If you don't know what you're trying to understand or measure, you can easily run in circles before you even get started. Carefully define your hypothesis or research focus well in advance of starting your investigation. For instance, you may start with the question, "How do students best learn about current events?" You quickly realize, however, that this question is too vague to research. Start with your larger topic of interest and move towards a measurable objective. For example, you may narrow the focus of this question by asking, "Does a student's retention level increase when presented with a current event in the form of a video rather than a print article?"
Once you have specifically articulated the question that you need to answer, review the work of other researchers. This helps you understand what work on the subject has already been conducted. Even if someone has already asked a similar question, you can confirm or contradict his findings. You may also want to analyze his research and decide to focus on a similar, but slightly different topic. Understanding the strengths and deficiencies of prior research can prevent you from duplicating mistakes and allow you to take similar concepts even further.
When collecting data, you may most likely do so based on a limited population; then generalize your findings to a larger population. For example, if you are researching study habits of students, you may sample several thousand students to draw conclusions about all students in a given state or country. If you are drawing conclusions about a large population of people, it's important that your sample size is large enough to lead to accurate conclusions. Surveying ten students, for example, to draw conclusions about an entire school with 1,000 students would likely lead to inaccurate data.
In the same way that sample size matters, the subjects of a study matter as well. If the individuals or objects that make up a sample size do not represent the larger population, then your conclusions can be skewed. For instance, if you were trying to compute the average weight of all of the residents in your town, but you collected all of your data from individuals walking in and out of the local gym, your sample probably does not represent the entire town. One way to avoid such mistakes is to conduct a random sample, where everyone in the population has an equal chance of being chosen for the sample.