| [BACK]
|
490 Project Design
All investigations, whether in a physics lab or a criminal mystery novel, have discrete components arranged into a purposeful structure. So, too, must your academic research project. There are many options for project design, however. What follows is a sampler of some common designs that may inspire your own project, but you are free to employ or devise other research structures also (subject to instructor approval). Inductive vs. DeductiveThese are two opposing, but not mutually exclusive, approaches to investigation. Inductive research begins with specific observations (usually many of them), and through various rounds of reorganization, ultimately extracts a general conclusion. Deductive reasoning works in opposite fashion; a starting generalization receives support or refutation through numerous specific observations. Many research projects go through alternating phases of induction and deduction. Experimental vs. EmpericalExperimental design, often called "Scientific Method", examines differences occur between manipulated subjects and unmanipulated subjects, in which the researcher makes the manipulations. Empirical design, by contrast, attempts to identify order and patterns within observations of unmanipulated subjects. Several variants of these general designs are common. Case-Control Method is an experiment designed to concurrently run the same test procedure on both manipulated and unmanipulated subjects. Ideally, all conditions other than the manipulation are equal for both groups. The comparison of results reveals any differences that the manipulation causes. An example is when you re-inflate two flat tires, one of which you have treated with a special sealant, to test the idea ("hypothesis") that sealed tires hold air better. Cross-sectional Method is sometimes called the "snapshot" approach, where many unmanipulated subjects are instantaneously recorded at once. From the results general patterns may emerge about the subject population at the time of observation, revealing the current state of affairs. An example is the electronic survey that you completed for this course, which reveals to me demographic and academic patterns among you and your peers right now; these generalizations probably differ from what they once were or someday will be. Longitudinal Method is sometimes the next step from cross-sectional studies; either repeated "snapshots" or continuous observations through time reveal trends of change (or consistency). From the measured trends we may cautiously extrapolate where the condition may be in the future, or infer where it was at some time in the past. There are many fine projects that have utilized multiple "snapshots" of the same subjects over long periods of time. Simulation or Scenario MethodsThese designs evaluate "observations" that never truly occurred, but when conducted properly they are not falsehoods without value. They assume many guises, but among the more common uses of this technique are computer modeling and the conduct of drill exercises. Both require very careful observation protocols, both can follow any of the inductive, deductive, experimental, or emperical designs, and both usually require some of comparison to real events for validation purposes. An example of this method would be a computed projection--or drill--that suggests the extent and anticipated dollar-value damage from a hypothetical river flood in an urban area. Results might be useful in guiding future city planning. DATA ORIGINWe distinguish between primary data (that which you have collected first-hand, and which exists nowhere else) and secondary data (that which someone else compiled, but which you are using in an original analysis). Both are valid for research, although you must acknowledge secondary data sources in your project. You are free to use either for your research. POSING THE QUESTION...Make no mistake about it, this is the single toughest, but also most essential, aspect of research. You will probably spend a substantial portion of the first two weeks in this course agonizing over the precise wording and meaning of your research question. Essentially, it comes down to devising the specific question your research will address, and how you shall pose it in some test-able fashion. It is critical because it will constitute the core of your statement of the problem, upon which all other attributes of your work will revolve. Most initial research questions suffer from being too broad. I will drive you crazy by telling you repeatedly to "narrow it down" because of that. If it consoles you, the same problem also occurs for me--and most other professional researchers. "Hypothesis: a supposition; a proposition or principle which is supposed or taken for granted, in order to draw a conclusion or inference for proof of the point in question; something not proved, but assumed for the purpose of argument." -- Webster's New Universal Unabridged Dictionary One approach to formalize, and narrow, your research question is to pose it as an hypothesis. In this you create a statement of the condition you would like to test (Ha), and also an opposing "null hypothesis" (H0). An example is:
If, upon observation, the dog has fleas, then you have confirmed the hypothesis; if the dog does not have any fleas, then you have refuted the hypothesis. "Theory: a formulation of apparent relationships or underlying principles of certain observed phenomena which has been verified to some degree." -- Webster's New Universal Unabridged Dictionary In and of themselves most hypotheses are relatively worthless, because they only test very specific single questions. They become much more valuable, however, when multiple hypotheses test related issues that support or reject a general idea; theories usually encompass multiple related principles. One refutation, and the general idea is invalid (or must have retooling). We call general explanatory frameworks of this sort "theories". To quote Nobel physicist Louis Alvarez, "You can never prove a theory. You can only disprove them, which happens whenever you refute any hypothesis related to them." While you may or may not undermine a theory in the course of your research, you MUST at least present those theories relevant to your project, AND offer commentary about how well your findings fit into them. "Law: a sequence of events in nature or in human activities that has been observed to occur with unvarying uniformity under the same conditions." -- Webster's New Universal Unabridged Dictionary A step upward in rigor from theory is a "law". Theories may be adjustable with new evidence and testing, but laws are immutable--they must apply everywhere, everytime; no exceptions. They usually are much more specific than comprehensive theories. You probably will not articulate a new law, but if you find challenge to a theory you will likely have challenged a law also. Laws are very hard to find, and even harder to keep; that is why so few receive recognition. Even one exception, at any level, negates the entire law. That is why you have probably heard of a few Laws of physics, but you will never hear about any "Law of Evolution" or "Law of Plate Tectonics". Think very carefully about your research design. Try to assess its viability from many standpoints: data quality and availability, process time and facilities, subject numbers and representativeness, etc. Remember, You
are visitor # N. C. Heywood maintains this page, last updated 05JAN02. |