自我照顧表現測驗於學齡兒童之心理計量特性驗證

Abstract

背景：學齡兒童在8項自我照顧活動之表現影響其生活獨立、環境適應與人際關係。此外，男女兒童受生理發育與社會文化影響，自我照顧表現差異漸增且發展性別特定之自我照顧活動，如：使用衛生棉。因此，亟需評估學齡男、女童之自我照顧表現以確認後續協助資源。自我照顧評估工具須滿足2項特質：（一）完整評量8項活動。（二）題目設計考量性別因素。自我照顧表現電腦適性測驗 (Computerized Adaptive Test-Self Care, CAT-SC)之候選題庫符合上述2項特質。然而，CAT-SC之候選題庫為訪談形式且為4點量尺不利於照顧者評分以及廣泛應用。本研究修改CAT-SC之候選題庫為照顧者自填式問卷以及簡化為3點量尺而發展學齡兒童自我照顧表現測驗 (Self-Care Performance Assessment for Children, SCPAC)。本研究進一步驗證SCPAC之心理計量特性以確認其可準確評量男、女童之自我照顧表現、評量結果是穩定的，以及可靈敏偵測男、女童自我照顧表現之改變。 目的：本研究旨在驗證SCPAC 8向度與總分於男、女與整體學齡兒童之心理計量特性，包含：內在一致性、再測信度、隨機測量誤差、同時效度、區辨效度、預測效度、團體層級反應性與個別層級反應性。 方法：本研究問卷由兒童照顧者填寫。納入條件有三：兒童就讀國小、照顧者與兒童同住，熟悉其自我照顧表現，以及照顧者具國中以上學歷，能閱讀繁體中文與填寫問卷。照顧者於3個時間點填寫問卷。（一）初評：填寫SCPAC與文蘭適應行為測驗第三版 (Vineland Adaptive Behavior Scale Third Edition, VABS-3)之自我照顧向度。（二）再評（1星期後至2個星期間）：填寫SCPAC。（三）追蹤（6個月後）：填寫SCPAC與VABS-3之家庭生活與社區生活向度。在統計分析上，內在一致性使用Cronbach's alpha、再測信度使用Intraclass Correlation Coefficient (ICC)、隨機測量誤差使用Percentage of Minimal Detectable Change (MDC%)分析。同時效度使用Pearson r與VABS-3之自我照顧向度進行相關分析。預測效度使用Pearson r 與VABS-3之家庭生活向度以及社區生活向度進行相關分析。區辨效度使用獨立樣本t檢定與ROC曲線下面積 (Area under the ROC Curve, AUC)進行分析。團體層級反應性使用Standardized Effect Size (ES)與Standardized Response Mean (SRM)分析。個別層級反應性以MDC做切點，並計算顯著進步組佔總人數比例。同時效度、預測效度與個別層級反應性僅分析SCPAC總分。其他心理計量指標則對總分與8向度皆分析。 結果：本研究共招募341位兒童，其中男生212人 (62%)。SCPAC總分在3個族群於8項心理計量特性皆良好至優異。內在一致性α=0.98~0.99。再測信度 ICC=0.96~0.97。隨機測量誤差MDC=11.22~11.65與MDC%=9.02%~10.63%。同時效度r=0.95~0.96。預測效度r=0.59~0.66。區辨效度一般發展與落後組2組有顯著差異 (p<.05)且AUC=0.87。團體層級反應性ES=0.23~0.26與SRM=0.64~0.67。個別層級反應性在所有兒童情形下可偵測到20%~22%兒童顯著進步；有疾病診斷兒童情形下，可偵測到31%~60%兒童顯著進步。SCPAC 8向度，大部分向度在3個族群之心理計量特性為可接受至良好。少數向度於信度不佳，包含：確保個人安全向度在女生族群之內在一致性低 (α=0.65)。照顧身體部位、照料個人健康與確保個人安全向度在3個族群之隨機測量誤差大 (MDC =1.60~4.71；MDC%=30.41%~50.87%)。 結論：SCPAC總分與多數向度在3族群之心理計量特性良好，可真實反映學齡兒童自我照顧表現並作為療效驗證工具。少數向度信度不佳，包含：照顧身體部位、照料個人健康與確保個人安全。本研究建議確保個人安全向度可增加題數或納入更多不同表現之學齡兒童以驗證內在一致性。照顧身體部位、照料個人健康與確保個人安全3向度宜參考2次以上結果取平均，以反映兒童實際表現。

Background: The performance of school-aged children in 8 self-care activities affects their independence, environmental adaptation, and interpersonal relationships. Additionally, Due to physiological development and socio-cultural influences, the differences in self-care performance between boys and girls gradually increase, leading to the development of gender-specific self-care activities, such as using sanitary pads. Therefore, it is crucial to assess the self-care performance of school-aged boys and girls to identify the necessary support resources. A good self-care assessment must meet two criteria: (1) comprehensive evaluation of 8 activities, and (2) questions designed considering the influence of gender factors. The candidate item pool of the Computerized Adaptive Test-Self Care (CAT-SC) meets these two criteria. However, the candidate item pool of the CAT-SC is in interview format and uses a 4-point scale, which is not conducive to caregiver scoring and widespread application. This study modifies the CAT-SC candidate item pool into a caregiver self-administered questionnaire and simplifies it to a 3-point scale, developing the Self-Care Performance Assessment for Children (SCPAC). Furthermore, this study validates the psychometric properties of the SCPAC to ensure it accurately assesses the self-care performance of boys and girls, that the assessment results are stable, and that it sensitively detects changes in the self-care performance of boys and girls. Purposes: This study aims to validate SCPAC 8 subdomains and total score across boys, girls, and the overall school-aged population. The psychological measurement properties to be assessed include internal consistency, test-retest reliability, random measurement error, concurrent validity, discriminant validity, predictive validity, group-level responsiveness, and individual-level responsiveness. Methods: The study's questionnaire is completed by child caregivers. Three inclusion criteria are used: children in elementary school, caregivers living with children and knowledgeable about their self-care, and caregivers with education beyond junior high school, able to read Chinese characters and perform paper tasks. The questionnaire is given at 3 time points: (1) Initial test: Completing the SCPAC and the personal subdomain of the Vineland Adaptive Behavior Scale, Third Edition (VABS-3). (2) Retest (7 days later): Completing the SCPAC. (3) Follow-up (6 months later): Completing the SCPAC and the domestic and community subdomains of the VABS-3.In statistical analysis, Cronbach’s alpha, the intraclass correlation coefficient and the percentage of minimal detectable change (MDC%) were used to examine the internal consistency, test–retest reliability and random measurement error, respectively. Pearson’s r was used to examine the concurrent validity with the personal subdomain of the VABS-3 and to examine the predictive validity with the domestic and community subdomains of the VABS-3. For discriminant validity, independent-samples t-tests and the area under the receiver operating characteristic curve (AUC) were used. The group-level responsiveness was measured by the standardized effect size and the standardized response mean. The individual-level responsiveness was measured using MDC as a cutoff point, and calculated the proportion of individuals showing significant improvement. Concurrent validity, predictive validity, and individual-level responsiveness were assessed only for the total score of the SCPAC. Other psychometric indicators were examined for both the total score and the 8 subdomains. Results: 341 children were included, with 212 boys (62%). SCPAC total scores showed good to excellent psychological measurement properties across all three groups. The internal consistency ranged from α = 0.98~0.99, test-retest reliability from ICC = 0.96~0.97, and random measurement error from MDC% = 9.02%~10.63%. Concurrent validity (r = 0.95~0.96) and predictive validity (r = 0.59~0.66) showed high correlations with VABS-3 subdomains. Discriminant validity showed significant differences (p < .05) between the typically developing and delayed groups, with the AUC of 0.87. Group-level responsiveness was indicated by ES = 0.23~0.26 and SRM=0.64~0.67. Individual-level responsiveness showed significant progress in 20% to 22% of children, and in 31% to 60% of children with diagnosed conditions. SCPAC 8 subdomains generally showed acceptable to good psychological measurement properties across three groups. Only a few subdomains exhibited poor reliability, including: the “Looking after one’s safety” subdomain with low internal consistency in the girls' group (α=0.65). The“Caring for body parts”, “looking after one’s health” and “looking after one’s safety” subdomains showed large random measurement errors across all three groups (MDC=1.60~4.71; MDC%=30.41%~50.87%). Conclusion: SCPAC total score and most subdomains exhibited good psychological measurement properties across the three groups, making it a reliable tool to reflect school-aged children's self-care performance and serve as a validation instrument for therapeutic efficacy. However, a few subdomains showed poor reliability, including “Caring for body parts”, “looking after one’s health” and “looking after one’s safety”. This study suggests increasing the number of items for the “Looking after one’s safety” subdomain or including a more diverse sample of school-aged children with varying levels of performance to validate its internal consistency. For the “Caring for body parts”, “looking after one’s health”, and “looking after one’s safety” subdomains, it is recommended to have caregivers complete the questionnaire more than twice and average the results to reflect children's actual performance.