社交网络与主动上学交通方式:探究父母人际关系如何影响子女的通勤决策
《Journal of Transport & Health》:Social networks and active school transportation: Understanding how parents’ connections shape commuting decisions for their children
【字体:
大
中
小
】
时间:2026年09月30日
来源:Journal of Transport & Health 3.3
编辑推荐:
摘要
引言
家长的社会网络会影响家长为孩子所做的决策,而主动上学交通(AST)是一项重要但尚未充分利用的体力活动(PA)来源。然而,社会网络如何影响AST决策仍不清楚。本研究旨在探讨1)社会网络特征如何与环境因素和人口统计因素的AST因素相互作用,以及2)家长社会网络对AS
摘要
引言
家长的社会网络会影响家长为孩子所做的决策,而主动上学交通(AST)是一项重要但尚未充分利用的体力活动(PA)来源。然而,社会网络如何影响AST决策仍不清楚。本研究旨在探讨1)社会网络特征如何与环境因素和人口统计因素的AST因素相互作用,以及2)家长社会网络对AST决策的影响。
方法
通过电子邮件向中西部小学儿童的家长(N=531)发放问卷,收集数据。问卷基于社会生态模型的人际层面、社会资本理论和技术创新扩散理论,内容涵盖家长社会网络、AST障碍感知以及为孩子做出的AST决策。多方法分析策略包括描述性统计、层级线性回归分析和零膨胀二项回归模型,使用SPSS软件进行。
结果
允许AST的家长(n=150,回复率28%)表现出更高的寻求建议频率(OR=3.21,p=.028)和更大的网络规模(OR=5.40,p=.009),在控制环境因素的情况下。尽管距离仍然是AST的最强预测因素(OR=0.04,p<.001),但零膨胀二项回归模型表明,家长网络因素可以解释AST决策中显著的方差(Nagelkerke R2=0.69)。在网络特征的中介作用下,不允许AST的家长报告的总AST障碍感知水平更高(M=47.14,SD=7.17),高于允许AST的家长(M=41.38,SD=9.21)。
结论
家长社会网络似乎在塑造对AST障碍的感知和决策方面发挥作用。未来的AST干预措施若考虑家长的社会连接和信息共享模式,可能会更加有效。
1. 引言
儿童主动上学交通(AST)的急剧下降是一个重要的公共卫生问题,相关比率已从1969年的48%下降至不到13%(Pedestrian and Bicycle Information Center, 2025)。虽然AST的环境和基础设施障碍已有大量记录(Aranda-Balboa et al., 2020),但最新研究表明,家长的社会网络可能在塑造交通决策方面发挥关键作用(Tristram et al., 2025)。了解这些社会连接如何影响对障碍的感知以及最终关于儿童上学交通的决策,可以改变促进主动交通干预方式的思路。这一点至关重要,因为仅有25%的儿童达到每天60分钟中等至剧烈体力活动(MVPA)的建议量(World Health Organization, 2024);此外,AST每天可为儿童贡献16分钟的体力活动(Bassett et al., 2013),并提高达到60分钟建议量的可能性(Daly-Smith et al., 2011)。
根据社会资本理论(Coleman, 1988)和技术创新扩散理论(Rogers, 2003),家长关于儿童交通方式的决策发生在复杂的社会环境中,在这些环境中,信息、态度和行为通过与其他家长和社区成员的互动被塑造。社会规范和社会影响观点认为,家长在评估风险和做出关于儿童的决策时,往往依赖于通过可信赖的社会关系获得的信息(Mah et al., 2017; McDonald et al., 2010)。这些互动可能塑造对安全、独立和交通选项的感知,从而影响AST决策(Savolainen et al., 2024)。然而,这些社会网络影响主动交通决策的具体机制尚不清楚。
关于家长社会网络的研究表明,这些网络会对各种与儿童相关的决策产生影响,从学业选择到健康行为。家长在为孩子活动做出决策时,通常依赖其社会连接获取信息、建议和支持(Gesell et al., 2021; Mah et al., 2017)。这些网络发挥多种功能:提供关于安全和物流的实用信息、为决策提供情感支持,以及建立关于可接受育儿方式的社会规范(Camargo et al., 2020; Cusack, 2021; Gesell et al., 2021; Hesketh et al., 2017; Tristram et al., 2025)。然而,现有文献中关于这些网络在塑造主动交通决策方面的作用研究有限。
家长在社会网络中互动的频率和性质可能与主动交通决策尤为相关。经常与其他家长讨论交通选项、就安全问题寻求建议以及观察他人交通选择的家长,可能对AST的障碍和机会形成不同的感知(Hesketh et al., 2017)。此外,家长网络的结构——包括其规模、密度和连接的强度——可能既影响其获取信息的能力,也影响其采纳主动交通行为的可能性。
1.1. 家长社会网络、社会规范与决策
家长很少孤立地做出关于儿童日常活动的决策。关于上学交通的决策往往受到与家庭成员、朋友、邻居和其他家长的互动影响,这些人提供信息、情感支持和实际指导。从社会资本的角度来看,这些关系可以作为资源,帮助家长解读风险、评估选项并做出关于儿童出行能力的决策(Bronfenbrenner, 1979; Coleman, 1988)。
社会规范也可能影响AST决策。家长在决定步行或骑自行车上学是否可接受、安全或普遍时,可能会参考其他家庭的行为和期望。如果AST在邻里或学校社区中是可见的且得到社会支持的,家长可能更倾向于将其视为合理的选择。相反,如果开车被认为是主要或预期的行为,AST可能被认为不太可接受或不太安全(Ajzen, 1911; Cialdini et al., 1990)。
家长关于AST的决策通常涉及与交通、监管、距离、天气和物流相关的不确定性。在这些条件下,家长可能会向其社会网络中可信赖的成员寻求建议。先前的研究表明,家长支持、感知邻里条件和社会认知因素与儿童的AST行为有关(Faulkner et al., 2010; Larouche et al., 2019; Lubans et al., 2011; Weir et al., 2006)。然而,关于家长社会网络的结构和特征如何影响这些决策的研究较少。
1.2. 研究目的
本研究考察家长社会网络如何与环境因素和人口统计因素相互作用以影响AST决策。除了传统的AST障碍外,本研究还探讨这些障碍如何通过家长社会连接的视角被感知和解读。通过分析网络特征,如规模、互动频率和信息共享模式,本研究旨在更好地理解社会网络如何放大或缓解对主动交通的感知障碍。在本研究中,网络特征被操作化为网络规模(被指定为育儿建议来源的人数)和寻求建议频率(向每个指定对象寻求指导的可能性)。这些指标源自社会资本理论中的概念,用于理解家长人际关系中嵌入的资源和支持(Coleman, 1988),同时也作为技术创新扩散理论的指标,因为AST可能通过人际间的寻求建议和沟通传播,而非通过大众媒体渠道(Rogers, 2003)。网络规模反映了家长潜在接触社会嵌入资源的广度,而寻求建议频率反映了采纳此类行为的人际渠道强度。
此外,根据社会生态模型(SEM),本研究的焦点环境特征是到学校的距离,因为这在文献中普遍被认为是最重要的环境因素(Galán et al., 2024)。其他考察的环境因素包括交通流量和速度、不安全的交叉口、缺乏过街护、缺乏其他成人、缺乏其他儿童、人行道/自行车道状况、天气条件、暴力/犯罪以及缺乏自行车存放设施。人口统计因素包括儿童年龄和PA(体力活动)、家长年龄、性别、教育程度、家庭收入、就业状态和家中儿童数量。
本研究探讨了三个相互关联的问题:(1)允许和不允许AST的家长在社会网络特征上有何差异?(2)在塑造家长对障碍的感知方面,网络因素与环境因素和人口统计因素相比影响如何?(3)社会网络特征如何与其他因素共同预测AST决策?本研究的目的是探讨1)社会网络特征如何与环境因素和人口统计因素的AST因素相互作用,以及2)家长社会网络对AST决策的影响。
2. 方法
本研究是制定如何利用社会生态模型(SEM)指导混合方法探索性顺序设计以推进安全上学路线基础设施升级战略的一部分(Creswell and Plano Clark, 2018)。采用基于社会资本理论(Colemaen, 1988)和技术创新扩散理论(Rogers, 2003)的横断面问卷调查设计,收集家长自我报告的AST社会网络、实践、障碍、信念、PA以及对孩子PA的感知。使用描述性统计、层级线性回归和零膨胀二项回归分析结果(图1)。
图1. 主动上学交通社会网络分析研究设计。
2.1. 参与者
本研究的参与者(N=531)来自美国俄亥俄州东北部一个正在进行安全上学路线基础设施升级的郊区学区的家长。
2.2. 测量工具
问卷包含56个条目,大部分为固定应答项,包括李克特量表条目,并设有一个开放式选项以提供详细信息(附录1)。目前尚未发现现有调查工具能同时结合(1)社会网络分析、(2)AST障碍、(3)为孩子做出的AST决策以及(4)家长和儿童的PA。因此,本研究所使用的工具基于社会生态模型(SEM)(Galán et al., 2024)、社会资本理论(Coleman, 1988)和技术创新扩散理论(Rogers, 2003),其中包含使用或改编自经过验证的社会网络分析、AST和PA测量工具的项目。
社会网络条目采用以自我为中心的(命名生成器)方法,遵循Adams(2020)和Schaefer(2019)的原则,基于社会资本理论(Coleman, 1988),家长最多列出五名他们寻求育儿相关建议的人。对于每个指定的人,家长报告了关系类型和沟通频率(结构网络属性),以及向此人寻求关于儿童上学交通建议的可能性(功能性、AST特定)以及寻求一般育儿建议或尝试新的育儿方式的可能性(功能性、一般育儿),使用六点李克特量表评分(1=非常不可能,6=非常可能)。网络规模计算为指定的人数,寻求建议频率计算为所有指定对象在AST特定条目上的平均可能性评分。
AST障碍的测量改编自Fyhri等(2011)和Panter等(2010)的研究,内容包括距离、交通、课外活动、交叉口安全、过街护存在与否、路线上的暴力/犯罪、沉重的书包、路线上是否有其他成人或儿童、人行道/自行车道的存在和质量、以及学校是否有地方存放自行车,使用5点李克特量表(从非常不同意到非常同意)。
关于家长为孩子做出AST决策的测量包括家长居住距离孩子学校的距离,以及在过去的学年中,孩子是否在学校学年的任何时候步行或骑车去学校或放学回家。这些测量改编自由Evenson等(2011)验证的"安全上学路线家长调查"。
家长PA的测量来自"运动作为生命体征"倡议,包括家长每周几天以及每天多少分钟进行PA,该测量工具经Coleman等(2012)验证。儿童PA的测量工具最初由Prochaska等(2001)在筛查工具中开发,已在包括全国儿童健康调查等调查中得到验证和使用,内容包括孩子每周有几天体力活动总时长至少达到60分钟(Sisson et al., 2010)。
2.3. 程序
在数据收集之前,已获得National University的伦理批准。数据收集于2024年秋季进行。通过Qualtrics平台,由俄亥俄州东北部学校管理部门向正在接受安全上学路线基础设施升级学校就读学生的家长发送邀请和调查链接。该小学在AST距离内(两英里内)的学生数量最多。随后几周发送了提醒邮件,并在社区Facebook群组中发布社交媒体帖子鼓励参与。完成问卷的平均时间为10分钟。
2.4. 数据分析
为检验研究目的、影响AST的因素,采用多步分析法进行了系统分析。分析策略从描述性比较逐步推进到使用层级线性回归和零膨胀二项回归的预测建模,以便理解变量之间的直接关系和交互效应。
以下是分析过程的逐步说明,以及该分析如何回应本研究目的。
首先,对AST(主动上学交通)家庭与非AST家庭的人口学特征及组间差异进行了检验,以比较人口学因素。这一初始分析采用独立样本t检验比较所有测量变量的均值,包括家长年龄、儿童年龄、家庭构成和网络特征。采用Cohen's d计算效应量,以判断所观察差异的实际意义。
其次,对AST的阻碍因素进行了检验,以确定塑造该样本中家长阻碍感知的环境因素。采用独立样本t检验分析综合阻碍得分和各项具体阻碍条目。该分析有助于识别区分AST家庭与非AST家庭的具体顾虑,同时通过仔细考察效应量和置信区间来控制I类错误。
第三,实施分层线性回归分析以探究研究目的的第一部分:确定网络特征如何与环境因素和人口学因素相互作用,从而影响家长关于AST的选择。该方法包含两个递进的模型。模型1纳入基本人口学变量和距离变量,以建立对总阻碍变异的基线理解。模型2加入网络特征变量,以考察其对解释阻碍感知的额外贡献。这种分层方法可在控制人口学影响的同时评估网络因素的独特贡献。
最后,为预测研究目的的第二部分,采用零膨胀二项回归模型,以更好地理解家长社会网络对AST决策的影响。选择该模型是因为因变量中存在大量零值(91.3%为非主动通勤者),并且该模型能够同时对多个预测变量对二元结果的影响进行建模。该模型纳入了所有此前考察过的变量,从而能够在控制其他因素的同时识别最具影响力的预测变量(Feng, 2020)。
2.5 统计考量
所有分析均使用SPSS 29.0版完成。采用双侧检验,显著性水平α = 0.05。报告效应量和置信区间,以提供关于实际意义、估计精度、模型拟合度以及组内样本量的信息(Verbeke, 2001)。
在进行零膨胀二项回归之前,先进行了标准二项回归分析,以确保预测零值与实际零值相匹配。在结果解释中考虑了各组的样本量差异(13例主动通勤者对比137例非主动通勤者),特别是在评估最终预测模型的分类准确率方面。这种方法论路径使分析能够从组间比较逐步推进到预测建模,同时兼顾数据结构上的局限。
3. 结果
3.1 样本人口学特征与组间比较分析
分析共纳入150名提供完整数据的家长(占应答者的98%),总应答率为28%。样本在AST方面表现出明显差异:13名家长(8.7%)报告其孩子步行或骑车上学,而137名家长(91.3%)表示孩子不进行AST。人口学分析揭示了这两组之间若干显著模式和差异。
距离成为最突出的区分因素,AST家庭居住地点离学校显著更近(M = 1.38, SD = 0.51),而非AST家庭则较远(M = 3.11, SD = 0.89),t(19.74) = 10.86, p < .001,呈现大效应量(d = 3.42, 95% CI [1.94, 4.87])。儿童年龄也存在有意义差异,主动通勤者平均年龄较小(M = 7.69, SD = 2.18),而非主动通勤者较大(M = 8.82, SD = 1.68),t(13.37) = 1.82, p = .091, d = 0.65。家长特征与家庭构成的差异较为细微。主动通勤者家长的年龄略大(M = 41.69, SD = 3.17),而非AST家长为M = 40.93, SD = 6.02,但该差异未达统计显著性(p = .652)。两组的家庭规模相近,AST家庭平均有2.13名儿童(SD = 0.83),非AST家庭平均有2.19名儿童(SD = 1.02)。
对育儿网络与支持系统的分析显示,参与模式存在细微差异。AST家长在多个维度上表现出略高的网络参与度:寻求建议的频率(M = 3.97, SD = 1.29 vs. M = 3.63, SD = 1.59)、网络任务支持(M = 2.91, SD = 1.36 vs. M = 2.76, SD = 1.51),以及采纳新育儿做法的可能性(M = 3.37, SD = 1.45 vs. M = 3.23, SD = 1.50)。然而,这些差异均未达到统计显著性。两组的身体活动模式相似,AST儿童每周进行60分钟以上身体活动的天数为6.08天(SD = 1.89),非AST通勤儿童为6.07天(SD = 1.73)(见表1)。
表1. 人口学特征与组间差异分析
| 特征 | 主动通勤者(n = 13) | 非主动通勤者(n = 137) | 均值差异 | t值 | p值 | 效应量(d) | d的95%置信区间 |
|---|---|---|---|---|---|---|---|
| 环境因素:距学校距离 | 1.38 (0.51) | 3.11 (0.89) | ?1.73 | 10.86 | <0.001 | 3.42 | [1.94, 4.87] |
| 儿童特征:年龄(岁) | 7.69 (2.18) | 8.82 (1.68) | ?1.13 | 1.82 | 0.091 | 0.65 | [0.00, 1.29] |
| 身体活动(天/周) | 6.08 (1.89) | 6.07 (1.73) | 0.01 | ?0.01 | 0.991 | 0.01 | [?0.54, 0.56] |
| 家长/家庭特征:家长年龄(岁) | 41.69 (3.17) | 40.93 (6.02) | 0.77 | ?0.45 | 0.652 | ?0.13 | [?0.68, 0.42] |
| 家庭儿童数 | 2.13 (0.83) | 2.19 (1.02) | 0.07 | ?0.18 | 0.396 | 0.725 | [?0.80, 0.45] |
| 网络特征:网络规模 | 4.03 (1.50) | 3.98 (1.88) | 0.05 | 0.29 | 0.778 | 0.06 | [?0.49, 0.61] |
| 建议频率 | 3.97 (1.29) | 3.63 (1.59) | 0.34 | ?0.87 | 0.395 | 0.24 | [?0.31, 0.79] |
| 任务支持 | 2.91 (1.36) | 2.76 (1.51) | 0.15 | ?0.39 | 0.705 | 0.11 | [?0.44, 0.66] |
| 网络采纳可能性 | 3.37 (1.45) | 3.23 (1.50) | 0.14 | ?0.32 | 0.747 | ?0.09 | [?0.64, 0.46] |
注:括号内为标准差。效应量(Cohen's d)解释标准:0.2 = 小效应,0.5 = 中等效应,0.8 = 大效应。CI = 置信区间。所有p值均为双侧检验结果。
3.2 主动交通阻碍因素分析
对阻碍因素的分析显示,AST家庭与非AST家庭在总体阻碍感知和具体顾虑方面均存在显著差异。不允许AST的家长其总阻碍得分显著较高(M = 47.14, SD = 7.17),而允许AST的家长则为M = 41.38, SD = 9.21,t(13.39) = 2.19, p = .047,呈中等至大效应量(d = 0.78)。距离相关阻碍成为两组之间最突出的差异。非AST家长对距离的阻碍评分显著高于AST家长(M = 3.44, SD = 1.43 vs. M = 1.92, SD = 1.61),t(13.82) = 3.62, p < .001, d = 1.05。这与两组实际距离的差异相吻合。
安全与交通相关阻碍在非AST家长中一致表现出更高的担忧程度。交通状况在非AST家长中的评分显著更高(M = 4.51, SD = 0.80 vs. M = 3.85, SD = 0.80),交通速度担忧也是如此(M = 4.36, SD = 0.77 vs. M = 3.85, SD = 0.80)。路口安全顾虑呈现类似模式(M = 4.20, SD = 0.96 vs. M = 3.92, SD = 1.04),但该差异未达统计显著性。
基础设施与监管方面阻碍也存在明显差异。非AST家长对缺少交通引导员的担忧更高(M = 4.07, SD = 1.13 vs. M = 3.46, SD = 1.45),对路上没有其他成年人的担忧更高(M = 3.88, SD = 0.99 vs. M = 3.15, SD = 1.28),对其他孩子步行或骑车的缺失也有更高担忧(M = 3.74, SD = 1.10 vs. M = 3.15, SD = 1.28)。
后勤方面阻碍呈现不同的担忧模式。课后活动对非AST家长构成显著更大的阻碍(M = 3.09, SD = 1.26 vs. M = 2.08, SD = 1.19)。沉重的书包问题也被非AST家长评为更为严重(M = 3.28, SD = 1.11 vs. M = 2.54, SD = 1.20)。天气相关问题与自行车停放问题在两组间的差异较小(见表2)。
表2. 阻碍因素按通勤状况的分析
| 阻碍类别与类型 | 主动通勤者(n = 13) | 非主动通勤者(n = 137) | 均值差异 | t值 | p值 | 效应量(d) |
|---|---|---|---|---|---|---|
| 距离与位置:距学校距离 | 1.92 (1.61) | 3.44 (1.43) | ?1.51 | 3.62 | <0.001 | 1.05 |
| 安全与交通:交通流量 | 4.00 (0.91) | 4.51 (0.80) | ?0.51 | 2.16 | 0.032 | 0.63 |
| 交通速度 | 3.85 (0.80) | 4.36 (0.77) | ?0.51 | 2.31 | 0.022 | 0.67 |
| 不安全路口 | 3.92 (1.04) | 4.20 (0.96) | ?0.28 | 0.99 | 0.325 | 0.29 |
| 基础设施与监管:缺少交通引导员 | 3.46 (1.45) | 4.07 (1.13) | ?0.61 | 1.82 | 0.071 | 0.52 |
| 缺少其他成年人 | 3.15 (1.28) | 3.88 (0.99) | ?0.73 | 2.47 | 0.015 | 0.72 |
| 缺少其他儿童 | 3.15 (1.28) | 3.74 (1.10) | ?0.59 | 1.83 | 0.069 | 0.53 |
| 人行道/自行车道状况 | 4.38 (0.96) | 4.34 (0.89) | 0.04 | ?0.16 | 0.872 | ?0.05 |
| 后勤阻碍:课后活动 | 2.08 (1.19) | 3.09 (1.26) | ?1.01 | 2.78 | 0.006 | 0.81 |
| 沉重的书包 | 2.54 (1.20) | 3.28 (1.11) | ?0.74 | 2.29 | 0.023 | 0.66 |
| 天气条件 | 3.23 (1.01) | 3.16 (1.03) | 0.07 | ?0.25 | 0.806 | ?0.07 |
| 犯罪与安全:暴力/犯罪担忧 | 2.00 (1.73) | 1.79 (1.09) | 0.21 | ?0.62 | 0.537 | ?0.18 |
| 自行车停放:缺少自行车停放处 | 3.69 (0.95) | 3.28 (0.92) | 0.41 | ?1.54 | 0.125 | ?0.45 |
注:括号内为标准差。所有阻碍项均按1–5量表评分(得分越高表示感知到的阻碍越大)。效应量(Cohen's d)解释标准:0.2 = 小效应,0.5 = 中等效应,0.8 = 大效应。所有p值均为双侧检验结果。
3.3 线性回归分析
进行分层线性回归分析,以考察AST总感知阻碍的预测变量,变量分两个模型依次纳入。模型1包含初始模型变量,即人口学变量和距离变量,解释了总阻碍7.9%的变异(R2 = 0.079,校正后R2 = 0.034,F(7,142) = 1.75, p = .103)。在该模型中,距学校距离是唯一显著的预测变量(β = 0.249, t = 2.99, p = .003),表明距离每增加一个单位,总感知阻碍增加1.91个单位。
第二个模型纳入育儿网络变量,使解释的变异上升至10.8%(R2 = 0.108,校正后R2 = 0.023,F(13,136) = 1.27, p = .241)。解释变异的增量(ΔR2 = 0.029)在统计上并不显著。在该扩展模型中,距学校距离仍保持其作为最强预测变量的显著性(β = 0.257, t = 2.97, p = .004)。新加入的网络变量中,没有任何一个达到统计显著性(见表3)。
表3. 总阻碍的分层线性回归结果
| 变量 | 模型1 β | t | p | 模型2 β | t | p |
|---|---|---|---|---|---|---|
| 第一块:人口学与距离变量 | | | | | | |
| 距学校距离 | 0.249 | 2.989 | 0.003 | 0.257 | 2.968 | 0.004 |
| 家长年龄 | 0.064 | 0.752 | 0.453 | 0.056 | 0.626 | 0.532 |
| 家庭儿童数 | 0.035 | 0.432 | 0.666 | 0.017 | 0.201 | 0.841 |
| 家长性别 | 0.091 | 1.074 | 0.284 | 0.073 | 0.788 | 0.432 |
| 就业状态 | 0.090 | 1.092 | 0.277 | 0.103 | 1.202 | 0.231 |
| 家长身体活动 | ?0.045 | ?0.546 | 0.586 | ?0.040 | ?0.477 | 0.634 |
| 第二块:网络变量 | | | | | | |
| 网络规模 | — | — | — | 0.092 | 0.776 | 0.439 |
| 交流频率 | — | — | — | ?0.089 | ?0.461 | 0.646 |
| 任务支持 | — | — | — | ?0.078 | ?0.505 | 0.614 |
| 主动交通讨论 | — | — | — | 0.092 | 0.547 | 0.585 |
| 建议频率 | — | — | — | ?0.238 | ?0.883 | 0.379 |
| 采纳可能性 | — | — | — | 0.296 | 1.338 | 0.128 |
| 模型统计量 | | | | | | |
| R2 | 0.079 | F = 1.748, p = 0.103 | | 0.108 | F = 1.266, p = 0.241 | |
| ΔR2 | — | | 0.029 | ΔF = 1.266, p = 0.241 | | |
注:报告的是标准化β系数。显著预测变量(p < .05)以粗体显示。
3.4 零膨胀二项回归分析
用于检验AST预测变量的零膨胀二项回归模型展现出稳健的预测能力(χ2(12) = 55.37, p < .001),Nagelkerke R2为0.693,解释了AST约69.3%的变异。该模型具有极高的分类准确率,总体正确识别96.7%的样本。具体而言,模型对非主动通勤者的分类准确率为99.3%(137例中正确识别136例),对主动通勤者的分类准确率为69.2%(13例中正确识别9例)。这种分类准确率的差异反映了样本的内在分布特征。
距学校距离成为模型中最强的负向预测变量(OR = 0.040, p < .001, 95% CI [0.013, 0.159]),距离每增加一个单位,主动通勤的可能性降低96%。网络规模显示出第二强的效应(OR = 5.400, p = .009, 95% CI [1.338, 11.482]),表明拥有更大支持网络的家长更可能允许孩子进行AST,其可能性高出五倍以上。儿童年龄与AST呈现明显的负相关关系(OR = 0.624, p = .111, 95% CI [0.412, 0.944]),提示年龄越大AST的可能性越低,尽管该效应未达统计显著性。其他变量与主动通勤均呈非显著关联。
3.5 网络影响
家长网络中的建议寻求频率成为显著的阳性预测变量(OR = 3.214, p = .028, 95% CI [1.134, 9.112]),表明更频繁寻求建议的家长允许孩子进行AST的可能性高出三倍以上。其他网络特征表现出不同的影响:……与AST行为呈非显著关联(见表4)。
表4. 零膨胀二项回归结果
| 预测变量 | 优势比 | 标准误 | Wald值 | p值 | 95% CI下限 | 95% CI上限 |
|---|---|---|---|---|---|---|
| 环境因素:距学校距离 | 0.040 | 0.939 | 11.765 | <0.001 | 0.013 | 0.159 |
| 人口学变量:儿童年龄 | 0.624 | 0.296 | 2.535 | 0.111 | 0.412 | 0.944 |
| 家长教育程度 | 1.300 | 0.744 | 0.124 | 0.725 | 0.298 | 5.674 |
| 家庭收入 | 1.222 | 0.446 | 0.201 | 0.654 | 0.510 | 2.927 |
| 网络特征:网络规模 | 5.400 | 0.642 | 6.908 | 0.009 | 1.338 | 11.482 |
| 建议频率 | 3.214 | 1.085 | 4.831 | 0.028 | 1.134 | 9.112 |
| 任务支持 | 0.669 | 0.935 | 0.185 | 0.667 | 0.107 | 4.184 |
| 交流频率 | 0.635 | 0.595 | 0.584 | 0.445 | 0.198 | 2.038 |
| 采纳可能性 | 0.253 | 0.902 | 2.316 | 0.128 | 0.043 | 1.484 |
模型拟合统计量:
- 对数似然值:?33.056
- Nagelkerke R2:0.693
- 总体分类准确率:96.7%
- χ2(12) = 55.371, p < .001
注:CI = 置信区间。
4. 讨论
4.1 网络对主动交通的影响
本研究表明,在控制环境与人口学变量之后,家长社会网络可能对AST决策具有实质性影响(Mah et al., 2017)。网络规模和建议寻求频率均为显著预测变量,提示家长与他人的联系越紧密、越积极地寻求指导,其允许孩子进行AST的可能性可能越大。本研究对文献作出了若干创新贡献。第一,它首次考察了家长社会网络如何影响AST决策。第二,它采用了多方法分析路径,以区分社会、环境和人口学因素在家长决策中的相对重要性。最后,它提出了利用社会网络促进AST的干预实践洞见。理解社会网络在塑造交通决策中的作用,或将根本性地改变AST的推广方式。与其仅关注环境改造或个体行为改变,干预措施若考虑家长的社会联系与信息分享模式,可能更为有效。本研究可为此类基于网络的干预提供实证基础。
4.2 社会网络规模与建议寻求的主导作用
社会网络规模(OR = 5.400)与建议寻求频率(OR = 3.214)的预测力凸显了网络参与在AST决策中的核心作用。身处更大网络中的家长可能更多地接触允许AST的榜样,并有更多机会协调共享责任,例如组建步行小组或共同监管(Hecker et al., 2025)。同样,寻求建议的行为可能反映出对替代性做法的开放态度,以及愿意采纳同伴中观察到的行为(Larouche et al., 2019)。综合来看,这些结果提示社会网络不仅提供信息支持,还充当传播AST相关规范的机制(Walsh et al., 2021)。
4.3 环境因素的持续作用
尽管社会网络特征具有影响,环境阻碍仍然是AST的最强决定因素。距学校距离是最显著的预测变量(OR = 0.040),表明邻近性仍是限制步行或骑车可行机会的关键(Smith et al., 2020)。然而,环境因素与社会因素之间的相互作用值得关注。身处支持性网络中的家长可能将基础设施挑战视为不那么难以逾越,或能找到克服挑战的创造性策略,例如与其他家长协调组队步行(Hesketh et al., 2017)。这些结果强调,看待AST决策不应仅局限于基础设施问题,还应将其视为场所条件与社会影响之间动态交互作用的结果(Zaragoza et al., 2020)。
4.4.
Barrier perception patternsDifferences in perceived barriers between parents who allowed and did not allow AST reveal the critical mediating role of social networks. Non-AST parents reported significantly higher total barrier scores, particularly regarding distance, traffic, and logistical concerns. Yet, parents with larger or more engaged networks appeared to perceive these barriers as less severe. Importantly, this analysis suggests that networks have a differential impact depending on barrier type. For example, logistical challenges such as coordinating schedules were more easily reframed when parents had peers to share responsibilities, whereas weather-related concerns showed little moderation by networks. This suggests that while networks cannot eliminate structural challenges, they can reduce the psychological weight of certain obstacles by offering practical solutions or alternative framing. These findings highlight the potential for interventions that target parental perceptions rather than solely focusing on physical modifications (Cusack, 2021).
4.5. Safety and infrastructure concerns
Traffic-related safety concerns were significant differentiators between active and non-AST groups. While these perceptions include safety risks, social support may alter how parents assess and respond to these dangers. For example, parents who know others walking the same route may feel reassured about traffic safety or may organize group strategies that mitigate risks (Molina-García et al., 2025). Similarly, concerns about infrastructure and supervision, such as the absence of crossing guards, safe crossings, or dedicated bike lanes, were less daunting for parents embedded in active, engaged networks. Larger networks were associated with fewer infrastructure-related concerns, suggesting that collective solutions such as "walking school buses" or rotating supervision responsibilities can directly offset the absence of formal safety measures. This underscores the potential of collective, community-based approaches to address safety and infrastructure barriers.
4.6. Demographic and social network interactions
Demographic patterns also intersect with social network dynamics in shaping AST decisions. Younger children were more likely to engage in AST, consistent with increased parental involvement with younger children, though this relationship was moderated by network characteristics. Parents of younger children who sought advice or had larger networks were more likely to permit active commuting, suggesting that networks can help counterbalance developmental concerns by providing reassurance or supervision strategies. Parent age also played a role, with older parents relying more on established community ties, while younger parents reported using digital networks or school-based groups to inform decisions. Socioeconomic indicators, including education and income, were not significant predictors on their own but may shape the quality and diversity of parental networks. For instance, higher-social economic status (SES) parents may engage with professional peer groups, while lower-SES parents may rely more heavily on neighborhood or school-based ties – or vice versa -potentially depending on the ability to have a single household income. These patterns indicate that while demographic factors alone do not determine AST, they shape how networks are formed and utilized, warranting further exploration (Rashidi et al., 2025).
4.7. Scientific contribution and research gap
The primary scientific contribution of this study is the examination of parental social networks as a previously understudied factor influencing AST decisions. Previous research largely emphasizes environmental and infrastructural determinants of AST, including distance from school, traffic conditions, sidewalk availability, and perceived safety (Faulkner et al., 2010; Larouche et al., 2019; Lubans et al., 2011; Weir et al., 2006) and substantial evidence regarding environmental, logistical, and safety-related determinants as barriers and facilitators (Larouche et al., 2014; Lubans et al., 2011; Panter et al., 2010; Weir et al., 2006). In contrast, comparatively little attention has been devoted to understanding how parents form perceptions of those barriers (Bronfenbrenner, 1979; Coleman, 1988).
Further, while parental attitudes are recognized as important predictors of transportation behavior (Ajzen, 1991; Faulkner et al., 2010), limited research has directly examined the role of parental social networks in shaping those attitudes and decisions. Although these factors consistently predict AST, they do not fully explain why parents living in similar neighborhoods and facing comparable environmental conditions often make different AST decisions for their children. This represents an important scientific gap as existing models have devoted relatively little attention to the influence of parents' social relationships and information networks on AST decision-making (Bronfenbrenner, 1979; Coleman, 1988; Cialdini et al., 1990). By examining parental social network characteristics alongside traditional environmental and demographic predictors, this study extends current understanding of AST decisions and behaviors and provides evidence that social network processes might contribute additional explanatory value.
Therefore, the novelty of the present study lies in the integration of parental social network characteristics and AST research. Specifically, this study demonstrates that network size and advice-seeking frequency are significant predictors of AST participation. These findings suggest that transportation decisions are socially embedded behaviors influenced by interpersonal relationships rather than solely by physical environmental conditions.
4.8. Study limitations
Several limitations warrant consideration. While the hierarchical linear regression analysis to investigate part 1 of the purpose(to determine the combined influence of how network characteristics interact with environmental and demographic factors to shape parents' choices about AST) achieves recommended statical power (Hox, 2010), to predict part 2 of the study purpose (to better understand the influence of parental social networks on AST decisions), the zero-inflated binomial regression model includes a small number of AST families (n = 13) therefore limiting statistical power and may restrict the generalizability of findings. This model was intentionally selected due to the significant preponderance of zeros in the outcome variable (91.3% non-active commuters) and its ability to simultaneously model the influence of multiple predictors on a binary outcome zero-inflated binomial including uneven groups (Verbeke, 2001). Therefore, while the zero-inflated regression model accounts for this imbalance, interpretation of the results should be included with caution as larger and more diverse samples are needed to validate results (Feng, 2020).
Second, the cross-sectional design prevents causal inference, leaving open the question of whether strong networks encourage AST or whether AST parents subsequently build stronger networks. Third, while social network measures were carefully constructed, they may not capture the full complexity of parental relationships. For example, the study did not include digital or weak-tie networks such as online parenting forums, teachers, or extended family, which may also influence decisions. Finally, the sample's geographic scope constrains external validity, as unique features of the built environment may shape network–barrier interactions in ways not generalizable to other contexts. These limitations suggest caution in interpretation but also highlight avenues for refinement in future studies.
4.9. Practical implications
The findings support the potential of network-based interventions and suggest that efforts to increase AST should extend beyond infrastructure improvements alone. While sidewalks, bicycle lanes, crossing guards, and traffic-calming strategies remain important components of AST initiatives (Larouche et al., 2019; Panter et al., 2010), schools and communities may also benefit from initiatives designed to strengthen parent-to-parent communication and support networks (Bronfenbrenner, 1979; Coleman, 1988). Programs such as walking school buses, bicycle trains, parent ambassador initiatives, neighborhood walking groups, and school-sponsored transportation committees may create opportunities for information sharing and social reinforcement that encourage AST behaviors (Sallis et al., 2006; McDonald et al., 2014).
4.10. Policy recommendations
Policies that deliberately foster parent-to-parent communication, such as schools requiring resources for and availability of walking groups, community "bike trains," or ambassador programs, may amplify social influence in favor of AST (Ross and Wilson, 2021). School-level initiatives, such as requiring parent teacher association -led safety patrols or structured walking school buses, can supplement infrastructural improvements by embedding safety strategies within social networks. At the community level, communication policies to connect parents across neighborhoods could help normalize AST practices and provide shared solutions to persistent barriers. By aligning infrastructural investments with policies to cultivate parental networks, schools and communities might create multi-level policies that address both physical and social determinants of AST.
4.11. Theoretical implications
The findings within this study are consistent with social capital theory (Coleman, 1988) and broader social network theory, which posits that decision-making is shaped by the flow of information and norms within social connections (Tristram et al., 2025). In this context, parents who have larger, more engaged networks appear to encounter more supportive information regarding safety and logistics, thereby reducing their perceived risks and encouraging AST (Gesell et al., 2021). Further, findings within this study support diffusion of innovations theory (Rogers, 2003) and the proposition that parental AST decisions occur within a broader social context. Consistent with previousAST research, parents appear to rely on information and advice obtained through their social networks when evaluating AST options and AST-related risks (Faulkner et al., 2010; Weir et al., 2006). The significant relationships observed between network size, advice-seeking behaviors, and AST decisions suggest that social networks may function as a mechanism through which perceptions of barriers are developed, maintained, and potentially reduced at the interpersonal level of the SEM (Galán et al., 2024;McDonald et al., 2014; Panter et al., 2010). These results extend existing active school transportation literature by emphasizing the importance of social environments in addition to physical environments (Larouche et al., 2019).
4.12. Future research and applications
Future research should adopt longitudinal designs to trace how parental networks evolve over time and how changes in these networks influence AST decisions (Camargo et al., 2020). Such designs could clarify causal pathways and identify critical periods for intervention. Additionally, qualitative work could provide deeper insight into the mechanisms through which parents determine transportation decisions within their networks, including how they balance safety concerns with developmental benefits. Researchers should also examine potential threshold effects, for instance, whether a minimum network size or frequency of advice-seeking is necessary to meaningfully offset environmental barriers such as distance.
5. Conclusion
The findings within this study suggest that parental social networks might be a helpful a consideration in shaping both perceptions of AST barriers and decisions, highlighting the potential value of network-based interventions for promoting AST. While accounting for limitations and including that findings are exploratory in nature, the results could suggest that future interventions might be more effective if they explicitly consider and leverage parents' social connections and information-sharing patterns, rather than focusing solely on environmental modifications or individual behavior change.
CRediT authorship contribution statement
Carissa R. Smock: Writing – review & editing, Writing – original draft, Validation, Supervision, Software, Resources, Project administration, Methodology, Investigation, Data curation, Conceptualization. Eugene Fitzhugh: Writing – review & editing, Validation, Supervision, Methodology, Conceptualization. Courtney Hamby: Writing – review & editing, Writing – original draft, Validation, Methodology, Conceptualization. Tyler Prochnow: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Software, Resources, Project administration, Methodology, Formal analysis, Conceptualization.
Funding sources:
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Carissa R. Smock|Eugene Fitzhugh|Courtney Hamby|Tyler Prochnow
Department of Leadership, Management, & Human Capital, College of Business, Engineering, and Technology, National University, 9388 Lightwave Ave, San Diego, CA, 92123, USA