| 摘要: | 本研究旨在編製一份具良好信、效度的「大學生心理健康篩選量表」，並建立量表常模。主要是想朝篩選性評估的方式進行，使量表可以運用於針對一般大學生的評估，避免落於診斷性評估。透過量表的檢測可以瞭解大學生的潛在困擾特質，藉此篩選出在常模中需要高關懷的對象（高於90％），以及可能是高危險群的對象（高於99％），以供給導師及學校諮商輔導單位做為留意、關懷及追蹤的對象。本研究以自編的「大學生心理健康篩選量表」做為研究工具，以大學生做為研究樣本，其中356位預試樣本，1078位正式施測樣本與常模樣本（當中包含32位曾接受精神科或心理相關門診治療樣本，與495位未接受精神科或心理相關門診治療樣本，兩者做為建構效度樣本）。本研究資料建立專家內容效度，並以SPSS統計軟體進行試題分析、內部一致性Cronbach α係數及折半信度考驗；以因素分析、團體差異分析（採t考驗）及敏感度分析來進行建構效度考驗，研究結果如下：
一、大學生心理健康篩選量表內容：總量表題目共計153題，分為六個分量表，分別為「憂鬱傾向」26題、「焦慮傾向」28題、「自傷傾向」29題、「衝動行為」24題、「精神困擾」29題，以及「效度指標」17題。
二、量表信度考驗：1、內部一致性信度：各分量表Cronbach α係數介於.81至.93之間，總量表α係數為.97，顯示本量表具有良好的內部一致性信度。2、折半信度：各分量表的折半係數介於.83至.90之間，總量表折半係數為.95，顯示本量表雖然量表題目減半，依然具有良好信度。
三、量表效度考驗：1、專家內容效度：量表題目經指導教授及4位專家學者，根據題目所屬定義內涵進行審查，以建立量表內容效度。2、建構效度：本研究利用因素分析法考驗各分量表的因素內涵，結果大致符合原設計構念；另外，以團體差異分析（採t考驗）考驗是否接受精神科或心理相關門診治療的大學生在量表得分的差異，結果顯示曾接受精神科或心理相關門診治療的大學生，除了「衝動行為」分量表外，其餘各分量表與總量表的得分平均數均顯著高於一般大學生；並以敏感度分析來檢驗量表的敏感度，結果顯示具有良好的敏感度，符合理論所說，具有良好建構效度。
四、量表常模建立：本研究分別建立總量表及六個分量表的原始分數與百分等級常模對照表。
最後根據研究的發現，提出具體的建議，以提供教育單位及相關專業人員進行協助及進行研究之用。
This research focuses on constructing a reliable and valid University Student's Mental Health Screening Scale (USMHSS) ,and establishing the norm of scale. The goal of constructing this scale is to go on in the way of the screening assessment, but not diagnosing assessment. It is expecting that the scale can be applied for assessing general university students. Thereafter, the university students' potential worries; the highly-concerned students (higher than 90%) and the highly-risk students (higher than 99%)can be detected by using this scale. According to the scores and explanations, mentors and counselors at the university could follow up these students who need special care. The tool of this study applies the University Student's Mental Health Screening Scale which was created by the researcher. Objects of this study are university students, which include 356 tryout samples, and 1078 formal and norm samples (including 32 people who had had the psychiatric or psychological therapy before and 495 people who hadn’t. Both of these data are constructing validity samples.) The expert content validity was used in this study. The SPSS statistical software was used to analyze items and to test the reliability through Cronbach α coefficient and Split-half coefficient. Moreover, the SPSS was used to test the constructing validity through factor analysis, group-difference analysis (using t-test), and sensitivity analysis. Results of this study are presented as the following:
1.University Student's Mental Health Screening Scale contains the total of 153 items,and 6 subscales. In addition, there are 26 items of “depression-tendency” questions; 28 items of “anxiety-tendency” questions; 29 items of “suicide-tendency” questions; 24 items of “impulse behavior” questions; 29 items of “mental disturbance” questions, and 17 items of “validity indicator” questions.
2.The scale reliability: (1)Cronbach α reliability: α coefficient is from .81 to .93 for each subscale, and α coefficient shows .97 for total scale. It shows a great reliability. (2)Split-half reliability: r coefficient presents .83 to .90 for each scale, and r coefficient is from .95 for the total scale. It shows a great reliability even after reducing the number of questions by half.
3.The scale validity: (1)Expert content validity: Questions were examined by adviser and 4 expert researchers and scholars to ensure the validity of the test. (2)Constructing validity: This study tests the factor content of each subscale by applying factor analysis, and results correspond with the original concept in creation. Besides, the group-difference analysis (using t-test) was used to detect the differences between students who accepted the psychiatric or psychological therapy and those who didn’t. The result reveals that students who accepted relevant outpatient services scored higher on each subscale(except “impulse behavior” scale) and total scales than those who had not accepted relevant outpatient services did. This research tests the sensitivity of each subscales and total scales by using the sensitivity analysis. The result reveals that this study has good sensitivities and fit the hypothesis of the theory which means the constructing validity is well.
4.Constructing the norm of scale: This study establishes the table of percentile rank for the row scores of the total scale and six subscales.
In the conclusion, according to the results of this study, the concrete suggestion were provided to the used of educational and professional research. |
