The causal forest (CF) model is a causal inference method through the random forest algorithm to estimate a conditional average treatment effect given covariates. This study introduced three different causal forest-based models for multilevel data and compared the average treatment effect estimates of each model under a variety of simulated conditions. Specifically, the simulation conditions considered in the current study were: diverse types of covariate selection, ratios of treatment to control cases (1:1, 1:3), number of groups (100, 200, 500), and group size (20, 40, 50). Three causal forest models accounting for multilevel data structure were introduced: a causal forest model including propensity score estimated by the multilevel logistic regression model (model 1), a causal forest model reflecting the multilevel structure in the subsampling (model 2), and a causal forest including both propensity score estimated by the multilevel logistic regression model and the clustering of the multilevel data structure (model 3). The relative bias and mean squared error of the average treatment effect estimates by the three causal forest models were compared. With a few exceptions, the model that applied propensity scores estimated via a multilevel logistic regression to the causal forest generally exhibited relatively better treatment effect estimation performance. Including covariates related to both the treatment and outcome variables also resulted in relatively higher precision of the average treatment effect estimates. The current study further suggested that securing at least 200 groups is necessary for stable estimation of the average treatment effect.

This study examined the relations between therapist perceptions of psychodynamic therapy (PDT) process, adherence to PDT principles, and interaction structures formed between therapist and patient. A graduate student therapist treated a 28-year-old man diagnosed with compulsive sexual behavior disorder and borderline personality disorder with PDT (N = 52 sessions). Following each session, the therapist completed the Psychotherapy Process Q-Set (PQS) to capture his perceptions of session process. Independent coders completed the PQS from session audio-recordings. Preceding each session, the patient completed the Outcome Questionnaire to measure psychological distress. The therapist was most discrepant on items pertaining to his own attitudes and behavior (including interventions), indicating countertransference. Therapist agreement with the independent coders was positively correlated with PDT session adherence and negatively correlated with untherapeutic interaction structures (i.e. reciprocal patterns of interaction) identified in a previous study. Simulation modeling analysis revealed that increased therapist agreement and psychological distress independently preceded increases in a potentially therapeutic interaction structure observed one session later. Therapist agreement was associated with PDT session adherence and interaction structures, suggesting routine monitoring to protect treatment fidelity and therapeutic interaction structures.

Objectives This study investigates the structural relationships among adolescents' relationships with teachers and peers, self-esteem, grit, and academic engagement. It also examines the mediating roles of self-esteem and grit in these relationships and compares their relative effects. Methods Data from 2,252 first-year middle school students, collected in the fifth wave (2022) of the Korea Children & Youth Panel Survey 2018 (KCYPS 2018) by the National Youth Policy Institute, were analyzed. Structural equation modeling with SPSS 23.0 and Mplus 8.11 was used to assess these relationships and mediating effects. Results First, adolescents' relationships with teachers directly positively influenced academic engagement. Second, these relationships positively affected self-esteem, as did relationships with peers. Third, both relationships with teachers and peers had positive effects on grit. Fourth, relationships with teachers partially mediated academic engagement through self-esteem and grit, while relationships with peers fully mediated it. Fifth, grit had a larger mediating effect on the relationship between teacher relationships and academic engagement, whereas self-esteem had a larger effect on the relationship between peer relationships and academic engagement. Conclusions The study highlights that relationships with teachers and peers, self-esteem, and grit significantly impact adolescents' academic engagement. To enhance academic engagement, improving the quality of these relationships and strengthening self-esteem and grit are essential.

The purpose of this study is to compare the predictive performance and variable selection accuracy of two machine learning methods—MERF and glmmLasso—that can account for the structure of longitudinal data. To achieve this, a simulation study was conducted with conditions varying in the number of predictors (25, 50, 100), number of repeated measurements (3, 4, 6), and sample sizes (500, 1,000, 2,000, 4,000). The performance of MERF and glmmLasso was evaluated using root mean squared error (RMSE), variable selection accuracy rate, error rate, relative bias, and bias. The results are as follows. First, glmmLasso consistently outperformed MERF in predictive performance across all conditions investigated under the current study. Second, MERF showed increasing RMSE with more predictors and smaller sample sizes, whereas glmmLasso maintained stable RMSE. Third, for both methods, RMSE increased with the number of repeated measurements when the sample size was relatively small. Additionally, in glmmLasso, relative bias was observed under some conditions when the sample size was 500. However, with a sufficiently large sample size, RMSE differences across the different numbers of repeated measurements were minimal. Fourth, in MERF, variable selection accuracy rate improved with fewer predictors, fewer numbers of repeated measurements, and a larger sample size. Based on these findings, practical guidelines for applying MERF and glmmLasso methods while considering the structure of longitudinal data and suggestions for future research were provided.

This study aimed to identify key predictors of changes in career maturity during adolescence, analyze changing patterns, and empirically test the effects of key predictors. Using the 10th, 13th, and 16th waves of the Korean Welfare Panel Study (n=310), penalized regression model and multilevel growth model were applied to account for the hierarchical structure of the longitudinal data. Results showed that academic performance, psychosocial school environment, school bonding, self-esteem, peer attachment, friends’ peer attachment, learning guidance experience, and participation in hobby or talent-related activities positively influenced career maturity, whereas average daily TV viewing time and internet overdependence (deviant behavior) had negative effects. Parents’ experience with child counseling or group programs positively influenced career maturity, while experience with tuition support and satisfaction with social acquaintances had negative effects. Multilevel growth model revealed that adolescents’ career maturity decreased over time. When variables selected by LASSO were included as predictors, most showed consistent signs and significance levels with the LASSO results; but depression and anxiety, and parents’ employment insurance enrollment emerged as additional positive factors only in the multilevel growth model. Based on these findings, educational strategies to promote career maturity among adolescents are discussed.

This study was conducted to classify latent classes of emotional problems in adolescence and to explore and test key predictors of change in emotional problems. To this end, latent class growth analysis was applied to data from the first through fifth waves of the middle school students’ panel of KCYPS 2018. SEM forest was used to explore the top ten most important variables explaining change in emotional problems, and the three-step approach was applied to examine how key predictors affect the classification of latent classes. The main findings are as follows. First, three latent classes in emotional problems during adolescence were classified: low-level decrease followed by increase (20.5%), middle-level maintenance (58.8%), and high-level maintenance (20.7%). Second, SEM forest analysis revealed that six variables had been studied in previous research while four other variables were newly identified. Third, happiness, self-esteem, and grit (consistency of interest) were associated with lower levels of emotional problems, while academic helplessness (lack of active performance), smartphone dependence, and parenting attitudes (inconsistency) were associated with higher levels of emotional problems. Based on the results, implications and suggestions for reducing emotional problems among adolescents are discussed.

Objectives This study was conducted to verify the effect of in-school work experience on job satisfaction in the first job among college graduates and to explore how the purpose of work experience while in school and the rele-vance of the major affect job satisfaction in the first job. Methods To do this, causal forests were applied to the 2020 Graduates Occupational Mobility Survey (GOMS) to examine the average treatment effect of in-school work experience on the job satisfaction of the first employment. Additionally, heterogeneous treatment effects were explored based on personal backgrounds, col-lege factors, job preparation, and characteristics of the first job among college graduates. Furthermore, hier-archical regression analysis was applied to control for variables that were significant in the heterogeneous treat-ment effects and to verify the effects of the purpose of work experience and the relevance of the major on job sat-isfaction in the first job. Results First, the average treatment effect of work experience during college had a positive effect on job sat-isfaction in the first job. Second, the heterogeneous treatment effect of college location was significant, with some regions showing a decreasing effect of in-school work experience on job satisfaction in the first job. Third, livelihood type work experience has a negative effect on job satisfaction in the first job. Fourth, the relevance of work experience to the major had a positive effect on job satisfaction in the first job. Conclusions Based on the findings, implications and policy implications of the study were discussed.

Adolescence is a stage during which individuals develop social adaptability through meaningful interactions with others. During this period, students gradually expand their social networks outside the home, forming a sense of community. The aim of the current study was to explore the key predictors related to sense of community among Korean high school students and to develop supportive policies that enhance their sense of community. Accordingly, random forests and SHapley Additive exPlanations (SHAP) were applied to the 7th wave (11th graders) of the Korean Education Longitudinal Study 2013 data (n = 6,077). As a result, 6 predictors positively associated with sense of community were identified, including self-related variables, “multicultural acceptance,” “behavioral regulation strategy,” and “peer attachment,” consistent with previous findings. Newly derived variables that predict sense of community include “positive recognition of volunteering,” “creativity,” “observance of rules” and “class attitude,” which are also positively related to sense of community. The implications of these results and some suggestions for future research are also discussed.

This study aims to examine the research trends related to achievement standards-based assessment. To this end, a total of 87 papers published in KCI-listed academic journals from 2011, when the ‘Advanced Plans for Academic Administration in Secondary Schools’ was announced, to March 2024 were analyzed through keyword frequency analysis, co-occurrence analysis, LDA topic modeling, and semantic network analysis. The main findings are as follows. First, research on achievement standards-based assessment revolved around three main topics: ‘Teaching and learning design (Topic 1)’, ‘Development of achievement standards and teacher perception (Topic 2)’, and ‘Introduction of High school credit systems (Topic 3)’. Second, Topic 2 showed a decrease in research proportion, while Topic 3 showed an increase. Third, topics in achievement standards-based assessment research were found to be interconnected around ‘Curriculum’ and ‘School’. Fourth, keywords such as ‘learning’, ‘curriculum’, ‘class’ in Topic 1; ‘development’, ‘operation’, ‘school’ in Topic 2; ‘operation’, ‘high school credit system’, ‘curriculum’ in Topic 3 were identified as top keywords in terms of both degree centrality and betweenness centrality. Based on the analysis results, implications for future research directions on achievement standards-based assessment systems were suggested.

The purpose of this study is to examine domestic research trends related to vocational high schools. To achieve this, we conducted frequency analysis, co-occurrence word analysis, topic modeling based on the latent dirichlet allocation (LDA) algorithm, and semantic network analysis on the Korean abstracts of 642 papers related to vocational high schools published in KCI-indexed journals between May 12, 2010, when the Policy for Advancement of High School Vocational Education was announced, and December 31, 2023, spanning approximately 13 years. The analysis revealed the following. First, vocational high school research can be classified into three topics: post-graduation employment support (Topic 1), curriculum operation (Topic 2), and career-related factors (Topic 3), with Topic 3 covering the largest proportion of research. Second, we observed changes in the proportion of topics over time in response to vocational high school policy changes. Third, the keyword networks of Topic 1 and Topic 2 were connected through teachers and operation, while Topic 1 and Topic 3 were linked through the keyword employment. Lastly, keywords such as teachers, support, and employment in Topic 1; curriculum, teachers, and education process in Topic 2; and career, satisfaction, and university in Topic 3 ranked within the top five in both degree centrality and betweenness centrality. Based on the main findings, the implications are discussed for future directions and research tasks to foster academic advancement in the field.

본 모의실험 연구에서는 잔차의 비정규성이 교차분류 다중소속 무선효과 모형 모수추정에 어떠한 영향을 미치는지 살펴보고자 하였다. 이를 위해 잔차 분포(정규 분포, 균일 분포 및 카이제곱 분포), 집단 내 상관 계수(0.1, 0.2 및 0.3), 집단 수(20, 50 및 100), 집단 크기(10 및 20), 교차분류 비율(20% 및 40%), 다중소속 비율(10% 및 20%) 조건이 고려되었다. 다양한 조건 하에서 교차분류 다중소속 무선효과 모형 추정 시 상대편의와 평균제곱근오차를 확인하였다. 분석 결과, 잔차의 비정규성은 특히나 집단 수가 상대적으로 작고 (즉, 50 이하일 때), 집단 내 상관계수가 커질 때, 교차분류와 2 수준에서 다중소속이 발생한 임의효과 분산 추정에 더 큰 영향을 미치는 것으로 나타났다. 또한 2 수준 잔차의 분포가 균일 분포(즉, 비정규적이나 대칭적인 분포)에 비해 카이제곱 분포(즉, 심하게 치우친 분포)일 때 상대편의성이 커지는 것으로 나타났다. 95% 신용구간의 경우 카이제곱 분포를 적용하였을 때 명목수준 보다 작게 나타난 반면, 균일 분포를 적용한 경우에는 명목수준보다 크게 나타났다. 본 연구결과는 교차분류와 상위수준에서 다중소속이 발생하는 다층자료의 맥락효과를 분석하기 위해 교차분류 다중소속 다층모형을 적용하는 데 있어 유용한 정보를 제공할 것으로 기대한다.This simulation study investigated how residual non-normality affects cross-classified multiple membership random effects modeling estimates. The conditions manipulated in the study were residual distribution type (normal, uniform, and chi-squared), intra-unit correlation coefficients (0.1, 0.2, and 0.3), number of groups (20, 50, and 100), average group size (10 and 20), cross-classification rate (20% and 40%), and multiple membership rate (10% and 20%). The relative bias and the root mean square of the parameter estimates were evaluated. The results indicated that non-normal residuals had a larger impact on the cross-classified multiple membership random effect variance component estimates, especially when the number of groups was relatively smaller (i.e., 50 groups or less) and that the intra-unit correlation coefficient increased. Specifically, the degrees of relative bias were larger when the level-two residuals followed chi-squared distribution (i.e., a severely skewed distribution) compared to uniform distribution (i.e., a non-normal but symmetrical distribution). The coverage rates of the level-two variance component were lower than the nominal 95% when residuals followed chi-squared distribution, while they were higher when residuals followed uniform distribution. These findings can be useful in applying cross-classified multiple membership random effects modeling to account for the contextual effects of cross-classified and multiple higher-level units.

본 연구는 청소년의 그릿 및 학업무기력 수준에 따른 잠재프로파일을 분류하고, 잠재프로파일 분류에 미치는 학생 개인, 부모, 교사 및 친구 변인의 영향력을 학교급별로 검증하기 위해 수행되었다. 이를 위해 한국청소년정책연구원이 수집한 한국아동·청소년패널조사 2018의 중학교 1학년(초4패널, 4차년도) 자료와 고등학교 1학년(중1패널, 4차년도) 자료에 3단계 접근법을 활용한 잠재프로파일 분석(Latent Profile Analysis)을 실시하였다. 주요 연구 결과는 다음과 같다. 첫째, 잠재프로파일 분석 결과, 중학생과 고등학생의 그릿 및 학업무기력은 모두 3개의 잠재프로파일로 도출되었으며, 이를 ‘높은 그릿 및 낮은 학업무기력 집단’과 ‘중간 그릿 및 학업무기력 집단’, ‘낮은 그릿 및 높은 학업무기력 집단’으로 명명하였다. 둘째, 중학생과 고등학생의 그릿 및 학업무기력 잠재프로파일 분류에 미치는 영향을 미치는 주요 변인을 살펴본 결과, 중·고등학교에 공통적으로 진로호기심, 진로관심, 우울, 주의집중, 협동심(팔로워십, 갈등문제해결), 스마트폰 의존도, 교사관계(신뢰성), 친구관계(부정적 관계)가 유의한 영향을 미치는 것으로 나타났다. 다음으로 중학교 1학년의 잠재프로파일 분류에 있어 나타난 특징적인 영향변인으로는 학업성취 만족도, 교사관계(접근가능성)가 확인되었으며, 고등학교 1학년의 경우 자아존중감이 특징적인 영향변인으로 나타났다. 연구결과를 바탕으로 시사점 및 청소년의 그릿 증진 및 학업무기력 감소를 위한 지원방안을 논의하였다.

The purpose of this study is to identify the latent profile classification of adolescents’ school adjustment and the transition patterns of latent profiles after school-level transitions, to explore the variables that affect them. To do this, we applied latent transition analysis with 3-step approach to Korean Education Longitudinal Study 2013 data from the 3th and 6th wave (7th and 10th grades). The main results are as follows. First, four latent profiles commonly appeared across the 7th grades and the 10th grades: ‘average-level adjustment group’, ‘high-observance of rules average-level adjustment group’, ‘high-level adjustment group’, and ‘low-level adjustment group’. Second, as a result of examining the transition patterns of latent profiles for school adjustment, the overall rate of maintaining the same group was the highest. Third, the higher academic self-concept, social self-concept, self-management, parent-child interaction, interactive teaching methods, teachers’ enthusiasm and achievement pressure, the higher the probability of belonging to a latent profile with relatively higher mean of school adjustment in the 7th grades. Fourth, the higher individualized teaching methods, interactive teaching methods, teachers’ enthusiasm and achievement pressure, the higher the probability of belonging to a latent profile with relatively higher mean of school adjustment in the 10th grades. Fifth, academic self-concept and self-management had statistically significant effects on the transition of latent profiles. Based on the findings, implications for improving adolescents’ school adjustment were discussed.

본 연구는 고등학생의 학업무기력(능동수행 결여, 학습동기 결여, 긍정정서 결여, 능동수행 결여) 관련 주요 예측변인을 탐색하기 위해 수행되었다. 이를 위해 한국아동·청소년패널조사 2018(Korean Children and Youth Panel Survey 2018; KCYPS 2018)의 중1패널 4차년도(고등학교 1학년) 자료에 랜덤포레스트를 적용하였고, 주요 예측변인의 중요도와 학업무기력의 관계를 파악하기 위해 SHAP 분석을 실시하였다. 주요 연구결과는 다음과 같다. 첫째, 학업무기력 하위요인에 공통적으로 도출된 주요 예측변인으로, ‘학습시간-학원 및 과외 시간’, ‘학업열의-헌신’, ‘진로호기심’, ‘진로관심’, ‘자아존중감’, ‘협동심-집단촉진·갈등문제해결’, ‘교사관계-신뢰성’이 학업무기력 변화와 부적인 관계가 있는 것으로 나타났으며, ‘우울’, ‘스마트폰 의존도’, ‘친구관계-부정적 관계’는 학업무기력 변화와 정적인 관계가 있는 상위 변인으로 도출되었다. 둘째, 학업무기력 하위요인에 따라 특징적으로 나타난 변인을 살펴보면, ‘부모양육태도-거부’와 ‘부모양육태도-강요’는 통제신념 결여에서만 도출되었으며, 정적인 관계를 갖는 것으로 나타났다. 학습동기 결여의 경우, ‘여가시간-스마트폰을 가지고 노는 시간’과 ‘학업열의-활기’가 주요 변인인 것으로 나타났으며, ‘여가시간-스마트폰을 가지고 노는 시간’은 학습동기 결여와 정적인 관계가 있는 것으로 나타났고 ‘학업열의-활기’는 부적인 관계성을 보이는 것으로 확인되었다. 긍정정서 결여에서는 ‘신체증상’과 ‘학교생활 만족도’, ‘교사관계-민감성’이 상위 변인에 포함되었으며, 모두 긍정정서 결여와 부적인 관계가 있는 것으로 나타났다. 능동수행 결여의 경우 ‘전 교과 성취도의 주관적 평가’와 ‘그릿-노력지속’이 주요 변인인 것으로 확인되었고, 부적인 관계를 보이는 것으로 나타났다. 이와 같은 연구결과를 바탕으로 고등학생의 학업무기력를 감소를 위한 방안과 후속연구를 위한 제언을 제시하였다.This study was designed to explore major predictors related to four sub-factors of academic helplessness in high school students: lack of locus of control, lack of learning motivation, lack of positive emotions, and lack of active performance. To do this, random forest was applied to 4th wave data (10th graders) of the Korea Children and Youth Panel Survey 2018 (KCYPS 2018), and Shapley addictive explanations (SHAP) analysis was conducted to explore the importance of major predictors and the relationship between academic helplessness and the predictors. The main results are as follows. First, the predictors commonly derived from the sub-factors of academic helplessness were “learning time−academy and tutoring time”, “academic engagement−dedication”, “career curiosity”, “career interest”, “self-esteem”, “cooperation−group promotion and conflict resolution” and “teacher relationship and reliability” which had a negative relationship with academic helplessness. However, “depression”, “smartphone dependence”, and “friend relationship−negative relationship” were derived as top variables that had a positive relationship with academic helplessness. Second, “parenting attitude−rejection” and “parenting attitude−force” were derived only from lack of locus of control. “leisure time−playing with smartphones” and “academic engagement−energy” were derived only from lack of learning motivation. In the case of lack of positive emotions, “physical symptoms”, “school satisfaction” and “teacher relationship−sensitivity” were included in the top variables, and “subjective evaluation of all subject achievement” and “grit−continuation of effort” were the main variables that were derived only from lack of active performance. Based on these findings, implications for reducing academic helplessness and suggestions for future research are discussed.

본 연구는 중학생의 학교적응 변화양상을 살펴보고, 학교적응 변화에 영향을 미치는 주요 예측변인을 도출하는데 목적이 있다. 이를 위해 한국교육종단연구2013(KELS2013) 3-5차년도(중1-중3) 데이터에 종단자료의 특성을 반영할 수 있는 머신러닝 기법인 SEMforest(Structural Equation Model forest)를 적용함으로써, 중학생의 학교적응 변화에 영향을 미치는 핵심적인 학생 및 부모, 학교 변인을 광범위하게 탐색해보고자 하였다. 주요 연구결과는 다음과 같다. 첫째, 중학생의 학교적응은 학년이 올라감에 따라 감소하는 것으로 나타났으며, 초기치와 변화율은 통계적으로 유의한 개인차가 있는 것으로 확인되었다. 둘째, 학생 변인 모형에서는 ‘자아개념’, ‘공동체의식’, ‘교사특성(교사열의)’, ‘수업방식(상호작용, 개별화)’, ‘진로성숙도’, ‘자기관리’, ‘정신건강’, ‘학생의견반영 정도’, ‘행동조절전략(노력조절)’이 학교적응 변화를 예측하는 중요도 높은 변인으로 도출되었다. 부모 변인 모형에서는 ‘학업적 지원’, ‘학교 만족도(학교 만족도, 학교의 안전, 전반적 만족도)’, ‘전반적 정보 제공’, ‘자녀와 함께 하는 활동’이 선택되었으며, 학교 변인 모형에서는 ‘컴퓨터 보유 현황(교원용)’, ‘총 학급 수’, ‘총 교원 수’가 상위 중요도 변인으로 도출되었다. 이상의 연구결과를 바탕으로 중학생의 학교적응을 돕기 위한 교육적 시사점을 제시하였다.

본 연구는 초등학생의 그릿 변화궤적과 사회적 지지(부모지지, 친구지지, 교사지지)의 동시효과 및 지연효과를 탐색하는 데 목적이 있다. 이를 위해 한국청소년정책연구원에서 실시한 한국아동・청소년패널조사 2018(Korean Children & Youth Panel Survey 2018; KCYPS 2018)의 초4 패널 1차년도～3차년도(초4～초6) 자료를 사용하였다. 초등학교 3년간의 그릿 변화 양상과 사회적 지지의 동시효과 및 지연효과를 탐색하기 위한 방법으로 잠재성장모형(Latent Growth Model)을 적용하였고, 이에 대한 연구결과는 다음과 같다. 첫째, 초등학생의 학년이 증가할수록 그릿은 감소하는 것으로 나타났다. 둘째, 사회적 지지(부모지지, 친구지지, 교사지지)가 1차(초4)～3차(초6) 시점에서 그릿에 정적 영향을 미치는 동시효과가 남・여학생 집단에서 모두 나타나는 것을 확인하였다. 또한, 남학생 집단에서는 1차(초4)와 2차(초5) 시점의 교사지지가 2차(초5), 3차(초6) 시점의 그릿에, 여학생 집단에서는 1차(초4) 시점의 교사지지가 2차(초5) 시점의 그릿에 유의한 영향을 미치는 지연효과를 확인하였다. 본 연구에서는 이러한 결과를 바탕으로 초등학생의 그릿은 학년이 올라감에 따라 감소하는 경향이 있지만, 남・여학생 집단에서 모두 사회적 지지의 그릿에 미치는 동시효과 및 교사지지의 지연효과가 있음을 밝힘으로써 부모, 친구 및 교사 역할의 중요성을 강조하였다. 후속연구로 장기적인 자료 축적을 통해 그릿에 대한 종단연구가 추가적으로 수행될 필요가 있음을 제시하였다.This study investigated the changing trajectory of grit and the contemporaneous and lagged effects of three sources of social support (parental, peer, and teacher support) in elementary school students. For this purpose, data from Waves 1 through 3 (4th through 6th grades) of the Korean Children and Youth Panel Survey 2018 (KCYPS 2018), conducted by the National Youth Policy Institute, were used. Latent growth modeling was applied to analyze the changing trajectory of grit and the contemporaneous and lagged effects of social support over the three years. The results are as follows. First, students’ grit appeared to decrease through the elementary school years. Second, the contemporaneous effects of the three sources of social support on grit were positively significant in each grade. In addition, significant lagged effects of teacher support in the 4th and 5th grades and on grit in the 5th and 6th grades were found in the male group, while significant lagged effects of teacher support in the 4th grade and on grit in the 5th grade were found in the female group. In sum, elementary school students’ grit tends to decrease as grade level increases. However, the importance of social support from parents, peers, and teachers is confirmed by revealing the contemporaneous and/or lagged effect of social support on the elementary male and female students’ grit. It is suggested that additional longitudinal research on grit be conducted.