Abstract:

This study was motivated by a discussion how the results of state maturity influence on academic undergraduate performance (average mark) and the number of years students take to complete their undergraduate studies (study duration).

Objectives of state maturity: completing high school education for high school students and range for enrolment in higher education institutions in the Republic of Croatia for all categories of gateways.

There are three compulsory subjects on the state maturity: Mathematics (MATH), Croatian (CRO) and foreign (FORL) language. MATH and CRO can be deployed on two levels. the higher level, being more demanding than the lower level exam. For this reason, the percentage of exam resolution achieved on the multi-level exam multiplied by the coefficient of 1.6.

Were analysed data using data from a questionnaire among 257 graduates from 2016 to 2018. Since our students have not remembered the total score just the mark and the level of exam, we have multiplied higher level marks with 1.6.

A multivariate linear regression was used to analyse if and how marks of subjects from state maturity predict undergraduate study achievement: average mark and duration of study (with logarithmic transformation - lnaverage_mark and lnduration). In order to consolidate these results, we used canonical correlation as a variation on the concept of multiple regression and correlation analysis. We analyse the relationship between two sets of variables. First one is maturity marks MATH, CRO and FORL. The second set of variables is undergraduate performance: duration and average mark of study. The significance level of 5% was considered statistically significant and results are underline.

Correlations between MATH (r=0.27), CRO(r=0.25) and FORL(r=0.18) with average mark of study are positive. Duration of study has negative correlation with MATH (r=-0.15) which suggests that as students achieve better MATH, they will study shorter. Results for CRO (r=-0.06) and FORL(r=-0.01) do not indicate that. Multivariate regression equations are:
lnaverage_mark=0.968+0.026 MATH+0.014 CRO+0.016 FORL
lnduration=1.537-0.035 MATH-0.005CRO+0.008 FORL
and show similar results as correlation analysis but the model with duration as dependent variable is nor adequate (good) model F(3,207)=1.69, while model with average mark of study as dependent variable is adequate model F(3,206)=9.48 although R2 is 12.1%. Mark of MATH on maturity is statistically significant predictor of average mark and duration of undergraduate study. The higher MATH the higher average grade of the study and the shorter duration of the study .

Correlation between duration and average mark of study is negative (r=-0.52) which suggests that students with a better mark study shorter.

Results of canonical correlation shows that the first canonical correlation is 0.35, which would appear to be substantially larger than any of the between-set correlations. The first canonical variable (standardized coefficients) for undergraduate performance shows most weight on average mark of study (1.06) than duration of study (0.16). The first canonical variable for maturity marks shows similar importance of MATH and CRO (0.57 and 0.51) while FORL has lower impact (0.34).

We can conclude that assessments of state maturity are good predictors for success in undergraduate study, especially MATH.

Keywords:

State maturity, study mark, study duration, multivariate regression, canonical correlation.