The college admission process, based on fair principles and methods, is a means to select students that colleges require. Therefore, current elements of the college admission process must accurately predict the potential scholastic aptitude of the st...
The college admission process, based on fair principles and methods, is a means to select students that colleges require. Therefore, current elements of the college admission process must accurately predict the potential scholastic aptitude of the students colleges seek to obtain. However, whether current admission factors demonstrate a high predictive validity remains a controversial question, since preceding research proposes mixed results on the influence and the predictive power of these factors. This study has, therefore, attempted to provide suggestions in creating a highly valid college admission process system in the future, by conducting an integrated analysis on the effectiveness of admission factors that influence college student academic achievement, and comparing the effect size of current admission factors to draw a general conclusion.
A meta-analysis on the research outcomes that provided analysis results of the correlation between admission factors and college student academic achievement in Korean college students admitted after the College Scholastic Ability Test (CSAT, also abbreviated as Suneung) was first conducted in 1994. It revealed that the average effect size of admission factors such as CSAT, school records, interview, and essay examination all showed meaningful static effectiveness on the level. However, unlike school records and interview factors, the effectiveness of the CSAT and essay examination in the admission process was slight, and did not surpass the rate of 0.1, the cut-off score of the ‘small effect size’ that Cohen(1988) proposed. This means that current admission factors, except for school records and interview, are rather ineffective measuring instruments in predicting college student academic achievement. On the other hand, this also shows that school records and interview can be recognized as factors with a relatively high predictive validity among admission factors in that they each showed a moderate effect size.
This correlates with the results of preceding researches that school records have a stronger predictive effectiveness on college student academic achievement than the CSAT. In terms of predictive validity, this result shows the common belief that the CSAT – as a nationwide standardized test – is the fairest and most objective evaluation method is not in accordance with the reality.
If we were to assess the appropriate reflection ratio of admission factors based on the study results above, we would have to give weight to school records, interview, essay examination, and CSAT in decreasing order, according to the relative effect size. In particular, the actual reflection ratio of school records should be increased so as to be utilized as credible information in predicting college student academic achievement. However, overall reconsideration of current admission factors will still have to be made, since all of these factors, including school records, each have an effect size rate of less than 0.2. A new mechanism to record and assess learners’ cognitive and affective traits in detail on a high school level is also called for.
In this study, a general conclusion was drawn on the effectiveness of admission factors in the Korean college admission process, by attempting to put together and meta-analyze the effectiveness of admission factors on predicting college student academic achievement, and then checking their effect sizes. Also, in terms of research methodology, this study compensated for the defects of the traditional testing method for the regulation effect by utilizing both the analog to ANOVA and meta-regression analysis.
Practically speaking however, this study discloses the possibility of publication bias because meta-analysis could only be conducted on published researches. Also, due to the pursuit of the simplicity of research models as well as the insufficient accumulation of research that provides the variables of research properties, the lack of diversity and detail in moderator variable analysis leaves much to be desired. If related researches putting diverse variables into consideration is sufficiently accumulated in the future, the creation of a more stable regression model will become possible through further research.