E-Mail Network Patterns and Body Language Predict Risk-Taking Attitude

Round 1

Reviewer 1 Report

In this paper, authors compared risk-taking attitudes assessed with the Domain-Specific Risk-Taking survey with individual e-mail networking patterns and body language measured with smartwatches. However, I have some suggestion for paper improvement as follows.

1. The main contribution is not clear in introduction. Authors should discuss their main contribution in the form of bullets. In addition, authors should include existing research work about social network and sensors data handling (‘An intelligent healthcare monitoring framework using wearable sensors and social networking data’, ‘A secure remote health monitoring model for early disease diagnosis in cloud-based IoT environment’, A smart healthcare monitoring system for heart disease prediction based on ensemble deep learning and feature fusion, and ‘The two-phase scheduling based on deep learning in the Internet of Things’.
2. Figure 1 is not discussed properly. In addition, the caption of figures should be self-explanatory, and clearly explaining the figure. Extend the description of the mentioned figures to make them self-explanatory.
3. Equation 1 is not properly presented.
4. The extra space between paragraphs should be removed.
5. Figure 2 and 3 are blurred, their quality should be improved.

Author Response

Rebuttal Letter – Reviewer 1

In this paper, authors compared risk-taking attitudes assessed with the Domain-Specific Risk-Taking survey with individual e-mail networking patterns and body language

measured with smartwatches. However, I have some suggestion for paper improvement as follows.

We thank the reviewer for an accurate summary of the paper. We next offer point-by-point response to each of the thoughtful comments the reviewer shared with us.

1. The main contribution is not clear in introduction. Authors should discuss their main contribution in the form of bullets. In addition, authors should include existing research work about social network and sensors data handling (‘An intelligent healthcare monitoring framework using wearable sensors and social networking data’, ‘A secure remote health monitoring model for early disease diagnosis in cloud-based IoT environment’, A smart healthcare monitoring system for heart disease prediction based on ensemble deep learning and feature fusion, and ‘The two-phase scheduling based on deep learning in the Internet of Things’.

We thank the reviewer for this great suggestion to improve our manuscript. We have carefully summarized the main contributions of our results, and they are as follows.

• Combining network theory, text mining and body-sensing technology, we introduce a novel interdisciplinary research method for analyzing one’s attitude of risk-taking.
• We validate the utility of the proposed method through two different empirical studies, from E-mail archives and Happimeter sensing system, respectively. That is, a strong correlation is found between the empirical signals and individual risk-preference in the different domains of the DOSPERT survey.
• Through the empirical evidence, we quantify significant predictors for one’s attitude of risk-taking based on tribal language features, emotionality, network structure metrics and body sensors, which provide valuable information for decision makers and managers to support an increase in ethical behavior of the organization’s members.

We have included the above contents in the revised manuscript and highlighted them to make it easy for the reviewer to track. Besides, we have added the relevant researches you provided in the revised Introduction.

2. Figure 1 is not discussed properly. In addition, the caption of figures should be self-explanatory, and clearly explaining the figure. Extend the description of the mentioned figures to make them self-explanatory. 

We thank the reviewer for this constructive comment. We have extended Figure 1’s cation and revised the captions of other figures for better clarity.

3. Equation 1 is not properly presented.

We thank the reviewer for this comment. Equation 1 have been removed in the revised manuscript. Instead, we describe it in a more detailed manner in the text.

4. The extra space between paragraphs should be removed.

All the extra spaces between paragraphs have been removed in the revised manuscript.

5. Figure 2 and 3 are blurred, their quality should be improved.

Thanks for this comment. We have provided all figures as individual PDF files in vector-graph format with the best resolution to avoid the blurring.

Reviewer 2 Report

This manuscript describes approaches to predict risk-taking attitude. After reviewing the manuscript,  I have some concerns about this manuscript:

1. The email network. the authors chose 32 out of 912 accounts to do further analysis. They are more active individuals. My first concern is the representative of these individuals. Because the more active individuals perhaps have higher risk-taking attitudes. Is the conclusion of the email network analysis still established for these not so active individuals?
2. The paper does not provide how do the authors get the six domains （general，ethical, financial,...), what are these domain meaning?
3. In table 3, some coefficients are very small, for example, the Degree centrality of Social domain is about ~0.003. Is this small value enough to obtain the perdictor conclusion?
4. The figure2 and figure3 are confusing: the perdict values vs real values. I am not sure how the perdict values are obtained. If the perdict values are good fitting the real values, the slope of straight line is about 1, but some of them are not. 
5. What are the sample size (number of Participants) of each domian of the two analysis (email network and Body sensor). Is the samples large enough to do the statistical analysis?
6. ethics concern. This research need to obtain personal data. does the research obey any ethics codes?

Author Response

Rebuttal Letter – Reviewer 2

This manuscript describes approaches to predict risk-taking attitude. After reviewing the manuscript, I have some concerns about this manuscript:

We thank the reviewer for his/her time and effort to review our work. We next offer point-by-point response to each of the thoughtful comments the reviewer shared with us.

The email network. the authors chose 32 out of 912 accounts to do further analysis. They are more active individuals. My first concern is the representative of these individuals. Because the more active individuals perhaps have higher risk-taking attitudes. Is the conclusion of the email network analysis still established for these not so active?

We thank the reviewer for this good point. The participants chosen in this study are a relatively small sample, but representative of a wider population of academics participating at the conference. The reason the 32 are most active is because they interacted most with the mailbox owner. We do not think that this influences their risk-taking attitude. We agree that a larger sample would be desirable, and will try to replicate the experiment in the future with a larger population.

The paper does not provide how do the authors get the six domains (general, ethical, financial, ...), what are these domain meaning?

We thank the reviewer for this comment. The 5 domains of life (ethical, financial, recreational, health and social) used in this work are specified by the DOSPERT test [29]. For instance, sample items include “Having an affair with a married man/woman” (ethical), “Investing 10% of your annual income in a new business venture” (financial), “Engaging in unprotected sex” (health), “Disagreeing with an authority figure on a major issue” (social), and “Taking a weekend sky-diving class” (recreational).

In this work, we get the scores of these five domains by asking every participant to complete the DOPSET test. Besides, to capture the general dimension of risk-preference, we introduce an additional ‘general’ domain as the average value in above five domains.

In table 3, some coefficients are very small, for example, the Degree centrality of Social domain is about ~0.003. Is this small value enough to obtain the predictor conclusion?

We thank the reviewer for pointing this out. In this work, we fit the data by the ordinary least regression (OLS) which is invariant. So, for invariant methods there is no real need for standardization or normalization. This means that the magnitude of each fitting coefficient is subject to the scale of the variable. If the scale of a specific value is extraordinarily large (e.g., the degree centrality), its corresponding coefficient will be small. But that does not affect our conclusion.

The figure2 and figure3 are confusing: the predict values vs real values. I am not sure how the predict values are obtained. If the perdict values are good fitting the real values, the slope of straight line is about 1, but some of them are not.

We thank the reviewer for this comment. As we explained in the last comment, the predicted value of risk-preference score in each domain is obtained by employing OLS to structured data which are extracted from email network and body signals. One can observe from Figure 2 and Figure 3 that the predicted value obtained by the proposed method is consistent with the true value in most domains (i.e., the slope approaches to 1), while in some domains, the prediction performance is slightly poor, implying that more data and more complicated algorithms may need to be considered. This could be a good project in the future.

What are the sample size (number of Participants) of each domain of the two analysis (email network and Body sensor). Is the samples large enough to do the statistical analysis?

We thank the reviewer for pointing this out. Each participant has six separate domains of data, which are obtained from their DOSPERT results. Thus, the sample size of each domain is equal to the number of total participants, i.e., N = 34 for study A and N = 15 for study B. We believe this is enough to conduct numerical analysis. Nevertheless, to prove the validity generalization of our results, a large-scale experiment involving more participants from a wider range of backgrounds should be done in the future. We have included above explanation in the Discussion section.

ethics concern. This research needs to obtain personal data. does the research obey any ethics codes?

We thank the reviewer for this comment. All the experiments conducted in this work have been approved by the Institutional Review Board (IRB) for the Massachusetts Institute of Technology (MIT). We have made the corresponding acknowledgment in the revised manuscript.

Reviewer 3 Report

The paper approaches a very interesting topic that theoretical has the potential to become even more and more appealing for different organizations, in the following years. However, there are a few issues worth considering before the paper can be accepted for publishing.

In the Introduction section, it would be helpful to better underline the novelty brought by this research in comparison to what has been previously investigated and written by other researchers.

As it contains the theoretical framework of the research, Figure 1 (from lines 62-63) should be explained in more detail. In addition, ending the Introduction section with a figure may not be the best option.

Line 275 – the limitations of the study (as there are plenty of them) should be thoroughly described in a separate section before Conclusions.

As stated by the authors, the paper evaluates individuals’ risk-preference. Therefore, both in the Introduction and mostly in the Conclusion section, it would be quite important to emphasize more the relevance of the findings from a practical (pragmatic) perspective. The only brief reference on this matter is made in the Abstract section on lines 17-18, which is not enough to justify the conducted research.

Author Response

Rebuttal Letter – Reviewer 3

The paper approaches a very interesting topic that theoretical has the potential to become even more and more appealing for different organizations, in the following years. However, there are a few issues worth considering before the paper can be accepted for publishing

We are delighted to hear that the reviewer appreciates the importance of the paper, and thinks the paper is interesting. We next offer point-by-point response to each of the thoughtful comments the reviewer shared with us.

In the Introduction section, it would be helpful to better underline the novelty brought by this research in comparison to what has been previously investigated and written by other researchers

We thank the reviewer for this constructive comment. To the best of our knowledge, there is little literature to apply social network analysis to access individuals’ attitude toward risk. Our work is among the first empirical efforts in solving this problem. We have carefully summarized the novelty of our research in the form of bullets, and they are as follows:

• Combining network theory, text mining and body-sensing technology, we introduce a novel interdisciplinary research method for analyzing one’s attitude of risk-taking.
• We validate the utility of the proposed method through two different empirical studies, from E-mail archives and Happimeter sensing system, respectively. A strong correlation is found between the empirical signals and individual risk-preference in the different domains of the DOSPERT survey.
• Through empirical evidence, we quantify significant predictors for one’s attitude of risk-taking based on tribal language features, emotionality, network structure metrics and body sensors, which provide valuable information for decision makers and managers to support an increase in ethical behavior of the organization’s members.

The above contents are highlighted in the Introduction section in the revised manuscript.

As it contains the theoretical framework of the research, Figure 1 (from lines 62-63) should be explained in more detail. In addition, ending the Introduction section with a figure may not be the best option.

We thank the reviewer for pointing this out. We have added the explanation of Figure 1 to make it self-explanatory. In addition, we have provided a new paragraph describing the organization of this paper, as the ending of the Introduction section.

Line 275 – the limitations of the study (as there are plenty of them) should be thoroughly described in a separate section before Conclusions.

We thank the reviewer for this comment. The discussion of the study’s limitation has been moved to a separate Discussion section before Conclusions.

As stated by the authors, the paper evaluates individuals’ risk-preference. Therefore, both in the Introduction and mostly in the Conclusion section, it would be quite important to emphasize more the relevance of the findings from a practical (pragmatic) perspective. The only brief reference on this matter is made in the Abstract section on lines 17-18, which is not enough to justify the conducted research.

We thank the reviewer for this constructive comment. We have revised the Introduction and Conclusion section to emphasize the practical implication of our work. All changes are highlighted to make it easy for the reviewer to track.

Round 2

Reviewer 1 Report

Thank you for addressing my comments. 

Reviewer 2 Report

The revised manuscript made a good improvement, I recommend it to be published in Future Internet.

Reviewer 3 Report

Most of the suggestions I made have been addressed. Therefore, I believe the paper has been improved and can now be accepted for publishing.
However, I would still have just one remark. The text from lines 64-68 (right after Figure 1) should be placed as normal text, not as a description of the figure.