the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 09 Nov 2022
Geoscientist’s views about science communication: predicting willingness to communicate geoscience
Abstract. The main barriers to science communication are common in different fields and they are widely identified in the literature. Studies focused on specific scientific communities framed science communication as an activity with the specificities of each context and field. In this study, we analysed geoscientists' representations and attitudes about communication to understand which factors can have significant impact on prediction of public engagement and that can explain the frequency/intensity of communication. The results pointed that factors such as professional experience, recognition by the institution, lack of financial support, personal satisfaction and geoscientific as an area of expertise, have a significant effect on their public engagement.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(938 KB)
-
Supplement
(1463 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(938 KB) - Metadata XML
-
Supplement
(1463 KB) - BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2022-1160', Rolf Hut, 22 Dec 2022
The authors present a study into the motivation of geoscientists to engage in science communication efforts, which is a topic of great interest to the readership of GC. However, in its current form the manuscript (and possibly the analyses) needs major revisions to fully inform other scientists on the science that the authors have done and its implications.
On the survey designThe authors have done a survey among geoscientists in Portugal and have provided the list of questions of their survey. The survey respondents are asked to report on how many ‘science communication activities’ they have carried out, but no definition of what is considered a science communication activity is presented, or at least: not to us in the article, or in the supplementary material. The second question does provide a list of potential activities to clarify what could be considered science communication, however in Q11 interaction with media is not listed as a type of science communication, whereas in Q43 it is. Since for many (notably the more senior scientists) the main ‘science communication’ they do is answering journalists' questions when asked, but not pro-active activities, this unclarity leads to a problem when interpreting the results of the survey. If a part of the survey responds did, and another part did not, consider answering questions from journalist to be part of ‘science communication’ than this would mean that the answer to “how many science communication activities did you do” (Q12), the main question that is used as target variable in the results, would differ between these groups. If the authors did communicate to the survey respondents what the definition of ‘science communication activities’ was that they used, then they have to report that in the article. If they did not communicate this to the survey respondents they either have to take this into account when interpreting their results, or they have to redo the survey. The authors do hint at this a little bit in line 318 where they mention that more senior scientists ‘may receive more demands’, implicating that the authors use a definition of ‘science communication activities’ that does include being interviewed, ie. not pro-active activities.
On the methodologyThe authors use fairly standard methods to analyze their data (which is good). However, a major explanation missing from the manuscript is how they determined their factors. Did they decide on the factors a priori based on the background (literature) presented in section 2? If so: how are the factors constructed (mathematically) from the answers to the questions? If they got the factors from a factor analyses of the data, please provide the results of that factor analyses. (I’m guessing it is an a-priori factorisation, which is perfectly fine, but needs to be explained). Furthermore, the authors divided the results of Q12 up in three classes: “very active, active and inactive”. They provide the boundaries for these classes in line 268, but they fail to justify why those boundaries where chosen as they were, nor did they present a sensitivity analysis on this choice
On the interpretation of the resultsThe authors present “area of expertise” as a significant influence on the amount of science communication activities done. However: “Geological and Energy Resources” were not listed in section 3.1 (lines 193 - 197) This leads to wonder: is that a category with very few people in it? If so: how does that affect the results? If one of the 4% of respondents that identified as ‘professional science communicators’ happened to be in that field, it would explain those results. The covariance between those categories can not be ignored.
In line 195 the authors state that ‘the sample slightly overrepresented scientists with a geology degree’, but in line 181 they state their is no data on the number of geoscientists in Portugal so it is unclear to me where they base their conclusion of overrepresentation on, given that the total number (ie, the frame) is not known.
Lines 211 through 243 provide a text based overview of the raw response per question. This should be presented in some visual way, potentially in the appendix. For questions where high correlation between answers are important to understand the results, some overview of the correlation / cross responses should be presented, for example between ‘area of expertise’ and ‘professional category ‘ (Q5 and Q6). Especially the responses to the survey questions that together contribute the factors used in answering RS1 should be presented.
Finally, a main Achilles heel not touched upon by the authors is that even in anonymized survey responses people are known to give socially desirable answers. Combined with the lack of a clear definition of science communication activities, this could easily lead to an overestimation of the amount of science communication reported by the survey respondents. The authors should reflect on the impact this has on their results.
On Open Science best practicesThe Copernicus publication guidelines on data sharing state that both data used to obtain the results and software used in the analyses should be shared alongside a publication. While those guidelines are written with the more classic earth science fields in mind, this does not exclude social sciences methods like surveys. The raw results of the survey might not be share-able due to legal reasons (GDPR), but there are good best practices available to anonymize survey responses and share them together with the article. See for example https://dmeg.cessda.eu/Data-Management-Expert-Guide/5.-Protect/Anonymisation
Sharing the data with the research allows other scientists, including the reviewers, to look for (other) patterns in the data collected and to check the analyses done by the authors. Without this data, the current submission violates the Copernicus guides on Open Data and should not be accepted.
ConcludingReading back this review I recognize that this might feel overly negative to the authors and I want to assure that my review is written with the purpose of making the manuscript better and getting this valuable data and analyses published. The insights that can be gotten from the data collected by the authors is great and should be shared with the scientific community. I hope that the comments above can be taken into account on the road to making this a great publication.
Rolf Hut
Citation: https://doi.org/10.5194/egusphere-2022-1160-RC1 -
AC1: 'Reply on RC1', Joana Rodrigues, 09 Jan 2023
We sincerely thank Referee Rolf Hut for the valuable suggestions and constructive comments, which we found extremely useful to improve our manuscript.
- On the survey design
Undoubtedly, we consider interaction with journalists as ‘science communication activities’ and the survey includes specific questions about journalists and media, such as Q43, Q17, Q37.11 (not analysed under the scope of the present study, with the exception of Q37.11). In Q11 were listed activities carried out directly with the ‘broad public’ and nonspecific for particular targeted publics, such as the media. Furthermore in Q13 the respondents reported the groups with whom they have engaged, where journalists were included. We will include in the appendix table the analysis of this question assessing the frequency of contact with journalists (Q13.1, Q13.2), where it is possible to conclude that it is very small. Participants were not given any definition of ‘science communication activities’ in the survey. We did not intend to condition each personal notion/perception of this activities. Each respondent reported the activities they assume as such and which they consciously carry out as ‘science communication’.
We agree that it may be relevant to add this circumstance in the text in section 4.2, before line 270:
To access this variable participants were asked how many science communication activities had they carried out in the previous year. As no prior detailed explanation of 'science communication activities' was given, each respondent reported the activities they assume as such and which they consciously carry out as science communication.- On the methodology
It was not conducted a sensitivity analysis in this study. We agree that it is important to mention in the text, on section 3, that when designing the questionnaire we did a-priori factorisation and the choice was made from a strategic point of view, considering the literature review, based on previous questionnaires such as Hartz and Chappell (1997), Wellcome Trust (2000), Royal Society (2006) and Liverman & Jaramillo (2011).
- On the interpretation of the results
In section 3.1 (lines 193 - 197) are listed the geoscientists’ degree area (Q4): Geology; Biology and Geology / Environmental Sciences / Environmental Education; Geological Engineering / Mine Engineering; Biology; Geophysics/ Meteorology / Oceanography / Physics and Geography / Aerospace Engineering.
The early training (Bachelor's) area does not always correspond to the area of work and research, therefore area of expertise was introduced as a different indicator (Q5): Geological and Energy Resources; Internal Geodynamics, Geophysics, Petrology and Geochemistry; External Geodynamics and Palaeontology; Geoconservation and Geotourism; History and Education; Environment, Environmental Geology and Engineering Geology. This detailed information will be included, in supplementary material, in the summary of the results of the demographic and descriptive analysis.There is no data on the number of geoscientists in Portugal, but considering that the geosciences include several areas, besides geology, and that 70% of the respondents are geologists, we believe that the proportion of the other geoscientists is over represented.
Geology (125 respondents), Biology and Geology/Environmental Sciences/Environmental Education (28 respondents), Geological Engineering/Mine Engineering (8 respondents), Biology (8 respondents), Geophysics/Meteorology/Oceanography/Physics (6 respondents), Geography / Aerospace Engineering (4 respondents). This detailed information will be included, in supplementary material, in the summary of the results of the demographic and descriptive analysis.We agree that a table that compiles the results of the descriptive analysis (lines 211 – 243), provides an easier and better overview, clarifying for example the two previous discussed topics. We will submit it as supplementary material.
We also agree that for questions where high correlation between answers are important to understand it is relevant to include an overview of these correlations. As it is a significant amount of information, we will submit it as supplementary material.
We are aware that a social desirability bias always exists in these studies and that is also why we mention in the conclusions that more robust results may be obtained through further studies. We agree that social desirability bias should be addressed in the text, around line 361:
In surveys like this where people are asked to provide personal opinions and report personal practices it is necessary to take into account that people may be led to answer according socially desirable ideas. To try to reduce this bias and obtain more truthful answers, participants were informed that their identities were protected and the results would be appropriately anonymized. At the same time, online self-administered questionnaires, without the presence of other people, as this case, also contribute to reduce social desirability bias.- On Open Science best practices
We will include in section 3, together with the methodology, the information regarding the software used for statistical analysis. Under this research project we are currently still preparing the publication of other studies with all these unpublished data as also the submission of the full data collected with this survey to a specialised journal for open research data. For this reason, we consider to submit only the variables analysed in the scope of this study.
Once again, we thank Referee Rolf Hut for the constructive insights. We hope that our answers have clarified all the comments.
The authors
Citation: https://doi.org/10.5194/egusphere-2022-1160-AC1 -
AC2: 'Reply on RC1', Joana Rodrigues, 09 Jan 2023
Publisher’s note: this comment is a copy of AC1 and its content was therefore removed.
Citation: https://doi.org/10.5194/egusphere-2022-1160-AC2
-
AC1: 'Reply on RC1', Joana Rodrigues, 09 Jan 2023
-
RC2: 'Comment on egusphere-2022-1160', Anonymous Referee #2, 02 Jan 2023
This study surveys Portugal geoscience majors and experts to explore potential predictors that have significant influences on geoscience communication activities, which is of great interest to GC. The most valuable parts of the manuscript are the detailed questionnaires about geoscience communication and the summary of potential predictors in past studies. However, the current manuscript has several severe problems in basic experiment design, and quantitative methods and lacks some critical information to evaluate the validity of analyses. A major revision is needed.
Major comments:- The total sample size is good (n=179), but the study has very weak control over the survey participants. There are at least six different groups with very different expertise and background, including geoscience students, professors, industry experts, researchers, and communicators. The sample size for each group is relatively small and probably cannot represent their populations (the authors also mentioned this in the paper at Ln186). The authors may want to give more statistical details about the individual groups. Different groups probably have very different opinions about geoscience communication, even the definition of geoscience communication. Thus analyzing their survey results separately will be more meaningful in my mind.
- The study used Chi-square tests to understand how different factors may affect geoscience communication activities (Research Question 1) and reported the chi-square values and p-values, but there is no information about the frequency expected. Especially where this expected frequency comes from. If it is from the literature review, then the authors should also explain how they deal with the different data collection standards, question design, and analysis of individual factors. Without this information, it is hard to evaluate the soundness of the test interpretation.
- The study applied multinomial logistic regression to evaluate the relationship between the frequency of geoscience communication activities and selected factors (e.g., professional experience), which is problematic. The frequency or intensity of the activity (from none to inactive to very active) is NOT an unordered categorical variable. The dependent variable of multinomial logistic regression has to be unordered nominals. The frequency of the activities is apparently ordinal (0, 1, 3-4, >4 in the last year). Thus, the authors may want to consider other models like the structure equation model.
Minor Comments:
The manuscript is hard to read, especially the introduction part.
The definition of public engagement should be defined in this paper because, in my mind, and I think for many researchers, public engagement and science communication are different in definition.
The title is too general and should indicate that the data is only from Portugal.
Geoscience communication is a very general word, and the authors give a wide range of activities in the survey items, but activities, such as speeches at geoscience conferences, are not typically considered as geoscience communication for public engagement purposes. The authors should give a more precise definition or scope of Geoscience communication in this study.
Overall, I think the study has potential, and I look forward to reading the revised version.
Citation: https://doi.org/10.5194/egusphere-2022-1160-RC2 -
AC3: 'Reply on RC2', Joana Rodrigues, 09 Jan 2023
We thank Referee 2 for the constructive and helpful comments, which were very insightful to improve our manuscript.
Major comments:
1. We are aware that the samples of the different professional groups are relatively small and may not represent accurately their populations. Because of that we haven’t processed their results separately. As the aim of the current work is to study the population in general, this question was included precisely to show that it is not a homogeneous population and that geoscience communication practitioners are not only researchers and academics.
Suggested by the Reviewer 1, we will include as supplementary material an overview through the results and correlations that will also present the professional groups distribution.
2. The expected frequency was calculated based on the assumption of independence between the two categorical variables being analysed and the chi-square test of independence was used to determine if there is a significant association between the two variables.
The analysis of individual factors was done by performing several independent chi-square tests to examine the relationship between the response variable and each predictor variable. This is a common approach for understanding how different factors may affect the response, as the reviewer well points out. However, as is well known, this approach has several limitations such as leading to an increased risk of Type I errors (i.e., false positives), because each test is evaluated separately, and the probability of making a Type I error increases with the number of tests performed. Furthermore, this approach does not allow for the analysis of multiple predictor variables simultaneously, so it is difficult to determine the relative importance of each predictor variable or to understand the combined effect of multiple predictor variables on the response. To deal with these well-known limitations, we decided to also use a multinomial logistic regression model, a method that seems to us to be more appropriate, as it allows the analysis of multiple predictor variables simultaneously and provides estimates of the relative odds of each level of the response variable occurring. This can help to better understand the relationship between the predictor variables and the response variable, and to identify the factors that are most influential in determining the response.
3. Both Structure Equation Model, SEM, and ordinal regression are statistical techniques that can be used to analyse relationships between predictor variables and an ordinal response variable. SEM is a more comprehensive model that allows for the analysis of multiple mediator and moderator variables, but this is not exactly the purpose of this study. On the other hand, ordinal regression models are primarily concerned with estimating the predictor and response variables, which is the aim of this study.
However, we chose to use a multinomial logistic regression model which has the advantage of having been specifically designed for analyzing relationships between predictor variable and a categorical response variable with more that two levels. We think it is a good choice where the research question involves examining the influence of predictor variables on the likelihood of different levels of the response variable occurring.
Another advantage of multinomial logistic regression is the simplicity of interpretation. It involves fitting separate logistic regression models for each level of the response variable, and the coefficients from these models can be used to estimate the relative probabilities of each level of the response occurring.
There are other reasons that led us to choose this procedure:
1) the goal is to identify the predictors of membership in the different categories of the response variable;
2) the categories of the response variable are not equally spaced, and ordinal regression assumes that the distance between levels of the response variable is equal;
3) the sample size is small, so a multinomial logistic regression model may provide more stable estimators.
Minor Comments:
- We will revise the introduction part to make it more readable.
- Indeed, public engagement and science communication are not synonymous. Science communication refers to a broad range of activities, but given that, in the current paradigm of science communication, scientists communicating are mostly required to promote public engagement, these terms are normally used interchangeably, even in the specific literature regarding science communication. But we agree that this may not be understood as such by everyone so we consider it to be pertinent to clarify these concepts, and we will do so in the section 1 Introduction
- The studies on this specific topic are very few, almost non-existent among international literature, with one exception made more than 10 years ago (Liverman & Jaramillo, 2011). At the same time, the main outcomes of the research confirm the patterns of scientists from other areas previously studied. We think that a more general title may boost more opportunities for further researches to develop strategies to suppress geoscience communication constraints worldwide.
- We agree that it is pertinent to include in the introduction a more precise definition about geoscience communication and its scope within this study.
Once again, we thank Referee 2 for the relevant questions raised which we hope we have fully answered.
The authors
Citation: https://doi.org/10.5194/egusphere-2022-1160-AC3
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2022-1160', Rolf Hut, 22 Dec 2022
The authors present a study into the motivation of geoscientists to engage in science communication efforts, which is a topic of great interest to the readership of GC. However, in its current form the manuscript (and possibly the analyses) needs major revisions to fully inform other scientists on the science that the authors have done and its implications.
On the survey designThe authors have done a survey among geoscientists in Portugal and have provided the list of questions of their survey. The survey respondents are asked to report on how many ‘science communication activities’ they have carried out, but no definition of what is considered a science communication activity is presented, or at least: not to us in the article, or in the supplementary material. The second question does provide a list of potential activities to clarify what could be considered science communication, however in Q11 interaction with media is not listed as a type of science communication, whereas in Q43 it is. Since for many (notably the more senior scientists) the main ‘science communication’ they do is answering journalists' questions when asked, but not pro-active activities, this unclarity leads to a problem when interpreting the results of the survey. If a part of the survey responds did, and another part did not, consider answering questions from journalist to be part of ‘science communication’ than this would mean that the answer to “how many science communication activities did you do” (Q12), the main question that is used as target variable in the results, would differ between these groups. If the authors did communicate to the survey respondents what the definition of ‘science communication activities’ was that they used, then they have to report that in the article. If they did not communicate this to the survey respondents they either have to take this into account when interpreting their results, or they have to redo the survey. The authors do hint at this a little bit in line 318 where they mention that more senior scientists ‘may receive more demands’, implicating that the authors use a definition of ‘science communication activities’ that does include being interviewed, ie. not pro-active activities.
On the methodologyThe authors use fairly standard methods to analyze their data (which is good). However, a major explanation missing from the manuscript is how they determined their factors. Did they decide on the factors a priori based on the background (literature) presented in section 2? If so: how are the factors constructed (mathematically) from the answers to the questions? If they got the factors from a factor analyses of the data, please provide the results of that factor analyses. (I’m guessing it is an a-priori factorisation, which is perfectly fine, but needs to be explained). Furthermore, the authors divided the results of Q12 up in three classes: “very active, active and inactive”. They provide the boundaries for these classes in line 268, but they fail to justify why those boundaries where chosen as they were, nor did they present a sensitivity analysis on this choice
On the interpretation of the resultsThe authors present “area of expertise” as a significant influence on the amount of science communication activities done. However: “Geological and Energy Resources” were not listed in section 3.1 (lines 193 - 197) This leads to wonder: is that a category with very few people in it? If so: how does that affect the results? If one of the 4% of respondents that identified as ‘professional science communicators’ happened to be in that field, it would explain those results. The covariance between those categories can not be ignored.
In line 195 the authors state that ‘the sample slightly overrepresented scientists with a geology degree’, but in line 181 they state their is no data on the number of geoscientists in Portugal so it is unclear to me where they base their conclusion of overrepresentation on, given that the total number (ie, the frame) is not known.
Lines 211 through 243 provide a text based overview of the raw response per question. This should be presented in some visual way, potentially in the appendix. For questions where high correlation between answers are important to understand the results, some overview of the correlation / cross responses should be presented, for example between ‘area of expertise’ and ‘professional category ‘ (Q5 and Q6). Especially the responses to the survey questions that together contribute the factors used in answering RS1 should be presented.
Finally, a main Achilles heel not touched upon by the authors is that even in anonymized survey responses people are known to give socially desirable answers. Combined with the lack of a clear definition of science communication activities, this could easily lead to an overestimation of the amount of science communication reported by the survey respondents. The authors should reflect on the impact this has on their results.
On Open Science best practicesThe Copernicus publication guidelines on data sharing state that both data used to obtain the results and software used in the analyses should be shared alongside a publication. While those guidelines are written with the more classic earth science fields in mind, this does not exclude social sciences methods like surveys. The raw results of the survey might not be share-able due to legal reasons (GDPR), but there are good best practices available to anonymize survey responses and share them together with the article. See for example https://dmeg.cessda.eu/Data-Management-Expert-Guide/5.-Protect/Anonymisation
Sharing the data with the research allows other scientists, including the reviewers, to look for (other) patterns in the data collected and to check the analyses done by the authors. Without this data, the current submission violates the Copernicus guides on Open Data and should not be accepted.
ConcludingReading back this review I recognize that this might feel overly negative to the authors and I want to assure that my review is written with the purpose of making the manuscript better and getting this valuable data and analyses published. The insights that can be gotten from the data collected by the authors is great and should be shared with the scientific community. I hope that the comments above can be taken into account on the road to making this a great publication.
Rolf Hut
Citation: https://doi.org/10.5194/egusphere-2022-1160-RC1 -
AC1: 'Reply on RC1', Joana Rodrigues, 09 Jan 2023
We sincerely thank Referee Rolf Hut for the valuable suggestions and constructive comments, which we found extremely useful to improve our manuscript.
- On the survey design
Undoubtedly, we consider interaction with journalists as ‘science communication activities’ and the survey includes specific questions about journalists and media, such as Q43, Q17, Q37.11 (not analysed under the scope of the present study, with the exception of Q37.11). In Q11 were listed activities carried out directly with the ‘broad public’ and nonspecific for particular targeted publics, such as the media. Furthermore in Q13 the respondents reported the groups with whom they have engaged, where journalists were included. We will include in the appendix table the analysis of this question assessing the frequency of contact with journalists (Q13.1, Q13.2), where it is possible to conclude that it is very small. Participants were not given any definition of ‘science communication activities’ in the survey. We did not intend to condition each personal notion/perception of this activities. Each respondent reported the activities they assume as such and which they consciously carry out as ‘science communication’.
We agree that it may be relevant to add this circumstance in the text in section 4.2, before line 270:
To access this variable participants were asked how many science communication activities had they carried out in the previous year. As no prior detailed explanation of 'science communication activities' was given, each respondent reported the activities they assume as such and which they consciously carry out as science communication.- On the methodology
It was not conducted a sensitivity analysis in this study. We agree that it is important to mention in the text, on section 3, that when designing the questionnaire we did a-priori factorisation and the choice was made from a strategic point of view, considering the literature review, based on previous questionnaires such as Hartz and Chappell (1997), Wellcome Trust (2000), Royal Society (2006) and Liverman & Jaramillo (2011).
- On the interpretation of the results
In section 3.1 (lines 193 - 197) are listed the geoscientists’ degree area (Q4): Geology; Biology and Geology / Environmental Sciences / Environmental Education; Geological Engineering / Mine Engineering; Biology; Geophysics/ Meteorology / Oceanography / Physics and Geography / Aerospace Engineering.
The early training (Bachelor's) area does not always correspond to the area of work and research, therefore area of expertise was introduced as a different indicator (Q5): Geological and Energy Resources; Internal Geodynamics, Geophysics, Petrology and Geochemistry; External Geodynamics and Palaeontology; Geoconservation and Geotourism; History and Education; Environment, Environmental Geology and Engineering Geology. This detailed information will be included, in supplementary material, in the summary of the results of the demographic and descriptive analysis.There is no data on the number of geoscientists in Portugal, but considering that the geosciences include several areas, besides geology, and that 70% of the respondents are geologists, we believe that the proportion of the other geoscientists is over represented.
Geology (125 respondents), Biology and Geology/Environmental Sciences/Environmental Education (28 respondents), Geological Engineering/Mine Engineering (8 respondents), Biology (8 respondents), Geophysics/Meteorology/Oceanography/Physics (6 respondents), Geography / Aerospace Engineering (4 respondents). This detailed information will be included, in supplementary material, in the summary of the results of the demographic and descriptive analysis.We agree that a table that compiles the results of the descriptive analysis (lines 211 – 243), provides an easier and better overview, clarifying for example the two previous discussed topics. We will submit it as supplementary material.
We also agree that for questions where high correlation between answers are important to understand it is relevant to include an overview of these correlations. As it is a significant amount of information, we will submit it as supplementary material.
We are aware that a social desirability bias always exists in these studies and that is also why we mention in the conclusions that more robust results may be obtained through further studies. We agree that social desirability bias should be addressed in the text, around line 361:
In surveys like this where people are asked to provide personal opinions and report personal practices it is necessary to take into account that people may be led to answer according socially desirable ideas. To try to reduce this bias and obtain more truthful answers, participants were informed that their identities were protected and the results would be appropriately anonymized. At the same time, online self-administered questionnaires, without the presence of other people, as this case, also contribute to reduce social desirability bias.- On Open Science best practices
We will include in section 3, together with the methodology, the information regarding the software used for statistical analysis. Under this research project we are currently still preparing the publication of other studies with all these unpublished data as also the submission of the full data collected with this survey to a specialised journal for open research data. For this reason, we consider to submit only the variables analysed in the scope of this study.
Once again, we thank Referee Rolf Hut for the constructive insights. We hope that our answers have clarified all the comments.
The authors
Citation: https://doi.org/10.5194/egusphere-2022-1160-AC1 -
AC2: 'Reply on RC1', Joana Rodrigues, 09 Jan 2023
Publisher’s note: this comment is a copy of AC1 and its content was therefore removed.
Citation: https://doi.org/10.5194/egusphere-2022-1160-AC2
-
AC1: 'Reply on RC1', Joana Rodrigues, 09 Jan 2023
-
RC2: 'Comment on egusphere-2022-1160', Anonymous Referee #2, 02 Jan 2023
This study surveys Portugal geoscience majors and experts to explore potential predictors that have significant influences on geoscience communication activities, which is of great interest to GC. The most valuable parts of the manuscript are the detailed questionnaires about geoscience communication and the summary of potential predictors in past studies. However, the current manuscript has several severe problems in basic experiment design, and quantitative methods and lacks some critical information to evaluate the validity of analyses. A major revision is needed.
Major comments:- The total sample size is good (n=179), but the study has very weak control over the survey participants. There are at least six different groups with very different expertise and background, including geoscience students, professors, industry experts, researchers, and communicators. The sample size for each group is relatively small and probably cannot represent their populations (the authors also mentioned this in the paper at Ln186). The authors may want to give more statistical details about the individual groups. Different groups probably have very different opinions about geoscience communication, even the definition of geoscience communication. Thus analyzing their survey results separately will be more meaningful in my mind.
- The study used Chi-square tests to understand how different factors may affect geoscience communication activities (Research Question 1) and reported the chi-square values and p-values, but there is no information about the frequency expected. Especially where this expected frequency comes from. If it is from the literature review, then the authors should also explain how they deal with the different data collection standards, question design, and analysis of individual factors. Without this information, it is hard to evaluate the soundness of the test interpretation.
- The study applied multinomial logistic regression to evaluate the relationship between the frequency of geoscience communication activities and selected factors (e.g., professional experience), which is problematic. The frequency or intensity of the activity (from none to inactive to very active) is NOT an unordered categorical variable. The dependent variable of multinomial logistic regression has to be unordered nominals. The frequency of the activities is apparently ordinal (0, 1, 3-4, >4 in the last year). Thus, the authors may want to consider other models like the structure equation model.
Minor Comments:
The manuscript is hard to read, especially the introduction part.
The definition of public engagement should be defined in this paper because, in my mind, and I think for many researchers, public engagement and science communication are different in definition.
The title is too general and should indicate that the data is only from Portugal.
Geoscience communication is a very general word, and the authors give a wide range of activities in the survey items, but activities, such as speeches at geoscience conferences, are not typically considered as geoscience communication for public engagement purposes. The authors should give a more precise definition or scope of Geoscience communication in this study.
Overall, I think the study has potential, and I look forward to reading the revised version.
Citation: https://doi.org/10.5194/egusphere-2022-1160-RC2 -
AC3: 'Reply on RC2', Joana Rodrigues, 09 Jan 2023
We thank Referee 2 for the constructive and helpful comments, which were very insightful to improve our manuscript.
Major comments:
1. We are aware that the samples of the different professional groups are relatively small and may not represent accurately their populations. Because of that we haven’t processed their results separately. As the aim of the current work is to study the population in general, this question was included precisely to show that it is not a homogeneous population and that geoscience communication practitioners are not only researchers and academics.
Suggested by the Reviewer 1, we will include as supplementary material an overview through the results and correlations that will also present the professional groups distribution.
2. The expected frequency was calculated based on the assumption of independence between the two categorical variables being analysed and the chi-square test of independence was used to determine if there is a significant association between the two variables.
The analysis of individual factors was done by performing several independent chi-square tests to examine the relationship between the response variable and each predictor variable. This is a common approach for understanding how different factors may affect the response, as the reviewer well points out. However, as is well known, this approach has several limitations such as leading to an increased risk of Type I errors (i.e., false positives), because each test is evaluated separately, and the probability of making a Type I error increases with the number of tests performed. Furthermore, this approach does not allow for the analysis of multiple predictor variables simultaneously, so it is difficult to determine the relative importance of each predictor variable or to understand the combined effect of multiple predictor variables on the response. To deal with these well-known limitations, we decided to also use a multinomial logistic regression model, a method that seems to us to be more appropriate, as it allows the analysis of multiple predictor variables simultaneously and provides estimates of the relative odds of each level of the response variable occurring. This can help to better understand the relationship between the predictor variables and the response variable, and to identify the factors that are most influential in determining the response.
3. Both Structure Equation Model, SEM, and ordinal regression are statistical techniques that can be used to analyse relationships between predictor variables and an ordinal response variable. SEM is a more comprehensive model that allows for the analysis of multiple mediator and moderator variables, but this is not exactly the purpose of this study. On the other hand, ordinal regression models are primarily concerned with estimating the predictor and response variables, which is the aim of this study.
However, we chose to use a multinomial logistic regression model which has the advantage of having been specifically designed for analyzing relationships between predictor variable and a categorical response variable with more that two levels. We think it is a good choice where the research question involves examining the influence of predictor variables on the likelihood of different levels of the response variable occurring.
Another advantage of multinomial logistic regression is the simplicity of interpretation. It involves fitting separate logistic regression models for each level of the response variable, and the coefficients from these models can be used to estimate the relative probabilities of each level of the response occurring.
There are other reasons that led us to choose this procedure:
1) the goal is to identify the predictors of membership in the different categories of the response variable;
2) the categories of the response variable are not equally spaced, and ordinal regression assumes that the distance between levels of the response variable is equal;
3) the sample size is small, so a multinomial logistic regression model may provide more stable estimators.
Minor Comments:
- We will revise the introduction part to make it more readable.
- Indeed, public engagement and science communication are not synonymous. Science communication refers to a broad range of activities, but given that, in the current paradigm of science communication, scientists communicating are mostly required to promote public engagement, these terms are normally used interchangeably, even in the specific literature regarding science communication. But we agree that this may not be understood as such by everyone so we consider it to be pertinent to clarify these concepts, and we will do so in the section 1 Introduction
- The studies on this specific topic are very few, almost non-existent among international literature, with one exception made more than 10 years ago (Liverman & Jaramillo, 2011). At the same time, the main outcomes of the research confirm the patterns of scientists from other areas previously studied. We think that a more general title may boost more opportunities for further researches to develop strategies to suppress geoscience communication constraints worldwide.
- We agree that it is pertinent to include in the introduction a more precise definition about geoscience communication and its scope within this study.
Once again, we thank Referee 2 for the relevant questions raised which we hope we have fully answered.
The authors
Citation: https://doi.org/10.5194/egusphere-2022-1160-AC3
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 259 | 72 | 10 | 341 | 32 | 3 | 4 |
- HTML: 259
- PDF: 72
- XML: 10
- Total: 341
- Supplement: 32
- BibTeX: 3
- EndNote: 4
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Cecília Castro
Elsa Costa e Silva
Diamantino Insua Pereira
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(938 KB) - Metadata XML
-
Supplement
(1463 KB) - BibTeX
- EndNote
- Final revised paper