Research questions in academic inquiry

¶ … Democrats, Independents, and Republicans on current health care options, I would conduct a survey and analyze the results using the Chi-square test. The survey would simply ask people to declare their political party (Democrat, Republican, or Independent) and probe their opinion regarding which health care option they would prefer: (a) the current system with no changes and no guaranteed coverage, or (b) a single-payer, full-coverage system funded and run by a government agency.

The Chi-square test allows researchers to see whether the data collected is significantly different from that expected by chance. It does not assume an equal distribution in all Independent Variable categories (in this case, Democrat/Republican/Independent), but instead predicts the expected number of Dependent Variable values in each IV category (Maxwell & Delaney, 2004). Since in survey research, it would be unlikely to get an equal distribution of Republicans, Democrats, and Independent voters, I should select a statistical test in which this is not a drawback.

Maxwell, S., and Delaney, H. (2004). Designing Experiments and Analyzing Data, Second Edition. Mahwah, NJ: Erlbaum.

2.

Statistical significance and meaningful significance are often mistaken for each other in scientific publications. Statistical significance is a quantitative measure of the prevalence of an effect in a sample population compared to chance. In the social sciences, statistical significance is generally described as an experimental or quasi-experimental effect that is obtained despite being only likely to occur in 5% of cases, if random sampling were used. Meaningful significance, on the other hand, is not strictly quantifiable. However, by looking at the effect sizes in experimental research one can estimate how prevalent the effect is, or how much of the studied behavior the effect might be responsible for. If the study is not replicable, the infinitesimal smallness of its p-value (and thus, the size of its statistical significance) is of no concern for meaningful interpretations of the research. Meta-analyses and longitudinal studies can provide quantitative buttresses to statistical significance of individual studies by including them in a larger context of research. If a statistically significant result in an individual study is in accord with this larger context, then its significance cam be said to be meaningful in the description of the world.

3.

Correlational studies are one of the most conservative quantitative types of research, yet they are often interpreted in the most intellectually liberal fashion. Anyone who has taken a research methods class knows by heart the dictum, "correlation is not causation," and yet the use of correlational research by the media often draws causal conclusions that are not supported by the cited studies. In short, a correlational study is one that measures only the co-occurrence of one thing with another, without postulating any underlying mechanism or connection between the two. So, if there is a high correlation between obesity and Type 2 diabetes, it is tempting to think that there must be some underlying mechanism causally connecting the two -- insulin dynamics related to abundant adipose storage, for example. However, a correlation is not causal because it is not in itself a hypothesis. It can be formulated as one, but it must be tested by a more controlled method.

The main methodological difference between correlational studies and traditional scientific studies is that correlational research uses pre-existing data, usually from large-scale surveys, instead of formulating a hypothesis first, and then developing a study that will address this specific cause-effect scenario. For example, the famous "China Study" (Campbell, 2006) is correlational research -- it points out statistically significant co-occurrences of certain behaviors (dietary content, in this case) with certain health outcomes (prevalence rates of cancer or diabetes).

Campbell, T. Colin (2006). The China Study: The Most Comprehensive Study of Nutrition Ever Conducted and the Startling Implications for Diet, Weight Loss and Long-term Health. Dallas, TX: Benbella Books.

4.

Ethnographies and case studies are two important forms of research that have been central to the social sciences and to educational science in particular. The main distinction between them is one of scale, although this is not an absolute requirement. Ethnographies are generally done on larger populations or communities, and case studies (as the name suggests) on individual cases, or people. Ethnographic methods are drawn largely from anthropology and sociology, which means that ethnographies often have detailed background and methods sections outlining the researcher's theoretical stance so that the reader can more accurately inhabit the observer's position with respect to the data. One method that spans both types of research is genealogical interviewing, which I will use as an example.

Ethnographers that use genealogical interviewing would draw broad generalizations within the community of study about the kinship relationships uncovered by their interviews. They would endeavor to sample a broad segment of the community in order to generate a "thick description" of that community with respect to the genealogical technique. They might describe the kinds of kinship relationships that their subjects listed, and the frequencies with which they were listed. By contrast, case studies are more prevalent in psychology and the medical sciences. In case studies, individuals are the unit of analysis, and the method of data collection is applied to the individual. So, in our example, genealogical interview data from a case study would be analyzed with respect to each individual in order to see how their unique family structure or genetic background produced their current position and experiences.

5. The basic "rules of thumb" for formatting references in APA style are:

For two authors in a journal article, use both last names and the year of publication in any in-text citations (e.g. Smith & Jones, 1970).

For a citation of one individual interview, or a personal communication, list the interviewee's name and "personal communication" (e.g. Kasparov, personal communication). In the bibliography, list the date of the personal communication.

For citing a single-author article in the New York Times, use the form (Author, DATE) just as for a scientific journal article. In the bibliography, list the Times as the publication in which the article appeared, along with the volume, issue, and page numbers.

For a report with no single author, the first in-text citation may be of the form (Gonzalez et al., 2011) if the number of authors is greater than four. If there are more than two authors but fewer than five, all authors' names should be listed (e.g. Gonzalez, Choi, Singh, & Hemingway, 2011) in the first in-text citation; all subsequent in-text citations may use et al.

American Psychological Association (2009). Publication Manual of the American Psychological Association, Sixth Edition. Washington, DC: APA Press.

6.

When reading a research paper, there are two sections to which you should pay special attention: the hypothesis, and the methods by which this hypothesis is tested. When you examine a study's hypothesis (usually found in the abstract, or in some cases at the end of the introduction) you should ask yourself: Have the authors provided solid theoretical grounds for this hypothesis? Does their hypothesis make fully testable predictions -- i.e. is it falsifiable? Does the hypothesis pit two truly possible outcomes against each other, or is it testing a likely outcome against a "straw man" null hypothesis?