Primary research involves obtaining first-hand information about a phenomenon in the real world. This can produce either quantitative or qualitative data, and the means whereby it can be conducted are interviews, experiments, surveys, and ethnographic research. Almost all papers and theses in the sciences and social sciences center on primary research, and it is also much in demand in the business world.
When conducting primary research, the first thing you need to ask yourself is, “What do I want to find out?” This will help determine whether you should conduct quantitative or qualitative research, and the best type of research for this.
Quantitative research arrives at findings that can be expressed in numerical terms, like percentages and totals. It would be best for discovering something like “How many people in my college are in favor of abortion?” or “What sort of market could an ice-cream store on Elm Street realistically expect?” In order to work out the former, you could do a simple survey of a representative sample (more on that in a moment) of students. If you worded the survey as “Are you in favor of abortion?” and, of the 100 people who answer it, 37 say “Yes”, 52 say “No” and 11 say “Undecided”, you can – certain other factors being established – say that 37% of the people in your college are in favor of abortion.
If you wanted to work out what sort of market an ice-cream store on Elm Street could realistically expect, you could observe ice-cream stores in similar locations – Is it a busy or a quiet street? Is it located in a business, residential or leisure area? – and try to work out what percentage of passers-by purchase an ice-cream. You could then work out the average number of pedestrians on Elm Street, and you will arrive at a more or less clear idea of what the market is likely to be.
The extent of the clarity of this idea depends on a number of factors. Among the most crucial of these is the representativeness of the sample. When studying phenomena such as those indicated above, it is usually impossible, because of factors of time and expense, to take account of every person involved. To get around this, researchers sample a portion of the population under consideration. That is, they choose a group of the people (or animals, or plants, etc.) about which they want to discover something and observe trends in that group. Whether or not that group is representative, and, hence, whether or not you can generalize the findings derived from it to the larger population, depends on whether the group is large enough to accurately reflect the range and the diversity of the population as a whole. For example, if you asked two people at a given college (of, say, 5000 students) whether or not they were in favor of abortion, you could quite easily get two answers of either ‘no’. It’s obvious that this would not be substantial enough evidence on which to base a claim that 100% of the people in the college oppose abortion. If that was your entire sample, though, and you wanted a statistic, such would be finding, as a result of an insufficiently representative sample. This is the importance of sampling procedures and techniques. Data analysis also plays a crucial role, and there are numerous tests of relevance and significance you can conduct to ensure your data and findings are sound and valid.
Qualitative research arrives at findings relating to opinions, perceptions, attitudes, and so on. These can generally not be put into numerical forms, as they apply to inexact, subjective matters. Nevertheless, qualitative research is just as, if not more, important in the social sciences and business. If a bank wanted to discover what its customers thought about its service, and how it could change and improve it, or a researcher wanted to discover how teachers’ attitudes affect student’ perceptions, they would conduct a qualitative study. Such would revolve around discovering what the people involved think, and how they feel, about the issue in question, which could be discovered by way of questionnaires or interviews. Because such data is more nuanced and subtle than the sort of ‘yes/no’ answer required in, for example, the abortion study, researchers need to use more detailed methods. Likert scale questionnaires, on which respondents are asked to indicate their level of agreement with a given statement on a level from one to five, are common. Most people have seen this in restaurants or banks, on the “How was our service: 5-excellent 4-good 3-neither good nor bad 2-bad 1-awful” sort of questionnaires customers are asked to fill out. This is a good method because it enables a large sample to be assessed without the researcher needing to interview everyone involved and it provides relatively nuanced data that is focused on the issue at hand.
Interviews are more time-consuming, but provide much richer, deeper data. If a researcher was hoping to learn about NGOs’ relationship with governments in the Third World, a very good technique would be to interview directors or liaison officers involved in the field in order to get first-hand perceptions and insight. One would then go through the interviews very closely in order to establish commonalities, differences and salient insights into the phenomenon in question. The need for the researcher to interpret such data, as well as subtleties such as body language, tone of voice and implication, means that the data analysis proceeds according to quite different procedures to that of quantitative data.
This is just the beginning. There are many detailed and accessible resources online and in libraries. Here is a good place to start. Good luck with your research. If you need help with referencing, language, or making sure your project is as good as it can possibly be, get in touch with us.
