Manuel Sacramento Gonzalez Canche
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Manuel Sacramento Gonzalez Canche has a total of 24 co-authors. Here are the top co-authors by number of shared publications:
Chelsea Zhang
University of Pennsylvania, Philadelphia, USA
6 shared publicationsYunhao Liang
Shanghai American School, Shanghai, China
3 shared publications- Ji Yeon Bae
University of Pennsylvania, Higher Education and Critical Quantitative Modeling
2 shared publications Kaiwen Zheng
Harvard University, Cambridge, USA
2 shared publicationsAdam K. Atwell
Mitchell Community College
1 shared publicationAlex Monday
University of Georgia
1 shared publication
Latest Publications
- Open Access
- 8 June 2026
- Published by Elsevier in SSRN Electronic Journal
Intelligent Systems for Academic Research Integration: An Offline and Multimodal Brainstorming Partner for Scholarly Inquiry
Artificial intelligence in education (AIEd) has rapidly expanded through applications centered on tutoring, feedback, personalization, writing assistance, and human–AI collaboration. Recent work has begun to examine generative AI as a brainstorming partner, a writing assistant, and a complementary collaborator in educational contexts. Yet a major gap remains: the field has paid far less attention to the infrastructures required for scholars to think with AI while preserving data privacy, interpretive authority, and institutional trust. We address this gap with Intelligent Systems for Academic Research Integration (ISARI), a fully offline, open-source, multimodal brainstorming partner designed to support—not replace—scholarly inquiry. We advance three central arguments. First, AIEd discourse overemphasizes cloud-based tutoring and writing tools while under-examining research-facing infrastructure. Second, privacy, data stewardship, and trust must be treated as fundamental design variables, not afterthoughts. Third, a local AI thinking partner can broaden access to data science for qualitative and mixed-methods researchers, while helping quantitative scholars rigorously integrate qualitative evidence. Building on work in human–AI brainstorming and offline-first architectures, we argue that the next frontier in AIEd should include tools for interpretation, memoing, comparison, synthesis, and evidence-grounded writing—all without uploading sensitive materials. These infrastructures must also handle multimodal evidence: images, charts, diagrams, scanned pages, screenshots, temporal visualizations, and geo-contextualized displays, to remain methodologically useful. ISARI is available for macOS and Windows (https://cutt.ly/ltJuvTEL). Its smallest model runs on a computer with at least 8 GB RAM and 340 GB of storage. The tool has been peer-reviewed by Code Ocean (https://cutt.ly/3tJrslhS) to ensure quality, rigor, and safety.
- 8 June 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessIntelligent Systems for Academic Research Integration: An Offline and Multimodal Brainstorming Partner for Scholarly Inquiry
Artificial intelligence in education (AIEd) has rapidly expanded through applications centered on tutoring, feedback, personalization, writing assistance, and human–AI collaboration. Recent work has begun to examine generative AI as a brainstorming partner, a writing assistant, and a complementary collaborator in educational contexts. Yet a major gap remains: the field has paid far less attention to the infrastructures required for scholars to think with AI while preserving data privacy, interpretive authority, and institutional trust. We address this gap with Intelligent Systems for Academic Research Integration (ISARI), a fully offline, open-source, multimodal brainstorming partner designed to support—not replace—scholarly inquiry. We advance three central arguments. First, AIEd discourse overemphasizes cloud-based tutoring and writing tools while under-examining research-facing infrastructure. Second, privacy, data stewardship, and trust must be treated as fundamental design variables, not afterthoughts. Third, a local AI thinking partner can broaden access to data science for qualitative and mixed-methods researchers, while helping quantitative scholars rigorously integrate qualitative evidence. Building on work in human–AI brainstorming and offline-first architectures, we argue that the next frontier in AIEd should include tools for interpretation, memoing, comparison, synthesis, and evidence-grounded writing—all without uploading sensitive materials. These infrastructures must also handle multimodal evidence: images, charts, diagrams, scanned pages, screenshots, temporal visualizations, and geo-contextualized displays, to remain methodologically useful. ISARI is available for macOS and Windows (https://cutt.ly/ltJuvTEL). Its smallest model runs on a computer with at least 8 GB RAM and 340 GB of storage. The tool has been peer-reviewed by Code Ocean (https://cutt.ly/3tJrslhS) to ensure quality, rigor, and safety.
- Open Access
- 8 June 2026
- Published by Elsevier in SSRN Electronic Journal
Intelligent Systems for Academic Research Integration: An Offline and Multimodal Brainstorming Partner for Scholarly Inquiry
Artificial intelligence in education (AIEd) has rapidly expanded through applications centered on tutoring, feedback, personalization, writing assistance, and human–AI collaboration. Recent work has begun to examine generative AI as a brainstorming partner, a writing assistant, and a complementary collaborator in educational contexts. Yet a major gap remains: the field has paid far less attention to the infrastructures required for scholars to think with AI while preserving data privacy, interpretive authority, and institutional trust. We address this gap with Intelligent Systems for Academic Research Integration (ISARI), a fully offline, open-source, multimodal brainstorming partner designed to support—not replace—scholarly inquiry. We advance three central arguments. First, AIEd discourse overemphasizes cloud-based tutoring and writing tools while under-examining research-facing infrastructure. Second, privacy, data stewardship, and trust must be treated as fundamental design variables, not afterthoughts. Third, a local AI thinking partner can broaden access to data science for qualitative and mixed-methods researchers, while helping quantitative scholars rigorously integrate qualitative evidence. Building on work in human–AI brainstorming and offline-first architectures, we argue that the next frontier in AIEd should include tools for interpretation, memoing, comparison, synthesis, and evidence-grounded writing—all without uploading sensitive materials. These infrastructures must also handle multimodal evidence: images, charts, diagrams, scanned pages, screenshots, temporal visualizations, and geo-contextualized displays, to remain methodologically useful. ISARI is available for macOS and Windows (https://cutt.ly/ltJuvTEL). Its smallest model runs on a computer with at least 8 GB RAM and 340 GB of storage. The tool has been peer-reviewed by Code Ocean (https://cutt.ly/3tJrslhS) to ensure quality, rigor, and safety.
- 8 June 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessIntelligent Systems for Academic Research Integration: An Offline and Multimodal Brainstorming Partner for Scholarly Inquiry
Artificial intelligence in education (AIEd) has rapidly expanded through applications centered on tutoring, feedback, personalization, writing assistance, and human–AI collaboration. Recent work has begun to examine generative AI as a brainstorming partner, a writing assistant, and a complementary collaborator in educational contexts. Yet a major gap remains: the field has paid far less attention to the infrastructures required for scholars to think with AI while preserving data privacy, interpretive authority, and institutional trust. We address this gap with Intelligent Systems for Academic Research Integration (ISARI), a fully offline, open-source, multimodal brainstorming partner designed to support—not replace—scholarly inquiry. We advance three central arguments. First, AIEd discourse overemphasizes cloud-based tutoring and writing tools while under-examining research-facing infrastructure. Second, privacy, data stewardship, and trust must be treated as fundamental design variables, not afterthoughts. Third, a local AI thinking partner can broaden access to data science for qualitative and mixed-methods researchers, while helping quantitative scholars rigorously integrate qualitative evidence. Building on work in human–AI brainstorming and offline-first architectures, we argue that the next frontier in AIEd should include tools for interpretation, memoing, comparison, synthesis, and evidence-grounded writing—all without uploading sensitive materials. These infrastructures must also handle multimodal evidence: images, charts, diagrams, scanned pages, screenshots, temporal visualizations, and geo-contextualized displays, to remain methodologically useful. ISARI is available for macOS and Windows (https://cutt.ly/ltJuvTEL). Its smallest model runs on a computer with at least 8 GB RAM and 340 GB of storage. The tool has been peer-reviewed by Code Ocean (https://cutt.ly/3tJrslhS) to ensure quality, rigor, and safety.
- Top Cited PapersOpen Access
- 19 January 2026
- Published by SAGE Publications in International Journal of Qualitative Methods
- Vol. 25
Network Analysis of Qualitative Data (NAQD): An Analytic Framework and Software for Qualitative Data Science Visualization and Hypothesis Testing
Network analysis offers a powerful methodological approach for uncovering patterns and structures within complex relational data. This study argues that qualitatively or quantitatively coded textual information can be conceptualized as a network of relationships between participants and the codes applied to their classified narratives/contributions, thereby forming a natural relational database. The main argument is that data gathered through qualitative methods (e.g., interviews, focus groups) and subsequently coded/classified/labeled contain mathematical structures that can be retrieved, visualized, and statistically analyzed using network methods— even if we are not analyzing social relationships . Furthermore, when participant attributes are incorporated, the resulting structures enable statistical evaluation of the extent to which experiences differ across groups or are similarly distributed across our participants’ attributes. This study introduces the Network Analysis of Qualitative Data (NAQD) framework, which integrates quantitative, mathematical, and qualitative principles to analyze textual data—an approach whose potential for rigorous statistical hypothesis testing remains largely unrealized in common qualitative software. Alongside presenting this analytic framework, and with the goal of democratizing access to data science tools, we introduce a peer-reviewed, free, and no-code software tool that implements NAQD (see https://cutt.ly/2riYDEhH for the published version of the software). However, this software is also freely available for Mac ( https://cutt.ly/unZDUMq ) and Windows ( https://cutt.ly/KnL9frz ) platforms. Unlike commercially available software, NAQD provides community detection, hypothesis testing of group similarities via Quadratic Assignment Procedures (QAP), and fully interactive HTML network renderings. We illustrate all steps and outputs using replication data from a project on teacher training at Minority Serving Institutions.
- 19 January 2026
- Published by SAGE Publications in International Journal of Qualitative Methods
- Vol. 25
Top Cited PapersOpen AccessNetwork Analysis of Qualitative Data (NAQD): An Analytic Framework and Software for Qualitative Data Science Visualization and Hypothesis Testing
Network analysis offers a powerful methodological approach for uncovering patterns and structures within complex relational data. This study argues that qualitatively or quantitatively coded textual information can be conceptualized as a network of relationships between participants and the codes applied to their classified narratives/contributions, thereby forming a natural relational database. The main argument is that data gathered through qualitative methods (e.g., interviews, focus groups) and subsequently coded/classified/labeled contain mathematical structures that can be retrieved, visualized, and statistically analyzed using network methods— even if we are not analyzing social relationships . Furthermore, when participant attributes are incorporated, the resulting structures enable statistical evaluation of the extent to which experiences differ across groups or are similarly distributed across our participants’ attributes. This study introduces the Network Analysis of Qualitative Data (NAQD) framework, which integrates quantitative, mathematical, and qualitative principles to analyze textual data—an approach whose potential for rigorous statistical hypothesis testing remains largely unrealized in common qualitative software. Alongside presenting this analytic framework, and with the goal of democratizing access to data science tools, we introduce a peer-reviewed, free, and no-code software tool that implements NAQD (see https://cutt.ly/2riYDEhH for the published version of the software). However, this software is also freely available for Mac ( https://cutt.ly/unZDUMq ) and Windows ( https://cutt.ly/KnL9frz ) platforms. Unlike commercially available software, NAQD provides community detection, hypothesis testing of group similarities via Quadratic Assignment Procedures (QAP), and fully interactive HTML network renderings. We illustrate all steps and outputs using replication data from a project on teacher training at Minority Serving Institutions.
- Open Access
- 1 April 2025
- Published by SAGE Publications in International Journal of Qualitative Methods
- Vol. 24
Sentiment and Emotion Network Analysis—SENA: An Analytic Framework and No-Code Software to Integrate Natural Language Processing, Linguistic Analysis, and Data Science with Qualitative and Mixed-Methods Research
Sentiment and emotion analyses (SEA) provide us with the opportunity to gain objective and fast insights into the experiences of vast amounts of individuals. Despite SEA’s efficiency, two main limitations associated with its use remain: outcome aggregation and narrow applicability. Outcome aggregation means that outputs do not consider participants’ attributes nor allow for individual analyses of participants’ experiences and/or the degree to which these experiences may express consistent or contradictory emotions (i.e., emotional entropy). Narrow applicability means that SEA has primarily been employed in market research to understand customers’ satisfaction thus foregoing its applicability in qualitative and mixed methods academic research. With these opportunities and limitations in mind, our study was designed to offer an analytic framework that avoids outcome aggregation and expands the use of SEA beyond market research. As part of our study’s contribution, we offer a free to use and free to distribute software tool that implements the totality of our proposed analytic framework. This software does not require any programming proficiency and executes all required back-end processes locally—thus avoiding the need to upload our data to any servers. In our proposed framework, outcome aggregation is avoided via network modeling that also enables testing for potential differences in emotions distributions across groups. Based on this integration, we refer to the resulting analytic framework and software as Sentiment and Emotion Network Analysis—SENA. As part of our methodological and research design discussion, we also illustrate how to integrate SENA with machine learning text classification of documents or open-ended responses. This integration serves to leverage the power of artificial intelligence in studies relying on SENA. The databases used to showcase this analytic framework are provided so that researchers may interact firsthand with the SENA framework and its externally peer-reviewed software ( https://cutt.ly/BrhEnyfY ) available in Mac ( https://cutt.ly/QwhYruBr ) and Windows ( https://cutt.ly/YwhJJKvO ) operating systems.
- 1 April 2025
- Published by SAGE Publications in International Journal of Qualitative Methods
- Vol. 24
Open AccessSentiment and Emotion Network Analysis—SENA: An Analytic Framework and No-Code Software to Integrate Natural Language Processing, Linguistic Analysis, and Data Science with Qualitative and Mixed-Methods Research
Sentiment and emotion analyses (SEA) provide us with the opportunity to gain objective and fast insights into the experiences of vast amounts of individuals. Despite SEA’s efficiency, two main limitations associated with its use remain: outcome aggregation and narrow applicability. Outcome aggregation means that outputs do not consider participants’ attributes nor allow for individual analyses of participants’ experiences and/or the degree to which these experiences may express consistent or contradictory emotions (i.e., emotional entropy). Narrow applicability means that SEA has primarily been employed in market research to understand customers’ satisfaction thus foregoing its applicability in qualitative and mixed methods academic research. With these opportunities and limitations in mind, our study was designed to offer an analytic framework that avoids outcome aggregation and expands the use of SEA beyond market research. As part of our study’s contribution, we offer a free to use and free to distribute software tool that implements the totality of our proposed analytic framework. This software does not require any programming proficiency and executes all required back-end processes locally—thus avoiding the need to upload our data to any servers. In our proposed framework, outcome aggregation is avoided via network modeling that also enables testing for potential differences in emotions distributions across groups. Based on this integration, we refer to the resulting analytic framework and software as Sentiment and Emotion Network Analysis—SENA. As part of our methodological and research design discussion, we also illustrate how to integrate SENA with machine learning text classification of documents or open-ended responses. This integration serves to leverage the power of artificial intelligence in studies relying on SENA. The databases used to showcase this analytic framework are provided so that researchers may interact firsthand with the SENA framework and its externally peer-reviewed software ( https://cutt.ly/BrhEnyfY ) available in Mac ( https://cutt.ly/QwhYruBr ) and Windows ( https://cutt.ly/YwhJJKvO ) operating systems.
- Open Access
- 17 March 2025
- Published by SAGE Publications in International Journal of Qualitative Methods
- Vol. 24
GeoStoryTelling: An Integrative Multi-Media Methodology and No-Code, Cost-Free Software to GeoContextualize Ethnographic, Health, and Survey Evidence
Our life stories evolve in specific and contextualized places. Although our households may be our primarily shaping environment, our neighborhoods expose us to the immediate “real world” outside home. Indeed, the places where we are currently experiencing and have experienced life play fundamental roles in the development of our identities, beliefs, fears, perceptions of the world, and even in the (mis)understanding of our prospects of social mobility. Despite the analytic benefits that knowledge of the places and contexts where we experience life brings to better understanding our life stories and health outcomes, most qualitative, health, and mixed methods researchers continue to forego the analytic and elucidating power that geo-contextualizing our participants' narratives bring to our analyses, research findings, conclusions, and plans of action. From this view then, most qualitative, health, and mixed methods research findings and conclusions may have been ignoring the spatial contexts that most likely have shaped the experiences of research participants. The main reason for the underuse of these geo-contextualized stories is the requirement of specialized training in geographical information systems (GIS) and/or computer and statistical programming along with the absence of cost-free and user-friendly geo-visualization tools that may allow non-GIS experts to benefit from geo-contextualized outputs. To address this gap, we present GeoStoryTelling , an analytic framework and user-friendly, cost-free, multi-platform software that enables researchers to visualize their geo-contextualized qualitative narratives. The use of this peer-reviewed software (see https://doi.org/10.24433/CO.7385885.v3 ), available in Mac ( https://cutt.ly/E724wAx ) and Windows ( https://cutt.ly/J724f0C ) operating systems, does not require users to learn GIS nor computer/statistical programming to obtain state-of-the-art, and visually appealing and interactive maps. These resulting HTML visualizations are capable of integrating a variety of multi-media inputs that include text, image, videos, sound recordings/music, as well as hyperlinks to other websites—see https://cutt.ly/k7X9tfN . Since GeoStoryTelling relies on OpenStreetMap , its applications may be used worldwide, thus strengthening its potential reach to the qualitative, mixed methods, ethnographic, and health scientific communities, regardless of location around the world.
- 17 March 2025
- Published by SAGE Publications in International Journal of Qualitative Methods
- Vol. 24
Open AccessGeoStoryTelling: An Integrative Multi-Media Methodology and No-Code, Cost-Free Software to GeoContextualize Ethnographic, Health, and Survey Evidence
Our life stories evolve in specific and contextualized places. Although our households may be our primarily shaping environment, our neighborhoods expose us to the immediate “real world” outside home. Indeed, the places where we are currently experiencing and have experienced life play fundamental roles in the development of our identities, beliefs, fears, perceptions of the world, and even in the (mis)understanding of our prospects of social mobility. Despite the analytic benefits that knowledge of the places and contexts where we experience life brings to better understanding our life stories and health outcomes, most qualitative, health, and mixed methods researchers continue to forego the analytic and elucidating power that geo-contextualizing our participants' narratives bring to our analyses, research findings, conclusions, and plans of action. From this view then, most qualitative, health, and mixed methods research findings and conclusions may have been ignoring the spatial contexts that most likely have shaped the experiences of research participants. The main reason for the underuse of these geo-contextualized stories is the requirement of specialized training in geographical information systems (GIS) and/or computer and statistical programming along with the absence of cost-free and user-friendly geo-visualization tools that may allow non-GIS experts to benefit from geo-contextualized outputs. To address this gap, we present GeoStoryTelling , an analytic framework and user-friendly, cost-free, multi-platform software that enables researchers to visualize their geo-contextualized qualitative narratives. The use of this peer-reviewed software (see https://doi.org/10.24433/CO.7385885.v3 ), available in Mac ( https://cutt.ly/E724wAx ) and Windows ( https://cutt.ly/J724f0C ) operating systems, does not require users to learn GIS nor computer/statistical programming to obtain state-of-the-art, and visually appealing and interactive maps. These resulting HTML visualizations are capable of integrating a variety of multi-media inputs that include text, image, videos, sound recordings/music, as well as hyperlinks to other websites—see https://cutt.ly/k7X9tfN . Since GeoStoryTelling relies on OpenStreetMap , its applications may be used worldwide, thus strengthening its potential reach to the qualitative, mixed methods, ethnographic, and health scientific communities, regardless of location around the world.
- Open Access
- 17 March 2025
- Published by Springer Nature in Research in Higher Education
- Vol. 66 (3) , 1-46
The College Transfer and Articulation Network: How are These Statewide Policies and Bilateral or Dyadic Partnerships Structured Across the United States?
Every academic year, millions of college students change institutions before degree completion, confronting the challenge of validating credits across colleges. Despite state-level efforts to legislate strategies for smoother transfers, actual credit recognition relies on non-state-regulated bilateral partnerships that are more (i.e., articulation agreements) or less (general transfer partnerships) specific as policies or guides to avoid credit loss. This study comprehensively sheds light on the USA nationwide structure of transfer and articulation agreements by focusing on statewide policies and in-state and out-of-state informal bilateral partnerships as units of analyses. The spatial configuration of both statewide policies and institutionally driven partnerships enabled testing for economic spillovers as well as measuring whether distance is a factor that may impact the formation of these partnerships. Data were retrieved from CollegeTransfer.Net (N = 18,260 partnerships and 1163 colleges), the Education Commission of the States (118 statewide policies), the IPEDs, and the US Census Bureau. Findings at the state-level revealed economic spillovers in two of four statewide policies, highlighting greater structure of program-specific articulation agreements over general transfer partnerships (i.e., agreements that do not require program continuation). Regarding institutionally driven agreements, the analyses indicated that general partnerships were the most prevalent form, which, compared to more structured articulation efforts, may be less effective in the avoidance of credit loss. We also found that shorter distances are a significant but impractical partnership-forming factor, for the average distance reduction among partnering colleges is 30 miles across models. Combining state and institutional datasets, we found that neither individual nor combined statewide policies actively predict institutional partnership formation. All databases and code created (statewide policies: https://cutt.ly/uwHyvkWQ, institutionally driven agreements: https://cutt.ly/7wHtPkEA, replication codes: https://cutt.ly/JwGRmVDu, https://cutt.ly/EwG1VbaW) may be used in future analyses to address questions of transfer effectiveness and transferring financial costs, which although important go beyond the scope of our study.
- 17 March 2025
- Published by Springer Nature in Research in Higher Education
- Vol. 66 (3) , 1-46
Open AccessThe College Transfer and Articulation Network: How are These Statewide Policies and Bilateral or Dyadic Partnerships Structured Across the United States?
Every academic year, millions of college students change institutions before degree completion, confronting the challenge of validating credits across colleges. Despite state-level efforts to legislate strategies for smoother transfers, actual credit recognition relies on non-state-regulated bilateral partnerships that are more (i.e., articulation agreements) or less (general transfer partnerships) specific as policies or guides to avoid credit loss. This study comprehensively sheds light on the USA nationwide structure of transfer and articulation agreements by focusing on statewide policies and in-state and out-of-state informal bilateral partnerships as units of analyses. The spatial configuration of both statewide policies and institutionally driven partnerships enabled testing for economic spillovers as well as measuring whether distance is a factor that may impact the formation of these partnerships. Data were retrieved from CollegeTransfer.Net (N = 18,260 partnerships and 1163 colleges), the Education Commission of the States (118 statewide policies), the IPEDs, and the US Census Bureau. Findings at the state-level revealed economic spillovers in two of four statewide policies, highlighting greater structure of program-specific articulation agreements over general transfer partnerships (i.e., agreements that do not require program continuation). Regarding institutionally driven agreements, the analyses indicated that general partnerships were the most prevalent form, which, compared to more structured articulation efforts, may be less effective in the avoidance of credit loss. We also found that shorter distances are a significant but impractical partnership-forming factor, for the average distance reduction among partnering colleges is 30 miles across models. Combining state and institutional datasets, we found that neither individual nor combined statewide policies actively predict institutional partnership formation. All databases and code created (statewide policies: https://cutt.ly/uwHyvkWQ, institutionally driven agreements: https://cutt.ly/7wHtPkEA, replication codes: https://cutt.ly/JwGRmVDu, https://cutt.ly/EwG1VbaW) may be used in future analyses to address questions of transfer effectiveness and transferring financial costs, which although important go beyond the scope of our study.
- 10 March 2025
- Published by Taylor & Francis in The Journal of Higher Education
- Vol. 96 (6) , 1114-1156
The Community College Transfer and Articulation Network
- 10 March 2025
- Published by Taylor & Francis in The Journal of Higher Education
- Vol. 96 (6) , 1114-1156
The Community College Transfer and Articulation Network
- Open Access
- 26 January 2025
- Published by SAGE Publications in International Journal of Chinese Education
- Vol. 14 (1)
To live or not to live at school? Boarding school impact on standardized academic outcomes for left-behind, rural, and low-income students in China
China's education system is striving to improve educational quality and address disparities. Recent policy initiatives have focused on providing better opportunities for left-behind students—those from low-income and rural backgrounds whose parents have sought employment elsewhere. One of these policies was the establishment of schools in rural areas that offer boarding education. Though early studies on this policy have raised/found safety and health concerns, there is little evidence of whether boarding education impact rural students and low-income students differently than their urban and non-low-income peers. Relying on non-self-reported, longitudinal academic outcomes (China Education Panel Survey) of a nationally representative sample of Chinese students, we offer quasi-experimental estimates (via doubly robust propensity score weighting and Oster’s sensitivity checks) of the impact of boarding education on Chinese, Mathematics, and English outcomes. We found strong evidence of positive impact of boarding education on Chinese and Mathematics, particularly among low SES and rural students. However, for English performance and the results for high SES and urban students, boarding education was not better than commuting from home. Our discussion elaborates on the mechanisms behind these findings and emphasizes the significance of our research findings for policymakers and its relevance to educational practices. Replication code/data access: https://cutt.ly/meS3atYc .
- 26 January 2025
- Published by SAGE Publications in International Journal of Chinese Education
- Vol. 14 (1)
Open AccessTo live or not to live at school? Boarding school impact on standardized academic outcomes for left-behind, rural, and low-income students in China
China's education system is striving to improve educational quality and address disparities. Recent policy initiatives have focused on providing better opportunities for left-behind students—those from low-income and rural backgrounds whose parents have sought employment elsewhere. One of these policies was the establishment of schools in rural areas that offer boarding education. Though early studies on this policy have raised/found safety and health concerns, there is little evidence of whether boarding education impact rural students and low-income students differently than their urban and non-low-income peers. Relying on non-self-reported, longitudinal academic outcomes (China Education Panel Survey) of a nationally representative sample of Chinese students, we offer quasi-experimental estimates (via doubly robust propensity score weighting and Oster’s sensitivity checks) of the impact of boarding education on Chinese, Mathematics, and English outcomes. We found strong evidence of positive impact of boarding education on Chinese and Mathematics, particularly among low SES and rural students. However, for English performance and the results for high SES and urban students, boarding education was not better than commuting from home. Our discussion elaborates on the mechanisms behind these findings and emphasizes the significance of our research findings for policymakers and its relevance to educational practices. Replication code/data access: https://cutt.ly/meS3atYc .