The idea of data being a mixture of signal and noise is perhaps one of the most fruitful and fundamental ideas of statistics. To enable future mathematics teachers to educate students to become statistically literate, we propose an integrative approach connecting central topics of school mathematics with the signal-noise idea. A course on modeling functional relationships-a core topic in any mathematics curriculum-confronts students with the signal-noise idea when looking at the deviation between model and data. We provide empirical evidence that students of such a course acquire implicitly important statistical thinking skills.