Primary research is research where you collect the information yourself, rather than using information collected by others. Primary research can be any form of information gathering, from searching through raw data (such as information collected on sales over the years) to some form of experiment.
If you plan to do interviews, first consider what sort of information you wish to obtain. Since interviews can be very time-consuming both to set up and to do, is there another way to get the same information?
What role will the interviewees play:-
● As experts on a management issue (who are often very comfortable with the give and take of the interview process)?
● As “potentates” who oversee an area and are of interest because of their position (who may be very comfortable if you are studying a decision process that was successful, but quite defensive and even hostile if the decision did not work out)?
● Or as ordinary people affected by a management change (and who may be quite uncomfortable with interviews both because they are not used to them and because of fear their comments may get back to their boss?
If you do decide to interview large numbers of people, you might consider a group session rather than individual interviews. Two contrasting ways of organizing groups are focus groups and nominal group technique.
● Focus groups are often used by marketing departments to develop in depth consumer attitudes on some issue. They depend on free-flowing discussion (for example, how do you use a product and what do you like and dislike about it?). One danger is that the discussion may be dominated by one individual or by a faction, so that other opinions don’t come out. At its best, the members of the group grow more confident to express their ideas as the group progresses. For this reason, focus group designers often try to make the groups fairly homogeneous, since then people may feel more comfortable in being candid. (For example, if an organizational change resulted in winners and losers, you might have separate focus groups for the winners and the losers.)
● Nominal group technique (NGT) tries to limit the pressure a group can exert against divergent opinion, while still utilizing group interaction. It starts by listing nominations, sometimes called brainstorming (for example, what possible new products can we think of). The next step is to clarify and eliminate duplicates. Finally the members are asked to individually rank the top nominations. The rankings are tabulated either in the session or afterwards.
Surveys are surprisingly difficult to do well. First, it is hard to write good questions. Next comes the difficulty of getting responses. Finally it is often hard to tell what the results mean and whether those from a sample reflect the broader population.
Professional survey designers spend a great deal of energy worrying about their target population and whether the sample selected reflects the population in an unbiased fashion.
A famous example of sample bias is contained in the 1936 Literary Digest survey that confidently predicted that Alf Landon would defeat Franklin D. Roosevelt, based on a sample size many times larger than those used in present day polls. The fallacy was that it was based on a telephone survey done at a time when many people did not own telephones. This would still not have been a problem if telephones had been randomly distributed among supporters of both political parties. Unfortunately for the Literary Digest, most people without phones were poor and much more likely to vote Democratic than their more prosperous neighbours. Having destroyed its credibility, the Literary Digest folded shortly thereafter.
Thus it is wise to consider potential bias both in the selection of a sample and in the likely response rate. If you are surveying about experience with a new management technique, are you most likely to hear from those who have had success and want to brag about that success–or conversely those who want to take the opportunity to complain about how worthless the technique proved to be?
There are several ways of administering surveys:-
● Self-administered, such as passing out a survey to you class and asking them to fill it out in class. This is a good way to get a high response rate so long as the group matches your targeted population.
● Mail questionnaire. Here the issue is getting them returned. In marketing a 5% return rate is considered good. Some people try bribes (promise to hold a drawing for a prize) or guilt (include a dollar bill). Consider whether to include a stamped envelop.
● Telephone survey. If you can get through and convince the person this is a good use of time, this can be effective. There is growing resistance to phone surveys, reflecting both increased demand on people’s time and abuses in the use of surveys. For example, pseudo-surveys, which purport to be surveys but which are really aimed at selling a product or influencing an election, are increasingly popular.
● Personal. Interviews, particularly if they follow a set script, can be a type of survey.
● Combination of techniques. Often some combination of techniques is most effective. For example, one student achieved a very high return rate by first sending out a letter describing the purpose of his research. He followed the letter with a phone call asking for a commitment to participate. Those who agreed to complete the survey were then sent the survey itself.
Obviously all these activities come at a cost, primarily in time, but also in money for postage and printing and so forth. Before undertaking a survey, it is very useful to make some estimates of the time and money cost.
Survey questions are generally divided into two classes: closed-ended and open-ended:-
● Closed ended questions allow a limited number of answers. Examples include yes/no questions (“Do you shop at store x?”), check all that apply (“Put checks by the stores where you shop”), scale (“Rate store x on a scale of 1 to 10″), and forced choice (“You have ten points. Distribute them among the following stores.”).
● Open-ended questions allow for a wide variety of responses (“What is the most important thing you look for in choosing a store?”)
Closed-ended questions are generally far easier to tabulate. Usually they are also easier to respond to and will lead to a higher return rate. Open-ended questions often work best in personal interviews.
Bad questions can lead to bad data. Here are some examples:-
● Pseudo-data, where you push people for opinions about things they have never thought about. For example, one computer magazine surveys its readers about what they think about various software and equipment that many them may have never used. Encouraging people to skip questions may help avoid pseudo-data.
● Leading questions, phrased in such a way that the right answer seems obvious. For example, the question “who are you planning to vote for in the upcoming election?” assumes the responder plans to vote.
● Loaded questions, where one answer seems far more virtuous than another. Political pollster have long been aware of the importance of how questions are phrased. For example, the question “do you support the right of parents to choose their children’s school?” tends to get much more support than “do you support the use of tax money for religious and for-profit schools?” even though both questions may refer to the same proposal.
● Double-barreled questions, combining two questions in one. These are often done in a misguided attempt to reduce the number of questions.
● Jargon, unique to a particular field, which may not be clear to the person responding.
The order in which questions are asked is also important. Start with easy questions first. Otherwise the person may get discouraged right at the beginning. Save sensitive questions (age is generally considered one) for the end. As much as possible look for logical transitions from one topic to another.
Avoid questions sensitizing the person to later questions. For example, if you are really interested in a particular management technique, it would be a mistake to first ask a series of questions about that technique and then ask a general question about which management techniques they find most useful.
Finally, it is important to pretest the survey to make sure the questions are clear and are interesting. You will probably find that questions that seemed clear to you are confusing to others. Also it is likely your questionnaire will take too long to complete.
Many students do a case study, either as the whole project or to help illustrate a thesis. Typically these case studies are based on a series of interviews. A survey may also be helpful in developing a case study.
A case study is anecdotal in that there is usually no attempt to develop any statistical significance. Rather, it assumes by going into depth about what happened in one organization will lead to understanding not only of that organization but to insight into others.
© Copyright for this article belongs to Milwaukee School of Engineering
This document was re-printed with the kind permission of Bruce Thompson. Original Source of the article is located here: http://people.msoe.edu/~thompson/thesis/primary.htm