為更系統化瞭解並釐清親職化與心理健康結果之間的關係，本研究在「臺灣期刊論文索引系統」、「臺灣博碩士論文知識加值系統」、「中國知網數據庫」、Pubmed、Wiley、ProQuest、EBSCO、PsycARTICLES等中英文資料庫檢索同時測量親職化與心理健康結果的相關研究，共納入51篇中英文期刊論文與碩博士學位論文、230個效果量，涉及研究參與者15,614名。本研究突破過去統計方法之限制，改採三層級後設分析模式進行資料分析，結果發現：（1）親職化與心理健康負向結果之間存在顯著性正相關。（2）來自於東方文化的研究所發現之親職化與心理健康結果的相關係數顯著低於來自於西方文化的研究。（3）親職化中「不公平感」與心理健康結果的相關係數，顯著高於其他類型親職化。（4）親職化與內化問題的相關係數，顯著高於其與外化問題行為的相關係數。（5）其他變項（如研究對象性別、年齡段、親職化測量工具、出版年份、出版狀態等）均未見顯著性調節效果。作為第一篇納入華人親職化研究的三層級後設分析，本研究結果提示實徵研究與實務工作者應特別關注「不公平感」與「文化脈絡」在親職化發生歷程中的獨特效應，避免對親職化「過度病理化」假設。最後，針對親職化現象在實務應用與未來研究方向，提出討論與建議。

Since its introduction by Boszormenyi-Nagy and Spark (1973) in the 1970s, parentification has become a ubiquitous topic in the fields of family therapy and family psychology. In empirical research, parentification is regarded as a risk factor threatening the psychological development of children (Nuttall et al., 2019). However, some studies have indicated the absence of a significant correlation between parentification and maladaptive mental health outcomes (Hamin, 2006; Hsu, 2007), whereas other studies have revealed a negative correlation between the two constructs (Li, 2021; Chang & Kuo, 2000). To determine the long-term consequences of childhood parentification, that is, the relationship between childhood parentification and adult mental health, Hooper, Decoster, et al. (2011) conducted a meta-analysis of the relationship between self-reported childhood parentification and adult psychopathology. Their results revealed a significant positive correlation between self-reported childhood parentification and adult psychopathological factors, such as depression, substance use, anxiety, posttraumatic stress disorder, personality disorders, and eating disorders, although the correlation coefficient was small. However, although they used various methods to collect relevant studies covering the period from 1984 to 2010, they included only 12 empirical studies and 12 effect sizes in their meta-analysis. Among all moderators included in the analysis, they focused on parentification type (emotional parentification and undifferentiated parentification) and ethnicity (White Americans and Black Americans). Given the universality of parentification in the context of family, adopting psychopathology as the only outcome variable may hinder the inclusion of studies in meta-analyses and reduce the generalizability of meta-analytical results. Over the past decade, many empirical studies have been conducted worldwide on parentification. Researchers in relevant fields have comprehensively explored the types, measurements, and research participants of parentification. Particularly, many studies have been conducted on parentification in China. In this study, we collected both English and Chinese articles on parentification and used a three-level meta-analytic model to comprehensively investigate the relationship between parentification and maladaptive mental health. We also investigated the moderators that affect the degree of correlation between parentification and maladaptive mental health to obtain further insights into this topic. We searched the following computerized databases from their inception until December 2021: Index to Taiwan Periodical Literature System, National Digital Library of Theses and Dissertations in Taiwan, China National Knowledge Infrastructure, PubMed, Wiley Online Library, ProQuest, EBSCO (Academic Search Complete), and PsycArticles. From these databases, we identified Chinese and English articles whose titles or abstracts paired terms related to parentification (parentif* or role reverse or role conflict or boundary resolution or boundary disorder or boundary problems or young care*) with terms related to mental health (mental health or internaliz* or depressi* or anxi* or somatization or distress). Using this search strategy, we identified a total of 1,081 potential articles for our meta-analysis. After excluding duplicates, we obtained 621 articles. After reading the titles and abstracts of these articles, we further excluded studies with unrelated themes (409 articles), theoretical studies (28 articles), and qualitative studies (76 articles). Finally, after reading the full texts of these articles, we excluded 57 articles, including studies with repeated samples (3 articles), studies with incomplete statistical data (24 articles), studies with other evaluation criteria (8 articles), and studies whose statistical results could not be converted into product error correlation coefficients (22 articles). Ultimately, a total of 51 journal papers, theses, and dissertations (12 Chinese and 39 English) meeting the data selection criteria were included in the meta-analysis. In this meta-analysis, studies are primarily coded in accordance with the following four aspects: sample characteristics (e.g., gender, age group, family context, cultural context), publishing characteristics (e.g., published or unpublished and year of publication), parentification characteristics (e.g., parentification type and measurements), and mental health outcomes (e.g., internalized and externalized behavioral problems). In this study, two coders (the first and third authors) independently coded the included studies. To eliminate the effect of selection bias on the results of the meta-analysis, inconsistencies were resolved through discussion with the corresponding author. Ultimately, a total of 230 independent effect sizes were obtained from 51 studies. Before conducting the meta-analysis, each correlation coefficient was transformed into a normally distributed Fisher's z value. After the analysis, these z values were converted back into correlation coefficients. The R software (version 4.1.1; R Foundation for Statistical Computing, Vienna, Austria) was used to identify the relationship between parentification and maladaptive mental health outcomes. The results were as follows. (1) A significantly weak positive correlation was observed between parentification and maladaptive mental health outcomes (r = .144). Random sampling variation (Level 1), within-study effect size variation (Level 2), and between-study effect size variation (Level 3) accounted for 7.381%, 34.538%, and 58.081%, respectively, of the overall effect size variations, with a statistically significant difference observed between Level 2 and Level 3 (p < .001). (2) Family context had a significant moderating effect on the relationship between parentification and maladaptive mental health outcomes (p = .076). Specifically, the correlation coefficient between parentification and mental health outcomes was significantly larger among participants with a clinical family context (i.e., parents or siblings with long-term and chronic physical and mental diseases at home; r = .229)) than among those with a normal family context (r = .104). Cultural context also exerted a significant moderating effect on the relationship between parentification and mental health outcomes (p < .001). Specifically, the correlation coefficient between parentification and mental health outcomes (r = -.048) in studies on the eastern culture was significantly smaller than that in studies on the western culture (r = .192). (3) Parenting type had a significant moderating effect on the relationship between parentification and mental health outcomes (p < .001). Specifically, the correlation coefficient between unfairness and mental health outcomes (r = .279) was significantly larger than that between other types of parentification and mental health outcomes (r = .067 to .150). (4) Different mental health outcomes had a significant moderating effect (p = .018). Among these outcomes, the correlation coefficient between parentification and internalization (r = .177) was significantly larger than that between parentification and externalization (r = .085). However, the correlation coefficient between parentification and other mental health outcomes (e.g., total score and other problems) did not vary significantly from that of internalization. (5) No significant moderating effect was observed for other potential variables, such as gender, age, measurements, publication year, and publication status. Given the aforementioned results, we suggest that clinicians should not only evaluate the degree and duration of parentification (East & Weisner, 2009) but also differentiate between parentification types and their effects, particularly in terms of the psychological perception of parentification-related unfairness. To comprehensively and accurately evaluate the family history and parentification experience of individuals in the context of the Chinese culture, the sensitivity of cultural values must be maintained. This approach is the important feasible means for avoiding the hypothesis of "overpathologization" based on western theories (Kuperminc et al., 2009). Empirical research on parentification should consider the psychological aspects of parentification represented by perceived unfairness as a construct independent of the behavior of parentification (e.g., instrumental or emotional parentification). It should also utilize a second-order model to include the two types of parentification while developing or revising parentification scales. Relevant variables such as family and cultural contexts should be included to examine the process and mechanism of parentification in more detail. In this study, we addressed the limitations of traditional meta-analyses and used a three-level meta-analytical model to examine empirical research on parentification. The results of this meta-analysis may serve as systematic evidence for researchers and practitioners in this field.