中文版親子二人互動記錄系統之信效度建立：以台灣樣本為例

Psychometric Properties of Chinese Version of the Dyadic Parent-Child Interaction Coding System in a Taiwanese Sample

曾楷文(Kai-wen Tseng)

指導教授 ： 陳怡群

國立中正大學/社會科學院/臨床心理學研究所/碩士(2013年)

摘要

目的：本研究主要目的即在於中文化國外臨床與研究上常來用評估父母教養行為的親子互動行為觀察工具—親子二人互動記錄系統第三版（Dyadic Parent-Child Interaction Coding System – Third Edition, DPICS-III），並以台灣兒童及其照顧者建立信效度。
方法：66名臨床轉介有情緒行為問題的兒童（年齡介於三歲到十一歲間），或（和）照顧者有教養技巧上的困難之親子參與本研究。本研究首先將收集到之資料，分成兒童決裂性行為問題達顯著組及未達顯著組兩組，再以此兩組資料分別與Bessmer（1996）研究中符合對立反抗行為疾患的美國臨床及非臨床兒童進行比較。其次，再以DPICS-III中文版所評估的目標行為，包括：九項父母言語行為（分別為具體和非具體稱讚、回應、行為描述、負面語句、直接和間接命令、問題，及中性語句）和三項兒童反應行為（分別為服從、不服從，及無機會服從），與一系列由照顧者填寫的量表，進行信效度之檢驗。
結果：本研究發現在比較DPICS-III中文版所評估的目標行為時，無論是上述哪一組的比較，台灣和美國兒童對照顧者命令的服從度並無差異，但台灣和美國照顧者除了問問題的頻率之外，其餘言語行為的表現皆有顯著差異。在信度部分，則發現父母行為的百分比同意度介於94.3~99.3%之間、kappa值則介於 .70~ .91之間，兒童行為的百分比同意度介於92.5~96.9%之間、kappa值則介於 .67~ .73之間。在區辨效度方面，以DPICS-III中文版觀察到照顧者的負面語句，能區辨兒童在1½-5歲或6-18歲兒童行為檢核表─父母報告表格（Child Behavior Checklist For Ages 1½-5 / 6-18, CBCL 1½-5/6-18）的整體行為問題量尺和艾伯格兒童行為量表（Eyberg Child Behavior Inventory, ECBI）的問題量尺是否達臨床顯著，照顧者的回應率能區辨兒童的CBCL 1½-5/6-18外化行為問題是否達臨床顯著；而以DPICS-III中文版觀察到兒童對照顧者命令的服從率，能區辨其在ECBI嚴重度和問題量尺上是否達臨床顯著。其次，在聚合效度方面，以DPICS-III中文版觀察到照顧者的負面語句與其報告的ECBI問題量尺、親職壓力量表（Parenting Stress Index, PSI）的兒童分量尺，以及衝突策略量表兒童父母版（Conflict Tactics Scale: Parent-Child Version, CTSPC）的身體襲擊量尺間有顯著正相關，而照顧者的命令總數與CTSPC的身體襲擊量尺間亦有顯著正相關，照顧者的回應率與CTSPC的身體襲擊和心理攻擊量尺間則有顯著負相關；而以DPICS-III中文版觀察到兒童對照顧者所有命令的服從率，與ECBI嚴重度和問題量尺、PSI全量表、PSI兒童和父母分量表間皆有顯著負相關。
討論：本研究發現台灣和美國樣本之父母在與兒童互動時的言語有多項差異，顯示父母言語行為在跨文化上的不同。整體而言，本研究證實DPICS-III中文版具有良好至傑出程度的觀察者間信度，且部分行為有適當的聚合及區辨效度。因此，初步檢驗結果證實DPICS-III中文版是一套具有適當信效度的行為觀察工具，建議日後研究者或臨床工作者在評估父母教養行為時，能使用此工具來評估親子互動關係。最後，文中亦針對本研究的研究限制和未來方向、研究貢獻做進一步的說明。

關鍵字

親子互動 ； 行為觀察記錄系統 ； 信效度

並列摘要

Purpose: The purposes of this study were to translate a behavioral observation coding system into Chinese to assess the quality of parent-child interactions (i.e., Dyadic Parent-Child Interaction Coding System – Third Edition, DPICS-III), and to examine its psychometric properties in a Taiwanese sample.
Methods: Sixty-six clinical- referred caregiver-child dyads participated in this study because children displayed emotional and behavioral problems, and/or their caregivers had difficulty in parenting. The data first were divided into two groups that with and without clinically elevated disruptive behavior symptoms which were used to compare to the clinical and non-clinical samples in the study of Bessmer (1996), respectively. Next, the nine parents' behaviors (i.e., Labeled Praise, Unlabeled Praise, Reflection, Behavioral Description, Negative Talk, Direct Command, Indirect Command, Question, and Neutral Talk), and three child's behaviors (i.e., Compliance, Noncompliance, and No Opportunity for Compliance) of the Chinese version of DPICS-III were used to examine its psychometric properties with a series of caregiver-reported measures.
Results: The results of this study showed no significant differences when compared to the compliance rates between Taiwanese and American children, but the frequencies of caregivers’ behaviors between the two samples were significantly different except the frequency of caregivers’ Question. In terms of reliability, the results indicated that the percent agreement ranged from 94.3% to 99.3% for parent behaviors, and from 92.5% to 96.9% for child behaviors; the Cohen's kappa coefficients ranged from .70 to .91 for parent behaviors, and from .67 to .73 for child behaviors. With regard to discriminant validity, the Negative Talk could differentiate between group with and without clinically elevated symptoms that were assessed by the Child Behavior Checklist For Ages 1½-5/6-18 (CBCL 1½-5/6-18) Total Problem and Eyberg Child Behavior Inventory (ECBI) Problem scales. The Reflection rates could differentiate between group with and without clinically elevated symptoms that were assessed by the CBCL 1½-5/6-18 Externalizing Problem scale. The ratio of children’s compliance to caregiver’s commands could differentiate between group with and without clinically elevated symptoms that were assessed by ECBI Intensity and Problem scales. Regarding convergent validity, the correlation analyses showed that the frequencies of Negative Talk were positively correlated with the ECBI Problem, Parenting Stress Index (PSI) Child Domain, and Conflict Tactics Scale: Parent-Child Version (CTSPC) Physical Assault scales; The frequencies of Command Total were positively correlated with the CTSPC Physical Assault subscale; Reflection rates were negatively correlated with the CTSPC Physical Assault and Psychological Aggression subscales; Children’s Compliance rate was negatively correlated with the Intensity and Problem subscales of the ECBI and the Total Scale, and Child and Parent Domain subscales of the PSI.
Discussion: The frequencies of the Taiwanese caregivers’ behaviors assessed by the Chinese version of DPICS-III in this study are different from that in the sample of American caregivers in the study of Bessmer (1996), suggesting that there may exist cultural differences in parent behaviors. In general, the Chinese version of DPICS-III has excellent to good reliability, and acceptable discriminant and convergent validity. When evaluating parenting practices, it is recommended to use the Chinese version of DPICS-III to assess the quality of parent-child interaction in Taiwanese samples for future clinical utility. Limitations and strengths of this study as well as suggestions of further research are discussed.

並列關鍵字

parent-child interaction ； behavioral observation coding system ； psychometric properties

參考文獻

王琳（2008）。親職認知對管教行為的影響：探討兒童衝突的管教。中華心理衛生學刊，21，245-272。

吳佳玲（2008）。台灣親職壓力相關研究之回顧與分析。台南科技大學通識教育學刊，7，63-90。

吳齊殷、黃鈺婷（2010）。青少年初期身心健康變化及其動態影響變因之討論。中華心理衛生學刊，23，535-562。

沈瓊桃（2005）。兒童知覺的雙重家庭暴力經驗與其適應行為之相關研究。中華心理衛生學刊，18，25-64。

林文瑛、王震武（1995）。中國父母的教養觀：嚴教觀或打罵觀？本土心理學研究，3，2-92。