CAJ | 학술논문

探讨了当相关系数对变量之间相关程度不易进行度量或者无法度量的情况下，是否可以采用关联规则方法来考察变量之间的相互关系。从理论上分析了关联规则方法在心理测量中应用的可能性，并运用Apriori算法挖掘频繁项集，对来华留学生跨文化适应性数据进行分析，将关联规则方法分析结果与Pearson相关、Spearman相关结果进行比较，发现当使用关联规则方法进行相互关系分析时，能够更加简易、有效地获得变量间的相关规则。研究结果表明关联规则方法可以获得变量各个水平具体的支持度和置信度，推导出多变量之间的相关规则，并且对于处理非线性的多变量以及多分类变量的相关是非常有用的方法。
탐토료당상관계수대변량지간상관정도불역진행도량혹자무법도량적정황하，시부가이채용관련규칙방법래고찰변량지간적상호관계。종이론상분석료관련규칙방법재심리측량중응용적가능성，병운용Apriori산법알굴빈번항집，대래화류학생과문화괄응성수거진행분석，장관련규칙방법분석결과여Pearson상관、Spearman상관결과진행비교，발현당사용관련규칙방법진행상호관계분석시，능구경가간역、유효지획득변량간적상관규칙。연구결과표명관련규칙방법가이획득변량각개수평구체적지지도화치신도，추도출다변량지간적상관규칙，병차대우처리비선성적다변량이급다분류변량적상관시비상유용적방법。
When the correlation between variables is difficult to measure by correlation coefficient, the association rules can be used to examine the relationship between the two variables. The possibility of applying the association rules in psychological measurement was analyzed in theories, then the a priori algorithm for mining frequent item sets was used to analyze cross - cultural adaptation data of foreign students. The formal questionnaire was trans- lated into English, Japanese and Korean, then distributed to 11 provinces including 16 universities. Eventually, we obtained 651 valid questionnaires. The formal questionnaire was divided into two parts： adaptation influencing factors and adaptation level, in which adap- tation influencing factors included demographic factors, social and cultural influencing factors, the school environment influencing fac- tors and personality psychological characteristics. The Cronbaeh＇s consistency coefficients of each part of the formal questionnaire was larger than . 7. We used the data mining software Weka to implement the a priori algorithm, and set the minimum support ＂minsup＂ to ~ 2, with the lowest confidence level being not less than. 6. We then extracted information, and used the lift rate to evaluate the rules. Comparing the results between association rules, Pearson correlation and Spearman correlation, we find that using the association rules can be effective to get the rules between variables. The results show that association rules can gain the support and confidence of each level of the variables; it can deduce the rules between the multi - variables ; it is very effective for dealing with the correlation in non - linear multi - variables and multi - categorical variables. This is the first time to introduce the association rules method to psycho- logical measurement; it is a very useful method for analyzing relationship between data.