구조적 인과 모형에 기반한 교수역량에 대한 인과 경로 분석: TALIS 2018 자료를 활용한 한국, 영국, 핀란드 간 비교를 중심으로

양혜원,백순근 한국교육평가학회 2022 교육평가연구 Vol.35 No.4

Based on the structural causal model, this study derived a causal graph that shows the causal relationship between the factors predicting the teaching competency of lower secondary school teachers in South Korea, the UK(England), and Finland. Also, it compared and analyzed the causal path to each country’s teaching competency. To this end, the data of lower secondary school teachers and principals, who participated in TALIS 2018, in Korea, the UK(England), and Finland were analyzed. First, the top 20 factors that predict teaching competency by each country were extracted by applying the mixed-effect random forest technique in consideration of the multi-layer structure of the data. Then, the causal graphs were derived by applying the causal discovery algorithm based on a structural causal model with the extracted predictors. As a result, there were common factors and discrimination factors in the top 20 predictors extracted from each national data, and the causal paths to teaching competency were compared and analyzed in the context of each country based on the causal graph by country. In addition, in the field of education, the possibility of using causal inference based on structural causal models was discussed, and the limitations and implications of this study were presented.

비실험 자료로부터의 인과 추론: 핵심 개념과 최근 동향

최영근,유동현 한국통계학회 2019 응용통계연구 Vol.32 No.2

Causal questions are prevalent in scientific research, for example, how effective a treatment was for preventing an infectious disease, how much a policy increased utility, or which advertisement would give the highest click rate for a given customer. Causal inference theory in statistics interprets those questions as inferring the effect of a given intervention (treatment or policy) in the data generating process. Causal inference has been used in medicine, public health, and economics; in addition, it has received recent attention as a tool for data-driven decision making processes. Many recent datasets are observational, rather than experimental, which makes the causal inference theory more complex. This review introduces key concepts and recent trends of statistical causal inference in observational studies. We first introduce the Neyman-Rubin's potential outcome framework to formularize from causal questions to average treatment effects as well as discuss popular methods to estimate treatment effects such as propensity score approaches and regression approaches. For recent trends, we briefly discuss (1) conditional (heterogeneous) treatment effects and machine learning-based approaches, (2) curse of dimensionality on the estimation of treatment effect and its remedies, and (3) Pearl's structural causal model to deal with more complex causal relationships and its connection to the Neyman-Rubin's potential outcome model. 과학적 연구에서 핵심적인 연구 주제 또는 가설은 대부분 인과적 질문(causal question)을 포함한다. 예를 들어, 전염병 예방을 위한 치료법의 효과 연구, 특정 정책의 시행으로 인한 효용(utility)의 평가에 대한 연구, 특정 사용자를 대상으로 노출된 광고의 종류에 따른 광고의 효과성에 대한 연구는 모두인과 관계(causal relationship)의 추론이 요구된다. 이러한 인과 관계를 다루는 통계적 인과 추론(statistical causal inference)의 주요 관심사 중 하나는 모집단에 일종의 개입(정책 혹은 처치)을 적용한 후 개입의 효과를 정확하게 추정하는 것이다. 인과 추론은 임상실험과 정책결정에서 주로 이용되었으나, 이른바 빅데이터 시대의 도래로 가용한 관측자료가 폭발적으로 증가하였고 이로 인하여 인과 추론에 대한 잠재적 응용가치와 수요가 지속적으로 증가하고 있다. 하지만 가용한 대부분의 자료는 임의실험 기반의 자료와 달리 개입이 임의로 분배되지 않은 비실험 관측자료이다. 따라서, 본 논문은 비실험 관측자료로부터 개입의 효과를 추정하기 위한 인과 추론의 핵심 개념과 최근의 연구동향을 소개하고자 한다. 이를 위하여 본문에서는 먼저 개입의 효과를 Neyman-Rubin의 잠재 결과(potential outcome) 모형으로 나타내고, 개입의 효과를 추정하는 여러 접근법 중 특히 성향점수(propensity score) 기반 추정법과 회귀모형 기반 추정법을 중점적으로 소개한다. 최근 연구동향으로는 (1) 평균 효과 크기 추정을 넘어선 개인별 효과 크기의 추정, (2) 효과크기 추정에 있어서 자료 규모의 증대로 인한 차원의 저주가 야기하는 난제들과 이에 대한 해결방안들, (3) 복합적 인과관계를 반영하기 위한 Pearl의 구조적 인과 모형(structural causal model) 및 잠재 결과 모형과의 비교의 3가지 주제로 구분하여소개한다.

판의 경계에 대한 고등학생들의 정신모형 분석

박수경 한국지구과학회 2009 한국지구과학회지 Vol.30 No.1

본 연구의 목적은 판의 경계에 대한 그리기와 면담의 결과를 바탕으로 정신모형의 유형별 분석 준거를 도출하고 이를 근거로 고등학생들의 정신모형을 밝히는데 있다. 이를 위하여 고등학교 2학년 학생을 대상으로 충돌형, 섭입형, 발산형 판의 경계에 대한 그리기 과제와 개별 면담을 병행하여 실시하고 그 결과를 정성적으로 분석하였다. 본 연구의 결과, 판의 경계에 대한 고등학생들의 정신모형의 유형은 소박한 모형(naive model), 불안정 모형(unstable model), 인과적 모형(causal model), 개념적 모형(conceptual model)의 4가지로 분류되었다. 이러한 정신모형에 대한 분석 준거는 판과 맨틀을 구분하는지, 판과 하부 맨틀의 이동을 파악하는지, 판의 경계 지형의 특징을 인식하는지, 맨틀의 대류와 지형의 특징을 관련짓고 있는지 등의 요인이다. 판의 경계에 대한 ‘소박한 모형’과 ‘불안정 모형’을 지닌 학생들은 맨틀의 대류와 지형의 생성원인을 관련짓지 못한 반면, ‘인과적 모형’과 ‘개념적 모형’을 지닌 학생들은 판의 경계에서 맨틀 대류와 지형의 생성과의 인과 관계를 인식하고 있었다. 정신모형 유형에 따라 지식 획득의 방법에 관한 학생들의 믿음을 조사한 결과, 소박한 모형과 불안정 모형을 지닌 학생들은 외부적인 권위에 의존하는 경향을 나타내었다. The purpose of this study was to derive the criterions of each type of mental models on the plate boundaries and to investigate high school students' mental models on these concepts. The 11th grade student participants were requested to draw the collisional, convergent, and divergent boundaries and were interviewed individually. The drawings and the data gathered through the interviews were analyzed qualitatively. The mental models on the plate boundaries were classified as 'naive model', 'unstable model', 'causal model', and 'conceptual model'. The criterions for analyzing the mental models were the differentiations of the lithospheric plates and the mantle, the explanations of the motion of the plates and lower mantle, the demonstrations of topographical features of the plate boundaries and the causal relationships between the mantle convection and the topographical features. The findings revealed that the students holding 'the naive model' and 'the unstable model' were unable to relate the mantle convection and the three boundaries. In contrast, the students holding 'the causal model' and 'the conceptual model' were able to explain that the mantle convection causes the three boundaries. Also, the types of epistemological belief were different depending on their mental models. Students holding the naive model and the unstable model tended to rely upon the external authorities.

Dynamic causal modeling for calcium imaging: Exploration of differential effective connectivity for sensory processing in a barrel cortical column

Jung, Kyesam,Kang, Jiyoung,Chung, Seungsoo,Park, Hae-Jeong Elsevier 2019 NeuroImage Vol.201 No.-

<P><B>Abstract</B></P> <P>Multi-photon calcium imaging (CaI) is an important tool to assess activities of neural populations within a column in the sensory cortex. However, the complex asymmetrical interactions among neural populations, termed effective connectivity, cannot be directly assessed by measuring the activity of each neuron or neural population using CaI but calls for computational modeling. To estimate effective connectivity among neural populations, we proposed a dynamic causal model (DCM) for CaI by combining a convolution-based dynamic neural state model and a dynamic calcium ion concentration model for CaI signals. After conducting a simulation study to evaluate DCM for CaI, we applied it to an experimental CaI signals measured at the layer 2/3 of a barrel cortical column that differentially responds to hit and error whisking trials in mice. We first identified neural populations and constructed computational models with intrinsic connectivity of neural populations within the layer 2/3 of the barrel cortex and extrinsic connectivity with latent external modes. Bayesian model inversion and comparison shows that interactions with latent inhibitory and excitatory external modes explain the observed CaI signals within the barrel cortical column better than any other tested models, with a single external mode or without any latent modes. The best model also showed differential intrinsic and extrinsic effective connectivity between hit and error trials in the functional hierarchy. Both simulation and experimental results suggest the usefulness of DCM for CaI in terms of exploration of hierarchical interactions among neural populations observed in CaI.</P> <P><B>Highlights</B></P> <P> <UL> <LI> We propose a dynamic causal model for calcium imaging for neural populations. </LI> <LI> The effective connectivity constitutes a convolution-based neural state model. </LI> <LI> A dynamic calcium ion concentration model is proposed as an observation model. </LI> <LI> Connectivity for sensory processing in a barrel column was identified by DCM. </LI> <LI> The results suggest advantages of DCM for CaI in studying neural interactions. </LI> </UL> </P>

Analyzing the Causal Relationship between Exports, Imports, and Economic Growth : The Case of Sub-Saharan African Countries

Denu Lemma, Tsegaye,Tae-hwan, Yoo,Jin-soon, Park 경성대학교 산업개발연구소 2016 산업혁신연구 Vol.32 No.1

본 연구는 1961년부터 2010년까지 사하라 이남 아프리카 8개 국가를 대상으로 무역과 경제성장의 관계를 벡터오차수정모형(Vector Error Correction Model)과 그랜저 인과관계(Granger Causality) 검정을 이용해 분석하였다. 실증분석 결과 대상 국가들은 수출입과 경제성장간에 쌍방향 또는 단방향의 인과관계를 보여주었다. 특히 콩고, 가나, 남아프리카공화국과 잠비아에서는 수출입과 경제성장간에 장기 인과관계가 존재했다. 그랜저 인과관계 검정에서는 콩고, 에티오피아, 가나와 세네갈에서 수출주도형 경제성장 가설이 입증된 반면, 보츠와나, 가나, 케냐, 세네갈과 잠비아에서는 수입주도형 경제성장 가설이 입증되었다. 또한 일반충격반응(Generalized Impulse Response) 함수를 이용한 시뮬레이션에서는 수출입을 통한 거시경제 충격이 에티오피아, 케냐와 남아프리카공화국의 국내총생산을 증가시키고 있음을 보여주었다. 결론적으로 수출뿐만 아니라 수입 또한 사하라 이남 아프리카 국가들의 경제성장에 유의미한 역할을 하고 있음을 확인했다. This study examines the link between trade and economic growth in eight Sub-Saharan African (SSA) countries, using data for the 1961–2010 period. We analyze the corresponding causal relationship using the augmented Cobb–Douglas production function and performing the Granger causality test based on the vector error correction model. The empirical results indicate that this causal relationship differs across the countries. Specifically, a long-run causal relationship between exports, imports, and economic growth is apparent in the case of Congo, Ghana, South Africa, and Zambia. In addition, the Granger causality test results for Congo, Ethiopia, Ghana, and Senegal indicate support for the export-led growth hypothesis, while those for Botswana, Ghana, Kenya, Senegal, and Zambia, for the import-led growth hypothesis. Moreover, the simulation result using the generalized impulse response function demonstrates that the GDP positively responds to a sudden favorable macroeconomic shock through exports and imports in Ethiopia, Kenya, and South Africa. Overall, these results imply that not only exports, but also imports could play a significant role in stimulating economic growth in SSA countries.

경향점수 추정 모형에 따른 매칭 및 인과효과 검정 결과 비교 : 다층자료의 모의분석

송승원(Seungwon Song),강상진(Sang-Jin Kang),이규민(Guemin Lee) 한국교육평가학회 2015 교육평가연구 Vol.28 No.3

이 연구는 경향점수의 추정과 인과효과 검정에서 분석 자료의 다층구조가 미치는 영향을 밝히고자 하였다. 동일한 모집단에서 반복 표집하여 생성한 모의표본 자료를 분석하는 실험을 통하여, 다층구조를 무시한 로지스틱 모형과 다층구조를 반영하는 로지스틱 다층모형으로 각각 추정한 경향점수에 어떠한 차이가 있는지 밝히고, 추정된 경향점수를 바탕으로 진행되는 경향점수매칭 및 인과효과 검정결과에서 어떠한 차이가 있는지 알아보았다. 연구 결과, 로지스틱 다층모형으로 추정된 경향점수가 로지스틱 모형으로 추정된 경향점수보다 더 큰 변산을 갖는 것으로 나타났으나, 해당 경향점수를 기준으로 매칭된 처치집단과 통제집단의 사례 수는 로지스틱 모형의 경우보다 더 작았다. 매칭 후 경향점수, 학교 및 학생수준 공변수의 처치집단과 통제집단 동등성 정도는 학생수준 공변수의 경우 로지스틱 모형이, 학교수준 공변수의 경우 로지스틱 다층모형이 두 집단의 분포를 보다 유사하게 만드는 효과가 있었다. 마지막으로 실시한 인과효과 검정 결과, 로지스틱 다층모형의 경우 대다수의 데이터 세트에서 처치효과가 없었다고 분석된 반면, 로지스틱 모형의 경우 100개의 데이터 세트 모두 처치효과가 있다고 분석되었다. 따라서 인과효과를 분석하고자 하는 비실험연구의 데이터가 위계적 구조의 자료인 경우, 두 모형은 전혀다른 결론에 도달한다는 증거가 발견되었다. 이 연구의 결과에서는 로지스틱 모형과 다층로지스틱 모형의 선택은 분명하지 않으며, 두 모형에 의한 교차타당화 작업이 요청된다. 또한 경향점수 추정에서 모형의 적절성을 평가할 수 있는 모의실험이 필요하다. This study examined the effect of multilevel structure of data on the estimation of propensity score and causal inference. By analyzing multilevel data sets repeatedly sampled 100 times from the statistically recovered population data, we investigated the differences of matching and those of causal inference when the logistic multilevel- and regression model were applied in estimating the propensity score. The results indicated that the distributions of propensity scores, and the number of matched data sets all differ when the propensity scores were estimated by the two models. In the case of matching students level covariates, logistic regression model performed better than logistic multilevel model in matching the covariate distributions between treatment and control groups. However, in case of school level covariates, we found the opposite results, in which multilevel model performed better. The results also showed two models influenced differently on the causal inference. Therefore this study reports that two models lead to different conclusion in causal inference via propensity score matching under the circumstance when the data are hierarchical. We suggest that cross validation is needed when causal inference is made based on propensity score matching in the analysis of multilevel data. Also, it is necessary to conduct a simulation study in order to evaluate the adequacy of a model to estimate the propensity score.

이중몰입의 구조방정식모형에 관한 연구

이병훈(Lee Byoung-Hoon),유제민(Yoo Jhe-Min) 한국사회학회 2002 韓國社會學 Vol.36 No.6

이 연구에서는 구조방정식모형(Structural Equation Modeling)의 실증분석기법을 활용하여 이중몰입의 개념구성과 인과관계에 대한 심층적인 검증을 시도하고 있다. 우선, 이중 몰입의 개념구성에 있어 1요인 모형(① 혼성모델(mix model), ② 확장모델(escalating model), ③ 대립모델(confronting model))과 2요인 모형 (① 독립모델 (independent model), ② 회사→노조 일방향모델(company-to-union one-way model), ③ 노조→회사 일방향모델(union-to-company one-way model), ④ 상호작용모델(interactive model))로 구분하여 7개 세부 개념모델에 대한 타당도를 검토하였다. 그 결과, 2요인 개념모형이 1요인 모형 보다 전체적으로 타당도가 높으며, 또한 2요인 모델중에서는 회사몰입→노조몰입의 일방향 작용모델이 가장 우수하다는 점이 입증되었다. 이중몰입에 대한 영향요인들로서 종업원들의 직무만족과 회사의 인사조직관리 및 노조의 활동방식을 포괄하는 인과경로모형을 분석함에 있어서는 회사 인사조직관리의 선행 변인들이 직무만족의 매개ㆍ조절을 거쳐 회사몰입과 노조몰입에 영향을 미치는 것이 밝혀졌다. 아울러, 노조활동에 대한 조합원들의 만족도가 노조몰입에만 긍정적으로 작용하고 있음이 드러났다. This Study aims to examine conceptual construct and causal path of dual commitment by using the analytical method of Structural Equation Modeling (SEM). In verifying the conceptual construct of dual commitment, seven models, comprised of three “One Factor” models (① mix model, ② escalating model, ③ confronting model) and four “Two Factor” models (① independent model, ② company-to-union one-way model, ③ union-to-company one-way model, ④ interactive model), are hypothetized, and the SEM validity of those models are compared. The result of comparison indicates that the validity of the “Two Factor” models are overall better than that of the “One Factor” models, and that the company-to-union one-way model has the highest validity among the seven models. The causal path analysis, which includes predictive variables, such as job satisfaction, human resource management, managerial styles, organizational management, and the operation of the labor union, to affect company and union commitment, shows that those variables of managerial schemes have positive effect on both company and union commitment only with the moderation of job satisfaction, and also that workers’ satisfaction with union activity makes a positive impact on union commitment.

Causal Relationship between Income Inequality and Economic Growth

조성원 한국자료분석학회 2019 Journal of the Korean Data Analysis Society Vol.21 No.3

This paper investigates the causal relationship between economic growth and income inequality using the time-series data of Korea. Prior to conducting the causality test, properties of the time-series variables are carefully examined using the unit root test and the cointegration test, and an error correction model is constructed to test the direction of causality. The quarterly data covers the period from 1990 to 2017. Income quintile ratio is used as a measure of income inequality. The unit root test result indicates that the variables are first-order integrated. The cointegration test result, however, shows that there is a cointegrating relationship between the two variables. Based on these results, an error correction model is constructed to test the causal relationship. The result of the Granger causality test based on the error correction model indicates a uni-directional causality running from economic growth to income inequality in the short-run. In the long-run, however, there is a uni-directional causality running from income inequality to economic growth. Estimation result of the cointegration equation shows that income inequality positively affects economic growth in the long-run.

Causality of Roundwood Import and Macro-economic Variables

김동준 한국산림과학회 2004 한국산림과학회지 Vol.93 No.6

This study confirmed econometrically the causality of roundwood import and macro-economic variables using Korean data. One national product variable and one demand sector variable were selected among macro-economic variables. National product variable was represented by gross domestic product(GDP), and demand sector variable was represented by building permit area(BPA). The relationships between roundwood import and GDPㆍBPA were represented by bivariate vector autoregressive model. The causality of roundwood import with GDPㆍBPA was measured by the causality test of Granger.There appeared to be evidence of the causal relationship between change in roundwood import and change in GDPㆍBPA. That is, change in GDPㆍBPA causes change in roundwood import in Korea. The causality test value of roundwood import with BPA was larger than that with GDP. That is, roundwood import can be more efficiently explained by demand sector variable than national product variable in Korea.

A method for solving causality conflicts in vehicle powertrain modeling

김남욱,이웅,정해성,박도현,김덕진 대한기계학회 2019 JOURNAL OF MECHANICAL SCIENCE AND TECHNOLOGY Vol.33 No.4

Model-based systems engineering is becoming important and is being more widely used as the computing power is improved, because it can be extensively utilized for evaluating performances of vehicles with less effort and time than real-world tests. In this study, a portbased model technique is reviewed, especially for investigating the model connection between the transmission output and the final-drive input in vehicle powertrain models. Commercialized tools that model vehicle powertrains provide well-organized processes for users to connect conventional component models, but a confliction problem of connecting ports frequently arise when users develop their own models and integrate them into a vehicle system. In this study, several ideas to address the conflict issue have been reviewed, and an idea is implemented to solve the port conflict of the vehicle powertrain, and discussions about the method are provided. The idea is very useful especially when the powertrain model uses effort and flow signals to connect sub-component models. Simulation results show that the method produces accurate results and is fast enough, which requires additional calculation time less than 1 %, compared to an alternative method, and it is proved that it guarantees numerical stability in computations. The idea has been widely used in powertrain models, but the feasibility and the numerical stability have not been carefully investigated in the previous studies, so this study would be helpful for proving a basis to model users who develop new sub-component models.