.

First, the paper makes - at its core - a causal claim that social identification with other protesters motivates protests. The authors are circumspect at times in their emphasis on non-causal language. But, overall, the paper veers towards causal claims at many junctures, including the first line of the discussion section, and the interpretation of the survey results as speaking to theories that presume a causal relationship between social identification and participation.

The issue here is that this finding is fundamentally endogenous. We have no evidence to discriminate between the world in which social identification predicts participation OR the world where people have more social identification BECAUSE they participate in a protest. I was surprised only to find a casual aside on this issue at the end of the discussion section. Instead, this is a foundational point and it speaks directly to the theoretically interesting result the paper presents.  While the others do present a specification with a few controls, these are minimal and don't cover the universe of confounders we should expect to see in the context of political advocacy. The fragility of some of the other results (including coefficient sign flips) only underscores the deep endogeneity concerns a reader should have.

There are some strategies that the authors could perhaps take to partially correct for this - they could add what additional controls might be available in the survey. They could also try and match on observables to try and reweight control units to facilitate a more apples to apples comparison between strikers and non-strikers. As is, we just don't understand the processes through which respondents entered into one or both categories. Of course, even this can't solve the knotty problem of selection on unobservables.

At the same time, the question is sufficiently interesting that the survey might have offered descriptive statistics about youth climate strikers, an understudied but inherently important population. yet, the sampling strategy for the survey remained opaque to me, either with how the authors recruited their first sampling frame and then with the snowball sampling component which shaped the strikers (but not non-striker?) respondents recruitment. The authors could, for instance, do a much better job of explaining how their sample compares to the broader universe of Swiss youth. I'd have loved a table like Table 1 that compares the sample to this broader universe, perhaps conditional on cohort. It would also be great if there were additional covariates (household size, religion, etc.) that could be assessed against Swiss census data.

The authors are correct that these concerns about representative sampling are not a major concern when measuring internal psychological mechanisms - but they matter absolutely for the ability to leverage the survey to make descriptive claims, and the differentiated selection process into strikers and non-striker sampling processes further confuses things.

This leaves the paper in a bit of a middle no mans land - it does not have a clear ability to make causal claims- but also does not characterize its sample in a way that allows for descriptive statistics that could make an empirical contribution on their own without having to rely on causal argumentation.

 

Author Response

(Please note that we uploaded a more reader friendly version of our responses as a PDF file).

Dear reviewers

We thank you all very much for taking the time to read our manuscript so carefully and for your thoughtful and constructive feedback. We know that for many people pressure from work and private lives have increased during the current COVID-19 pandemic and that it is difficult to meet all our obligations. Against this backdrop, we especially appreciate the effort you put into your reviews.

A general comment that is probably best addressed here is that we needed to update the Pseudo R2s. During the summer, an error has been detected and fixed in the R package DescTools (see https://github.com/AndriSignorell/DescTools/issues/19). This resulted in somewhat smaller explained variance in the models.

 

Reviewer 3

This paper explores the psychological motivations for youth climate strikers. Youth-led climate strikes around the world are an extremely important development - and this study is one of the first that I am aware to study participant motivation with large-n empirical survey data.

The paper has several strengths. In particular, the paper offers an unusually crisp and clear theory section that appropriately situates the study within relevant academic literatures. It dissects the psychological bases for collective action with precision, and avoids some of the imprecise and often generic concepts that make other studies  difficult to follow. Moreover, it also presents an interesting review of age and generational differences in motivations (section 1.3); this is a literature I was not familiar but the authors introduce it with a lot of clarity.

Empirically, the factor analysis was also competently done and I have no concerns in how its psychological variables were constructed.

However, I also have two intersecting concerns about the papers' empirics and, jointly, they leave me with a much more mixed assessment of the paper.

First, the paper makes - at its core - a causal claim that social identification with other protesters motivates protests. The authors are circumspect at times in their emphasis on non-causal language. But, overall, the paper veers towards causal claims at many junctures, including the first line of the discussion section, and the interpretation of the survey results as speaking to theories that presume a causal relationship between social identification and participation.

The issue here is that this finding is fundamentally endogenous. We have no evidence to discriminate between the world in which social identification predicts participation OR the world where people have more social identification BECAUSE they participate in a protest. I was surprised only to find a casual aside on this issue at the end of the discussion section. Instead, this is a foundational point and it speaks directly to the theoretically interesting result the paper presents.  While the others do present a specification with a few controls, these are minimal and don't cover the universe of confounders we should expect to see in the context of political advocacy. The fragility of some of the other results (including coefficient sign flips) only underscores the deep endogeneity concerns a reader should have.

Thank you for highlighting this important issue. We had already tried to be careful about our language. However, your comment shows that we should pay even more attention to this. Accordingly, we have rephrased several passages (incl. abstract, interpretation of the results and beginning of the discussion section) to better reflect the cross-sectional nature of the data. More specifically, we have replaced ‘predictive power’ with ‘explanatory power’.

We do acknowledge the reviewer’s point about cause and effect (i.e. does social identity lead to participation, or the other way around?). They are of course correct that our data cannot of itself answer this question, however, we do have confidence in the analysis carried out given that it is grounded in a well-established theoretical approach, namely the SIMCA. It may of course be that there is a bidirectional relationship between social identity and participation, to which we also now refer. That is, we now explicitly discuss the possibility of reversed or bi-directional causality between social identity and participation (lines 456–465).

There are some strategies that the authors could perhaps take to partially correct for this - they could add what additional controls might be available in the survey. They could also try and match on observables to try and reweight control units to facilitate a more apples to apples comparison between strikers and non-strikers. As is, we just don't understand the processes through which respondents entered into one or both categories. Of course, even this can't solve the knotty problem of selection on unobservables.

Thanks for these suggestions. Unfortunately, the questionnaire did not include any additional variables that we could use as controls. And while it is true that there may well be other variables that influence the likelihood of people participating in the strikes, we feel that the use of two validated models allows us to capture some of the more important explanatory variables. In addition, the use of clearly defined measures which together account for a large proportion of variance gives us confidence that the analysis is sound.

At the same time, the question is sufficiently interesting that the survey might have offered descriptive statistics about youth climate strikers, an understudied but inherently important population. yet, the sampling strategy for the survey remained opaque to me, either with how the authors recruited their first sampling frame and then with the snowball sampling component which shaped the strikers (but not non-striker?) respondents recruitment. The authors could, for instance, do a much better job of explaining how their sample compares to the broader universe of Swiss youth. I'd have loved a table like Table 1 that compares the sample to this broader universe, perhaps conditional on cohort. It would also be great if there were additional covariates (household size, religion, etc.) that could be assessed against Swiss census data.

Thanks for highlighting these potential issues. We now describe the recruitment strategy and its underlying rationale in more detail (lines 195–205). We also discuss the representativity of the sample relative to German-speaking people of the same age in Switzerland (lines 207–216). Unfortunately, we did not assess any additional covariates that could be compared against the census data.

The authors are correct that these concerns about representative sampling are not a major concern when measuring internal psychological mechanisms - but they matter absolutely for the ability to leverage the survey to make descriptive claims, and the differentiated selection process into strikers and non-striker sampling processes further confuses things.

We agree that descriptive statistics about the youth strikes are on their own interesting. However, the present research was not set out to meet this goal; this research is not interested in making descriptive claims; nor are the data suitable to making such claims.

This leaves the paper in a bit of a middle no mans land - it does not have a clear ability to make causal claims- but also does not characterize its sample in a way that allows for descriptive statistics that could make an empirical contribution on their own without having to rely on causal argumentation.

Author Response File: Author Response.pdf

Reviewer 4 Report

This is an original and very well built and structured paper. The paper features a acute and relevant social issue, with a scientific soundness. The scientific rigor of the paper is proven in its method, large sample N=4057, great!!!), techniques and discussion of the results. Just point out two remarks: the first is linked to the title and the second is a relevant reference. It is suggested that in the title it appears the place (Switzerland town) and, maybe the authors include the German canton. There is an article within a Sustainability Special Issue, entitled Global Environmental Policy and Governance in Sustainability, see: https://www.mdpi.com/journal/sustainability/special_issues/global_environmental_policy_and_governance_in_sustainability. There is a research about the youth narrative on climate change [Heejin Han & Sang Wuk Ahn (2020). Youth Mobilization to Stop Global Climate Change: Narratives and Impact, Sustainability 2020, 12, 4127; doi:10.3390/su12104127].

The pair of remarks are invaluable for the rigorous and brilliant research shown in the paper.

Author Response

(Please note that we uploaded a more reader friendly version of our responses as a PDF file).

Dear reviewers

We thank you all very much for taking the time to read our manuscript so carefully and for your thoughtful and constructive feedback. We know that for many people pressure from work and private lives have increased during the current COVID-19 pandemic and that it is difficult to meet all our obligations. Against this backdrop, we especially appreciate the effort you put into your reviews.

A general comment that is probably best addressed here is that we needed to update the Pseudo R2s. During the summer, an error has been detected and fixed in the R package DescTools (see https://github.com/AndriSignorell/DescTools/issues/19). This resulted in somewhat smaller explained variance in the models.

Reviewer 4

 

This is an original and very well built and structured paper. The paper features a acute and relevant social issue, with a scientific soundness. The scientific rigor of the paper is proven in its method, large sample N=4057, great!!!), techniques and discussion of the results. Just point out two remarks: the first is linked to the title and the second is a relevant reference. It is suggested that in the title it appears the place (Switzerland town) and, maybe the authors include the German canton

We have now included Switzerland in the title.

There is an article within a Sustainability Special Issue, entitled Global Environmental Policy and Governance in Sustainability, see: https://www.mdpi.com/journal/sustainability/special_issues/global_environmental_policy_and_governance_in_sustainability. There is a research about the youth narrative on climate change [Heejin Han & Sang Wuk Ahn (2020). Youth Mobilization to Stop Global Climate Change: Narratives and Impact, Sustainability 2020, 12, 4127; doi:10.3390/su12104127].

The pair of remarks are invaluable for the rigorous and brilliant research shown in the paper.

Thank you for suggesting this important article! We have now have included in our article (introduction: lines 42 and 46; discussion: line 391).

Author Response File: Author Response.pdf

Reviewer 5 Report

The article investigates the psychological variables behind the participation of young people in environmental demonstrations. More precisely, it contrasts two streams of literature, namely public risk perceptions and research on collective action, that were previously proposed to explain participation in strikes.

I am grateful to be a reviewer of this paper. I found the topic very interesting and extremely relevant. And I commend the authors for having collected this data with such a delicate sample.

 

My personal evaluation of the paper is positive and I think it is important to publish these results.
Below, you can find my comments. In the first section, I suggest additional analysis that could enrich your paper.
In the second section, I offer food for thought that can be investigated in future research. In other words, I do not expect the comments in the second section to be answered in the revision of the article, but they may, at the discretion of the authors, be discussed in the discussion.

 

1) Additional analysis:
- Have you tried running a cluster analysis? While it is of extreme value to find explanatory variables applicable to the entire population investigated, it is possible that different segments may have different motivations to participate in strikes. Clusters analysis could allow you to highlight these differences and therefore have a more accurate picture of why young people participate in these events. Perhaps the different theories referred reflect the motivations of different segments.
- In addition, I would advise you to analyze the data also in a non-aggregated form. The scales used are very heterogeneous, anger and pride are mixed with other emotions (e.g., happiness, satisfaction [supplementary material]), in addition, emotions and "efficacy belief" are directed to the actions of private individuals, but also politicians. Although the factorial analysis justifies the aggregation of different items, I hypothesize that a finer analysis can offer even richer results, or perhaps highlight differences between segments.

 

I think that the two additional analyses can offer high added value, with little effort.

 

2) Miscellaneous comments:
Regarding the implementation of the study, it would have been extremely interesting to have a continuous measure of participation and not dichotomous.
As mentioned in your article, young people form their own character and are therefore generally more likely to try new things. With a dichotomous response it is not possible to distinguish whether participation was occasional or repeated. In the first case, young people may have participated in one and only one strike to try something new or to show themselves to others. In the second case, this participation shows a deep interest and involvement. Disregarding this difference may have confounded your results.

In the same direction, it would be of extreme interest to investigate the link between participation in strikes and other actions for the environment. Contrasting hypotheses emerge from the literature. It is possible to predict consistency in engagement in pro-environmental behaviors (pro-environmental values literature). However, participation in strikes could also result in less engagement in other pro-environmental actions (moral credits, negative spillover effect literature), especially if young people blame others for the situation.

 

Finally,  I was happy to read that the authors discussed the issue of randomness for their study. The direction of the effects is indeed open to different interpretations and it would be very interesting to test the Hs with an experiment.

Author Response

(Please note that we uploaded a more reader friendly version of our responses as a PDF file).

Dear reviewers

We thank you all very much for taking the time to read our manuscript so carefully and for your thoughtful and constructive feedback. We know that for many people pressure from work and private lives have increased during the current COVID-19 pandemic and that it is difficult to meet all our obligations. Against this backdrop, we especially appreciate the effort you put into your reviews.

A general comment that is probably best addressed here is that we needed to update the Pseudo R2s. During the summer, an error has been detected and fixed in the R package DescTools (see https://github.com/AndriSignorell/DescTools/issues/19). This resulted in somewhat smaller explained variance in the models.

Reviewer 5

 

The article investigates the psychological variables behind the participation of young people in environmental demonstrations. More precisely, it contrasts two streams of literature, namely public risk perceptions and research on collective action, that were previously proposed to explain participation in strikes.

I am grateful to be a reviewer of this paper. I found the topic very interesting and extremely relevant. And I commend the authors for having collected this data with such a delicate sample.

 My personal evaluation of the paper is positive and I think it is important to publish these results.

Below, you can find my comments. In the first section, I suggest additional analysis that could enrich your paper.

In the second section, I offer food for thought that can be investigated in future research. In other words, I do not expect the comments in the second section to be answered in the revision of the article, but they may, at the discretion of the authors, be discussed in the discussion.

 

1) Additional analysis:

- Have you tried running a cluster analysis? While it is of extreme value to find explanatory variables applicable to the entire population investigated, it is possible that different segments may have different motivations to participate in strikes. Clusters analysis could allow you to highlight these differences and therefore have a more accurate picture of why young people participate in these events. Perhaps the different theories referred reflect the motivations of different segments.

Thank you for suggesting this interesting idea. Learning more about possible segments within the sample could indeed advance our understanding of why different groups of people do or do not participate in strikes. However, after careful consideration and discussion, we have agreed that this would go beyond the scope of the present research. We think it should better be addressed in a separate article.

- In addition, I would advise you to analyze the data also in a non-aggregated form. The scales used are very heterogeneous, anger and pride are mixed with other emotions (e.g., happiness, satisfaction [supplementary material]), in addition, emotions and "efficacy belief" are directed to the actions of private individuals, but also politicians. Although the factorial analysis justifies the aggregation of different items, I hypothesize that a finer analysis can offer even richer results, or perhaps highlight differences between segments.

We recognize the potential benefit of analyzing patterns on the level of individual items. As you rightly point out, some constructs were operationalized using a broad range of contents, which could result in different relationships with participation in strikes. To examine this possibility, we have added a table to the Supplementary Materials that reports the bivariate correlations between each item and strike participation. We have also added a passage to the manuscript to make readers aware of this additional analysis. There are some differences but these are relatively small (apart from the most proximal severity item).

I think that the two additional analyses can offer high added value, with little effort.

We thank you very much for these suggestions and we feel truly sorry that we could only implement one of them within the context of this manuscript.

2) Miscellaneous comments:

Regarding the implementation of the study, it would have been extremely interesting to have a continuous measure of participation and not dichotomous.

As mentioned in your article, young people form their own character and are therefore generally more likely to try new things. With a dichotomous response it is not possible to distinguish whether participation was occasional or repeated. In the first case, young people may have participated in one and only one strike to try something new or to show themselves to others. In the second case, this participation shows a deep interest and involvement. Disregarding this difference may have confounded your results.

This is an interesting point. We have actually included something that comes close to what you suggest: We asked participants how often they had participated in strikes with the following four answer options: ‘0’, ‘1’, ‘2’, ‘3’, ‘4 or more times.’ However, we decided to dichotomize this variable into ‘not participated’ and ‘participated.’ The reason for this was mainly that not every participant in our research had the same number of opportunities to participate in strikes because of the time when they took the survey and because the frequency of strikes differed across different geographical regions. To illustrate, those who responded to our survey relatively early had fewer opportunities to strike than those who did so later. And for those who lived in a place where many strikes took place it was relatively easier to strike than for those who lived somewhere with fewer strikes. As a consequence, using a dependent variable with more than two levels would probably have introduced more noise than precision.

In the same direction, it would be of extreme interest to investigate the link between participation in strikes and other actions for the environment. Contrasting hypotheses emerge from the literature. It is possible to predict consistency in engagement in pro-environmental behaviors (pro-environmental values literature). However, participation in strikes could also result in less engagement in other pro-environmental actions (moral credits, negative spillover effect literature), especially if young people blame others for the situation.

We couldn’t agree more. This is a very interesting question and avenue for future research that we are currently thinking about.

Finally,  I was happy to read that the authors discussed the issue of randomness for their study. The direction of the effects is indeed open to different interpretations and it would be very interesting to test the Hs with an experiment.

Thank you!

Author Response File: Author Response.pdf

Round 2

Reviewer 3 Report

The authors have more clearly caveated their conclusions, and now offer a more clear exposition of their methods and sample.