From the 29 studies reviewed, 11 were published in 2017, 9 studies were published in 2016, 3 studies in 2015 and 6 in 2014. There were no studies from the year 2013 and 2012 meeting the inclusion criteria. Twenty five studies were cross-sectional (86%). There was also 1 longitudinal (prospective) study, 1 study with a pre-post design, 1 study including 2 studies (1 cross sectional and 1 longitudinal study), and 1 genetically sensitive sample research on twin teenage student pairs. All studies used quantitative methods, except for one cross-sectional study that used mixed methods for data analysis. Twenty six studies used regression analyses (90%), amongst other statistical analyses.

The sample size ranged from 105 to 4813 participants. In all the studies participants were students attending several levels of education (primary school to

university) since student status was part of the inclusion criteria. In the majority of studies (52%) students attended college or university, and in 15 out of the 29 studies the majority of participants were females. Most studies used student samples from the US, but there were also samples from Korea, Canada, Australia, Turkey, Peru, Austria and the UK. The US samples consisted of White Caucasian, Latino/Hispanic, Black Americans and Asian/Asian Americans.

The reviewed studies addressed all three research questions. Some of them addressed more than one question. Grit has been found to associate with a range of positive outcomes in education, such as academic performance and retention in education. Its two dimensions were found to correlate with different educational variables. Positive psychology variables were found to predict grit in the educational context.

Grit and positive educational outcomes. Sixteen studies investigated grit as a unidimensional factor that relates to and can predict several educational outcomes. Most studies found a positive link between grit and academic performance reflected in measures such as GPAs, and other positive educational outcomes, such as retention, to education.

Note: Acronyms/Abbreviations: CFA = Confirmatory Factor Analysis, EFA = Exploratory factor analysis, GCSE = U.K.-wide national exam administered at the end of compulsory schooling, Grit-PE = the “perseverance of effort” dimension of grit, Grit-CI = the “consistency of interest” dimension of grit, GPA = Grade Point Average, ICC = Intraclass Correlation, SEM = Structural Equation Modelling.

Grit considered as a dimension of character also provided consistent results. Tucker-Drob et al. (2016) found that character measures (among which grit) were significantly correlated with the achievement and knowledge outcomes in 89% of twin/triplet pairs of 3^rd to 8^th grade students. Park et al. (2017) explored a tripartite taxonomy of character (interpersonal, intellectual and intrapersonal) in three related studies on 4th to 8th graders. Results showed that only intrapersonal character (which includes grit) was related to higher grades (study 3) and was a significantly predicted class participation and academic achievement (study 1) as well as higher student GPA (study 2).

Strayhorn (2014) who explored the role of grit in the academic success of Black male collegians’ at predominantly White institutions, concluded that grittier Black males achieved higher grades than their less gritty same-race male peers, and that grit was a positive predictor of college grades, explaining almost as equal variance as high school GPA and ACT score. Similarly, Akos and Kretchmar (2017) found that self-report grit total score significantly predicted GPA in undergraduate students, and Pate et al. (2017) found that grit-total score was an independent predictor of high academic performance in a sample of student pharmacists in the US. Grittier Taiwanese students tended to self-report higher academic performance and increased academic satisfaction in the study by Lin & Chang (2017) , while grit was significantly positively associated with academic achievement only for non-citizen Latino first-generation college students in the study by O’Neal et al. (2016) . In two studies, the positive relationship between grit and academic performance was mediated by deliberate practice in Korean students (Lee & Sohn, 2017) and by math confidence (for students with lower math confidence the positive effect of grit was stronger) in university students in Canada (Flanagan & Einarson, 2017) .

Four studies explored the predictive capacity of grit beyond personality and intelligence in terms of academic performance or achievement, and three in terms of retention in education. Grit predicted unique variance in performance on the Academic Diligence Task (as a behavioral measure of academic diligence, focused on time spent on the task or on productivity of problems solved) in high school seniors, after controlling for agreeableness (Galla et al., 2014) . Complementary were the results of the study by Dumfart and Neubauer (2016) , who investigated whether non-cognitive traits have incremental predictive capacity over intelligence and conscientiousness in terms of school achievement. Grit correlated moderately with academic achievement criteria, such as GPA, science and language scores. However, only together with other variables, rather than alone, could it account for almost half the variance in adolescent school achievement. Similarly, grit added little phenotypically or genetically to the prediction of academic achievement beyond traditional personality factors, especially conscientiousness in a sample of twins (Rimfeld et al., 2016) . Even though grit correlated positively with school satisfaction, school recognition and decreased rule violation, it did not correlate with GPA and academic honors; neither did it explain additional variance in any school outcome compared to conscientiousness in the study by Ivcevic and Brackett (2014) .

Interesting was the finding that grit is linked with retention in education. Eskreis-Winkler et al. (2014) found that grit was positively correlated with school motivation and academic conscientiousness, and it remained a significant predictor of graduation, as a measure of retention. Grit predicted retention over and beyond established context-specific predictors of retention (e.g., intelligence) and demographic variables (Eskreis-Winkler et al., 2014) . In the aforementioned study by Lee and Sohn (2017) , grit significantly predicted career preparation behavior and major-career relatedness when Big Five personality traits were controlled for. Complementary were the findings by Akos and Kretchmar (2017) where self-report grit total score inversely predicted change in major in a sample of first-year undergraduate students.

Grit was also related to a variety of educational outcomes apart from academic performance. In the study by West et al. (2016) which examined relationships between non-cognitive skills and academic achievement in charter and district school students, grit was positively related with math and ELA test scores, as well as attendance and behavior at the student level. Grit was also positively related to academic-self efficacy at primary and secondary school level, but only with satisfaction with school at secondary school level (Oriol et al., 2017) .

Overall, results regarding the first research question are homogeneous. Grit is linked with various positive educational outcomes. It is positively correlated mainly with academic performance and achievement, although correlations do not reach strength. However, rarely does it show incremental predictive validity over other predictors (such as personality) regarding achievement, but it has satisfactory predictive capacity regarding retention in education.

The two facets of Grit and educational outcomes. Thirteen studies explored grit in terms of its two facets and found that they associate differently with educational outcomes. In the reviewed studies there was no consensus regarding the way in which the two facets of grit correlate with academic performance. Yet, results on the variables predicted by each of the two facets of grit were quite consistent.

High performers reported higher Grit-PE and Grit-CI scores compared to average or low performing Taiwanese students (Lin & Chang, 2017) and student pharmacists in the US (Pate et al., 2017) . Both Grit-PE and Grit-CI showed weak correlations with the GCSE of twin teenage students in the UK (Rimfeld et al., 2016) , positive correlations with estimated grade (Weisskirch, 2016) as well as academic engagement and productivity (Hodge et al., 2017) in undergraduate students, and were associated with a lower expectation of changing majors (Bowman et al., 2015) . Dixson et al. (2016) who examined talented students found that Grit-PE positively, but moderately, correlated with academic achievement, whereas Grit-CI displayed a same strength negative correlation. However, Dixson et al. (2017) found that Grit-PE and Grit CI showed non-significant correlations with academic achievement of African American high school students, and neither comprised a significant predictor.

A number of studies found that each facet of grit is associated with different variables. Grit-PE was negatively correlated with academic maladjustment (Hwang et al., 2017) and positively associated with GPA (Bowman et al., 2015; Farruggia et al., 2016) , achieved grade (Weisskirch, 2016) , motivation, academic self-efficacy, and time management (Farruggia et al., 2016) . It also correlated positively with college satisfaction, faculty-student interaction, and co-curricular engagement, while Grit-CI was related to less intent to change careers (Bowman et al., 2015) , procrastination, time and study environment management strategies (Wolters & Hussain, 2015) . Another interesting finding was that being the first family member to attend university was linked to higher levels of student Grit-PE (Hodge et al., 2017) .

Generally, Grit-PE was found to be a more powerful predictor of academic performance and retention compared to Grit-CI. In the study by Hodge et al. (2017) in Australian university students, Grit-PE, but not Grit-CI, showed a small significant contribution to the prediction of academic productivity measured in perceived academic performance. In the same study, both Grit-PE and Grit-CI significantly predicted academic engagement, but Grit-PE contributed more than twice compared to Grit-CI. Similar results come from the following studies. Only Grit-PE predicted GPA compared to Grit-CI in undergraduate university students (Akos & Kretchmar, 2017) , grades in both high school and college student samples (Muenks et al., 2017) , GCSE in high school students in the UK (Rimfeld et al., 2016) , as well as higher GPA, greater academic adjustment, and intent to persist in college (Bowman et al., 2015) . Wolters and Hussain (2015) found that Grit-PE predicted academic achievement as well as all self-regulated learning indicators (such as self-efficacy, cognitive, metacognitive, motivational, time and study environment management strategies), and was the only significant predictor of estimated grade reported by undergraduate students in the study by Weisskirch, 2016 ). Perseverance significantly predicted academic achievement and consequently retention, reflected in progression to the next year of studies (Farruggia et al., 2016) . Hwang et al. (2017) found that only Grit-PE had a positive indirect effect on GPA scores and it mediated the negative relationship that conscientiousness and self-control had with academic maladjustment. The reviewed studies only showed that Grit-CI was a significant negative predictor of the likelihood of undergraduate University students to change majors (Akos & Kretchmar, 2017) .

In conclusion, findings agree that perseverance of effort and consistency of interest function as two separate variables, with Grit-PE being a stronger positive predictor of academic performance compared to Grit CI.

Variables predicting grit in education. The reviewed studies that investigated the factors that predict grit in education were seven. Four studies revealed that factors related to hope, meaning and connection predict grit. Hill et al. (2016) found that purpose in life and positive affect were unique positive predictors of students’ grit. Vela et al. (2015) found hope to be the strongest positive predictor of grit in Latino students, and search for meaning the strongest negative one, while first generation Latino students’ families and communities were found to play an important role in grit (O’Neal et al., 2016) . Lin and Chang (2017) also found that perceptions of family influence and Big Five personality traits (apart from extraversion) were significant predictors of Taiwanese students’ grit, with conscientiousness being the strongest one.

Factors associated with learning also emerged as predictors of grit in three studies. Akin and Arslan (2014) found that learning (mastery) approach goal orientation was the strongest positive predictor of grit, while learning-avoidance and performance approach/avoidance goal orientations significantly negatively predicted grit. Findings by Yeager et al. (2014) revealed that a self-transcendent purpose for learning predicted greater personal meaningfulness of schoolwork, grit, academic self-control, and performance on boring math problems of the Academic Diligence Task, and increased both the attempt towards deep learning from such a task, and course grades for all students, but especially for low-performers. Additionally, a self-oriented, intrinsic motive for learning was a significantly weaker predictor of reported grit compared to a purpose for learning. Finally, Weisskirch (2016) found that self-esteem and general strategies for learning predicted both facets of grit, while attitudes toward lifelong learning also predicted Grit-PE.

In summary, the few studies that have investigated predictors of grit in education have shown that both positive psychology variables such as hope, meaning and connection, and learning factors, such as learning approach-avoidance goal orientation and a self-transcendent purpose for learning, predict grit.

The current systematic review comprises the first attempt to gain an insight into the antecedents and outcomes of grit in education. Even though literature on this field is still at its infancy, we managed to spot twenty nine journal articles that provided valuable information on the topic. Its importance lies in the fact that it unveiled a number of factors that can predict grit. Hence, it provided the first step towards a more in depth exploration of the processes and characteristics that underlie grit and keep individuals going towards pursuing their goals.

A number of limitations of this review need to be addressed though. First of all, selection of studies entails two drawbacks. The emerging interest in the role of grit in education was manifest in unpublished research papers, such as doctoral dissertations, and working papers that were not included in the review due to rigorous inclusion and exclusion criteria. Furthermore, the majority of studies selected used self-report scales and questionnaires, amongst which the Grit Scale, which might have biased the selection process, therefore, restricting generalizability of findings. Specifically, the Grit Scale is a self-report questionnaire that has received much criticism not only by researchers but also from its developers. Duckworth (2007) admits that the scale is susceptible to social desirability bias, entails retrospective reflection since participants have to respond to items by thinking of their usual past behavior, and finally, does not take into consideration situational variables that may affect successful performance (Duckworth et al., 2007; Duckworth, & Yeager, 2015) . It seems to be a measure that identifies and indicates individual differences, ignoring environmental and social influences to success. All these, although beyond the scope of this review, raise many issues about whether research on grit is directed to learners of the same level tested against the same outcome measures, or to individuals who belong to minorities, have access to limited opportunities, and may be vulnerable to discrimination due to exam performance. As Shechtman, DeBarger, Dornsife, Rosier and Yarnall (2013) claim, individuals are at risk of being accused of poor performance due to lack of grit, instead of lack of critical support and opportunities in their educational environment. This last point could be one new insight into the measurement of grit, which is linked with the antecedents of the construct as well.

Moreover, the review was vulnerable to sample bias. Grit was found to predict performance in some samples but not in others, indicating that such new constructs should be assessed in a wider range of samples before assuming generalization (Dixson et al., 2016) . Some studies aimed at specific samples (talented students, particular schools, students with high GPAs) and the majority of studies used samples where females outnumbered males. Furthermore, students from ethnic minorities comprised a large proportion of the samples studied. All these factors raise important ethnic group, cultural and gender issues. Ethnic minority students may carry the stigma or stereotype and may put maximum effort to achieve their academic goals, hence scoring higher in self-reports of grit (e.g. Strayhorn, 2014; Weisskirch, 2016 ). Empirical evidence suggests that women attribute academic success to effort and are more motivated by academic rewards, whereas men link their success to their ability (Lindo, Sanders, & Oreopoulos, 2010) , so gender differences in grit are yet to be further explored.

Additionally, there were scarce studies that explored long-term goals, which is the actual object of reference of the grit concept according to Duckworth et al. (2007) . Most of the studies used cross-sectional designs, addressing outcomes within a small time frame rather than investigating grit for long-term goals. Prospective studies investigated participants over short periods of time, most studies focused on standardized test scores rather than actual learning processes, and no study explored whether goals were meaningful to participants instead of perceived as merely an endpoint in the educational process.

Findings indicate that there is ground for research and practice regarding grit and its antecedents in education. It seems that individual factors are not the only ones that contribute to the occurrence of grit; situational and environmental factors also play an important role. Hence, by understanding the factors that can make individuals grittier, researchers and educators can work on specific programs and projects to enhance grit, not only in learners but in the entire educational community. Targeted actions to nurture a positive educational environment with support networks and strong relationships among members could offer a sense of belonging and foster hope in an effort to keep individuals’ efforts focused on their long-term educational goals.

Future research should consider a broad range of issues. First of all, updated or supplementary instruments need to be developed to measure grit more accurately, and clarify its structure and its relation to conscientiousness. More focus should be directed on the “perseverance of effort” as a distinct facet and its role in educational outcomes. This will facilitate a deeper understanding and will allow for the development of adequate interventions, such as the employment of character strengths to nurture and strengthen grit (Perez, 2015) . Studies should also use repeated measurements and collect scores on exams across time instead of only relying on final exam scores to measure performance in educational contexts, in an effort to assess learning and acquired knowledge.

Additionally, longitudinal studies should explore whether its two facets remain positively correlated in the face of long-term adversity, situational changes or personal goal modification. It could also be investigated whether grit remains the same across all educational settings, e.g. whether individuals report the same level of grit in school or in computer or foreign language education, and whether it is retained after education completion, e.g. in relevant work environments. It would also be interesting to investigate grit of individuals who choose to pursue a career in the Academia, e.g. doctoral students and researchers. Grit should be investigated in separate groups of students, since different levels of grit may appear in talented students, ethnic minority students, students of different educational levels, or even students that come from collectivistic and individualistic cultures. Additionally, grit could be explored in cases of learning disability as well as in terms of setting variation, e.g. differences in grit between learners of regular versus Montessori school settings, or even distance learners. Finally, qualitative studies could help us find out how grit is perceived by different cultures and ethnic groups, especially in multicultural classes, and explore in more depth whether group dynamics and positive relationships in the educational context may foster grit.