This study aimed to find out the effect of learning model and form of assessment toward inferential statistical achievement after controlling numeric thinking skills. This study was quasi experimental study with 130 students as the sample. The data analysis used ANCOVA. After controlling numeric thinking skills, the result of this study show that: (1) the inferential statistical achievement of the students group which follow SCSS learning model is higher than the group which follow conventional learning inferential, (2) statistical achievement of students group which got performance assessment is higher than conventional assessment, (3) there is interaction effect between learning model and form of assessment toward the students' statistical achievement, (4) in SCSS learning model, the students' inferential statistical achievement which got performance assessment is higher than the group which got conventional assessment, (5) in conventional learning model, the students' inferential statistical achievement which got performance assessment are lower than the group which got conventional assessment, (6) in performance assessment, the students' inferential statistical achievement which follow the study with SCSS learning model is higher than those which follow conventional learning model, and (7) in conventional assessment, the students' inferential statistical achievement which follow SCSS learning model is lower than the group which follow conventional learning model.