Essentially, in social research there is no limit to the unit of analysis, which is what or whom you are studying. What is your unit of analysis for your research question?

For example, my unit of analysis would be each individual student that participates in a sex education curriculum, whether it is abstinence-only or safe-sex.  This is because they are the unit of observation – I am observing whether they become pregnant when they are 19 years old, or earlier.

Are there other potential causes that could be impacting your social phenomena? This does not mean that your independent variable is not valid, it just demonstrates the complexity of social phenomena.

For example, the following may also be related to teen pregnancy:  socioeconomic status, poor parental supervision, poor self-esteem, victim of sexual abuse, or substance abuse could all be contributing factors.

Is your research more appropriate for a cross-sectional study or longitudinal study?

For example, my study is going to be a longitudinal study because I will be following students from when they take sex education until they complete their 19th year of age.

In your experiment or study, who will be your experimental group (receives your proposed intervention) and who will be your control group (does not receive your proposed intervention)? How will you obtain these groups?

For example, in a state that prescribes abstinence-only sex education curriculum, I will ask for schools that would be willing to pilot the safe-sex curriculum, this group will be the experimental group.  The schools that continue to use the abstinence-only curriculum will be the control group.  An equal number of students for each group will be sought.

The problem of internal validity refers to the possibility that the conclusions drawn from the experimental results may not accurately reflect what has gone on in the experiment itself. The threat of internal invalidity is present whenever anything other than the experimental stimulus can affect the dependent variable.  The text discusses history, maturation, testing, instrumentation, statistical regression, selection biases, experimental mortality, and demoralization.   Which ones of these are potential problems for your study?

For example, in my study maturation may be a potential problem because the individuals continually grow and change and may grow wiser because it is a longitudinal study, which may impact the outcome.  Selection biases could also have an impact, the populations could be different in rural versus urban school settings.  Experimental mortality because participants may drop out of the study or move to a different school district.