The Effect of Generation Z’s Understanding of Free Will and Fatalism on Their Political Choice: A Field Study

Round 1

Reviewer 1 Report

The authors test the effect of different items that constitute fatalism, namely predetermination, superstition, personal control and luck, on voters' preference for political parties and political leader
political party and political leader preferences. Although it is exciting to read a paper on fatalism and political preference where there is undoubtedly a research lacuna, the analysis
cannot be published mainly because of a series of methodological issues, as outlined below. However, if the authors conduct a severe re-writing and delve deeper into the fundamentals of regression analysis, I would be willing to review their article as a new contribution.     

- What is the contribution of the study? I was expecting to read how the study contributes to relevant past research under section 6
but the authors merely repeat the title of their study and provide some further information on how they measured fatalism- that should be moved to another section.
-Section 7. is redundant as the authors have repeatedly stated their research question—same applied to section 8. 'Universe' as it could be integrated into a section where the authors discuss all matters related to their data e.g. variables, measurement etc. Equally, section 7.1 could be omitted as the authors just repeat the obvious.

 
Quantitative analysis.

There are several problems related to the analysis presented. First, a sample of 488 respondents can hardly be representative of the Turkish population. The former can be easily seen if one compares the measurements of central tendency between their sample and the population (For example, it is evident that the men are underrepresented in the sample compared to women). Also, other variables in the sample are pretty off compared to reality. For example, we can see from the distribution of participants' presidential preference that the support for Kılıçdaroğlu support  (to mention one example) in the sample has been underestimated.
Therefore, an analysis that does not apply proper weights that account for imbalances in vital demographic factors (e.g., gender, age, education, and SES) leads to biased results. The authors can try different weighting techniques and choose the one that is the most proper for their analysis.
Suppose the authors can find another representative sample of the Turkish population. In that case, my advice is to go for the 'cem' technique, where the representative sample will assume the role of the control and the author's sample will assume the role of the treated group.
See: 'cem: Coarsened Exact Matching in Stata' by Blackwell et al. (2010)

Methods:

The info in Table 7 that provides results from a likelihood ratio test can be integrated into Table 8. The authors claim that the likelihood test showed that "the model was appropriate" could just be deleted as the test compares the model with predictors versus a null model with only a constant.    
 
Multinomial Logistic Regression Analysis is employed when the dependent variable is a categorical variable with more than two categories that cannot ranked, and not as the authors write, "In cases where there are more than two dependent variables"

There is no discussion on the standard errors (Are they clustered? It would make sense to cluster the SE on the level of voters' preferences)

However, the most severe limitation of the current manuscript is that they test the effect of independent variables (predetermination, superstition, personal control and luck) on political party and political leader preferences, without yet controlling for the effect of other factors in their regression model. The former is odd, and there can be no trust that the few coefficient estimates that were reported to be statistically significant would continue to be so should appropriate controls be introduced in the regression model. Subsequently, for the analysis to have any meaning, the authors should control for the effect of both demographic (e.g. age, gender, education, region of residency (rural/urban ) etc.) as well as other idiosyncratic political-related characteristics (e.g. ideology, pid, political interest etc.)

Minor:

I would move the discussion on the models of the study that now appear after the results before them.
 
I would keep the discussion on fatalism closer to political preferences.

I think a native proofreader would improve the flow of the arguments but in general I do not detect any serious issue regarding the quality of English .

Author Response

Reviewer#1

Changes Made:

The authors test the effect of different items that constitute fatalism, namely predetermination, superstition, personal control and luck, on voters' preference for political parties and political leader, political party and political leader preferences. Although it is exciting to read a paper on fatalism and political preference where there is undoubtedly a research lacuna, the analysis
cannot be published mainly because of a series of methodological issues, as outlined below. However, if the authors conduct a severe re-writing and delve deeper into the fundamentals of regression analysis, I would be willing to review their article as a new contribution. 

Thank you

    
- What is the contribution of the study? I was expecting to read how the study contributes to relevant past research under section 6 but the authors merely repeat the title of their study and provide some further information on how they measured fatalism- that should be moved to another section.

Thank you. Changes Made

-Section 7. is redundant as the authors have repeatedly stated their research question—same applied to section 8. 'Universe' as it could be integrated into a section where the authors discuss all matters related to their data e.g. variables, measurement etc. Equally, section 7.1 could be omitted as the authors just repeat the obvious.

Thank you. Changes Made

Quantitative analysis.

There are several problems related to the analysis presented. First, a sample of 488 respondents can hardly be representative of the Turkish population. The former can be easily seen if one compares the measurements of central tendency between their sample and the population (For example, it is evident that the men are underrepresented in the sample compared to women). Also, other variables in the sample are pretty off compared to reality. For example, we can see from the distribution of participants' presidential preference that the support for Kılıçdaroğlu support (to mention one example) in the sample has been underestimated.
Therefore, an analysis that does not apply proper weights that account for imbalances in vital demographic factors (e.g., gender, age, education, and SES) leads to biased results. The authors can try different weighting techniques and choose the one that is the most proper for their analysis.
Suppose the authors can find another representative sample of the Turkish population. In that case, my advice is to go for the 'cem' technique, where the representative sample will assume the role of the control and the author's sample will assume the role of the treated group.
See: 'cem: Coarsened Exact Matching in Stata' by Blackwell et al. (2010)

We cordially thank to reviewer 1 for valuable comments.

Our study focused on the political preferences of Generation Z young adult voters. In this context, the gender variable was not considered as a determining variable. In addition, no studies have been found on this subject in Turkey that reveal that the gender variable has an impact on political preferences.

Political choices are a sensitive issue in Turkey. For this reason, many of the participants refrained from sharing their political preferences. During the data collection process, a total of 20 universities and 2000 people were reached in seven regions of Turkey. In order to obtain realistic answers from the participants during the data collection phase, participation in the scale was implemented on a voluntary basis. Due to the sensitivity of the subject, feedback was received from a total of 550 participants. After removing the erroneous data, the analysis continued with 488 participants. From the literature review and calculations, it was concluded that a minimum of 385 participants was sufficient for the sample size. Therefore, it was decided that the sample size of 488 people was sufficient. In addition, the sample size was decided by taking similar studies as a reference.

The unequal distribution of men and women is a limitation of the study. The reason for this is the difficulty of the subject we are working on. As I mentioned before, Turkish voters are very reluctant to share which party they voted for. For this reason, I kindly request you to consider the sample size and gender distribution within the scope of the research limitation.

Table 3 shows the political leader preferences of the participants. Looking at the distribution in the table, it can be thought that it does not fully coincide with the 2023 election results. There are logical explanations for this. First of all, do not forget that this study was applied to young voters between the ages of 18-26. It is not possible to know which age group voted for which party or leader in the 2023 elections. Naturally, there is no criterion that can be compared to understand how much the data in the table corresponds to reality. The results given in Table 3 largely reflect the vote share Erdogan received in the 2023 elections. However, the results of Ogan and Kılıçdaroğlu do not match the election results. As it is known, Sinan Ogan carried out his election campaign mostly through social media, not national channels. For this reason, it mostly made propaganda aimed at young people and reached them more. It is known that Ogan's vote in the Turkish public is largely from young people. In addition, AKP and MHP (parties that support Erdoğan) have youth structures and carry out activities aimed at young people. Naturally, it is known that Erdogan and Ogan have the potential to get votes from young people. This study may have found an interesting result that reveals that Kılıçdaroğlu is not supported enough by the young segment. Considering this aspect, the table presented may be of value from which all political parties and leaders can draw meaningful conclusions, rather than being biased or erroneous.

We haven't experienced the method you suggested (Coarsened Exact Matching) before. For this reason, literature was used when creating the sample and was used in the calculation formulas. In our calculation, the minimum size was determined as 385. Therefore, it was decided that 488 participants were sufficient.

However, your comments were taken into account and the sample was expanded. While creating the sample, stratified sampling method was applied in order to represent the Turkish population.

Methods:
The info in Table 7 that provides results from a likelihood ratio test can be integrated into Table 8. The authors claim that the likelihood test showed that "the model was appropriate" could just be deleted as the test compares the model with predictors versus a null model with only a constant.  

Thank you. Changes Made

Multinomial Logistic Regression Analysis is employed when the dependent variable is a categorical variable with more than two categories that cannot ranked, and not as the authors write, "In cases where there are more than two dependent variables"

Thank you

There is no discussion on the standard errors (Are they clustered? It would make sense to cluster the SE on the level of voters' preferences)

Thank you. Changes Made

However, the most severe limitation of the current manuscript is that they test the effect of independent variables (predetermination, superstition, personal control and luck) on political party and political leader preferences, without yet controlling for the effect of other factors in their regression model. The former is odd, and there can be no trust that the few coefficient estimates that were reported to be statistically significant would continue to be so should appropriate controls be introduced in the regression model. Subsequently, for the analysis to have any meaning, the authors should control for the effect of both demographic (e.g. age, gender, education, region of residency (rural/urban ) etc.) as well as other idiosyncratic political-related characteristics (e.g. ideology, pid, political interest etc.)

Thank you. Changes Made

Minor:

I would move the discussion on the models of the study that now appear after the results before them.

Thank you

I would keep the discussion on fatalism closer to political preferences.

Thank you

Reviewer 2 Report

I was enthusiastic to read this paper because I think the idea of fatalism among Generation Z and their political choices is an interesting area of research. The authors did a good job of outlining the theoretical background to fatalistic ideas and also detailing how things may be different for Generation Z. Unfortunately, the study is not designed to demonstrate that Generation Z is different than previous generations, nor does it focus on understanding reasons why political choices may be influenced by fatalism. There are many promising aspects of this paper but ultimately it requires further work to achieve its goal.

I have several comments about the paper.

First, the paper abstract sets out that the project is intended to understand how young adult fatalistic tendencies influence their political preferences. The authors undertake this with a limited sample of young people (18-26). I have several questions about the sample. Why was the sampling purposive? Why were only young people sampled, rather than the general population which would allow for comparisons? How do the demographic characteristics of the sample compare to the population (even the young population)? There is certainly a gender discrepancy. Further, why was the study conducted face to face – is the concern about random answers that great? The findings of this paper are interesting but I am left wondering a) how reliable are these results, given that the sample is not representative, and b) are young people the same as others? What can be said about young people’s political choices that is meaningful since we do not know whether it is the same or different for others? The authors touch on this on p.19 when they discuss the attitudes of young people in terms of the "dichotomous relationship type in Turkish politics", but not having comparisons with older individuals makes this less convincing.

My second comment is about the structure of the paper. I started reading expecting a discussion of how the political parties use fatalism in their appeals or what kinds of relationships are known to exist in the Turkish public already related to these attitudes. Instead, the authors spend considerable time discussing values and generational change. While this is interesting, and relevant to their overall discussion, it does little to provide the necessary background for understanding the specific results about political choices. If younger people hold different world views than their elders what does that mean for politics? How can I understand the political tendencies brought about by seeing the world differently? Is the expectation that young people will be more likely to prefer certain parties? Hypotheses along these lines would be very helpful and would provide the reader with the tools needed to interpret the results. In the discussion section there are some comments about the relationship between fatalistic tendencies and right-wing parties/leaders that is very interesting, but this relationship was never adequately theorized so as a reader I could not put its importance into context. I also could not judge whether there is a change or continuation from previous generations.

My last comments are a bit smaller:

1)    Why use the subscales rather than the full fatalism scale in the analyses? What is learned from specific dimensions of fatalism rather than the concept holistically? Here it would be most valuable to understand whether young people who hold certain perspectives are more or less likely than others to have certain political preferences, and even whether the composition of fatalistic tendencies differs between generations.

2)    Why was the baseline category changed in the regressions? Was it to facilitate interpretation?

3)    The mention of locus of control is interesting and perhaps useful for connecting fatalist tendencies to political preferences and/or vulnerability to specific appeals.

There were some sentence fragments in the paper and some typos (what looked like mistakes in editing) that should be cleaned up.

Author Response

Reviewer#2

Changes Made:

I was enthusiastic to read this paper because I think the idea of fatalism among Generation Z and their political choices is an interesting area of research. The authors did a good job of outlining the theoretical background to fatalistic ideas and also detailing how things may be different for Generation Z. Unfortunately, the study is not designed to demonstrate that Generation Z is different than previous generations, nor does it focus on understanding reasons why political choices may be influenced by fatalism. There are many promising aspects of this paper but ultimately it requires further work to achieve its goal.

Thank you

I have several comments about the paper.

First, the paper abstract sets out that the project is intended to understand how young adult fatalistic tendencies influence their political preferences. The authors undertake this with a limited sample of young people (18-26). I have several questions about the sample. Why was the sampling purposive? Why were only young people sampled, rather than the general population which would allow for comparisons? How do the demographic characteristics of the sample compare to the population (even the young population)? There is certainly a gender discrepancy. Further, why was the study conducted face to face – is the concern about random answers that great? The findings of this paper are interesting but I am left wondering a) how reliable are these results, given that the sample is not representative, and b) are young people the same as others? What can be said about young people’s political choices that is meaningful since we do not know whether it is the same or different for others? The authors touch on this on p.19 when they discuss the attitudes of young people in terms of the "dichotomous relationship type in Turkish politics", but not having comparisons with older individuals makes this less convincing.

Thank you

Only young voters were selected in the sample. Because research organizations in Turkey predict that young voters will display a different political attitude compared to the general population. At the same time, evaluations are made that the young population is moving away from traditional values. The issue we are curious about in the research is this: Is the level of fatalism high among young voters and does the change in the level of fatalism affect political preferences? As a result of the research, it was evaluated that the young voter base will not cause a change as expected in the elections to be held in the near future.

Voters in Turkey are generally anxious about sharing their political preferences. Therefore, data was collected with this method to ensure that the participants gave realistic answers.

1. a) how reliable are these results, given that the sample is not representative, and

While the sample was not determined, only those who voted in the 2023 elections were preferred. For this reason, purposeful sampling type was deemed common.

Additionally, in the "Sample" section, the number of young people in Turkey was given and it was decided that the sample size was sufficient for both women and men. I will definitely take your comments on this matter into consideration. However, since the research results were similar to the 2023 election results, we thought that the representativeness of the sample was sufficient.

1. b) Are young people the same as others? What can be said about young people’s political choices that is meaningful since we do not know whether it is the same or different for others? The authors touch on this on p.19 when they discuss the attitudes of young people in terms of the "dichotomous relationship type in Turkish politics", but not having comparisons with older individuals makes this less convincing.

The young population differs from other generations. This situation is explained in the background of the article. However, there is a widespread belief that the differences in the general tendencies of the young population will also be reflected in their political preferences. For this reason, the research was conducted only on young people.

My second comment is about the structure of the paper. I started reading expecting a discussion of how the political parties use fatalism in their appeals or what kinds of relationships are known to exist in the Turkish public already related to these attitudes. Instead, the authors spend considerable time discussing values and generational change. While this is interesting, and relevant to their overall discussion, it does little to provide the necessary background for understanding the specific results about political choices. If younger people hold different world views than their elders what does that mean for politics? How can I understand the political tendencies brought about by seeing the world differently? Is the expectation that young people will be more likely to prefer certain parties? Hypotheses along these lines would be very helpful and would provide the reader with the tools needed to interpret the results. In the discussion section there are some comments about the relationship between fatalistic tendencies and right-wing parties/leaders that is very interesting, but this relationship was never adequately theorized so as a reader I could not put its importance into context. I also could not judge whether there is a change or continuation from previous generations.

Thank you; Changes made.

My last comments are a bit smaller:

• Why use the subscales rather than the full fatalism scale in the analyses? What is learned from specific dimensions of fatalism rather than the concept holistically? Here it would be most valuable to understand whether young people who hold certain perspectives are more or less likely than others to have certain political preferences, and even whether the composition of fatalistic tendencies differs between generations.

Thank you

The fatalism scale, which has sub-dimensions, was used in the study. Because the concept of fatalism is defined in more than one way in the literature. For example, different fatalistic attitudes are defined, such as an individual being fatalistic at the level of superstition or simply believing in fate. This distinction is explained in the section on the concept of fatalism.

• Why was the baseline category changed in the regressions? Was it to facilitate interpretation?

Thank you

The regression analysis baseline category has been changed. The aim here was to ensure a more meaningful interpretation of the results. The perspective was expanded by looking at it from two different perspectives.

• The mention of locus of control is interesting and perhaps useful for connecting fatalist tendencies to political preferences and/or vulnerability to specific appeals.

Thank you

Round 2

Reviewer 2 Report

Thank you for considering my comments.

The restructuring of the purpose of the paper is helpful. It is now clearer that you are not evaluating whether fatalistic tendencies exist more or less than for others (which would require comparison to older individuals) but simply looking to see what the levels are and whether there are political effects. I am still not fully comfortable with just accepting that the revealed patterns match those expected for other generations, as discussed by the authors. Have other studies been done that show similar relationships with fatalist tendencies and parties/leaders for others? Perhaps a way around this would be to link fatalist tendencies more explicitly to political ideology earlier in the paper - right now it appears after the results. But I do buy that fatalist attitudes are more traditional, and so likely more conservative, and that left-wing attitudes are more positivist. Making that argument about political ideology could help to set up the results better. 

With the data you have, can your research questions be examined as they are currently worded? How can you look at change in tendencies with a single dataset?

For the sample description, it was initially misleading that all participants were university students. In the current wording, it appears that you sampled from across the country. How were individuals contacted, and what kind of process led to their selection? What was the sampling frame? Was a firm used for contact, or random telephone numbers, or addresses, or something else? If you could provide details about the process of contact it would clarify whether this was a probability or non-probability sample. Also, what were the stratification categories used? Tables 1, 2 and 3 would benefit from another column that shows the population statistics (from a census or elsewhere) to facilitate comparisons with the sample proportions.

Some of the text on p.15 was not in English so I could not read it.

I appreciate that the contribution of this paper is to show that even though the young generation is very different on many dimension, their political preferences are still structured in traditional ways.

The writing is pretty clear but there are some issues with grammar that should be addressed for an English-speaking audience. There are several sentence fragments that are distracting for reading. In one spot the authors write "In fatalism, or in other words, fatalism, the outcome ...", which does not make sense. A thorough review by a native English speaker would be helpful.

Author Response

Editor

Social Sciences-MDPI

Manuscript ID: socsci-2757250

Type of manuscript: Article

Title: The Effect of Generation Z's Understanding of Free Will and Fatalism on Their Political Choice: A Field Study

Dear Editor,

We have received the results of the second-round evaluation process of our article titled "The Effect of Generation Z's Understanding of Free Will and Fatalism on Their Political Preferences: A Field Research", which was prepared to be published in your Social Sciences Journal.

We are happy that the evaluation process of our article continues for publication in your Journal. We would like to thank the Referee 2 for their valuable comments and corrections.

We reviewed the evaluation letter received in the second round. We thank referee 2 for his new comments. Our responses and corrections are shown below (in blue) in this rebuttal letter.

Our corrections to Reviewer #2 are highlighted in blue on the revised paper. Further details are given below. If you have any question, please contact me by email.

With warmest personal regards,

REVIEWER#2

Changes Made:

• The restructuring of the purpose of the paper is helpful. Thank you. Changes Made.

• It is now clearer that you are not evaluating whether fatalistic tendencies exist more or less than for others (which would require comparison to older individuals) but simply looking to see what the levels are and whether there are political effects. I am still not fully comfortable with just accepting that the revealed patterns match those expected for other generations, as discussed by the authors. Have other studies been done that show similar relationships with fatalist tendencies and parties/leaders for others? Perhaps a way around this would be to link fatalist tendencies more explicitly to political ideology earlier in the paper - right now it appears after the results. But I do buy that fatalist attitudes are more traditional, and so likely more conservative, and that left-wing attitudes are more positivist. Making that argument about political ideology could help to set up the results better. Thank you. Changes Made

• With the data you have, can your research questions be examined as they are currently worded? How can you look at change in tendencies with a single dataset? Thank you

In our research, fatalistic tendencies were determined in the young generation. As stated in the purpose of the study, it is a matter of curiosity whether young generations have a tendency to believe in fate. For this reason, the tendency towards fatalism in young people was examined and whether these tendencies had a significant relationship with political preferences. Of course, other variables affecting political preferences can be addressed by conducting more comprehensive studies. However, the fact that similar studies have not been conducted in the literature is among our limitations. In this respect, this study will provide a new perspective to other researchers. At the same time, how much the findings presented in this study match reality will gain meaning with the results of other studies to be conducted from now on. Although the data obtained in this study are limited, they are valuable findings that are meaningful for Turkish politics and Generation Z.

• For the sample description, it was initially misleading that all participants were university students. In the current wording, it appears that you sampled from across the country. How were individuals contacted, and what kind of process led to their selection? What was the sampling frame? Was a firm used for contact, or random telephone numbers, or addresses, or something else? If you could provide details about the process of contact it would clarify whether this was a probability or non-probability sample. Thank you. Changes Made.
• We contacted our colleagues working at universities in seven regions of Turkey. Contact was made with academics at a total of 20 universities. The scale was sent to our colleagues for data collection. Our colleagues sent the scale to students' WhatsApp groups. It is very difficult to ask questions about political views and get realistic answers in Turkey. This. For this reason, only 630 healthy data could be obtained from 2000 students.
• Also, what were the stratification categories used? Tables 1, 2 and 3 would benefit from another column that shows the population statistics (from a census or elsewhere) to facilitate comparisons with the sample proportions. Thank you. Changes Made
• Some of the text on p.15 was not in English so I could not read it. Thank you. Changes Made
• I appreciate that the contribution of this paper is to show that even though the young generation is very different on many dimension, their political preferences are still structured in traditional ways. Thank you.