ISSN 0439-755X
CN 11-1911/B

心理学报 ›› 2023, Vol. 55 ›› Issue (10): 1637-1652.doi: 10.3724/SP.J.1041.2023.01637

• 研究报告 • 上一篇    下一篇


陈必忠1, 黄璇2, 牛更枫1, 孙晓军1(), 蔡志慧1   

  1. 1青少年网络心理与行为教育部重点实验室; 华中师范大学心理学院, 武汉 430079
    2北京师范大学心理学部, 北京 100875
  • 收稿日期:2022-09-05 发布日期:2023-07-26 出版日期:2023-10-25
  • 通讯作者: 孙晓军, E-mail:
  • 基金资助:

Developmental change and stability of social anxiety from toddlerhood to young adulthood: A three-level meta-analysis of longitudinal studies

CHEN Bizhong1, HUANG Xuan2, NIU Gengfeng1, SUN Xiaojun1(), CAI Zhihui1   

  1. 1Key Laboratory of Adolescent Cyberpsychology and Behavior, Ministry of Education; School of Psychology, Central China Normal University, Wuhan 430079, China
    2Faculty of Psychology, Beijing Normal University, Beijing 100875, China
  • Received:2022-09-05 Online:2023-07-26 Published:2023-10-25


鉴于社交焦虑的高患病率和广泛的不良后效, 大量理论和纵向研究对社交焦虑的发展进行了探讨, 但观点和研究结果存在矛盾。为厘清争议, 准确刻画社交焦虑平均水平变化和稳定性的年龄趋势, 采用三水平模型的方法, 对来自173项纵向研究的192个独立样本(N = 170197)进行元分析。结果显示: (1)在平均水平上, 社交焦虑从学步期至青年期呈逐步下降趋势, 仅青春中期有小幅上升。(2)在稳定性上, 社交焦虑在学步期和学前期较高, 在小学儿童期降至低谷后逐渐回升, 青年期稳定在较高水平。(3)社交焦虑的平均水平变化不受研究特征、样本特征和变量特征的影响。(4)社交焦虑的稳定性受文章语言、大洲、文化、性别和测量方式的调节。(5)社交焦虑的稳定性随时间间隔变长呈先快后慢的下降趋势。研究首次利用元分析技术刻画了社交焦虑从学步期至青年期的发展趋势, 其中发展轨迹总体上支持了人格成熟论, 稳定上则呈现出较强的特质属性, 结果为探索社交焦虑的毕生发展和干预提供了新视角。

关键词: 社交焦虑, 平均水平变化, 稳定性, 纵向研究, 三水平元分析


Given the high prevalence and its extensive possible adverse outcomes, a large number of theoretical and longitudinal studies have explored the development of social anxiety, but the research findings are inconsistent or even contradictory in preschool childhood, late childhood, and adolescence. In addition, there is still debate between trait theory and state theory of social anxiety, and there is also a lack of study on the age trend of social anxiety stability.

To clarify the above controversies and accurately characterize the age trend of the mean level and the stability of social anxiety, this study conducted a meta-analysis of longitudinal studies involving social anxiety. After pre-registering the study protocol on PROSPERO, we searched literature in six databases (CNKI, Wanfang Data, Web of Science, ProQuest, PubMed, and EBOSCO). In addition, we also backtracked the references cited in previous meta-analyses and reviews. Ultimately, a total of 192 independent samples (N = 170, 192) from 173 longitudinal studies were included in the current meta-analysis. In order to quantify the trajectory of social anxiety more precisely, we divided the sample into 11 age groups according to the mean age of the sample between two adjacent measurement waves, and estimated the mean-level change and stability for each age group. The standardized mean difference (d) between two adjacent waves is used to estimate mean-level change, whereas the correlation coefficient (r) is used to estimate stability. Science most of the included studies reported multiple effect sizes, and these multiple effect sizes were most likely dependent, which violated the basic assumption of independent effect sizes in the conventional meta-analytic methods. We applied the three-level meta-analysis approach to handle such data-independency among effect sizes.

Results showed that: (1) The mean level of social anxiety showed a gradual decline from toddlerhood to young adulthood, with only slight increase in mid-adolescence. (2) In terms of rank-order stability, social anxiety rose slowly from toddlerhood to preschool childhood, then swiftly dropped to a low point in elementary childhood, recovered gradually after mid-adolescence, and stabilized at a high level in young adulthood. (3) The mean-level change of social anxiety was not affected by the study characteristics, the sample characteristics, and the variable characteristics. (4) The rank-order stability of social anxiety was moderated by written language, continent, culture, gender, and assessment mode. (5) The stability of social anxiety was a logarithmic function with time lag. Specifically, with the increase of time lag, the stability declined first quickly and then slowly, and almost reached a plateau after 6 years. (6) The results of moderator test for publication status, Egger's test, and Begg's test indicated the absence of publication bias in this meta-analysis.

This study makes a valuable contribution in characterizing the age-specific trends and stability of social anxiety from toddler to young adult by using the meta-analytic method. We conclude from this study that, in terms of mean level, the trajectory of social anxiety generally supports the personality maturation hypothesis. For stability, similar to personality traits, social anxiety tend to be a trait rather than a state construct. Overall, this study provides a new perspective for exploring the lifelong development of social anxiety.

Key words: social anxiety, mean-level change, rank-order stability, longitudinal studies, three-level meta-analysis