Continuous assessments encourage sustained learning that contributes to deeper knowledge retention and

academic success. Investigating continuous assessments, such as quizzes and tests, is important in predicting

final scores because they provide a reliable measure of students’ learning progress across the semester. The

present study examined the extent to which quiz scores, and test scores were able to predict final scores in a

software and the formulated model provides useful insights for educators in identifying early indicators of

ultimately enhancing the overall quality of instruction.

multiple linear regression)

The importance of having a good grasp of the English language in particular amongst English language majors

in the university is increasing as institutions participate in the race to achieving global ranking. Central to this

is the role played by grammar in English Language Studies where Advanced Grammar is a compulsory

predictive framework between formative assessments which act as predictors and the outcome which is

summative assessments. Prior studies have highlighted the usefulness of such models in mathematics (Darman

et al., 2019), in overall first year performance (Dagdagui, 2022), and language education forecasting

achievement (Khiang & Cho, 2019). However, relatively few studies have examined advanced grammar

courses specifically, despite their significance in language proficiency development. This study aims to address

that gap by investigating the predictability of grammar quizzes and test scores on the final exam among first

semester English language major undergraduates enrolled in Advanced Grammar course at UiTM. Specifically,

it aims to (i) develop a multiple regression model to predict final scores and (ii) compare predicted outcomes

with actual results. Findings from this research provide evidence for the effectiveness of continuous

assessment as a reliable

The Role of Assessment in Language Education Proficiency in English grammar is a cornerstone of success for

English language majors, particularly in the context of competitive global higher education (Kasim & Sukarno,

2024). Within this domain, advanced grammar courses are often challenging, requiring consistent practice and

robust assessment mechanisms to track mastery of complex structures. Assessment, therefore, is not merely an

endpoint but an integral part of the learning process. It serves to diagnose understanding, provide feedback,

and ultimately, certify competence. As noted by Rodríguez Rincón, Munárriz and Magreñán Ruiz (2024),

continuous assessment is valued for its ability to motivate students and provide an objective basis for final

foundational knowledge and skills that are ultimately measured through summative assessments. Effective

learning environments strategically use formative tasks to prepare students for summative success.

A growing body of research supports the notion that performance on continuous, formative assessments can

significantly predict final academic outcomes. For instance, Dagdagui (2022) used multiple linear regression to

demonstrate that high school performance and admission test scores could predict 67.3% of the variance in

first-year university students' grades. This finding is reinforced in a specific subject context by Darman et al.

(2019), who successfully predicted students' final grades in a mathematics module using scores from

continuous assessments. Most directly relevant to the current study, Madsen (2020) investigated a university-

level grammar course and found that results from continuous exercises were a strong predictor of final exam

variable (e.g., final exam score) (Uyanık & Güler, 2013). Its application extends beyond education to fields

like biological assays, underscoring its robustness for quantitative prediction (Jarantow et al., 2023). When

applying MLR, it is critical to test its key assumptions, including linearity and the absence of

multicollinearity—a condition where predictor variables are too highly correlated, which can be detected using

statistical measures like the Variance Inflation Factor (VIF) (Shrestha, 2020). Studies by Khaing and Cho

(2019) have effectively utilized MLR in educational contexts, validating it as a suitable method for this study.

Despite the established predictive validity of continuous assessments and the proven utility of MLR, a specific

gap remains concerning advanced grammar courses in English language programs. While studies like

Madsen's (2020) exist, they are relatively scarce compared to research in mathematics (Darman et al., 2019) or

development.

This study employs multiple linear regression analysis to see the relationships between the quiz and test

(independent variables) and the final grammar scores (dependent variable) obtained from 223 students who

enrolled in an advanced English grammar course. In addition, the analysis also shows the variation of the

independent variables accounted for in the dependent variables (final grammar scores) and also produces

predictive models (Uyanik & Güler, 2013). The predictive model for this study is as follows:

+ β

X

where

X

: Test scores

Multiple regression can effectively estimate the relationship between the independent variable(s) and the

dependent variable, provided that their association is linear. Test of linearity was used in determining the

linearity between these variables. Table 1 and Table 2 below show that the linearity exists between each

variable (quiz and test) and the dependent variable (the final grammar scores).

Table 1 Linearity test (Final grammar scores* Quiz)

Between

Deviation from Linearity

1112.56

38

16749.9

222

scores. Based on the analysis in Table 6, the multiple linear regression model can be expressed as:

Predicted Final Grammar Scores=18.298 + 1.237(Quiz) + 1.042 (Test)

Thus, it can be concluded that quiz and test can significantly predict students’ final grammar scores, with

regression coefficients of 1.237 and 1.042, respectively.

Using multiple linear regression in developing regression model to predict the outcomes has been conducted in

many studies, eg. Dagdagui (2022) and Khiang and Cho (2019) in predicting the students’ academic

performance, Darman et al. (2019) in predicting Math scores, and Oflaz (2019) in predicting speaking skills.

Therefore, this technique is deemed applicable to be employed in predicting the final grammar scores of the

students.

Table 7 compares the predicted and observed final grammar scores for 15 selected students. The relatively

Using multiple linear regression, this study examined the extent to which the results of formative assessments

between the two assessment types explaining 75.3% of the variance. The close alignment between actual and

predicted scores, with differences generally below two marks, highlights the reliability of the regression model

as a practical tool for forecasting academic outcomes. These findings are consistent with those outlined in

literature, it also emphasises the importance of continuous assessment not only as a means of tracking student

progress but also as a reliable indicator of their achievement in the summative assessment. This model is

indeed a valuable framework that helps to identify underperformers, it enables educators to undertake remedial

education. In conclusion, this study shows that multiple regression is an effective method for predicting

grammar achievement, confirming the strong predictive value of continuous assessments and supporting data-

driven, student-centred teaching practices. The study while acknowledging its limitations in scope, proposes

that similar studies in the subject of English Language courses particularly in the area of Grammar to enhance

the generalisability of the findings. Future research could incorporate additional independent variables such as

You can dedicate this section to give recognition and acknowledgement to those involved in your project

1. Dagdagui, R. T. (2022). Predicting students’ academic performance using regression analysis. American

Journal of Educational Research, 10(11), 640–646. https://doi.org/10.12691/education-10-11-2