Planning a Research Project: Things You Need to Know

Planning a Research Project: Things You Need to Know

 

 

 

Display your skills

Start with existing research

How to generate a weak project

How to generate a great project

Find out what is known already

Plan your experiment

Impress your instructor ...

...but keep it simple

Write the report

"Ancient of Days" (William Blake) scanned by Mark Harden, at Artchive.

These notes are relevant to any course that requires students to carry out a research project. It may also be useful even if students do not actually carry out the research. For example, the first few sections apply also to course where a research proposal is required

Display your skills

The research project is an opportunity for you to demonstrate your skills as a research psychologist. The more impressive your demonstration, the better will be your grade for the project.

One way to evaluate your own project is to review all the major principles you have learned in the course and ask, "Have I demonstrated that I understand this principle". For example, one of the first things you learned (we hope) is that all good research is founded in theory. Have you made the theoretical foundation for your study clear? Another principle was, avoid confounding extraneous variables with the independent variables. What steps did you take to avoid such confounding?

Start with existing research

We don't expect undergraduates to make major contributions to the advancement of psychology. It's difficult to think of a good study that has not already been done by someone, somewhere. Nevertheless, if you start with a recently published research article, you can usually think of an interesting extension of that research.

A good way to begin is by perusing recent issues of the journals Psychological Science and Current Directions in Psychological Science. You can surely find an article there that interests you, and which suggests a simple experiment that you could do for a class project. There are other journals in specific areas such as memory, thinking, social psychology, etc., that could serve as a starting point, but the advantages to the two recommended journals are that they are cover many different topics, they are (usually) quite readable, and the articles can (usually) be counted on to represent high quality research.

How to generate a weak project

A common method for generating project ideas is the "Wouldn't it be interesting" approach. For example, wouldn't it be interesting to find out if ...

people think blondes are more attractive than brunettes.
listening to classical music can reduce stress.
people do better on a test when they're in a good mood.
memory is better when items are presented in different colors.

Well, actually, no, it would not be interesting. At least, not unless you have a good theoretical reason for asking these questions.

If you really want to pursue topics like these, find a body of research and theory from which the question emerges as a valuable next step. Then try the following suggestions.

How to generate a great project

Great projects always emerge from your reading of existing research. There is always a strong reason for addressing the research question, beyond the weak "it might be interesting..." excuse.

Once you have found a topic that interests you, there are several ways to generate an interesting project of your own. The relevance of the suggestions below will depend on the specifics of the topic - they don't all work for every topic.

Test the generalizability of a finding. You may be able to extend a research finding by showing that it applies to a different population of subjects, or that it works in a different context, or with a different task.

Look for interactions with some other variable. You may read an article showing that independent variable X has an effect on dependent variable Y. Can you think of a second independent variable that might alter this relationship?

Find a new way to test a theory. Perhaps you have read an article in which investigators test a theoretical explanation for some form of behavior. Can you think of another way to test the theory, using a different setting in which it's not clear if the theory would work or not?

Show that there's another explanation for a finding. You may read an article in which it seems to you that the authors missed a possible alternative explanation for their results. Can you design a test of your alternative explanation?

Apply a theory to the solution of a problem. Perhaps there's a practical problem that interests you - How can one alleviate test anxiety? How can one modify people's attitudes? How can one improve memory for new information? If you can find a theory that seems to address the problem, or if the problem itself is suggested by a theory, you may have a good project.

Use a theory to explain a common observation. There may be a common piece of folklore that seems to cry out for an explanation. You might want to (a) find out if, or when, common folklore is correct, and (b) test a possible theoretical explanation.

Resolve conflicting claims. You may find two or more existing studies that seem to make conflicting statements. Can you design an experiment to settle the conflict?

Compare two theories. You may sometimes come across two competing theories, and may be able to think of a setting in which the theories lead to different predictions.

Find a way to integrate two different topics. It may be hard to make an original contribution in one area, but there are many ways to combine topics in new and interesting ways. Find two recent articles that interest you, then generate a question that would relate one to the other.

Find out what is known already

Whatever your idea may be, you are obligated to find out what others have already done.

If your idea was generated by reading a recent article, then that article will contain references to some of the most relevant literature. At a minimum, you must check those references. In addition, you should know how to use data bases such as PsychInfo to search for relevant literature. Google Scholar can be a valuable resource when you begin your literature review. However, it is not comprehensive, and it should be backed up with PsychInfo.

Plan your experiment

The most important principle when planning your study is, it will take longer to get it up and running than you think it will. Don't wait.

Define your variables. The design of your experiment requires a careful operational definition of the independent and dependent variables. With regard to the former, decide whether to manipulate them between groups or within subjects (using matched subjects is pretty well ruled out by time constraints).

If you use a between-groups (independent groups) design, consider using a covariate to increase power.

Then think about manipulation checks (look it up!). With regard to the dependent variables, think about the reliability, validity, and range of variation of your measurements. The more you can use variables that have been used by other researchers, the less trouble you will have.

Remember that there must be at least one manipulated independent variable. You may include classification variables (sex, race, etc.) as well, but it is usually a mistake to use more than three independent plus classification variables combined.

You may also be tempted to measure every dependent variable you can think of. Again, this would be a mistake. Settle on the one or two that are most important.

Once you have identified your variables, you face the practical problem of how to manipulate and measure them. If you have followed the suggestions so far, your project is an extension of research reported by others. So, whenever possible, use procedures that are already well established in the literature.

Seek help and advice,but get started right away.

Run a pilot study. If, on the day of the experiment, you don't know if your procedure will work, you are inviting serious trouble.

Think about data analysis. Early in the planning stage you should think about data analysis. If you don't know what you are going to do with your data when they are collected, you have not completed your planning. Distinguish between measurements and counts, and between between-groups and within-subjects variables. This can be the key to the analysis. Look ahead to the information on analyzing data.

The other important consideration for data analysis is deciding exactly how each data point will be generated. Before you can carry out a t-test or ANOVA, your final data set must be in the form of a matrix, each row being a subject and each column being a value for a variable. Decide how you are going to fill in that matrix. In many cases, you will need to do some preliminary analysis (counting errors, calculating average response times, etc.) to generate a single data point. BCR can do much of this for you automatically.

Learn to live within the constraints. The constraints imposed by the class format may prevent you from doing something that would clearly improve your study. For example, it may be necessary to give one treatment on one day and a second treatment on a second day. This, of course, is a form of confounding.

A principle that applies in cases like this is, "Sins will be forgiven, provided they are confessed". In other words, your report should make clear that you recognize the weakness in your study, and that you had no alternative.

Impress your instructor ...

Imagine two projects, both of which were well grounded in theory, both of which used well controlled experiments, and for both of which the data analysis was complete and accurate. The first project involved a comparison of two independent groups on a single dependent variable. The second was a mixed design, with three levels of a between-subjects variable and three levels of a within-subjects variable. The first project used a simple t-test. The second used factorial ANOVAs of two dependent variables, where tests of the interactions were key to interpreting the results. Guess which project received the higher grade.

...but keep it simple

Imagine a third project in which there were several independent variables and dependent variables. The only way to test the research hypothesis turned out to involve a factor analysis of the dependent variables, followed by a multi-factor analysis of covariance. Never heard of them? That's the point. The student gave up in frustration.

Remember, the instructor will not be impressed by your attempting a project that is beyond your competence.

Two more rules of thumb: Don't waste time gathering data you will never use. But it's better to have data you never need than to forget to collect data you do need.

Write the report

If you followed the advice given in the first part of this tutorial, you have put together an important and interesting study. Now it's up to you to let the reader share your sense of involvement.

Until you write your report, you have may have been working on the project very closely with a partner. The reports, though, must be separate and independent. This does not mean you may not share ideas. By all means discuss the theoretical foundations for the study. By all means review the data analysis together, and discuss the meaning of the results. This is no different from other class projects - the ideas and data are to be shared. The reports, though, are individual products. When you begin writing, you and your partner must go your separate ways. See the notes concerning cooperation and Honor Code concerns.

Review the separate handout that contains suggestions for your report.