Learning loss due to online learning has caused students to experience a decrease in their ability to solve mathematical problems. One of the declines in students' mathematical problem-solving skills occurred in the statistics course. Statistics as a branch of mathematics has an important role in training students' skills in processing and analyzing data. The low ability of students to solve mathematical problems became a serious problem. So we need an appropriate learning model to overcome these problems. One of the learning models that can be used is hybrid learning. The subject of this study was the primary student study program from the Faculty of Education at UPH in Indonesia, with 110 samples divided into 54 control groups and 56 experimental groups. The control class was treated with full online learning, while the experimental class used hybrid learning. This research aimed to compare which type of learning is more effective, full-online or hybrid learning. The research method used in this study is quasi-experimental. Data were processed using SPSS, called the normality test with Shapiro-Wilk, and statistical tests using Mann-Whitney. The results showed that hybrid learning was more effective than full online learning. [For the full proceedings, see ED654100.]