The relationship between secondary education outcomes and academic achievement: a study of Finnish educational sciences students

ABSTRACT A multitude of research shows that the secondary education (SE) outcomes relate with university academic achievement (AA). More research is needed on SE outcomes subject-specifically. We examined the relationship between the Finnish matriculation examination (ME) results and AA, study success (grades) and study progress (credits), of educational sciences students (N = 1,214) who were pursuing a three-year bachelor’s degree in 2007–2017. The overall ME results explained 15% of the variance in study success and less than 1% in study progress. Success in mother tongue, mathematics, psychology and health education in the ME along with 1st- and 2nd-year study success related positively with higher study success throughout the three years. For the 3rd-year study progress, the results were more mixed, and progress in the 1st and 2nd year was statistically significant compared to the ME results. In conclusion, it is recommended to consider subject-specific ME outcomes regarding student admissions and curricula.


Introduction
The secondary education (SE) outcomes, such as high school grade point average (HSGPA), have been shown to predict well university students' academic achievement (AA) (see e.g.Richardson, Abraham, and Bond 2012).Field-specific prior knowledge is known to relate to high performance within a field in question (see e.g.Pinxten, Van Soom, and Peeters et al. 2019).A closer look at the literature on meritbased approaches to explaining AA reveals several gaps and shortcomings.First, the validity of a HSGPA has been questioned for not making visible subject-specific skills and knowledge (see e.g.Office of Qualifications and Examinations Regulation 2012).Second, the educational sciences are underrepresented in research settings compared to, for example, STEM (Science, Technology, Engineering and Mathematics) subjects and fields (see e.g.Corcoran and O'flaherty 2017).Third, there is less evidence on the relationship between SE outcomes and AA in the years between the 1 st year of university study and graduation (see e.g.Westrick, Lee, and Robbins et al. 2015).
Therefore, we examined students who studied educational sciences in 2007-2017 at a Finnish university (henceforth called 'students').We studied AA, both study success and study progress, in the three-year bachelor's degrees (BA) overall and year-specifically (the 1 st , 2 nd and 3 rd years) based on both overall and subject-specific results of the matriculation examination (ME) which is the national congruent test taken among SE students.To gain a longitudinal perspective, we examined the relationship between AA in the 1 st and 2 nd year with subsequent years of BA.

SE outcomes as one indicator at the gate to a university study
In some countries, youth study a wide range of different subjects (for Finland, see the Context section), while in others, youth study field-specific tracks already in SE (European Education and Culture Executive Agency [Eurydice] 2018).Hence, each applicant and enrolled student has an individual combination of SE outcomes, which can be considered their entry characteristics (Noyens, Donche, and Coertjens et al. 2017).
The admission criteria applied for admission to a university are different.Some criteria emphasise a background in a certain track (Vulperhorst, Lutz, and De Kleijn et al. 2017), a certain overall level of accomplishment in SE (e.g.HSGPA) (Steenman, Bakker, and Van Tartwijk 2016), or a certain level of success in standardised tests, e.g. the Scholastic Assessment Test (SAT) or American College Testing (ACT) (Noble and Sawyer 2002).Other admission systems open the doors for all applicants but grant study rights based on success in the first year of study (Brouwer, Jansen, and Hofman et al. 2016).Regardless of admission criteria, an admission process aims to identify and admit applicants who are most likely to complete the education (Noble and Sawyer 2002).
According to Noyens, Donche, and Coertjens et al. (2017), research on the transition to universities can be categorised in terms of phase, time, level or outcome.While the relationship of the SE outcomes with AA has been studied in other contexts, the Finnish educational sciences offer insight into an education system known for its highly selective student admission (OECD 2022), but which has considered the tests taken and grades attained in the national matriculation examination (ME) as an admission criterion in heterogeneous ways, if at all (Kunnari, Pursiainen, and Läärä et al. 2022).Students have broad possibilities for choosing subjects to include in the ME (Finnish National Agency for Education 2003); however, students in educational sciences have typical profiles (Kunnari, Pursiainen, and Läärä et al. 2022).

Evaluation of AA based on SE outcomes
The research field does not have a shared definition of AA (Andersson 2003).AA is usually measured with indicators for study success in terms of grades (study results in particular courses or a period of time) and study progress in terms of credits (e.g.number of European Credit Transfer and Accumulation System (ECTS) that a student has earned in a particular period of time) (see, e.g.Andersson 2003).However, Hattie (2009) identified 130 significant variables to explain AA in the context of schools, colleges, and universities.In this study, we concentrated solely on merit-based outcomes, i.e. grades in SE as well as grades and credits in universities.
A large number of studies in the broader literature have examined AA based on SE outcomes.Students with high grades in SE usually succeed better in university (see e.g.Corcoran and O'flaherty 2017).Also, the higher the success is in SE, the better the students perform during the 1 st year of study (Brouwer, Jansen, andHofman et al. 2016). De Koning, Loyens, andRikers et al. (2012) found that SE outcomes relate to higher achievements in the 1 st year at a high level and the 2 nd and 3 rd years at a moderate level in bachelor's studies.The 1 st year has also been identified as important for later study success (Allen, Mattern, and Ndum 2019).
Literature on the relationship of subject-specific knowledge with AA in university is less consistent.Prior knowledge (Dochy 1992) of STEM in SE was associated with higher success in the 1 st year of STEM studies (Pinxten, Van Soom, and Peeters et al. 2019), and psychology was associated with more successful completion of the 1 st year of psychology study (Diseth, Pallesen, and Brunborg et al. 2010).High success in a native language and mathematics in SE has been identified as related to high success in university studies across disciplines (De Koning, Loyens, and Rikers et al. 2012).
In Finland, discipline-specific studies on the relationship between the ME and study success of medical and law students (Kleemola and Hyytinen 2019;Lindblom-Ylänne, Lonka, and Leskinen 1999) have identified similar relationships between the ME and study success as international research on AA.Among students in educational sciences, success in the ME has been found to relate to study success in universities more than entrance tests (Kupiainen, Marjanen, and Ouakrim-Soivo 2018;Mankki, Räihä, and Joutsenlahti 2018).
The SE outcomes have been identified as relating less to study progress (De Koning, Loyens, and Rikers et al. 2012).Subject-specific ME results have been connected to field-specific study progress among law students (Kleemola and Hyytinen 2019) and students of social sciences and humanities (Kallio et al., 2018).Also, the 1 st -year study progress has been found to affect progress in later years across disciplines (Koster and Verhoeven 2017).
In this study, we approach the relationship of SE outcomes with AA in universities with a temporal comparison aspect (Wigfield, Eccles, and Möller 2020), where AA is examined considering previous achievement-related experiences (SE outcomes) (see Wigfield, Eccles, and Möller 2020, 658).The other way around, this also relates with 1983 expectancy-value theory and more precisely with the utility value of educational choices: SE students might have chosen the ME tests based on known subject matter in educational sciences in university (Wigfield and Cambria 2010).In this empirical study with registry data (Wolff, Helm, and Zimmermann et al. 2018) the temporal comparison aspect was utilised when investigating AA longitudinally at multiple waves (time and phase: Noyens, Donche, and Coertjens et al. 2017).Also, the educational outcomes was examined on two levels; ME results overall and subject specifically and AA year specifically and the 3 years overall (Office of Qualifications and Examinations Regulation 2012).

Research questions and objectives
Overall, there is scant research on educational sciences students' AA.This study investigated the relationship between SE outcomes and educational sciences students' AA.Students' ME results are interpreted as prior knowledge (see e.g.Dochy 1992) and is one entry characteristic (Noyens, Donche, and Coertjens et al. 2017) upon enrolment to university studies.We addressed the following research questions: (1) What kinds of entry characteristics did students possess based on the ME results?(2) How did entry characteristics relate to AA year-specifically (the 1 st , 2 nd and 3 rd years) of BA studies?(3) How did the 1 st -and 2 nd -year AA relate with AA in subsequent years of BA studies?

The Finnish general upper SE and the ME
The academically emphasised general upper SE culminates in the national ME and is the main path to Finnish universities (The Matriculation Examination Board 2020).The national curriculum (Finnish National Agency for Education 2003) includes 75 compulsory and optional courses (approximately 38 hours/course) in mother tongue, advanced or basic syllabus mathematics (henceforth called 'advanced mathematics' and 'basic mathematics'), advanced and basic syllabus languages, and subjects in humanities and natural sciences (Kunnari, Pursiainen, and Läärä et al. 2022).
Participants in this current study needed to accomplish within three consecutive times (e.g.spring -autumn -spring) the obligatory mother tongue and at least three other subjects in the ME.Previously, they had had to complete the compulsory and a number of optional courses in these subjects to receive the ME certificate (Finnish National Agency for Education 2003).The ME assesses whether students have reached an adequate level of maturity and the targeted learning outcomes defined in the national curriculum (Britschgi 2014).The assessment is based on seven grades: (descending in order) laudatur (L), eximia cum laude approbatur (E), magna cum laude approbatur (M), cum laude approbatur (C), lubenter approbatur (B), approbatur (A), and improbatur (I) (failed test) (The Matriculation Examination Board 2017).

Research-based educational sciences
The research-based Finnish educational sciences (Toom 2010) consist of theoretically emphasised courses and practical training in an everyday context (Tirri 2014).To be an eligible teacher, students need to, according to the European Bologna process, finish bachelor's and master's degrees in programmes designed to educate teachers.An exception is early childhood education, where BA is adequate for eligibility.The students at Finnish universities attain grades ranging from 1 (low) to 5 (high) or passed/failed, and study credits based on the ECTS for accomplished courses.A BA is composed of basic and intermediate studies.
The study was conducted at a faculty of education which offers study programmes that educate future teachers for kindergarten and primary schools (classes 1-9, ages 6-15), as well as professionals in education outside the teaching profession.The study programmes examined are intercultural teacher education, educational sciences, primary school teaching, artistic and practical subjects (including study programmes for music, handicrafts, and technology) and early childhood education (bachelor and master).

Data sources and participants
Originally, the registry data consisted of 1,579 students from 2006 to 2015.Data on BA achievements were obtained from the studied university and the ME results from the Matriculation Examination Board.The two datasets were combined by this university and analysed anonymously.
First, the data was limited to include three consecutive years of BA studies among enrolled and beginning students between 2007 and 2014; the data coverage extended until 2017 (N = 1,316).Enrolled students without any registered achievements during the three years (16 students) and students with no cumulative change in credits from the 1 st year forwards (37 students) were excluded from the data.No cumulative change in credits might indicate a drop-out and did not provide a coherent overall view for the analysis of the three-year BA studies.Students who had not received assessments on a scale of 1-5 (11 students) were excluded from the data.One student with no grade in mother tongue in the ME and one student with incoherent information on the year of enrolment (−3 gap years, most probably a registration error) was excluded.Extreme outliers in both the lower and upper ends of the 1 st , 2 nd and 3 rd study years (36 students) were not examined due to the credits being very different from the rest of the data (Field 2018) which may violate the assumptions (Tabachnick and Fidell 2007).After these restrictions and exclusions, the study population included 1,214 students, 83% of whom are women.

Dependent variables
Study progress was defined based on study credits gained year-specifically and the three-year period in BA.Student's study success in the courses included in the curriculum was assessed based on grades 1-5, during the same timeframe.

Independent variables
The MEs were considered based on overall success (similarly to the high school grade point average, HSGPA representing students' average success in a particular period of time).The tests selected for closer examination were obligatory mother tongue, advanced syllabus English (henceforth called 'advanced English'), mathematics (advanced, basic, or none), psychology and health education.Advanced English was chosen because most SE students take the test, and psychology and health education being the most popular tests in the humanities and natural sciences group (Kunnari, Pursiainen, and Läärä et al. 2022).Mathematics and mother tongue were considered important to examine due to their central role in explaining AA (see e.g.Allen, Mattern, and Ndum 2019).We transformed the grades into numerical values according to the following standard practice: I = 0, A = 2, B = 3, C = 4, M = 5, E = 6, and L = 7 (Lindblom-Ylänne, Lonka, and Leskinen 1996; The Matriculation Examination Board 2017).See Table 1 for descriptive statistics on the dependent and independent variables.
The selected ME tests were coded into four categories.The first category included grades I, A and B (low achievers), category two grades C and M (average achievers) and category three grades E and L (high achievers).The fourth category included those not taking a test on the subject in question (except for their mother tongue).For study success and progress, the first category included those who succeeded or progressed with one standard deviation below the mean (low success/progress), while the third category included those who succeeded or progressed with one standard deviation above the mean (high success/progress).Students between categories one and three were placed in the second category (average achievers).The fourth category covered those not gaining any credits (ECTS) or not attaining any grades.Mathematics was coded into seven categories.The low achievers in basic mathematics were included in category one, the average achievers in category two and the high achievers in category three.The low achievers in advanced mathematics were included in category four, the average achievers in category five and the high achievers in category six.The applicants who took neither advanced nor basic mathematics were included in category seven.The reference category was average achievers, and for mathematics average achievers in basic mathematics.
The overall ME results and the ME tests selected for closer examination correlated with each other (see Table 2).Also, the year-specific AA each correlated with the variable that considered three-year AA.These variables were considered to refer to the same constructs and were applied separately in different analyses (Field 2018).The correlations could not be computed for mathematics because the students had mostly taken either advanced or basic mathematics or neither of them.

Statistical analysis
Before conducting the statistical analysis, the relevant assumptions were tested.The large sample size was fitting for regression analysis and, after the exclusions of outliers, the values for skewness and homoscedasticity were within reference values (Tabachnick and Fidell 2007).
The distributions of the variables were summarised with descriptive statistics (RQ1).A simple regression analysis was applied to reveal the relationship between the average success in the ME (grades 0-7) and three-year AA (RQ2).Zero stands for failed tests (N = 11).Not taking a test in the ME was not considered in the analysis.To answer the second and third research questions, multiple linear regressions with categorical predictors were applied (Field 2018) using general linear model in SPSS.This rendered us to test the effects of several independent variables, individually and in combination, on a dependent variable.For RQ2, we examined the relationship of the categorical independent variables (mother tongue, advanced English, basic and advanced mathematics, psychology and health education) with year-specific AA in university.Similarly, for RQ3 we examined the relationship between the 1 st -and 2 nd -year study success and study progress with AA in the 3 rd year.

Tests taken and grades attained in the ME
The first research question on student's ME results is answered with descriptive results.On average, students' ME background (see Table 3) was relatively similar.Other tests in humanities and natural sciences were taken at most by 17%, intermediate Finnish or Swedish by 78%, advanced syllabi language by 0-6% and basic syllabi language by 0-10%.The average overall success in the ME was 4.6.The standard deviations (1.0 to 1.4) indicate that there was considerable variation for each examined ME test.

Relationship between prior knowledge and year-specific AA
We answered the second research question to understand the relationship between overall and subject-specific ME results and AA.A significant regression equation was found in the three-year study success (F(1, 1212) = 212.412;p < .001),with an R 2 of .149.Students predicted three-year study success was equal to 2.849 + 0.198 (the ME results) when the ME results were measured in grades 0-7.Students' three-year study success increased by 0.198 with each increase in the average grade in the overall ME results.
The relationship of subject-specific ME results with year-specific AA is reported in Table 4.The ME results explained 15% of the 1 st -year study success variation, 12% of the 2 nd -year study success and 7% of the 3 rd -year study success.In the 1 st year, all of the examined ME tests were statistically significant, in the 2 nd year all but advanced English and in the 3 rd year only mother tongue and mathematics.In general, the results show that in relation to the average performers, the higher the success was in all the examined ME results, for mathematics in advanced mathematics, students succeeded on average with higher grades throughout the BA.Mother tongue and average or high success in advanced mathematics have the highest Beta values.These values decreased in comparison to the reference group after the 1 st year.
Regarding study progress, a significant regression equation was found in the three-year study progress (F(1, 1212) = 10.38;p < .001),with an R 2 of .008.Students predicted three-year study progress would equal 154.65 + 4.006 (the ME results) when the ME results were measured in grades 0-7.Students' three-year study progress increased by 4.00 ECTS, with each increase in the average grade in the overall ME results.The relationship of subject-specific ME results with year-specific AA are reported in Table 5.The year-specific examinations of the tests chosen for a closer examination revealed that the R 2 remained low throughout the BA.Mathematics and health education in the 1 st year and mathematics in the 2 nd year were statistically  significant.In the 3 rd year, none of the examined ME tests were statistically significant.In relation with the reference group, students gained more credits throughout the BA if they succeeded with high grades in advanced mathematics, psychology, and health education.Low achievers in health education succeeded better in the 2 nd and 3 rd year in relation with the reference group.Also, low achievers in psychology compared to the reference group did succeed better in the 2 nd year.Table 4. Bachelor's degree students study success in the 1 st , 2 nd and 3 rd year by tests taken and grades attained of selected matriculation examination tests by level of performance.Point estimates (beta) and standard errors (SE) of the regression coefficients with 95% confidence intervals (CI) in contrast to the reference category as defined by average success, and for mathematics, average success in B-level mathematics.
tongue, mathematics, and the 1 st -year study success were statistically significant variables.For the 3 rd -year study success, the ME together with the 1 st -and 2 nd -year study success explained 20% of the variation, statistically significant variables being mother tongue and the 1 st -and 2 ndyear study success.For study progress, the ME results together with the 1 st -and 2 nd -year study progress increased the R 2 to 7% for both the 2 nd and 3 rd years.For the 2 nd -year study progress, mathematics and the 1 st -year study progress were statistically significant while for the 3 rd -year study progress, the 1 st -year study progress and the 2 nd -year study progress were statistically significant.Changes in the parameter estimates (B, SE, and 95% CI) were minor, and the main principle of higher success or progress in the independent variables was related with higher success or progress in the 2 nd and 3 rd study year.

Discussion
First, the descriptive results showed that besides the test in obligatory mother tongue, the most frequently taken tests were advanced English, basic mathematics, psychology, and health education.The highest average grades were attained in these tests, except for advanced English.The entry characteristics were similar to those accepted nationally for educational sciences programmes (Kunnari, Pursiainen, and Läärä et al. 2022).
Second, the linear regression analysis showed that the overall ME results explained approximately 15% of study success and less than 1% of study progress in the three-year BA studies.Both results were statistically significant; however, the coefficients of determination (referring to the linear regression analysis result, R 2 ) remained rather low which is in line with research internationally (Richardson, Abraham, and Bond 2012).Research on the ME has had similar results on the correlations (Kupiainen, Marjanen, and Ouakrim-Soivo 2018) and coefficients of determination (Mankki, Räihä, and Joutsenlahti 2018).Multiple linear regression analysis was applied to examine the relationship of the ME results with the three-year BA year-specifically.For study success in BA, the results show that higher success in all the examined ME tests was related with higher success throughout the BA.This result is in line with research results by, for example, Corcoran and O'flaherty (2017).For study progress, high success in advanced mathematics, psychology, and health education was related with better progress throughout the BA.This is in line with, for example, Kleemola and Hyytinen (2019) research results on subject-and field-specific research results.Also, even low success in psychology was indicative of better progress in the 2 nd year and low success in health education with better progress in the 2 nd and 3 rd year.
These results could be indicative of prior knowledge that gives the readiness to acquire new knowledge and, hence, supports progress in BA (Diseth, Pallesen, and Brunborg et al. 2010).The novel insight to AA in this study is the use of the ME results on two levels (overall and subject specific), the context (Finnish universities), and the longitudinal approach through a transition phase from SE to educational sciences.It is also noteworthy that the meaning of grades is different for each individual throughout the study path (Cachia, Lynam, and Stock 2018;Wigfield and Cambria 2010).Also, on a contextual level, the grades are employed differently depending on the countries education system (Powell & Arora-Jonsson, 2021).Hence, it was fruitful to study educational sciences due to its highly selective student admission.It is important to highlight that these results provide an additional insight to prior research on AA from the perspective of temporal comparison (Eccles and Wigfield 2020) on an individual level (De Clerq et al., 2021).Considering the vast strands of research on educational transitions and AA (Noyens, Donche, and Coertjens et al. 2017), these results reinforce the importance of subject-specific prior knowledge in psychology and health education (Diseth, Pallesen, and Brunborg et al. 2010; Office of Qualifications and Examinations Regulation 2012) which tell more specifically the individuals' interest and learning than the overall average of success.Prior research has shown that subject-specific prior knowledge in the ME relates with admission to educational sciences (Kunnari, Pursiainen, and Läärä et al. 2022) and with AA among law sciences students (Kleemola and Hyytinen 2019).Together with the research results that the ME is better predictors of AA than the entrance tests (Mankki, Räihä, and Joutsenlahti 2018), we consider the subject-specific results, along with mother tongue and mathematics (De Koning, Loyens, and Rikers et al. 2012), to be good indicators for AA in educational sciences, Also, for RQ3, the multiple linear regression analysis was applied.The main result is that towards the 3 rd year of study, the share of statistically significant variables in the ME declined.For study success in the 2 nd year, only mother tongue and mathematics remained statistically significant when the statistically significant 1 st -year study success was added to the analysis.For study success in the 3 rd year, the only significant variables were mother tongue along with the 1 st -and 2 nd -year study success.For the 2 nd -year study progress, mathematics and the 1 st -year study progress were statistically significant while for study progress in the 3 rd year, only the 1 st -year study progress and the 2 nd -year study progress were statistically significant.These results are in line with prior research, see e.g.Allen, Mattern, and Ndum (2019).Despite the weak coefficient of determination between the ME and study progress, subject-specific aspects of progression were identified.Following Westrick, Lee, and Robbins et al.'s (2015) suggestion, it was meaningful to examine the years between entry into university and graduation.Our results clarify the role of SE outcomes and BA studies in explaining especially study success among students in the educational sciences (cf.Corcoran and O'flaherty 2017).
A few limitations can be found in this study.Due to the lack of courses graded 1-5 in the 1 st year, we could not carry out a detailed analysis of students' study success.Courses graded passed, failed, accomplished, or other non-numerical indicators throughout BA studies were not included in this examination.The data also lacked information on precise points of graduation; this was not examined.A subject's level of difficulty and type of knowledge it provides should be examined to understand their share in explaining AA and learning in BA.

Table 1 .
Descriptive statistics of independent and dependent variables.
1 IV = independent variable, DV = dependent variable. 2d on numerical scoring of the grades from lowest to highest: I = 0, A = 2, B = 3, C = 4, M = 5, E = 6 and L = 7 (The Matriculation Examination Board 2017).3Includesonlygrades1-5 4 Missing values represent those students who did not complete the test in question.5Missingvalues for GPA represent courses evaluated based on passed/fail or other non-numerical assessments.

Table 6 .
Bachelor's degree students AA in the 2 nd and 3 rd year by study success and study progress in the 1 st and 2 nd year.Point estimates (beta) and standard errors (SE) of the regression coefficients with 95% confidence intervals (CI) in contrast to the reference category as defined by average AA.2 nd -year study success, adjusted R 2 13% 3 rd -year study success.Adjusted R 2 18%