이행적 인과 경로를 통한 원인 효과에 대한 해명: 구조 방정식에 토대한 인과 모형의 원인 효과 개념에 대한 평가와 대안

김준성 ( Joon Sung Kim ) 한국논리학회 2015 論理硏究 Vol.18 No.1

이 글에서는 상반된 이중 원인 효과를 해명하는 데 히치콕 (Hitchcock 2001a)이 제시한 두 가지 원인 효과를 비판적으로 평가하고 그 비 판을 상쇄할 대안을 제시한다. 이를 위해 다섯 가지를 논의한다. 첫째, 상반된 이중 원인 효과에 관한 사례를 소개하고 이에 대한 오테(Otte 1985)와 엘스 (Eells 1987)의 논쟁을 검토한다. 이 검토는 이중 원인 효과에 대한 이해와 그 와 관련해 논의할 히치콕, 카트라이트(Cartwright 1979, 1989, 1995), 엘스(Eells 1991, 1995)의 이론들을 비교, 평가하는 준비가 된다. 둘째, 구조 방정식 모형 에서 두 원인 효과에 착안하여, 히치콕이 제시한 원인 효과의 두 개념, 망 효과 와 요소 효과를 소개한다. 셋째, 이중 원인 효과를 해명하는 데 카트라이트와 엘스의 논쟁을 논의하면서, 동질적 부분 집단의 역할, 인과 상호 작용의 중요성 에 주목한다. 넷째, 이중 원인 효과를 해명하는 데 히치콕이 적용한 두 원인 효 과 개념을 비판적으로 검토하고, 엘스의 이론에 대한 히치콕의 비판을 다시 비 판한다. 다섯째, 구조 방정식 모형의 조절 효과 개념을 인과에 대한 확률 이론 에 응용하여, 이중 원인 효과에서 초래된 인과 상호 작용을 선언(disjunctive) 관계와 반 사실적 요인으로 형식화하여 보여준다. 조절 효과에 대한 논의는 최 근에 많은 주목을 받은 구조 방정식 모형에 토대한 인과 모형 이론들이 인과의 문제들을 해결하는 데 기여할 수 있다. 다른 한편으로, 선언 관계, 반 사실적 조건을 이용한 인과 상호 작용에 대한 논의는 통계적 방법론으로서 구조 방정 식 모형을 제고하는 데 도움을 줄 수 있다. In this paper, I pose a problem for Hitchcock`s arguments for two concepts of effects that are intended to explicate double causal effects, and put forth a theory that is intended not just to meet the problem but also to accommodate Hitchcock`s theory and Eells` theory both. First, I introduce an example of dual causal effects, and examine the accounts of Otte(1985) and Eells(1987) on how to explicate the dual effects. I show that their accounts of the dual effects help us understand the problem of dual effects and see how different it is for Cartwright(1979, 1989, 1995), Eells(1991, 1995), and Hitchcock(2001a) to meet the problem. Second, I introduce two concepts of effects on Hitchcock(2001a), that is, net effect and component effect that are allegedly analogous to two effects of structural equation model. Third, I reveal the significance of homogeneous subpopulation and causal interaction regarding the problem of dual effects while examining Cartwright`s theory and Elles` theory. Fourth, I critically examine the two concepts of effects on Hitchcock and argue against Hitchcock`s criticism of Eells` theory. Fifth, I take a moderator variable of structural equation model and a moderator effect into the probabilistic theory of causality, and formally generalize causal interaction due to the dual effects in terms of disjunctive relation and counterfactual conditionals. I expect my account of disjunctive relation and counterfactual conditionals to contribute not just to several problems the received theories of causal modelling confront but also to the structural equation models many people exploit as a promising statistical methodology.

대학생의 온라인상의 불법 성콘텐츠 접촉과 불법 성콘텐츠 유통의 원인요인과 통제요인의 상호작용효과 비교

이성식(Seong-Sik Lee),임형연(Hyeong Yeon Lim),신지민(Ji-Min Shin) 한국콘텐츠학회 2021 한국콘텐츠학회논문지 Vol.21 No.5

이 연구는 불법 성콘텐츠 접촉과 불법 성콘텐츠 유통의 그 원인요인들 및 통제요인들을 제시하고 그 요인들간의 상호작용효과를 각각 살펴보고 비교하였다. 원인요인으로는 낮은 자기통제력, 불법 성콘텐츠 경험 친구, 그리고 가부장가정 요인을, 그리고 통제요인들로는 도덕태도, 공식처벌, 기회제약을 다뤘다. 그리고 세 원인요인들의 영향에 대한 세 통제요인들의 조절 혹은 완충효과를 살펴보았다. 서울시 대학생을 조사대상으로 하는 분석결과 대체로 불법 성콘텐츠 접촉과 불법 성콘텐츠 유통에서 그 원인요인들의 작용은 유사해 낮은 자기통제력, 불법 성콘텐츠 경험 친구, 그리고 가부장가정 모두가 크게 유의미했고 그 중에서도 불법 성콘텐츠 경험친구와의 접촉의 영향력이 가장 컸다. 불법 성콘텐츠 접촉 및 유통에 대한 원인요인들과 통제요인들의 상호작용효과에 대한 결과를 보면 그 결과가 상이했다. 불법 성콘텐츠 접촉의 경우 원인요인들과 통제요인들과의 상호작용효과가 모두 유의미하지 않았다. 즉 불법 성콘텐츠 접촉에서는 원인요인 모두가 독립적으로만 유의미하게 작동했으며 그 작용을 어떠한 통제요인들도 통제하지 못했다. 그러나 불법 성콘텐츠 유통의 경우는 낮은 자기통제력과 도덕태도간의 상호작용효과, 불법 성콘텐츠 유통 친구와 도덕태도간의 상호작용효과가 유의미했고, 가부장가정과 공식처벌간의, 그리고 가부장가정과 기회제약간의 상호작용효과가 유의미했다. 그 결과의 함의를 논의한다. This study considers both causal and control factors for explaining both contact and distribution of sexually explicit material and tests their interaction effects for examining buffering effects of control factors in causal process. This study examines the effect of low self-control, differential association with peers and patriarchy family environmental factor as causal factors. It also considers three control factors such as morality. legal punishment, and lack of opportunity and test interaction effect between causal and control factors in both contact and distribution cases. Using data from college students in Seoul, results show that the effects of all causal factors are significant in both contact and distribution of sexually explicit material and the effect of differential association is the most significant. However, their interaction effects between causal and control factors are different across contact and distribution cases. In the contact case, all interaction effects are not significant. It means that causal factors have independent effects and control factors have no buffering effects. However, in the distribution case, the interaction effects between low self-control and morality, between differential association and morality, between patriarchy family and punishment, and between patriarchy family and lack of opportunity are significant. Empirical implications are discussed.

유효법인세율과 부채비율 사이의 인과관계 분석과 세법상 지급이자 손금불산입 규정의 유효성 평가

신승묘 ( Seung Myo Shin ) 한국회계학회 2005 회계학연구 Vol.30 No.4

그 동안 세무회계 분야의 많은 연구들에서 기업특성의 차이에 따라 유효법인세율의 차이가 관찰되는가를 실증분석하였다. 이중에서 전규안(1997)은 우리나라의 상장기업을 대상으로 실증분석한 결과 부채비율이 높은 기업일수록 유효법인세율이 높다는 결과를 얻었으며, 이에 대하여 부채비율이 높을수록 세법상의 지급이자 손금불산입 규정에 따라 손금불산입되는 지급이자가 많기 때문일 것으로 해석하였다. 만약 그렇다면 이는 세법상의 지급이자 손금불산입 규정이 당초에 의도했던 효과를 충분히 달성하지 못하고 있음을 시사한다. 그런데 기업은 가능하면 지급이자의 손금불산입 규정으로 인한 세제상 불이익을 회피하고자 할 것이 분명하며, 따라서 부채비율이 높은 기업일수록 유효법인세율이 높게 나타난 이유는 부채비율이 높을수록 손금불산입되는 지급이자가 많기 때문만이 아닌 다른 데도 있을 가능성이 존재하는 것으로 생각된다. 예컨대 높은 부채비율이 높은 유효법인세율을 야기할 뿐만 아니라 그와는 반대로 높은 유효법인세율이 높은 부채비율을 야기할 수도 있다. 이러한 가능성이 있음을 감안하여 본 연구에서는 통상적인 최소자승법(ordinary least square method)이 아니라 2단계 최소자승법(2 stage least square method)에 의한 회귀분석을 실시함으로써 유효법인세율과 부채비율 사이의 실질적인 인과관계를 실증하고 그 실증결과를 토대로 세법상 존재하는 지급이자 손금불산입 규정의 유효성을 평가해 보았다. 실증결과 높은 부채비율이 높은 유효법인세율을 야기한다는 인과관계가 존재하기는 하나 그 정도가 기존의 실증결과가 제시한 것보다는 덜 유의적이었으며, 이는 과도하거나 부적절한 부채의 사용을 억제한다는 취지에서 도입된 세법상의 지급이자 손금불산입 규정이 기존의 실증결과가 시사하는 것보다는 당초에 의도한 효과를 상당부분 달성하고 있음을 의미한다. There have been many empirical studies investigating whether firms` characteristics affect their effective tax rates. As one of them, K.A. Jeon (1997) studied the determinants of the effective tax rates of Korean listed firms. The study obtained empirical results demonstrating that the effective tax rates are higher for firms with a higher debt ratio, and concluded that the reason for this is that the amount of interest expense that is not deducted in calculating tax to be paid is larger for firms with a higher debt ratio due to the tax provision for limiting the tax deductibility of interest expense. If this is the case, it implies that the tax provision does not satisfactorily achieve its originally intended effectiveness. By the way, firms will evidently try to avoid the tax disadvantages of the tax provision for limiting the tax deductibility of interest expense. Therefore, there may be another reason why the effective tax rates are higher for firms with a higher debt ratio. For example, it is also possible that higher effective tax rate causes higher debt ratio. That is, the causality between debt ratio and effective tax rate can be bi-directional. Taking this possibility into account, this study empirically tests the de facto causality between a debt ratio and an effective tax rate by performing the 2-stage least square regression analysis instead of the ordinary least square regression, and, based on test results, evaluates the actual effectiveness of the tax provision for limiting the tax deductibility of interest expense. The test results show that higher debt ratio causes higher effective tax rate, but the degree of this causality is not as strong as that presented in the existing literature. This means that the tax provision for limiting the tax deductibility of interest expense achieves its intended effectiveness considerably better than is implied by the existing literature.

독일의 예방접종 이상반응 신고와 부작용에 대한 피해보상

차성안 한국사회보장법학회 2022 사회보장법학 Vol.11 No.2

In Germany, the causality review for adverse events following immunization (AEFI) reports is conducted by the PEI Institute at the federal level, and damage compensation is made by individual compensation managers in each state. The AEFI reporting process and damage compensation process are separated. It is a structure in which the person involved in the causality evaluation of the AEFI report cannot be involved in the judgment of the normative causal relationship related to damage compensation. To determine the causality between vaccination and AEFI, the WHO's Adverse Event Causality Evaluation Criteria are used. Regarding the normative causal relationship judgment, detailed guidelines are provided in the relevant rules. The Korean AEFI Guidelines also suggest damage compensation standards as well as causality review standards for AEFI reports. The causality criteria of the three categories of WHO should be introduced into the causality review criteria of Korea's AEFI management guidelines. In particular, the ‘indeterminate’ category should be formally introduced, but the attitude of uniformly regulating the above category in the guidelines as a type that has no causal relationship in damage compensation should be discarded. The criteria for judging causality in compensation for damages should be newly set in a form that can consider even normative causality, which is different from the criteria for reviewing causality between the vaccination and AEFI. It is necessary to refer to the contents of the rules related to the determination of causality in Germany's Act on Care for Victims of War. It is necessary to improve the situation where there are only two lawyers among the 15 committee members who play the most important role in the vaccination damage compensation hearing in Korea. At the damage compensation expert committee stage, it is necessary to ensure that the majority of members reflect the views of lawyers and ordinary citizens. In the long term, the above improvements are needed for all vaccinations, but in the short term, it is necessary to consider enacting a special law on compensation for damages from COVID-19 vaccination side effects. 독일의 경우 예방접종 이상반응 신고에 대한 인과성 검토는 연방 차원의 PEI 연구소에서 이루어지고, 피해보상은 각 주의 개별 보상담당자에 의하여 이루어진다. 이상반응 신고 절차와 피해보상 절차가 분리되어 진행된다. 이상반응 신고 인과성 평가에 관여한 사람이 피해보상과 관련된 규범적 인과관계 판단에까지 관여할 수는 없는 구조이다. 예방접종과 이상반응 사이의 인과성 판단은 WHO의 이상반응 인과성 평가 기준을 활용한다. 규범적 인과관계 판단 관련하여서는 원호의학규칙에서 상세한 지침을 제시한다. 한국의 이상반응 지침에서는 이상반응 신고에 대한 인과성 검토기준과 함께 피해보상 기준에 대해서도 제시하고 있다. WHO의 3가지 카테고리의 인과성 판단기준을 한국의 이상반응 관리지침의 인과성 검토기준에도 도입해야 한다. 특히 ‘불확실한(indeterminate)’ 범주를 정면으로 도입하되, 위 범주를 피해보상에서 인과관계가 없는 유형으로 획일적으로 지침에서 규정하는 태도는 폐기되어야 한다. 피해보상 기준은 이상반응 신고 인과성 검토기준과 다른 의미로 규범적 인과관계까지 고려할 수 있는 형태로 새로 정해 넣어야 할 것이다. 그 과정에서 독일 연방원호법 원호의학규칙의 내용이 참고될 수 있다. 한국의 예방접종 피해보상 심의에서 가장 중요한 역할을 하는 15명가량의 피해보상전문위원회의 위원들 중 변호사 등 법률가가 2명에 불과한 상황을 개선할 필요가 있다. 피해보상 전문위원회 단계에서는 법률가와 일반시민의 관점을 반영하는 자의 비중이 과반수를 차지하게 할 필요가 있다. 장기적으로는 모든 예방접종에 대하여 위와 같은 개선이 필요하지만, 단기적으로는 코로나19 예방접종 피해보상에 관한 특별법을 제정하는 것도 고려해볼 필요가 있다.

물리주의 인과론과 정신인과

박정희(Park, Jung Hee) 새한철학회 2018 哲學論叢 Vol.94 No.4

이 논문에서는 인과법칙의 특성에 대하여 고찰하면서 흄과 데이비슨, 그리고 김재권의 인과론에 관하여 살펴보았다. 흄은 인과관계에서 필연성이 필수 요소라는 점을 인정하지만, 결과의 관념을 원인의 관념에서 필연적으로 이끌어 낼 수 없기 때문에 인과적 필연성은 관념의 세계에 속하지 않는다고 본다. 데이비슨의 인과론은 사건 존재론에 근거하고 있는데, 정신적 사건과 물리적 사건 어느 것으로도 기술될 수 있는 개별 사건에 의해서 성립하는 인과관계이다. 데이비슨에 따르면 정신적인 것은 물리적인 것으로 환원할 수 없지만, 정신적 사건은 물리적 사건이기 때문에 정신적 사건은 원인이 될 수 있다. 김재권은 정신적 인과관계는 물리적 인과관계로 환원되고, 정신적 원인은 물리적 원인과 법칙을 통해서 환원적 관계로 인과적 역할이 가능하다고 주장한다. 그래서 물리 세계의 인과 폐쇄의 원리에서 볼 때 환원될 수 없는 정신적 사건과 정신적 사건의 인과적 효력을 주장하는 비환원적 물리주의는 유지될 수 없다는 것이 김재권의 입장이다. 따라서 그는 정신적 속성 때문에 정신적 사건이 물리적 사건을 야기했다고 할 수 없기 때문에, 정신적인 것의 인과적 역할을 주장할 수 없다고 데이비슨의 인과에 대해서 비판한다. 결론적으로 말하자면 흄은 인과관계에 있어서 필연성을 거부하는 그 이유에 대한 해명이 분명하지 못하여 설득력을 잃는데 반해, 데이비슨의 인과론은 흄의 사건 모형이론으로부터 영향을 받아서 그 나름대로 독특한 인과론을 전개하였다. 반면, 김재권은 물리 세계의 인과 폐쇄의 원리와 정신인과가 양립할 수 있는 가능성으로써 심리 물리 환원 법칙을 함축하고 있는 인과적 환원주의와 수반적 인과론을 제시하였다. In this paper, we discussed the characteristics of the causality and the causal theories of Hume and Davidson and Kim Jae-gwon. Hume acknowledges that necessity in causal relationships is essential, but sees causal necessity as not part of the world of ideology because it cannot necessarily derive the concept of consequences from the idea of the origin. Davidson`s causal theory is based on the theory of the existence of events, a causal relationship formed by individual events that can be described by both mental and physical events. According to Davidson, mental events can be caused because they are physical events. Kim argues that the cause and effect of the spirit is reduced to physical causality, and that the cause of the spirit is reduced through physical causes and laws. Therefore, Kim`s position is that the non-recoverable physical principle that claims the causal effect of psychological and mental events that cannot be reversed cannot be maintained in view of the principle of causal closures in the physical world. Therefore, he criticizes Davidson`s cause for causal roles in spirituality, because mental events do not result in physical events. In conclusion, while Hume`s explanation of why he refuses to accept the inevitable in causal relationship is unclear, Davidson`s causal theory was influenced by Hume`s case model and developed his own unique theory. On the other hand, Kim suggested causal reductionism and, that implies the law of mental-physical reduction as a compatible possibility of the principle of causality closure of the physical world and mental causality.

Exploring modern machine learning methods to improve causal-effect estimation

Kim, Yeji,Choi, Taehwa,Choi, Sangbum The Korean Statistical Society 2022 Communications for statistical applications and me Vol.29 No.2

This paper addresses the use of machine learning methods for causal estimation of treatment effects from observational data. Even though conducting randomized experimental trials is a gold standard to reveal potential causal relationships, observational study is another rich source for investigation of exposure effects, for example, in the research of comparative effectiveness and safety of treatments, where the causal effect can be identified if covariates contain all confounding variables. In this context, statistical regression models for the expected outcome and the probability of treatment are often imposed, which can be combined in a clever way to yield more efficient and robust causal estimators. Recently, targeted maximum likelihood estimation and causal random forest is proposed and extensively studied for the use of data-adaptive regression in estimation of causal inference parameters. Machine learning methods are a natural choice in these settings to improve the quality of the final estimate of the treatment effect. We explore how we can adapt the design and training of several machine learning algorithms for causal inference and study their finite-sample performance through simulation experiments under various scenarios. Application to the percutaneous coronary intervention (PCI) data shows that these adaptations can improve simple linear regression-based methods.

우로보로스 효과(Ouroboros Effect)와 그 네트워크의 인과적 특성 연구

홍영교 ( Young Kyo Hong ),김태연 ( Tae Yeon Kim ),최남희 ( Nam Hee Choi ),이만형 ( Man Hyung Lee ) 한국시스템다이내믹스학회 2013 한국시스템다이내믹스 연구 Vol.14 No.2

This article deals with a Causal-Loop analysis of the characteristics of Ouroboros effects. The meaning of the effects is that solutions to problems are usually intended as final fixes, but more often than not, while solving one problem, they generate more problems. It means ``to worsen through attempts to make better``. The Ouroboros effects are very dangerous because they are regarded as a kind of self-destructive mechanisms. While existing researchers recognize the risks of the effects, they are not able to solve the systematic problem but only to describe the symptom. Such a way to deal with the effects results in difficulty to clarify the cause-and-effect relationship within the system. Thus the authors are focusing on explaining the effects by SD (System Dynamics), which are considered as the sole methodology to do effectively so far since SD is a great tool to discover structural causes. To solve the self-destructive problem of the effects, the authors analyze the characteristics of several cases of the effects in elaborating the CLD (Causal Loop Diagram) for each case. The analysis reveals the structural characteristics of the effects and how to correct them towards maintaining the sustainability of systems.

오피스시장 주요 변수의 인과 효과 분석

금상수(Keum, Sang-Su) 한국주거환경학회 2014 주거환경(한국주거환경학회논문집) Vol.12 No.1

This empirical study analyzes the causal effect which is caused by relationship interactions among the Significant component variables that affect office rent in Seoul through path analysis. The results, exogenous variables (distance to subway, GFA) are affected by the endogenous variables (building grade, management services, rent) directly and indirectly, and endogenous variables are affected by the subsequent variables directly and indirectly. In addition, the path coefficients of the hypothesis path were analyzed to estimate the significance and the direct effects and indirect effects were calculated coefficients for a causal effect. In other words, rent in Seoul office market was analyzed to influence directly or indirectly with interaction of the Significant component variables : accessibility, GFA, managed services, building grade. This study is meaningful that fragmented and decomposed the results of previous studies and useful information to office market stakeholders.

Relationship between the leadership style and organizational effectiveness

Sang-Young Lee(이상영),Hae-Sool Yang(양해술) 한국컴퓨터정보학회 2017 韓國컴퓨터情報學會論文誌 Vol.22 No.10

In this paper, we propose the role of job crafting in the relationship between leadership and organizational effectiveness in voluntarily carrying out each member’s assigned tasks. This study surveyed the manufacturing, construction, service industries in Seoul and Gyeonggi province, identified the type of leadership they recognized, and empirically analyzed the organizational effectiveness of leadership. The purpose of this paper is to grasp the types of leadership acknowledged by the industry of manufacturing, construction, and service, and also to empirically analyze the organizational effectiveness of the leadership. The study measures the organizational effectiveness in terms of the job satisfaction, organizational commitment, and organizational citizenship behavior while classifying the leadership into coaching leadership, transformational leadership, and sensible leadership. In addition, the strictness of the analysis is imposed by estimating the simple least square model and ranking probit model. The results of the least square model is summarized as the following. Regardless of the different defining terms of organizational effectiveness, transformational leadership was shown to have the greatest organizational effectiveness. Sensible leadership positively effected job satisfaction whereas coaching leadership positively effected job satisfaction and organizational effectiveness. Compared to transformational leadership and coaching leadership, the impact of the sensible leadership was very much limited. The result of the ranking probit model is summarized as the following. First, sensible leadership had a positive impact on the member’s job satisfaction and organizational commitment. Second, regarding the organizational citizenship behavior, coaching leadership showed greater impact than transformational leadership. This results contradicts the results from the simple least square model. If similar studies were to be conducted in the future, two models and the results must be compared. Third, as the leadership score increases by 1 point, there is greater possibility of having more than 4 points for all job satisfaction, organizational commitment, organizational citizenship behavior. Lastly, the analysis proves that job crafting has the mediation effect between the leadership and organizational effectiveness.

Semi-automatic extraction of technological causality from patents

Kim, Hongbin,Joung, Junegak,Kim, Kwangsoo Elsevier 2018 COMPUTERS & INDUSTRIAL ENGINEERING Vol.115 No.-

<P><B>Abstract</B></P> <P>The goal of this study is to suggest a method to extract technological causalities from patents, which are formal documents that include a large amount and large variety of information about technology. The core of patents is composed of both inventive principles to solve problems, and purposes that the invention achieves by solving them. The principles and purposes can be understood as a concept of technological causality which is reusable knowledge as technological analogy. Because reading and understanding patent documents that generally consist of dozens of pages and have difficult and profound statement of technologies is hard even for technology experts, a method to extract technological causalities is needed. As a solution, this paper proposed a method to extract technological data from patents, to identify technological causes and effect relation from the extracted data and to calculate the representativeness of technological causes and effect. Based on this study, technology experts can be given a list of ranked alternatives for technological causes and effect. This study helps to analyze patent, and it finally contributes to new product development and technology opportunity discovery. To achieve the objectives, the proposed method included the characteristics of patents that are structured documents consisting of various particular fields that have each different contents and importance. And natural language processing technology is adopted to automatically extract meaningful data and to perform linguistic processing. The implementation and case study of the proposed method demonstrated how a prototype system can be developed and utilized.</P> <P><B>Highlights</B></P> <P> <UL> <LI> We proposed a method to extract technological data from patents. </LI> <LI> Causalities are condensed extraction of patent as a most abstractive expression. </LI> <LI> Extracted technological causalities can be reusable and applied to other domains. </LI> <LI> Causalities can be built a network consisting of nodes connected by causalities. </LI> <LI> Network users can find various principles or solutions to solve problems. </LI> </UL> </P>