本研究旨在探討國中資優生鑑定成績與學生入學後三年學業成就表現之相關。研究對象取自國立臺灣師範大學特殊教育輔導區內設置資優資源班的國中十五所 (含六類學科資優)。學生人數總計2637人，畢業於七十七至八十一學年度。樣本分為資優鑑定入選組及落選組，以比較兩組學生學業成就的差異。本研究亦調查教師對於鑑定效度的看法，取之與客觀化的數據資料做比較。樣本選取曾經參與資優生鑑定工作的教師及行政人員共82人。 本研究分為兩部分，第一部分以學生在國一參加資優鑑定的成績，在校三年的學業成就與高中聯考成績，分析不同鑑定工具結果與學業成就間之相關；第二部分以問卷調查方式，搜集教師對於資優鑑定工具及甄選方式的看法。研究結果發現： 1.在北區國中資優決選鑑定會中入選的學生，其三年學業表現及高中聯考成績顯著優於落選組學生，顯示北區囤中資優學生之鑑定工作能夠甄別具有學術性向的學生。 2.各種鑑定工具結果與學業成就間之相關在不同資優類別、不同學校及不同學年度有不一致的情形，唯共同的現象是：加權總分與學生三年學業成績反聯考成績間有高度的相關，可見採用加權總分有其優點，惟比重訂定的方式，各校可參考本研究所提供的實證資料加以調整。 3.各種鑑定工具以學校自編測驗成績 (如：小學成就測驗、國中各社段考及國文作文等成績) 與學生在校三年學業成就表現間之相關最高，准自編成就測驗與聯考成績間之相關情形在各學年度變異頗大，可能是因為學校考題真聯考命題年年不同，命題趨向之一致程度每年不同之故。 4.智力測驗以語文量表部分與學業成就之相關較高。在若干組別，智力與聯考成績間有顯著的相關存在，唯與在校學業表現相關較低。高級瑞文氏推理能力測驗與學業成就間之相關多未達顯著水準，甚至有顯著的負相關存在。 5.性向測驗與學生在校學業表現及聯考成績之相關各學年度變異頗大。雖然部分年度相關達到顯著水準，但多數年度相關不顯著。三種性向測驗以數學及生物性向與學業的相關較高，理化性向相關多不顯著。 6.各資優類組鑑定工具對學業表現預測的情形雖然各校、各年度變異情形頗大，但預測力較高的工具國文組為教師自編測驗及國文作文；英文組為英文段考、魏氏量表及教師自編測驗；數學組為數學段考、魏氏量表及數學性向；自然組為數學段考、生物段考及魏氏語文；數理組為數學段考、生物段考及魏氏語文；一般能力優異組為魏氏量表、國文段考及生物段考。 7.資優班教師對於各種鑑定工具之有效性看法與本研究實證結果不太一致。多數教師仍較肯定智力測驗的效度，研究者認為這也許是受長期以來國內在鑑定資優生時較看重智力測驗所影響；在決定資優人選的方法上，較多教師贊成採用多元截斷標準選取資優生，以其可選取各方面能力俱優的學生；對於決定的標準，教師多傾向於採用現階段所用之比重或截斷標準。研究者建議未來各校可參考本研究的結果，謂整校內鑑定的工具及標準。多元迴歸模式亦可嘗試採用，各校可參考歷年之鑑定資料重新訂定加權比重。

　　　　　Currently, the identification of the gifted students is based on the results of multiple instruments, it brings a critical issue, that is whether the gifted could be identified from a set of test scores with satisfactory, validity or not. The instruments used and the cutoff criteria made usually are the focus of the concern. In the past twenty years, there were few relate reports on the predictive validities of the instruments used to identify gifted students in Taiwan. No professional could accurately/answer the questions such as: Which test could better predict giftedness? or What cutoff level is the most resonable one to select the gifted? The researcher believed the two questions were urgently to be explored. Thus, the purposes of this study, were to analyze the correlationships between the identifying data and the academic records of the students after they entered the gifted programs. There were two groups in this study, one was identified as the gifted group, while the other was identified as non-gifted group, The identification work were at the 7th grade. Those students graduated from junior high schools seperately from 1988 to 1992. The sample were 2637 totally, selected from 15 schools engaged in gifted resource program in Northern Taiwan area. Their teachers were also sun, eyed on their openions about the validities of the identification tests used and the methods to decide who the gifted were. The test scores during identifying process of each student were taken as die independent variables, and the achievement records as well as the scores of ;he Entrance Examination to the Senior High School (EESHS) were taken as the dependent variables. The research results were listed following: 1. There were significantly differences of academic achievement between the gifted group and the non-gifted group, it indicated the identification of the past years could select out the academic 2. Among the different kinds of instruments, the teacher-made achievement test scores correlated with the school performances highest, they predicted the students' school achievements well. But for the correlation with the scores of EESHS, there was great varie9' among different academic years. 3. The verbal scale of WISC predicted the scores of the EESHS better than the performance scale of INISC, but generally speaking, 1(2 scores didn't correlate highly with school records. The Advanced Raven's Progressive Matrix Test showed quitely low correlation with students' achievements. 4. The academic aptitude tests didn't correlate well with school performences, either. The correlations changed a lot among academic years. The Mathematical and Biological Aptitude tests showed better predictivity than the Physical Aptitude rest. 5. The researcher found that the weighting scores of the tests correlated with the students' achievement very well. Using weighting scores to select the gifted probably is better than using any one of single rest to select the students. She suggested keep on using weighting score, but the weighting value should be modified according to the research results. 6. Most teachers affirmed the validity of intelligent test and reported prefer to use multiple- cutoff method to select the gifted. The researcher suggests the school teachers to adjust and improve current identification in the future. Multiple regression method is recommended as a better option to decide who the gifted are.