From Left-Behind Children to Youth Labor Migrants: The Impact of Household Networks, Gendered Migration, and Relay Migration in Southeast Asia
Round 1
Reviewer 1 Report
I have read the current manuscript with interest. The paper investigates the intergenerational transfer of migration patterns between migrant parents and their children, exploring which factors are likely predictors of migrant behaviour. The object of the article is very relevant and not rather new as there isn't a lot of research dealing with the topic especially for SEA and Thailand. The study applies a logistic regression to carry out the data analysis and reports the scores together with significance and pseudo R2 and log likelihood.
The manuscript would in general benefit from English revision as some sentences seem a bit off.
The hypotheses are clearly stated however, it would be useful to add a few sentences to introduce them rather than dropping them at the end of each section.
I particularly appreciated reading the conclusion, it is well articulated and supported both by results and relevant literature.
My only remarks, which would improve the understanding of results, concern sample sizes and R2.
1. All tables should report sample size (not only the percentages as in table one). For instance table 3, should give the exact sample size for the full, male and female sample.
2. You use pseudo R2 and it's reported values are very low, but do not comment that in the text. Low pseudo R2 values might have to do with high variability. Have you tried using McFadden (1974) formula? it employs the natural log
Author Response
Response to Reviewer 1
General Comment: I have read the current manuscript with interest. The paper investigates the intergenerational transfer of migration patterns between migrant parents and their children, exploring which factors are likely predictors of migrant behaviour. The object of the article is very relevant and not rather new as there isn't a lot of research dealing with the topic especially for SEA and Thailand. The study applies a logistic regression to carry out the data analysis and reports the scores together with significance and pseudo R2 and log likelihood.
Response:
Thank you for all your valuable comments and suggestions. Point-to-point response to your comments is listed as follows.
Comment 1: The manuscript would in general benefit from English revision as some sentences seem a bit off.
Response:
Thank you for the comment. We have carefully proofread the manuscript for grammar and language.
Comment 2: The hypotheses are clearly stated however, it would be useful to add a few sentences to introduce them rather than dropping them at the end of each section.
Response:
Thank you for your comments. We have added more sentences for each hypothesis with more consideration of the literature review (Line. 457-459, 701-703, and 975-976).
Comment 3: I particularly appreciated reading the conclusion, it is well articulated and supported both by results and relevant literature.
Response:
Thank you for your comments. We appreciate your interest.
Comment 4: All tables should report sample size (not only the percentages as in table one). For instance table 3, should give the exact sample size for the full, male and female sample.
Response:
Thank you for your comments. We reported the sample size of each model.
Comment 5: You use pseudo R2 and it's reported values are very low, but do not comment that in the text. Low pseudo R2 values might have to do with high variability. Have you tried using McFadden (1974) formula? it employs the natural log.
Response:
Thank you for your suggestions.
We agree that low pseudo R2 might indicate a high variability. We tested model fitness and reported McFadden R2. McFadden (1977, p.35) wrote that "values of .2 to .4 for ?2 represent an excellent fit.” The reported McFadden R2 in our manuscript range from 0.082 to 0.152. Although the statistics are comparatively small, we believe the models can explain certain degree of variations in youth migration, the variable of our interest. We also supplemented our interpretation by adding potential high degree of variability in the manuscript (Line. 1428-1430).
We further compared the values of R2 reported in our manuscript to other migration literature. The value of around 10% is commonly reported in studies on youth migration (e.g., Kandel and Kao 2001).
Reviewer 2 Report
Overall:
The authors present an interesting study that links “left behind” experiences with youth labor migration. There is an exciting opportunity to contribute to the literature exploring different frameworks for youth migration as well as linking distinct areas of research that are often not linked. However, there are a substantial number of details that are glossed over that are needed to evaluate the rigor of the methods as well as understand the objectives and arguments of the authors. A simple but illustrative example is the lack of clarity about what we mean by “youth” throughout the piece. Who are we talking about exactly? With the addition of these details and further attention to clarity in the manuscript as well as more contextualization to the Thai context, I do think this could make an important contribution. Please see below for more specific feedback and comments.
Abstract:
Conceptually, the authors distinguish between three possible patterns of migration. However, could they not all be happening at the same time? My question is, are these patterns necessarily distinct?
Line 14: When you say “migration engagement of mothers,” you are referring specifically to mothers who are migrating, correct? Can this be reworded for clarity?
Line 16: Whose return migration are you referring to here specifically when you say “non-parental”?
Introduction:
Line 26: Can this be made more specific with a statistic?
Line 29: What effects/arguments?
Line 34: What type of significant impact?
Line 36: What do you mean it is independent of parental or household influences?
Line 56: What do you mean when you say “the gender aspect?” and, you make an important point that there are multiple factors beyond gender – other aspects of people’s social locations – that are relevant. Is there space to take a more intersectional approach here?
Line 72: When you refer to migration incentives, I wonder about the language of “incentive.” Why did the authors use this language? Is it really an incentive? Perhaps other language (pull factor?) could be useful here.
Line 89: This sentence is incomplete.
Section 2.1: There are a lot of pieces summarized and defined in this paragraph, and the overall point is a bit jumbled. Suggest it be reviewed and revised for clarity and simplicity?
Line 168: In what ways do women exert agency in the migration decision-making process? How are migration decisions made at the household level, particularly in Thailand?
Lines 187-9: Not clear.
Section 3. What types of work are people doing when migrating for work to/from Thailand? What are gender differences? What are age differences? Suggest the addition of more specifics here.
Overall questions:
I agree that migration of young people cannot be disentangled from the household. What then, distinguishes migration of young people from other family members? Why is it important to look at their experiences specifically?
How are the authors defining “youth” or young people and why?
What other non-family network connections (e.g. friends) could have been considered?
Methods:
Line 265: What was the replicable protocol? Although it has been published previously, details on the sampling strategy used is needed to enable the readers to understand representativeness.
Line 273: So a second wave was conducted that followed up on a subset of participants from CHAMPSEA and CLAIM? From CHAMPSEA, there were 1030 households, but only 411 at the second wave. For CLAIM, it was 1456 households, followed by 465 households. How was the decision made to have only about 400 households from each of the previous studies? How were those subsets selected? How were the analyses conducted – were the samples simply combined? How exactly?
Line 276: It is unclear what you mean about the attrition rate of 20%. Where is the 20% coming from?
Line 279: What specific reasons for migration were included if education migration was excluded? Could they provide multiple reasons for migration?
Line 285: What relevant ethics boards?
Line 291: Why is it a different consideration? Please explain a bit here to illuminate your point.
Line 292: the way this is written is a bit confusing. I think you want to say “whether they were labor migrants or not (0=no, 1=yes).
Line 297: Are you really measuring gender dynamics, or simply the sex of parents and children?
Section 3.2.3. more detail about parental occupation and economic circumstances and how those were operationalized is recommended to be added.
Section 3.3. what do you mean that you “apply stepwise regression?” what type of regression models were used? I am a little confused about how you analyzed the data longitudinally. How was model fit assessed for analyses conducted?
Results:
Line 337: become labor migrants ever, or become labor migrants between wave 1 and wave 2?
Line 343-4: More explanation of this proportion you report is important to consider.
Table 1. I recommend only reporting to 1 decimal place. For control variables, are you reporting the reference category? It might be clearer to just write (Male) rather than (Gender (0=male)). For marriage history, what is included in the bucket of “having a marriage history”?
Line 369: I would recommend reconsidering the use of the language “gender dynamic,” as the results suggest simply that the effect of the gender of the parent or child is important.
Discussion:
Tables 2 and 3 seem to be presenting log odds – is there a reason why odds ratios and 95% confidence intervals are not presented?
Line 402: this missingness not at random is an important consideration for interpreting the results. Once the authors identified this association, what specific sensitivity analyses or sub-analyses did you do (perhaps looking at that group at wave 1 to see if the trends were consistent with those who were not lost to follow up?)
Discussion:
What qualitative research would the authors recommend to understand and unpack the hypotheses that were supported and refuted in their analyses? For example. The authors write “further studies can enhance understanding of the dynamics,” (line 465). What further studies? Similar point in reference to the effects of community and its potential migration culture – see likes 544-8. What studies should be done?
Line 476: I think there is an assumption here about migration decision-making that needs to be considered. The authors emphasize that migration of youth is influence by multiple factors at the household level. Is it really that youth labor migrants may not be motivated for relay migration?
How might destination, duration, frequency be important to consider in thinking about the analyses and results presented her? How might associations vary by these characteristics of migration?
Paragraph beginning on line 519: Can you elaborate more fully on this as a limitation? The point is important, but not exactly clear why is it presented as a limitation.
I would appreciate further contextualization of the findings to the Thai context in the discussion section. It is done a bit, but the section would be stronger if done more rigorously.
Please incorporate a discussion of limitations of the sampling method and generalizability.
Author Response
Response to Reviewer 2
General Comment: The authors present an interesting study that links “left behind” experiences with youth labor migration. There is an exciting opportunity to contribute to the literature exploring different frameworks for youth migration as well as linking distinct areas of research that are often not linked. However, there are a substantial number of details that are glossed over that are needed to evaluate the rigor of the methods as well as understand the objectives and arguments of the authors. A simple but illustrative example is the lack of clarity about what we mean by “youth” throughout the piece. Who are we talking about exactly? With the addition of these details and further attention to clarity in the manuscript as well as more contextualization to the Thai context, I do think this could make an important contribution. Please see below for more specific feedback and comments.
Response:
Thank you for your comments. Point-to-point response to your comments and suggestions is as follows.
Comment 1: Conceptually, the authors distinguish between three possible patterns of migration. However, could they not all be happening at the same time? My question is, are these patterns necessarily distinct?
Response:
Thank you for your comments. Taking into account your suggestions along with Reviewer 3, we have revised the argument.
We agree that these three possible patterns of migration are not completely distinct. However, the objective of the manuscript is to examine the effects of networked migration, gendered migration, and relay migration on the transition between the role of left-behind children and youth labor migrants. Therefore, we examined the application of three theoretical frameworks, respectively, to highlight their effects on this transition.
We also discussed the issues in the section of limitation in our manuscript (Line. 1854-1865).
“Thirdly, the present study investigated the transition from left-behind experiences to youth labor migration using three fundamental theoretical arguments without their potential interactive effects. It is possible that migration networks facilitate deci-sion-making processes on a gendered basis. It has been suggested that whilst men tend to be connected to relatively more extended networks such as workplace networks for practical support, women are more likely to be connected to family networks, which provide them with not only information and practical support but also social protection (Hoang 2011). We speculate that relay migration could also be gendered as a way of reproducing gendered divisions of labor in migration. Future studies, including those comparative, could consider more diversity regarding social position, such as ethnic or religious diversity, and adopt a more n intersectional approach to explore their interactive effects with gender on the transition.”
Comment 2: Line 14: When you say “migration engagement of mothers,” you are referring specifically to mothers who are migrating, correct? Can this be reworded for clarity?
Response:
Thank you for your comments. We referred to the phenomenon that mothers are migrating with fathers. To clarify, we revised the term as “mother-involved migration”. We also provided justification of such measurement in the section of Methods (Line. 1167-1170).
“We categorized mother-only migration with both-parents migration referring to the category as mother-involved migrants because the proportion of mother-only migration was extremely small in the sample (<4% at both waves).”
We further considered the difference between mother-involved migration and both-parent migration to rule out the possibility that the effect of mother-involved migration is due to the migration of both parents; and therefore, to highlight the effect of gendered migration. We examined single-parent migration and both-parent migration and did not find any significant associations.
Comment 3: Whose return migration are you referring to here specifically when you say “non-parental”?
Response:
Thank you for the question. We refer to the migrant members attached to the household of index child other than biological parents.(The referenced study differs from many household surveys as the index child is the reference point within the household and the one who is tracked over time). These household members migrated out at Wave 1 while returned at Wave 2. These members include grandparents, parents’ relatives, and other members attached to the household of the index child. We also added the details in the manuscript (Line. 1157-1159).
Comment 4: Line 26: Can this be made more specific with a statistic?
Response:
Thank you for your question. To the best of our knowledge, there are no global statistics regarding the number of left-behind children, but national data for specific countries are available. We added data from UNICEF and UNESCO as examples to illustrate the range of prevalence for left-behind children in a few Asian countries (Line. 27-28).
Comment 5: Line 29: What effects/arguments?
Response:
Thank you for your comment. Following your suggestions, we have revised the sentence as “Research on left-behind children focuses primarily on health, education, and employment outcomes; however, parental absence has compelling effects, as remittances sent by migrant parents provide more educational investment, whereas psychological distance may adversely impact child health.” (Line. 28-32).
Comment 6: Line 34: What type of significant impact?
Response: This is a relevant question. We revised the sentence as
“Their participation in global migration suggests that migration behaviors may be transmitted across generations, which may present unique challenges and opportunities in accommodating the new generation of labor migrants given their experiences of growing up in migrant households.” (Line. 35-38).
Comment 7: Line 36: What do you mean it is independent of parental or household influences?
Response:
Here we are referring to how the current scholarship on youth migration has developed as a separate strand without considering the household effect. To clarify, we have revised the sentence (Line. 39-41) and added more details in the following sentence,
“The existing literature provides a unique theoretical framework that explains their migration behaviors; however, there has been little research situating youth migration in the context of an entire household.” (Line. 41-43).
Comment 8: Line 56: What do you mean when you say “the gender aspect?” and, you make an important point that there are multiple factors beyond gender – other aspects of people’s social locations – that are relevant. Is there space to take a more intersectional approach here?
Response:
Thank you for your question.
We refer to the gender of both migrant parents and their left-behind children when considering the gender aspect in transgenerational migration. The objective of the sentence was to highlight the importance of gendered selection in migration, and we have revised the sentence as
“It is crucial to include a gender perspective, including the gender of both migrant parent and left-behind child, in any examination of transgenerational migration, especially given ongoing debates about migration, gender and social reproduction.” (Line. 114-117)
We also agree that a more intersectional approach should be adopted in future studies to explore the impact of gendered migration when studying the transition from left-behind children to youth labor migrants to highlight how gender roles intersect with other social position factors and are transmitted in the context of transgenerational migration (Line. 1881-1884).
Comment 9: Line 72: When you refer to migration incentives, I wonder about the language of “incentive.” Why did the authors use this language? Is it really an incentive? Perhaps other language (pull factor?) could be useful here.
Response:
Thank you for your comment. Although the term “migration incentives” is used in migration literature across different research contexts (Čajka, Jaroszewicz, and Strielkowski 2014; Heitmueller 2005; Stark and Taylor 1991) to describe reasons and motivations of migration, we agree that the use of migration incentives here might be inaccurate to delineate the reasons behind youth migration examined in our manuscript. We have revised the term to give a more precise description (Line. 131).
Comment 10: Line 89: This sentence is incomplete.
Response:
Thank you for your suggestion. We have revised the sentence (Line. 149-150).
Comment 11: Section 2.1: There are a lot of pieces summarized and defined in this paragraph, and the overall point is a bit jumbled. Suggest it be reviewed and revised for clarity and simplicity?
Response:
Thank you for your suggestion. We have revised the section for more clarity and simplicity.
Comment 12: Line 168: In what ways do women exert agency in the migration decision-making process? How are migration decisions made at the household level, particularly in Thailand?
Response:
Thank you for your question.
Women’s agency in the migration decision-making process is more prevalent among young girls or without entering marriage and/or reproductive age (Hoang 2011). Women may perceive destination societies as providing greater opportunities for earning a living, freedom, marrying, and obtaining an education, especially difficult circum-stances facing women in the sending society may motivate them to leave (Thorsen 2010).
We also added further consideration in the Discussion section to show how migration decisions are made at the household level and presented gender-specific expectations from migrant parents. (Line. 1737-1793)
Comment 13: Lines 187-9: Not clear.
Response:
Thank you for your comment.
We have revised the sentence as follows:
“The relay migration theory emphasizes the use of procedural explanations to explain youth's migration motivations or behaviors in migrant families and they follow the steps of their migrant parents as a replacement. Especially when parents, who constitute the core family structure, have migrated, their children may be forced to migrate in the future.” (Line. 715-719)
Comment 14: Section 3. What types of work are people doing when migrating for work to/from Thailand? What are gender differences? What are age differences? Suggest the addition of more specifics here.
Response:
Thank you for your question. We have added more specifics in this section (Line. 989-993) and highlighted the age of first migration intensity (Line. 1041).
Comment 15: I agree that migration of young people cannot be disentangled from the household. What then, distinguishes migration of young people from other family members? Why is it important to look at their experiences specifically?
Response:
Thank you for your questions.
Young people are increasingly demanding opportunities and solutions that are just, equitable and progressive in their society, which underscores the need to address the multifaceted challenges faced by young people (such as access to education, health, employment, and gender equality). When parents migrate one of the major reasons is to improve the life chances for their children, thus understanding how the experiences of youth may differ based on the migration of their parents can highlight the presence or absence of socio-economic mobility. There continues to be an absence of conclusive research which provides information about this mobility of the next generation who grow up within migrant parent households. This paper seeks, in part, to address this gap in knowledge.
Comment 16: How are the authors defining “youth” or young people and why?
Response:
Thank you for your question. We followed the definition of the United Nations to target people aged 15-24 years old. We also added this in the section of Introduction (Line. 39).
Comment 17: What other non-family network connections (e.g. friends) could have been considered?
Response:
We agree that friends are important components in migrant networks. We also mentioned it when comparing networks tied to men and women (Line. 1863-1866). Unfortunately ,the current study does not have that information available.
Comment 18: Methods: Line 265: What was the replicable protocol? Although it has been published previously, details on the sampling strategy used is needed to enable the readers to understand representativeness.
Response: Thank you for your question.
Thank you for your suggestion. We have introduced more details about the sampling strategy and also added a citation which can give our readers a more comprehensive understanding of the two datasets used in this study (Line. 1067-1063).
Comment 19: Line 273: So a second wave was conducted that followed up on a subset of participants from CHAMPSEA and CLAIM? From CHAMPSEA, there were 1030 households, but only 411 at the second wave. For CLAIM, it was 1456 households, followed by 465 households. How was the decision made to have only about 400 households from each of the previous studies? How were those subsets selected? How were the analyses conducted – were the samples simply combined? How exactly?
Response:
Thank you for your question.
Approximately 1,000 households were interviewed for the baseline study of CHAMSEA in each of the four Southeast Asian countries, and Thailand is one of them. Using a three-stage flexible quota sampling strategy, CHAMPSEA recruited 1,030 eligible households with children between the ages of 3 and 5 or 9 to 11 in Thailand. Using a similar study design, the CLAIM study recruited 1,456 households with a child between the ages of 8 and 15 in 2010. Households initially recruited at baseline were re-invited to participate in the follow-up study in 2019. The second wave specifically targeted around 50% of each of the two prior studies, which had youth in the same birth years 1996-1999. The combination of the samples allowed for specifically examining any potential differences between internal and international migration experiences of parents on the transition from left-behind children to youth labor migrants. Table 1 shows the sample size at two waves, and the follow up subsamples are in bold.
Table1. CHMPSEA and CLAIM dataset
|
Dataset |
W1 Cohort |
W2 Follow-up |
Attrition |
|
CHAMPSEA (N=1030) |
N=509 (Aged 3-5) |
N.A. |
N.A. |
|
N=521 (Aged 9-11) |
N=411 |
21% |
|
|
CLAIM (N=1456) |
N=860 (Early childhood) |
N.A. |
N.A. |
|
N=596 (Comparable cohort) |
N=465 |
22% |
Comment 20: Line 276: It is unclear what you mean about the attrition rate of 20%. Where is the 20% coming from?
Response:
Thank you for your question.
A total of 411 households were followed up by CHAMPSEA at Wave 2 from the baseline study of 521 households, with a 21% attrition rate. CLAIM followed up with 465 households from the 596 households at baseline, with a 22% attrition rate. The statistic has also been added to our manuscript (Line. 1085-1086).
Comment 21: Line 279: What specific reasons for migration were included if education migration was excluded? Could they provide multiple reasons for migration?
Response:
Thank you for your question. The objective of the study was to investigate the transition from left-behind children to youth labor migrants. Considering that their parents were also labor migrants, the study strived to shed light on the reproduction of labor migration between generations. Therefore, we only included those who migrated with the aim of full-time employment, indicating their status as a labor migrant.
The questionnaire consisted of two modules: a questionnaire for the household and a questionnaire for the sampled child. A household questionnaire was completed by the responsible adult, who had the most knowledge of the family's background. These questions included demographic characteristics of the household, a list of residential and non-residential family members, and a summary of the migration history of the father or mother. The index child was tracked along with migrant family members attached to the household in order to determine whether he or she migrated from the original household. The sampled children were asked about their development, education, and employment aspirations and experiences. We asked the children who did not live in their parental household to complete an additional interview via telephone or video call in which they answered questions regarding their: 1) migration history, if they had migrated as labor migrants; 2) current educational or employment status. This enabled us to distinguish between migrants who were driven by labor force participation or migrants who were driven by the pursuit of education. We also added the information in data operation (Line. 1098-1099).
Comment 22: Line 285: What relevant ethics boards?
Response:
Thank you for your comment. We have added more information on ethics boards (Line. 1084).
Comment 23: Line 291: Why is it a different consideration? Please explain a bit here to illuminate your point.
Response:
Thank you for your comment.
We theorize the pursuit of education and the participation in labor force are two divergent pathways for young people. Specifically, upon reaching the age of 14, compulsory schooling ends in Thailand, leaving adolescents with a choice between further education and employment. Some left-behind children benefit from financial support from their migrant parents for educational investment, increasing their human capital in comparison to their non-left-behind peers (Wassink and Viera 2021). Left-behind adolescents, as substitute labor for absent parents, may also need to become more involved in income-producing activities during a shortage of the working population (Skoufias and Parker 2006). When outmigration is prevalent, second-generation migrants often anticipate chain migration as labor migrants to help support their families after leaving home (Durand and Massey 2010).
Comment 24: Line 292: the way this is written is a bit confusing. I think you want to say “whether they were labor migrants or not (0=no, 1=yes).
Response:
Thank you for your comment. We have revised the sentence following your suggestion.
Comment 25: Line 297: Are you really measuring gender dynamics, or simply the sex of parents and children?
Response:
Thank you for your comment. We have revised the term throughout the manuscript. We also updated the title as “From left-behind children to youth labor migrants: The impact of household networks, gendered migration, and relay migration in Southeast Asia” for consistency.
Comment 26: Section 3.2.3. more detail about parental occupation and economic circumstances and how those were operationalized is recommended to be added.
Response:
Thank you for your comment. We have added details about parental occupation and household economic circumstances (Line. 1199-1211).
Comment 27: Section 3.3. what do you mean that you “apply stepwise regression?” what type of regression models were used? I am a little confused about how you analyzed the data longitudinally. How was model fit assessed for analyses conducted?
Response:
Thank you for your comment.
Stepwise regression is essentially a process by which variables are added to the forward selection process in order to obtain a satisfactory regression equation. By its most common definition, stepwise regression is a method of iteratively searching and comparing models to determine which independent variables, previously thought to be of some significance, are most strongly associated with a dependent variable (Henderson and Denison 1989). This technique typically involves exploring a number of alternative models, with variables added or dropped based on their statistical importance (Draper and Smith 1998).
Although two-waves of data were collected, the status of youth labor migration was only reflected at Wave 2. Therefore, we mainly used logit regression to assess the transition from left-behind children to youth labor migrants. We also clarified this in the section of analytical approach (Line. 1251).
We reported log likelihood and McFadden R2 to assess the fitness of logit models.
Comment 28: Line 337: become labor migrants ever, or become labor migrants between wave 1 and wave 2?
Response:
Thank you for your question. Our data reported the changes between two waves. We have revised the sentence for more clarity (Line. 1262).
Comment 29: Line 343-4: More explanation of this proportion you report is important to consider.
Response:
Thank you for your comment.
We have added more interpretation to supplement our explanation (Line. 1269-1271).
Comment 30: Table 1. I recommend only reporting to 1 decimal place. For control variables, are you reporting the reference category? It might be clearer to just write (Male) rather than (Gender (0=male)). For marriage history, what is included in the bucket of “having a marriage history”?
Response:
Thank you for your suggestion.
We have revised Table 1 to 1 decimal place.
For control variables, we reported the group of interest by showing the reference group in binary variables. For instance, Gender (0=male): 42.9% means that gender is a binary variable (0=male, 1=female), and the percentage of female in the sample is 42.9%. For further clarity, we have revised Table 1 to show the percentage instead.
For marital status, we have added more details about the conceptualization and operations in the section of control variables (Line. 1199-1203).
“Marital status. The marital status of the index child was recorded as a categorical variable (1=single, 2=living together, 3=married, 4=divorced). We further recoded the variable into a binary outcome to indicate whether they had a marriage history as the proportion of living together and divorced was comparatively small (0=single or living together, 1=married or divorced).”
Comment 31: Line 369: I would recommend reconsidering the use of the language “gender dynamic,” as the results suggest simply that the effect of the gender of the parent or child is important.
Response:
Thank you for your comment. We have revised the term throughout the manuscript and title as noted above.
Comment 32: Discussion:
Tables 2 and 3 seem to be presenting log odds – is there a reason why odds ratios and 95% confidence intervals are not presented?
Response:
Thank you for your comment.
We agree that presenting log odds and confidence interval is important in logistic regression. We also understand Social Sciences is an interdisciplinary journal, so we browsed the journal with a focus on migration studies. Many studies on migration present coefficients and standard errors (e.g., Wikle and Ackert 2022). In our study, Table 2 and 3 presented the results of logit regression; therefore, we reported regression coefficients and standard errors (S.E.) of logit models in line with these other papers.
Comment 33: Line 402: this missingness not at random is an important consideration for interpreting the results. Once the authors identified this association, what specific sensitivity analyses or sub-analyses did you do (perhaps looking at that group at wave 1 to see if the trends were consistent with those who were not lost to follow up?)
Response:
Thank you for your comment. We conducted an attrition analysis after taking into account both the characteristics of the child and the household. Our study examined the following characteristics of children: age, education, psychological well-being, and physical health. The characteristics of parents were analyzed based on their age, education, occupation, and migration status. For household characteristics, we examined household size (number of people), the number of young children and the elderly, as well as household wealth. It was, however, only the migration status of the parents that significantly predicted sample attrition, with households with migrant parents at Wave 1 being more likely to drop out. We also noted this limitation.
“As a consequence of this attrition, our study may underestimate the impact of parental migration.” (Line. 1652-1653)
Comment 34: Discussion:
What qualitative research would the authors recommend to understand and unpack the hypotheses that were supported and refuted in their analyses? For example. The authors write “further studies can enhance understanding of the dynamics,” (line 465). What further studies? Similar point in reference to the effects of community and its potential migration culture – see likes 544-8. What studies should be done?
Response:
Thank you for your comment.
We have added more details about the suggestion in the discussion (Line. 1727-1730, 1737-1793).
Comment 35: Line 476: I think there is an assumption here about migration decision-making that needs to be considered. The authors emphasize that migration of youth is influence by multiple factors at the household level. Is it really that youth labor migrants may not be motivated for relay migration?
Response:
Thank you for your comment.
As opposed to the theoretical foundation of relay migration, which suggests that the last generation of migrants returns while the new generation migrates as a replacement, our results do not provide robust evidence for the pattern of relay migration. As an alternative, two explanations were proposed. Firstly, the returned migrants may not have completely withdrawn from the labor market and may continue to perform economic activities for the household. Secondly, young migrants may have higher occupational aspirations than their parents, rather than following the occupational structure of their parents. We also added information from ethnographic research to support our argument (Line. 1836-1838).
Comment 36: How might destination, duration, frequency be important to consider in thinking about the analyses and results presented her? How might associations vary by these characteristics of migration?
Response:
Thank you for your comment.
We also examined whether the transition was associated with migration destination (for example, internal versus international migration), duration of parental migration, and migration history to record circulate migration. However, the results showed that the effects of these factors were not statistically significant. Furthermore, the objective of the study was to examine the three theoretical frameworks, the discussion on these factors might shift the major focus of our research.
Comment 37: Paragraph beginning on line 519: Can you elaborate more fully on this as a limitation? The point is important, but not exactly clear why is it presented as a limitation.
Response:
Thank you for your comment.
Migration decision-making is greatly conditioned by marriage and fertility status. In our sample, these young adults are mostly single without children, their migration decisions are largely influenced by individual factors, as well as interactions with their parents and other extended family members. Compared to their migrant parents that we captured after they had at least one child whose migration aspirations were mainly the investment of children’s education. , these youth were in the early stages of adulthood and having not yet faced other key transitions which may influence their migration decisions without considering childcare and other parental responsibilities.
We have added more details about the limitation for more clarification (Line. 1866-1870).
Comment 38: I would appreciate further contextualization of the findings to the Thai context in the discussion section. It is done a bit, but the section would be stronger if done more rigorously.
Response:
Thank you for your comment. We have added more discussion on the Thai context in different section (Line. 1737-1740, 1836-1839, and 1866-1867).
Comment 39: Please incorporate a discussion of limitations of the sampling method and generalizability.
Response:
Thank you for your comment.
We have added this point following your suggestion in the section on limitations (Line. 1894-1903).
Reviewer 3 Report
This paper analyzes if household members' migration experience affects the migration status of the youth in Thailand. This topic is essential to understand the migration decisions of the new generation in many developing countries which have experienced a large scale of international or domestic migration in recent decades. Still, it is understudied in the academic literature. Therefore, this paper can potentially make a meaningful contribution to migration study. However, the empirical analyses have severe weaknesses, and the conclusions are not convincing. Below I will delineate my concerns about the paper.
- The narratives in Section 2.2. "Gendered migration and family gender dynamics" only provide weak support for Hypothesis 2.1. and do not support Hypothesis 2.2. As argued in paragraph two on page 4, a mother's migration experience might encourage her daughter's migration. Therefore it is not proper to hypothesize, "The migration of young males into the labor force is more likely to be affected by 176 parental migration than that of young females."
- The concerns about the empirics:
- This paper makes no effort to address the migration selection of household members, implicitly assuming that their migration decisions are exogenous to the migration status of their children in the later years. This approach runs against the argument that migration is often a household decision, not an individual one, implying that parents' migration decisions can be related to factors determining children's migration status. For example, a mother may migrate out for work to earn additional income because her daughter is sick. The illness may prevent the daughter from migrating. In this case, the migration decisions of mother and daughter are correlated in a spurious sense. Therefore, one must address such endogeneity issues to obtain a causal inference from parent migration to child migration.
- The analysis sample only includes the index children (one child of each household is qualified for the study). The siblings' migration status information seems to be used to code other household members' migration status. It is not clear to me why the authors don't treat children's migration decisions within a household the same by separating an index child from the rest. As the siblings of the index children are not included as children of the parents, the sample selection could lead to an underestimate of the impact of parent migration.
- Theoretically and empirically, one cannot separate relay migration from network migration. A child can migrate out after a parent returns, which would be counted as "relay migration." But the child could still benefit from the network due to the parent's migration. Hence, the analysis of relay migration is weak.
Author Response
Response to Reviewer 3
General Comment: This paper analyzes if household members' migration experience affects the migration status of the youth in Thailand. This topic is essential to understand the migration decisions of the new generation in many developing countries which have experienced a large scale of international or domestic migration in recent decades. Still, it is understudied in the academic literature. Therefore, this paper can potentially make a meaningful contribution to migration study. However, the empirical analyses have severe weaknesses, and the conclusions are not convincing. Below I will delineate my concerns about the paper.
Response:
Thank you for your suggestion. We have addressed your comments and suggestions point-to-point as follows.
Comment 1: The narratives in Section 2.2. "Gendered migration and family gender dynamics" only provide weak support for Hypothesis 2.1. and do not support Hypothesis 2.2. As argued in paragraph two on page 4, a mother's migration experience might encourage her daughter's migration. Therefore it is not proper to hypothesize, "The migration of young males into the labor force is more likely to be affected by 176 parental migration than that of young females."
Response:
Thank you for your suggestion.
We agree that the current theoretical frameworks provide inadequate support for H 2.1. After consideration of your observations, we have revised the literature review to highlight the impacts of parental engagement in the migration decisions of children based on child gender and H2.1 hypothesis as follows:
“The gendered impact of parental migration is stronger for daughters than sons.” (Line. 697)
Comment 2: The concerns about the empirics:
This paper makes no effort to address the migration selection of household members, implicitly assuming that their migration decisions are exogenous to the migration status of their children in the later years. This approach runs against the argument that migration is often a household decision, not an individual one, implying that parents' migration decisions can be related to factors determining children's migration status. For example, a mother may migrate out for work to earn additional income because her daughter is sick. The illness may prevent the daughter from migrating. In this case, the migration decisions of mother and daughter are correlated in a spurious sense. Therefore, one must address such endogeneity issues to obtain a causal inference from parent migration to child migration.
Response:
Thank you for your comment.
We agree that the issue of selection bias is a critical point. In the two studies of CHAMPSEA and CLAIM, we determined the migration status of household members based on household roster and tracked migration history of both parents, which are frequently used to address selection bias in migration studies (e.g., Carletto, Covarrubias, and Maluccio 2011; Mu and de Brauw 2015).
The study design includes a ‘control’ group of same age children within the sample communities for comparison with the households with parent migrants. Specifically, we included children in both projects whose parents had not been migrants in the last six months and lived in the same communities.
In the revision we have sought to better clarify that the current study was intended to examine the relationship between household migration, both parental and non-parental, and child migration without establishing a causal relationship.
Comment 3: The analysis sample only includes the index children (one child of each household is qualified for the study). The siblings' migration status information seems to be used to code other household members' migration status. It is not clear to me why the authors don't treat children's migration decisions within a household the same by separating an index child from the rest. As the siblings of the index children are not included as children of the parents, the sample selection could lead to an underestimate of the impact of parent migration.
Response:
Thank you for your comment.
We agree that this is also a limitation derived from our sampling strategy. The index child was randomly selected if there was more than one child in each qualifying household that met the inclusion criteria after accounting for the quota design of child age and gender. Therefore, it is not entirely impossible that the sibling effect could be compromised in this study.
We also agree that the migration decision-making of parents is holistic, taking into account household circumstances and all children living in the household. However, we have gathered information on the migration history of index children only to ensure that they have been left behind once in the past because the purpose of the study is to examine the transition from left-behind children to youth labor migrants, which might understate the effects of parental migration.
We discussed the limitations and also proposed directions for future studies (Line. 1889-1894).
Comment 4: Theoretically and empirically, one cannot separate relay migration from network migration. A child can migrate out after a parent returns, which would be counted as "relay migration." But the child could still benefit from the network due to the parent's migration. Hence, the analysis of relay migration is weak.
Response:
Thank you for your comment.
We agree that we cannot completely rule out the possibility that children can benefit from their migrant parents even when these parents returned. In this study, we intentionally distinguished the effects of migrant networks and relay migration with the consideration of theoretical foundation.
Migrant networks refer to the social ties that connect individuals in sending and host contexts. Therefore, migrant networks were operationalized as current migrant household members in this study. Several studies also indicate that migrant networks might be significantly compromised in the case of return migration (Apsite‐Berina, Manea, and Berzins 2020; Aranda 2003; Duval 2004). Furthermore, the study also aims to offer an explorative lens by examining three theoretical frameworks separately to highlight the application of each of them.
We also agree that an intersectional approach will be of great help in future studies to enhance our understanding of the migration behaviors of left behind children and mechanism behind (Line. 1872-1875).
Round 2
Reviewer 2 Report
Thank you to the authorship team for their detailed responses to the concerns presented. I am happy to see the changes in language related to gendered migration away from gender dynamics as well as to see the additional details added to the context around migration in Thailand as well as in the methods and in the results and discussion sections. I have a few smaller comments based on my most recent read through that I would like to highlight, but overall the manuscript is much improved.
Language/Formatting:
Some of the sections in the new manuscript are now not numbered correctly (e.g., under 4. Data and Methods, the first section is 3.1. Data.) This should be updated.
With the revisions, there is a need for the authors to read the revised sentences for clarity. There are a number of examples of places where the sentences now do not make sense or have typos. For example, p. 17, line 668-671: spaces missing, adulthood has a typo, etc., but there are other places throughout as well.
Methods:
Thank you for the detailed description of the sample sizes and attrition via a table in the response to reviewers. It appears as though these sample sizes have been included in the revised manuscript, but the table is quite clear. I might suggest the authors consider incorporating this table into the methods to simplify the paragraph in 3.1.– is all that in the methods?
The authors used the words logistic and logit regression in the manuscript. Please clarify. In addition, while I defer to the journal guidelines as to whether log odds or odds ratios are appropriate, I would prefer to present odds ratios rather than log odds.
Thank you for adding further details about how stepwise regression was employed, but further clarity on this process is needed. Were any variables considered not retained in the final models? Please clarify how the authors assessed collinearity and model fit in the final models presented.
In the analytical approach section, further description of the analysis and what data from which waves is included in the final models would be helpful. The explanation in response to the authors is helpful, and further elaboration in the text would help clarify this for the readers.
Results:
Tables: I appreciate the revisions to the tables, but for table 2 and table 3, the reference categories do still need to be shown so the interpretation of the coefficients is clear. What I would suggest is that for the variable, for example “Child education,” the authors do something like: child education (upper secondary or above vs. secondary or below).
Author Response
please see attachment for detailed responses
Author Response File: 
Author Response.pdf
