本研究之目的乃在透過統合分析法(meta-analysis),對柯氏性格量表過去34年來所累積
的資料,做客觀計量性的分析,以了解其效度概化的程度.本研究經電腦檢索,溯洄法及柯永
河先生提供後,計蒐集到87篇有關柯氏性格量表的施測或研究報告,依本研究所設定的品質
評鑑標準及研究類型後,共選取其中52篇施測報告作為本研究的資料,然後將這52篇施測報
告依受試者的性質分為六類,以為本研究的自變項,分別為心理症病人,生理症病人,不同心
理症病人,犯人,特殊族群以及男女,最後以研究者與張沼澤所發展的電腦程式Meta-HC作為
分析的工具.研究結果顯示:在心理症病人與一般人的比較中,有八個分量尺的效果量在中
度水準以上;在不同心理症病人的比較中,有六個分量尺的效果量在中度水準以上;在生理
症病人與一般人的比較中,只有兩個分量尺的效果量在中度水準以上;在犯人與一般人的比
較中,有五個分量尺的效果量在中度水準以上;在特殊族群與一般人的比較中,有七個分量
尺的效果量在中度水準以上;在男女的比較中,只有兩個分量尺的效果量在中度水準以上.
由這些結果顯示:當以柯氏性格量表作為人格上的區辨工具時,以在心理症病人的效度最高
,其次依序為特殊族群,不同心理症病人,犯人;而當施測於生理症病人與男女之間,其區辨
力則稍嫌不足.本研究雖由縝密的研究設計而得到豐富的研究結果,但因現有資料的限制,
針對特定精神病的研究相當少, 為能獨自成為一自變項以為分析的單位,因此本研究便無
法對柯氏性格量表在不同特定精神疾病間的區辨力做一明確的判斷,只能籠統地說柯氏性
格量表在不同心理症病人間是具有區辨力的,也就是說目前所擁有的是測驗的訊息,但診斷
的功能則稍嫌不足.綜觀本論文的結果:顯示柯氏性格量表的效度概化程度頗高,區辨力頗
為理想,因此是一個值得進一步開發,研究與推廣使用的人格測驗.

This research aims at objective and statistical analysis on the investigat
ivedata Ko's Mental Health Questionnaire (KMHQ) in the past thirty-four yearst
o get some idea about the extent of KMHQ's validity generalization. Thi
s research collects eighty-seven surveys and research papers onKMHQ from compu
ter index, the ancestry approach and the information suppliedby Mr. Ke Yeong-h
e. Regarding to the quality rating standard and researchcategory set by this
research, fifty-two pieces are chosen to serveas the fundamental database of t
his research. Then according to the testers' characteristic the fifty-two sur
vey reports are classified into six independent variable categories for this
research. They are psychological-invalids, nonpsychological-invalids, physica
l-invalids, criminals, specialclans and sex. Last but not least, this resear
ch employ the computer programMeta-HC developed by Cang Jao-tzer and the resea
rcher as the analyzing tool. The research reveals that in the psychological
-invalids there have eight subtests weighted average effect size "d" reaching
mid and up, in the nonpsychological-invalids there have six subtests weighted
average effect size "d" reaching mid and up, in the physical-invalids there ha
ve only two subtests weighted average effect size "d" reaching mid and up, in
the criminals therehave five subtests weighted average effect size "e" reachin
g mid and up, in the special clans there have seven subtests weighted average
effect size "d" reaching mid and up, and in the sex there have only
two subtests weighted average effect size "d" reaching mid and up. Th
is showing that KMHQ's standard of validity generalization is pretty high, tha
t the distinction is obvious, and it is an effective distinguishable perso
nality test. Among the six independen
t variables, the weighted average effect size of the psychological-invalids i
s the highest, meaning that KMHQ's distinguishing capability being the strong
est. The next three independent variables are special clans, nonpsychologica
l-invalids and criminals. However, the effect size on physical-invalids and se
x only reach mid or low degree, showing that KMHQ's distinguishing capability
on physical-invalids and sex is not high. This research has quite abundant
result through deliberate research design; however, since the obtainable info
rmation is limited. For example, research papers on specific psychologic
al invalids are very few and thus cannot be singled out as one indepen
dent variable for analysis unit. As a result, this research cannot have a
bsolute judgement on KMHQ's distinguishing capability on specific psycho
logical invalids. What we can say is that, generally speaking, KMHQ certainl
y has distinguishing capability on no npsychological invalids. That is, prese
ntly handy information is the survey, but the diagnosis function is a bit ins
ufficient. To sum up, the utmost conclusion of this
essay is that, the extent of KMHQ's validity generalization is pretty hi
gh, its distinguishing capability is pretty idealized. Therefore, KMHQ is a
personality test that deserves to be further developed, studied as well as pop
ularly adopted.