Examining literature or collecting data?

Now that you have thought about which type of research you are the most interested in conducting, you need to decide which type of research paper you will pursue.

There are two broad types of papers, one in which you review literature, and one in which you collect your own data. Even for a paper in which you are collecting your own data, you will still need a literature review to support your hypothesis and to justify your topic of interest, but you do not necessarily need to collect your own original data in order to investigate and write about a topic.

There are many different ways to test your thesis, some of which require oversight of an Institutional Review Board (IRB), which will be discussed later on.

Theses using literature references:

Analytical paper - Exploratory or Descriptive- You go into this paper with a broad topic in mind without a specific hypothesis that you are testing for and providing evidence in support of. To use the example in the previous post, you may be interested in writing about which types of communities have higher rates of retail spending, but you have not come to a conclusion as to what type of community you expect will have higher rates of retail spending.
(Some additional advice on writing analytical papers will be provided later on)

Argumentative paper - Causal- Start reading about the subject first to develop your hypothesis that you will provide evidence for in the literature you have read. So, to continue with the example, you might find after reading several sources, that capitalistic and individualistic communities tend to have the highest rates of retail spending. You will then argue your point and draw conclusions about the how and why of this topic using the sources you have found as well as anything else you might find after you come up with your hypothesis.
(Additional advice on writing argumentative papers will be provided later on)

Source describing what these papers are: http://owl.english.purdue.edu/workshops/hypertext/ResearchW/types.html

Case Study - Descriptive or Causal- Similar to the two above, but in this case, we're talking about a source that is written about a specific individual, group, or occurrence. These are frequently used in the social sciences, so sorry to those in the hard sciences, but these don't usually apply to you. Case studies tend to offer much more detailed information (qualitative data) as opposed to statistical or numerical information (quantitative data). So, for this type of study, you might look at one event during which retail spending drastically increased/decreased in a population after some big event took place.

Theses involving data collection:

Observations - exploratory or descriptive - There are two basic types of observations, a Natural Observation, and a Laboratory Observation. You may collect both quantitative and qualitative data. You may count the number of shopping bags each person you observe is carrying or how many times each person walks into a particular store (quantitative) as well as which store each of the shopping bags are from or what kind of shoes each person is wearing (qualitative).

• As you might have guessed, a natural observation will take place out in the field (not literally a field, it just means "out in the world"), wherever you want to observe some phenomenon. You might go to the mall, the library, or some other public place to observe people, animals, or whatever else. You do not usually need to get approval from the IRB to do this type of study, as long as where you follow the proper ethical guidelines for conducting research (for human observation, not for animal observation).

• A laboratory observation, as you also may have figured out, will take place in a controlled environment that is not in the field. You will almost always need IRB approval for this because you will be placing animal or human subjects into an artificial environment which must follow ethical guidelines. The steps for gaining IRB approval will be discussed later on.

More in-depth advice on conducting observations will be provided later on!

Experiments - causal
There are three types of experiments that you may conduct depending on certain constraints and needs.

First, some basic terminology:

• Independent variable (IV)- This is what the researcher changes in order to observe differences in the
• Dependent variable (DV)- This is what is affected by the Independent variable.

(For example: you might want to see how sleep affects test performance. You could change the amount of sleep that someone might have before a test (IV) and the score of the test would be the DV.)

• Control variable- The researcher will keep certain things the same for all treatment groups. This ensures that you do not have as many confounds in your study.
• Control group- In true experiments, the researcher will have at least one group which does not receive any treatment. This allows the researcher to measure the effect of the IV on the DV.

Non-experiment - Yeah, I know what you're thinking, this is the dumbest name for an experiment, and truthfully, as the name suggests, this isn't really an experiment. The only thing truly experimental about this type of study is the terminology used. In this case, you assume that something is the IV and that something else is the DV. You will not have a control group. You expose your subjects to the IV(one level - meaning, there aren't different degrees of the IV) and measure the DV.

Quasi-experiment - Randomization is not used, and this is what sets this type of experiment apart from a true experiment. Usually, the quasi-experimental model uses participant-matching (to be discussed later on). In this case, you may have a pre-test for your experimental group and your control group, the experimental group is exposed to the IV and the contorl group is not, and then you give both a post-test.

True experiment - Random sampling and randomization are usually employed. Subjects are selected at random, and then placed into treatment or control groups at random. There may be one or more levels of the IV in addition to the control group. This usually includes a pre-test and post-test, but may only include a post-test (and the decision of whether to include a pre-test or not will be determined based on certain constraints).

So, what's the deal? Why might you use a quasi-experiment instead of a true experiment? It all depends on the population you're looking at. Participant-matching is when you take all of the subjects in a sample group and pair them up based on their similarity on some important traits. For example, it's usually a good idea, when testing for the effect of some drug on running ability, to pair participants up based on their current health status. To not do this runs the risk that one group is healthier or has a better running ability already than the other group. Obviously, if you are doing a laboratory experiment (takes place in a controlled environment), you will have the option to do one or the other, but in a natural experiment (takes place in the field), you probably won't have the option to use a quasi-experiment.


So now, it's time to chew on this information and decide what the best option for you is! Up next, I will discuss the literature options in greater detail and offer some advice on doing those, so stay tuned!