Regression is fairly simple, you have a set of data, you think that you found something that can be expressed via an equation in the data (in simple cases the example would be linear). Stupid example : "Number of murders = Base crime rate* Sexual frustration factor - gun control factor * number of guns present"
Now you try to see how good your modell (basically your equation) fits both your own data and the data of others. You now start to analyse the robustness (if you change this parameter within reasonable boundaries, does the outcome stay within reasonable boundaries) of your parameters. If everything is fine (so, your modell fits your data, it fits equivalent data from someone else and your parameters are fairly robust), you can now apply your equation to the future and say things like:
"According to my modell, having lots of sex reduces violence more than gun control."
Statistics are fairly straght forward, if you are not afraid of math that is.