The objective of this study was to examine the relationship of formative tests to early learning ability of students in the science learning style. This research used an experimental method with a 2 x 2 factorial design. The participants comprised all the students in class VII of the Islamic Junior High School State of Kolaka, a total of 343 students from 9 classes. Sampling was based on the use of a two-stage cluster sampling technique to randomly select four classes of students. The four classes were re-randomized, with two being given formative tests and two being given multiple choice tests. The t-test was used to measure the student learning outcomes. The results showed: (1) the use of formative tests in narrative form in the learning process can provide a measure of higher learning output compared with the use of formative multiple choice tests. (2) For those students who have a high initial ability, the use of formative tests can provide higher learning outcomes compared with the use of multiple choice tests. (3) For those students who have a lower initial ability, the use of formative tests in narrative form does not provide a different outcome to tests in multiple choice formats.