I recently presented at the National Science Teaching Association (NSTA) regional conference in Cincinnati, OH. My topic was "Connecting simulations to real-world examples." I think using simulations is a fantastic way of introducing, teaching, and reviewing concepts. They are also valuable tools for practicing data collection, analysis, and experimental design. For all the benefits of simulations in education, though, there are some negatives. By making connections between simulated data and real examples we can mitigate these drawbacks.
One potential problem with relying on simulations is that they often oversimplify natural phenomena in order to better help students understand the content. Real systems often have variables that, even if understood, cannot be fully controlled for. Students should understand the limitations of the simulations they are using. They don't necessarily have to understand the complexity of a system, but they should at least be aware that they are working with a simplified version.
Simulations also tend to have data that matches predicted trends. Biology Simulations always incorporates random components, so students get variations in their data, but it still typically turns out the way it's expected to. Again, this is highly useful to teaching content but misses some of the messy nature of real-world science. Data does not always lead to clear-cut conclusions...it is often messy, inconsistent, and inconclusive. It is important to understand that inconclusive data is still an important part of the scientific process.
Finally, relying on simulation too much prevents students from experiencing the hands-on nature of science. Simulations should supplement lab activities, not replace them.
My NSTA presentation introduces four approaches to making real-world connections to a simulation; a teacher-selected information source, analysis of data sets, student research, and a follow-up lab. Here is a link to my presentation slides, which include examples for using each of these approaches.