There is a standard format for all research reports, whether they be of the natural or social sciences. The ultimate goal is to test hypotheses, the predictions derived from one's theory and built upon the findings of others.
The structure of the research paper reflects the profound relationship between theory and fact. Facts do not speak for themselves. As Marvin Harris (Cultural Materialism 1979:7) observed, "facts are always unreliable without theories that guide their collection and that distinguish between superficial and significant appearances." On the other hand, theories without facts are meaningless. The premise of science (and what distinguishes it from dogma and armchair philosophizing) is the authority of experiment and observation over reason, ideology, and intuition.
Also reflected in a research paper is the cumulative nature of scientific knowledge. In the "Review of Literature" section one reports on the findings of others relevant to one's own inquiry. Your enterprise may be simply attempting to replicate others' findings using a different population at a different point in time. Or you may be addressing contradictory predictions derived from two different theories, or noting inconsistencies in findings of different researchers. In the "Discussion" section you will be communicating to future investigators: what did you find and where does one now go?
Finally, it's worth stressing that the evaluation of your paper will never be determined by whether or not your hypotheses are verified. It is important to remember that a hypothesis supported by the data does not mean that it is true as there conceivably is an infinite number of other theories that lead to the same prediction. Similarly, failure of support does not necessarily mean that your hypothesis is wrong: it may be hold true in some populations, you may have incorrectly measured your theory's concepts, your sampling may be flawed, etc. Philosopher Karl Popper, in fact, argues that science is not a method for verifying hypotheses. Instead, all that science can logically lead to is the falsification of hypotheses. In sum, negative results can be every bit as important as positive ones.
We expect clear and polished prose, not something that reads like some last-minute late Sunday night effort. In this era of word processing there is little excuse for misspellings. The outline of your papers should be organized in the following way (Here examples are given for a research question involving the effects of religiosity on Americans' attitudes toward euthanasia):
Describe what precisely you intend to show/argue and why (i.e., address the ever-lurking "So What?" question). Is your research problem addressing a significant social problem, or is it testing some theoretical hypothesis, such as the Marxist argument that high television viewing levels make people feel apolitical and powerless?
The success of any "science," whether it be natural or social, depends on asking the proverbial "right" question. What distinguishes good questions from bad? In part, good questions advance knowledge about significant issues, issues that are timely (e.g., why the growing homicide rates of American adolescents), that advance our ability to predict future events, that test theoretical hypotheses or resolve contradictory theoretical predictions. And what constitutes a good sociological question? First, are important issues even raised? Obviously why violent gangs appear in the poorer parts of some cities and not others is more important than whether blue or brown eyed children are more likely to prefer playing with a yo-yo. The issues raised ideally are timely, relevant to the problems or trends of the present time, and have broad applicability. Good questions are those allowing theories to be tested or, as when two theories make opposing predictions, be compared. Finally, and perhaps most importantly, a good sociological question is value-free. If, for instance, you are analyzing Americans' attitudes toward government intervention in family life (such as the government intervening when Christian Science parents refuse to give their children life- saving medical treatment), the question is not "Are there circumstances demanding government intervention?" This is a question for lawyers and political philosophers. Instead, a more appropriate question is "Which social groups are most likely to endorse government government intervention when parents, because of their religious beliefs, refuse to allow their seriously ill children to receive life-saving medical treatment."
In this section you should first grab the attention and interest of your readers, and secondly introduce the problem to be studied. You may consider using a rich illustration of the phenomenon you are studying. Remember, a research paper is not an essay nor a journalistic feature story. All assertions of fact must be documented. Be careful of any generalizations that you make. And strive to be value-free in your inquiry. A social science research paper is not an editorial piece wherein one espouses one's own beliefs.
What have others found regarding your research question? From their findings, coupled with your theory, develop a logical argument that leads to the statement of your hypothesis (this is your theoretical hypothesis expressed in terms of concepts). Your hypothesis/hypotheses should be the conclusion of this "Review" section. A good hypothesis is comparative, measurable, and falsifiable.
When writing this section weave the arguments and findings of others into your own argument--in other words, don't have a review simply bulleting the findings of others. Take extreme care to avoid over-generalizations and be sure to document your statements of fact. Do not cite work that has no bearing on your argument. Be sure to define key concepts. In the case of euthanasia, for instance, it is worth distinguishing its active versus passive forms. Example:
Over the past two decades in the wake of the highly-publicized stories of Karen Ann Quinlin, Paul Murphy, Nancy Cruzan, Dr. Jack Kevorkian, the Hemlock Society, and the legalization of euthanasia in the Netherlands, researchers have increasingly investigated the determinants of Americans' attitudes toward physician-assisted suicides (or "homicides," according to the perspective of some) and individuals' "right to die." Over the past fifty years, national surveys of Americans (NORC 1947, 1972-98; Harris 1973, 1981) show consistently increasing support for active euthanasia. To what extent is this due to the supposedly increasing secularization and moral relativism of American culture (Bellah 1989)?
In their analysis of the determinants of the American death ethos, Kearl and Harris (1981) found religiosity and education to be two of the strongest predictors of attitudes toward abortion, suicide, and euthanasia. Religion preserves moral tradition, and many faiths believe that the sanctity of life should always be honored and never violated. Here we will test the proposition that the more religious individuals are the more likely they oppose physician-assisted deaths of the terminally ill. We further suspect that this relationship between religiosity and euthanasia attitudes is not uniform across the social landscape. In particular, because of the greater moral conservativism researchers (e.g., Argyle 1993; Nelson 1979) have found among those in the lower classes, we hypothesize that because of this dampening effect of social class that the influence of religiosity on euthanasia attitudes will be greater among those of the middle and upper classes than among those from the low and working classes.
Underlying every theory is the issue of causality. What exactly does it mean to say that poverty "causes" crime, that cultural materialism "causes" moral decay, or that religiosity "causes" one's euthanasia attitudes? Just because two events historically occur simultaneously does not necessarily mean that one is influencing the other.
Describe the sample employed and the variables used to test your hypothesis. One should give just enough information here so that others can replicate your procedures (and hopefully come up with the same findings and conclusions as you did).
In this section you should address anticipated criticisms regarding internal (e.g., adequacy of operationalizing your theory's concepts) and external (e.g., generalizability of findings, sampling representativeness) validity. This matter of operationalization, of transforming concepts of one's theory into measurable variables is considerably more difficult than it may initially seem. How does one, for instance, measure religiosity? Does one discover how often people attend church or pray? Must highly religious Christians have read and understand the Bible? Can we say that one who spends three hours a day "capturing" cosmic forces in crystals is religious? Once one has made these decisions, the theoretical hypothesis is reexpressed as a statistical hypothesis. Example:
To test these ideas, the results of the 1998-2002 General Social Surveys (GSS) conducted by the National Opinion Research Center (NORC) are analyzed (Davis, Smith, and Marsden, 2003). These surveys are composed of random samples of non-institutionalized, English-speaking Americans 18 years of age and older.
To gauge individuals' attitudes toward euthanasia, responses to the following question (EUTHANASIA) were considered:
When a person has a disease that cannot be cured, do you think doctors should be allowed by law to end the patient's life by some painless means if the patient and his family request it? 1=ALLOW, 2=DON'T ALLOW, 3=DON'T KNOW.
As question phrasings influence response patterns, it is often worth critiquing the questions asked and possible inferences that might be made toward them. In the euthanasia question, for instance, note that nothing is said about individuals having terminal illness--only that they have an incurable disease. Not all incurable diseases are lethal. Did this fact influence response patterns or did respondents assume the question referred to terminal diseases?
To measure religiosity, responses to the following two questions were considered:
RELIGION: What is your religious preference? Is it Protestant, Catholic, Jewish, some other religion, or no religion?
HOW RELIG?: Would you call yourself a strong (PREFERENCE NAMED ABOVE) or not a very strong (PREFERENCE NAMED ABOVE)?
From these the variable RELITEN was constructed, where 1=STRONG, 2=SOMEWHAT, 3=NOT VERY (from HOW RELIG?) and 4=NO AFFIL. (from RELIGION). With this variable, individuals may argue that people with no religious affiliation may still be "religious" perhaps even more than those church or synagogue members who reported being "not very" religious.
It is predicted that more strongly religious one is the less likely one agrees that physicians should be allowed to end the lives of incurably ill persons.
In this section you present those results that specifically address your hypotheses.
First present and discuss the marginals, that is, the percentages of respondents falling into each response category of all of your categorical variables. For example, the marginals of the dependent variable EUTHANASIA might be that two-thirds of Americans approve. This summary statistic of EUTHANASIA provides a baseline for later analyses (e.g., in Group A, only 4 out of 10 approve of euthanasia, as opposed to 9 out of 10 in Group B). Example:
Responses to variable EUTHANASIA revealed that individuals were more than twice as likely to agree as oppose to disagree. Interestingly, although numerous professional books and journal articles on the subject detail considerable ambiguities and numerous moral shades of gray, only 5 percent of the public admitted not knowing where they stand on the issue.
Table 1: Marginals of EUTHANASIA
|CATEGORIES||FREQUENCIES||PERCENT OF TOTAL|
If your variables are continuous (e.g., raw age, years of schooling, precise income), present their mean values. For example, to the question of how much television individuals watch daily, respondents in our sample reported watching a mean of 2.87 hours (median hours=2.39 hours; standard deviation=2.04).
Next one separately considers the relationship between one's dependent variable and the primary independents variable of one's hypothesis. This relationship should be presented in table or graph formats (no stapled-on computer output please--"process" your findings). Example:
Table 2: RELIGIOSITY by responses to EUTHANASIA
gamma = -.50 Pr(chi-square=0) <.001
In Table 2 it is evident that the more religious one is the less likely one supports euthanasia (Gamma = -.50). Those having no religious affiliation are about two-thirds more likely ([86%-52%]/52%) to support hastened death than are those who reported being strongly religious.
Does the degree of religiosity similarly affect the euthanasia beliefs of those of different religious faiths? In the table below we observe how the more conservative the faith the lesser the percent of strongly religious members supporting euthanasia--and the greater the difference in support for euthanasia between the faith's strongly and not strongly religious members.
% EUTHANASIA = YES
|STRONG||NOT STRONG||% DIFF|
Finally, let's examine the interactive effects of social class and religiosity upon euthanasia attitudes. Comparing the percent differences in the bottom row of Table 4 below, observe how class differences are greatest among the strongly religious--generally, the higher the class the greater the agreement. Interestingly, among those not strongly religious, the upper and lower classes are the most conservative. Finally, from the right-most column, note how differences in religiosity produce the least difference in euthanasia support among those of the upper class.
|STRONG||NOT STRONG||% DIFF|
|% DIFF UPPER-LOWER||12.7||-0.3|
Given what you have found, what is the status of your hypothesis? Can it be improved upon (i.e., perhaps the hypothesis holds only among Catholics and not Protestants)? What should other researchers look for given what you have found? The discussion section can also feature related findings not directly related to the hypotheses being tested. For instance, the relationship between religiosity and euthanasia attitudes may be found to show more dramatic flip-flops among individuals who have/have not experienced a recent death of a significant other.
Since this is too often the only part of a paper that some individuals read it is important to reiterate what you intended to discover and what, in fact, you found.
At the end of your paper include (alphabetically by first authors' last names) all materials cited in your paper. Recommended format:
Baird, Robert M., and Stuart E. Rosenbaum. 1989. Euthanasia: The Moral Issues. Buffalo, N.Y.: Prometheus Books.
Davis, James A., Tom W. Smith, and Peter V. Marsden. GENERAL SOCIAL SURVEYS, 1972-2002: [CUMULATIVE FILE] [Computer file]. 2nd ICPSR version. Chicago, IL: National Opinion Research Center [producer], 2003. Storrs, CT: Roper Center for Public Opinion Research, University of Connecticut/Ann Arbor, MI: Inter-university Consortium for Political and Social Research [distributors], 2003.
Kearl, Michael. 2004. "The Research Paper." (WWW document).
Sherlock, Richard. 1986. "Selective Nontreatment of Newborns." Pp. 147-151 in Robert F. Weir (ed.), Ethical Issues in Death and Dying. New York: Columbia University Press.
Wood, John B. 1986-87. "The Birthday-Deathday Effect: Fact or Artifact?" Omega 17(4):321-326.