return elderly migration

This repository contains the R code and related files for a study on return migration intentions among older Syrian refugees in Turkey. The dataset includes socio-demographic and discrimination variables collected from individuals aged 60 and above. Key variables include legal status (temporary protection vs. citizenship), self-reported health, employment status, perceived discrimination (measured by a daily discrimination scale), and other control variables such as gender, marital status, language proficiency, and access to social services.The primary objective is to investigate factors influencing the likelihood of return migration. Logistic regression models were estimated using R (<code>glm()</code> with binomial family), and model performance was assessed through ROC curves, AUC, accuracy, and other classification metrics. Special consideration was given to class imbalance in the dependent variable (return intention: Yes/No), and weighted models were also explored.The repository includes:Raw dataset (<code>.sav</code> file)R scripts for analysisOutput plots (ROC, OR plots)Regression summaries and model diagnosticsFigures and tables suitable for academic publicationThis analysis contributes to the literature on forced migration, aging, and integration by highlighting how socio-demographic variables and daily discrimination interact in shaping return decisions among an understudied population.本仓库收录了针对土耳其境内老年叙利亚难民回返意愿展开研究所用的R代码及相关文件。本数据集收集自60岁及以上的个体，涵盖社会人口统计学特征与歧视感知相关变量。核心变量包括法律身份（临时保护身份与公民身份）、自我报告健康状况、就业状态、感知歧视（通过日常歧视量表测量），以及性别、婚姻状况、语言能力、社会服务可及性等其他控制变量。本研究的核心目标为探究影响回返移民意愿的各类驱动因素。研究采用R语言（<code>glm()</code>函数搭配二项分布族）构建逻辑回归模型，并通过受试者工作特征曲线（Receiver Operating Characteristic curve，简称ROC曲线）、曲线下面积（Area Under Curve，简称AUC）、准确率及其他分类指标评估模型性能。研究特别关注因变量（回返意愿：是/否）的类别不平衡问题，并探索了加权模型的应用场景。本仓库包含以下内容：原始数据集（<code>.sav</code>格式文件）、分析所用R脚本、输出图表（含ROC曲线、优势比（Odds Ratio, OR）图）、回归结果汇总与模型诊断结果，以及可直接用于学术发表的图表与表格。本研究通过揭示社会人口统计学变量与日常歧视感知如何共同作用，影响这一未被充分关注的老年群体的回返决策，为强迫移民、老龄化与社会融入领域的学术文献提供了新的研究视角。