Data Collection for Dissertation & Thesis Research
When collecting dissertation or thesis data, there are numerous things to consider. First, you must develop a good idea. Dissertation ideas can come from many places. Do a thorough library search in areas that interest you. Read others' dissertations. Don't waste time researching a topic you don't have an interest in. It's hard enough finishing a dissertation you ARE interested in!
After you have the idea, develop a good question. A good question is one that can be answered by your research, oftentimes using empirical methods. For your paper, you may be interested in looking at whether angels exist. But, how would your dissertation test that? How would your dissertation measure the presence of angels?
When thinking of a dissertation or thesis question, you must also think about the research design. A testable question regarding angels is "Do people believe in angels?" You know you CAN answer this question, but you need to determine HOW to answer this question. As an example, your survey might simply ask "Do you believe in angels? Circle one: Yes No". This is not an empirical question and it's not going to give you much data, but it's one that can be answered in a dissertation.
Developing an Empirical Question
An empirical question involves the manipulation of a variable. An empirical question could be "What influences a person's belief in angels?" You might conduct research and find several potential variables that are thought to influence the belief in angels. For example, other dissertations may have found that more women than men believe in angels. You might also find a dissertation that reported Catholics believe in angels more often than atheists do. A third research might have reported that people are more likely to believe in angels if the people in their social circles do.
So, how would you design an experiment to answer the question "What factors influence a person's belief in angels?" You could collect demographic data -- such as gender, religion, and social networks -- from several subjects, but your dissertation is not really adding anything to the knowledge base. What if you asked "Does watching a news story about a person's encounter with an angel influence their belief in angels?" This is an empirical question! Combine it with a bunch of demographic information from your subjects and you may have a good dissertation.
Answering an Empirical Question with Good Data Collection
To answer this question, you could randomly assign subjects to two groups. One group could be shown a news story about a person who claims they saw an angel. The subjects in this group, the experimental group, then complete a survey. The other group, the control group, does not see the news story, but completes the same survey.
What do you include in the survey? Good data collection involves collecting relevant data that adds to the body of knowledge. Knowing that people who believe in angels also eat spaghetti is not particularly useful nor important. The main thing to remember with data collection is to keep it simple, but important. And get the data you need the first time out! There's nothing more frustrating than realizing you could answer an important question if you added just one more variable.
For this research design, you could collect demographic information thought to be associated with your dependent variable (belief in angels). Age, gender, religion, ethnicity, social network, and frequency of church attendance may be just a few. But how are you going to measure these variables? Are they dichotomous variables? Is your survey going to use a Likert scale? Obviously, you need to start thinking about your statistics! What statistic(s) are you going to use? How are you going to analyze your data? Are you looking for a relationship between two variables? Or do you want to see if one variable predicts another? When you know how you are going to analyze your data, you will know how to measure your variables.
Once you have thought of a good idea with a good, answerable question, and you have designed a experiment to answer the question, and you know what you are going to do with your data once you have it (Whew!), then and only then do you begin to collect your data!
One more important point. In any research involving human subjects, research approval from a board is almost always required. This is an involved process and must be completed prior to collecting any data. Talk to your committee chair throughout your process, so that he/she can advise you about the ethical implications of your research. Be clear about your question, the theory behind your research, your experimental design, and the statistics you will use in analyses. Keep clear, concise, accurate records and your data collection should go smoothly.