eng Science and Education Publishing American Journal of Educational Research 2327-6150 2020-05-19 8 5 293 298 10.12691/education-8-5-11 EDUCATION20208511 article Friska Labora Christiani friskachristian@gmail.com 1 Pargaulan Siagian 1 M Mukhtar 1 Universitas Negeri Medan, Jalan Willem Iskandar Pasar V Medan Estate, Indonesia This study aims to determine (1) whether the mathematical connection ability of students taught with Autograph-assisted Guided Discovery Learning (GDL) models is higher than students taught with Autograph-assisted Problem Based Learning (PBL) models, (2) whether there is interaction between the Guided Discovery Learning (GDL) model and the Problem Based Learning (PBL) model with numerical ability on students' mathematical connection abilities. This type of research is quasi-experimental research with a population of all students of class XI Pancur Batu Private High School Methodist. This study uses two-way ANAVA. Then proceed with the Scheffe test to see whether the two groups are significantly different. Based on the statistical calculation of the Scheffe test and confirmed with SPSS, it can be concluded that (1) The ability of students' mathematical connections taught by Guided Discovery Learning (GDL) assisted by Autograph Software is higher than that of students taught with Problem Based Learning (PBL) assisted by Autograph Software, (2) There is an interaction between the learning model with numerical ability on the ability of students' mathematical connections. http://pubs.sciepub.com/education/8/5/11/education-8-5-11.pdf Guided Discovery Learning Problem Based Learning connection ability numerical ability Autograph