This is a complex question that has many answers but I will try to give a complete answer.
When designing an experiment you have to make sure
1) you have a representative, random sample - this means your participants are representative for the population you're taking them from (having random students from a class which are representative for students for example
2) you have a control and an experimental condition - you always need a control condition with which you can compare your experimental manipulation. Only in this case can you make sure that your results are the consequence of your experimental manipulation.
3) always test the desired variable before and after the experimental procedure
