The focus of this study was on comparing three groups of Algebra I 9th-year students: one group using linked representation software, the second group using similar software but with semi-linked representations, and the control group in order to examine the effects on students' understanding of linear relationships. Data collection methods included mathematics pre- and posttests, follow-up interviews, computer-based clinical interviews at the end of the treatment, classroom and lab observations, document analysis, and survey. The conclusion of this study was that semi-linked representations could be as effective as linked representations and that there was a role for each in different situations, at different levels, and with different mathematical concepts. (Contains 2 figures and 2 tables.)