When designing an experiment examining sex education's ability to encourage contraception use, you first need to figure out how to operationalize contraception use. In this case, it would be simplest to use the total sales of contraception mediums within the area. However, with this you run into the issue of whether or not people buy AND use contraceptive vs whether they just buy them. Thus, simply going with sales is an invalid method of operationalizing contraception use. Another method would be through administration of survey designed to assess attitudes toward and likelihood of contraception use before and after the sex education class. I draw emphasis to the fact that the questionnaire needs to be administered before and after the sex education class as this way the experiment is able to take on a pre/post design mitigating some issues that come with comparing independent groups.

Once we've found out how to measure likelihood of contraception use in our experiment, we need to ensure that an adequate control group is present for comparison. This group would act as a baseline comparison group so that we can measure how much of a change has occurred in the sex education group relative to that of the rest of the population. In this experiment, the control group would likely be comprised of a group of individuals sampled from the same population as those who are taking the sex education course. However, instead of taking a course on sex ed, the control group would be taking a class that is unrelated to sex ed (e.g., carpentry, modern lit, how to actually do your taxes, etc.). Please note, in order to ensure that the control group is, well, a valid control group, the same questionnaire on contraception that is administered to the sex education group is administered to the control group before and after their class.

Finally, we need to ensure that there is no bias with respect to the likelihood that an individual will end up in the control or the sex ed group. This can be simply done by randomly assigning experiment participants to either group. However, random assignment is not always possible, so try to stay flexible with your thinking here.

These are the basics of how one would go about designing an experiment to assess whether participating in sex education classes increases the likelihood that sex education attendees will use contraception.