7400.685-080 - Research Methods in FCS
School of Family and Consumer Sciences
Spring Semesters - Tuesday Evenings 5:20-7:55pm in 209 Schrank Hall South
Instructor: David D. Witt, Ph.D.
Planning Research: Performing the Review of the Literature
Before thinking about starting data collection or statistical analysis, the researcher has to possess the knowledge that exists in the form of past theoretical thinking and research findings.  There's just no easy way around this, the most difficult part of the research process. Because the enormity of coming to own the knowledge your discipline posesses on any given topic of inquiry is intellectually challenging  and a considerable time investment, your professors will spend large segments of their careers on a specialized topic of investigation. They will research that topic for a long time before turning to another, often related topic, in the next segment of their professional lives.  

While tedious and time consuming, a full literature review on the topic of interest guarantees the researcher that past mistakes will not re-emerge.  The purpose of the detailed literature review is to help the researcher visualize all previous research steps in the cumulative march of science towards a more unified understanding of the phenomena under study.  If successful, the literature review process will allow the next research step to logically fall into place.  The Literature Review is at the top of Wallace's Wheel - at Theory.  After the Lit. Review, deductive logic will take the researcher to the next point that hypotheses can be derived.

A complete literature review would consist of just about every theoretical line of reasoning and every research finding on topic that has come before the project at hand - and this can be an enormous pile of findings.A working literature review would consist of more immediate and strategic theory/research sufficient to generate hypotheses. Over time, if a researcher was interested in becoming expert and learned on topic, the goal would be to complete and continue adding to a working literature review over the life of his or her career.

There are a couple of very useful methods for organizing the literature review that will prove invaluable in actually writing up the past in literature review form. So from the outset, while using library resources, a well designed literature review form and toughts about constructing  summary tables.of theories, methodologies and findings are in order.

The Literature Review Form - Rather than traveling to the library (or sending an assistant) with copy card in hand to photocopy every possible page that might be important, thoughtful researchers will resort to a short hand form that, once completely filled, will serve in the place of the photocopy. This method makes a smaller footprint in the ecology and a smaller dent in the budding researcher's pocketbook. More importantly, using the form instead of copying whole articles, or downloading entire articles in electronic .pdf format, will provide a quick and fluid method for organizing findings.
Student should feel free to design their own forms, but here are some details that will prove useful:

Main Concern/Theoretical Position:

Research Methodology/Analytical Strategy:

Dependent (main) Variable of Interest:

Important Independent Variables Considered:

Description  of Findings/Assertions:


Your Overall Evaluation:

By using a Literature Review Form, with one journal article per sheet, the researcher can place the forms in any array for purposes of the literature review.   The order may be chronological, or sorted by any of the detailed categories, such as Theoretical Position, Methodology used, Dependent Variables Considered, Findings, and so on.   Writing the literature review will be aided by such organization.  Remember our purpose - to adequately review the literature to the extent necessary to generate hypotheses or research questions.

Two additional library resources will make quicker work of mining the journals for findings.
First, many journals - the main journals in the field of study, will periodically publish issues devoted solely to one topic, or will publish 5 or 10 year reviews of important topics of interest. While additional library research is always advisable, summary articles usually can be counted on to review the literature up to the date of publication. For example, the Journal of Marriage and the Family has, since 1960 or so, published literature decade reviews, thus establishing the important findings on an array of topics. If your topic falls withing established research traditions, it is likely you can count on finding summary articles that will cut your historical searches down considerably.

The second library resource, especially for science and social science is the online index known as the Web of Science: Social Science Citation Index.  Since the mid to late 1960s, these citation indexes have been cataloging research on thousands of journals. Most of the journal indexes are good for finding most of the articles on any particular topic.  The citation indexes do a marvelously innovative service, however, by allowing the capability of tracking the articles that cite any particular article or author.  So, for example, if when you find an important work that is central to your topic, the citation index will allow you to see all the research your central article cites (going back in time) and all the subsequent articles your central article is cited by (going forward in time).  Thus, the investigator is now able to track an article backwards and forwards, allowing the placement of a measure of worth of any particular research effort.

Another similar approach to referencing a central article you already have is to use Google Scholar at http://scholar.google.com/.  Google Scholar will allow searching for a specific author or title or topic. The number of citations, if present, for a specific article can be clicked and revealed.

After amassing Literature Review Forms, and sorting them, Summary Tables can then be constructed to show a vast amount of past research at a glance.  Table 6 below from Prof. Susan Witt's doctoral dissertation (1995) is a clear example of an effective use of the summary table.

By first summarizing all the theoretical literature on the topic of children's socialization, and keeping the summary columns consistent by topic, she was able to construct a summary table containing about 100 years of theory on the topic. She then was able to easily write the narrative portion of her literature review from this one summary table.

Table 6 above is an example of a theoretical summary table. A similar format can be used for reviewing relevant research methodologies as well. Instead of theories, what is summarized would be methodologies used in past research. To get at that idea, we need some definitions of methdological devices.

Some Important Methodological Devices

Hypothesis or Research Question - Research has to answer some kind of inquiry, of which there are two main types.
In reviewing the literature, the researcher has all the theoretical concepts well defined. However, concepts cannot be measured because they are ideas - not tangible things that can be measured. For example, bigotry is a concept that can't be measured adequately, but discrimination is an acton that can be measured.  Thus, if a person is thought to be a bigot, one way to assess the level of his/her bigotry is to count the number of times a specific act or acts of discrimination occur over a given period of time. 

In theoretical terms, Concepts are related to Concepts. But to get down to the real, concrete world of everyday things, and actually pose an hypothesis from the literature review, concepts have to be operationalized as variables that can be measured.
Logically, if Concept A is somehow related to Concept B, then Variable A will be similarly related to Variable B.

By using first literature review forms, followed by summary tables, the researcher can construct theoretical models based on past research to generate hypotheses or research questions. In Figure 1 to the right, the author was able to construct a model of all the researched factors that mold or influence the development of a child's self-concept. The researcher was able to find measures for each of the concepts residing in the category boxes labeled Parents, Peer Relations, Media, and School toward the end of the literature review. This figure is the result, in visual terms, of a very lengthy review, and allows the researcher to "see" the hypotheses.  It is a relatively simple matter of operationalization to rephrase the graphic hypotheses into relationships between variables - for example between print media and child's self-concept.

Depending on the level of development of a theory, the researcher may find that concepts have been operationalized in past research and are ready to be placed into relational hypotheses. Otherwise, the research will be burdened with the  task of operationalizing concepts for the topic of interest. 

When researching a well documented theory, hypotheses are always preferred over less precise research questions.  However, there is still a lot FCS researchers haven't studied, leaving research questioning a viable and necessary step in the evolution of science.  More on hypotheses and research questions in a moment, but first some notes on Variables.

A Variable is an operationalized concept that is measurable in some relatively concrete way, but there are many types of variables.  A Dependent Variable (also known as the Variable of Interest) is the measure the researcher is interested in exploring.  It is so named because it is dependent upon the values of other factors.

In the example here, a researcher is interested in the dependent variable Voting Record, and the literature review shows past researchers finding the independent factors of Occupation, Education and Income to be predictive of voting behavior.

It is hypothesized here that the variable voting record is dependent on the other three variables and not the other way around.  Logic would also have it as such.  People don't usually achieve high levels of education or income because they vote a certain way, but folks with high levels of education and income tend to be more conservative in their voting behavior, as a general rule.

This doesn't exhaust the discussion of variables.  It could be that another type of variable comes between Independent and Dependent variables.  To fully flesh out the theoretical picture, past research could suggest Intervening Variables (those which underlie the relationship between dependent and independent variables) or Moderator Variables (those which have no apparent logical reason for expanding our understanding of the relationship but somehow have influence on the relationship of interest).

What's the difference?  There are those with higher incomes who vote against school funding because they have no children or grandchildren in the local school system. In the example of voting, there are very rich, fairly well-educated people who vote a liberal agenda (famous rock musicians.  These would be Intervening factors.

There are also factors that have no logical or ostensible reason for posing influence, yet do so indirectly. Measures of intelligence often vary by (but not because of) cultural grouping (i.e., ethnicity, race, or gender).  Men aren't more or less intelligent than women because of their gender, but intelligence differences between men and women have been noted in research to the extent that research in on this topic would be deficient if gender was not made part of the research plan.

In terms of the actual measurement of variables, they can be either Continuous or Discrete in nature.  A continuous variable is one that represents a continuum of graduated values from low to high with gradations between major values. Age is such a variable (I'm 56.4 years of age this month), as is time (2 hours 3 minutes 15 seconds), weight (9 pounds 6 ounces0, height (56.786 inches), and so on.   A discrete variable has a finite number of whole number values. A family can have 1, 2, 3, 4 or more children, but there's no such thing as 3.5 children.  One can be married or not, pregnant or not, a parent or not - but graduated amounts of marriage, pregnancy, or parenthood just are not possible.

An important note here is that when initiating data collection, it is always preferred to collect continuous measures whenever possible.  A continuous variable can be converted (redefined later) to a discrete one, but the reverse is not possible.

In terms of measurement of variables, there are four possibilities:  Each level of measure adds a single dimension to the measure's qualities:
-Nominal level measures simply classify a measure with no greater or lesser values (i.e., gender, race, geographical location).
-Ordinal level measures classify and allow imprecise order (i.e., values of on or off, low-medium-high, married-not married).
-Interval level measures classify and allow order, and possess equal intervals between values of the measure (i.e., on a scale of 1 to 10, temperature in degrees, units of time).
-Ratio level measures provide classification, order and equal intervals and have the additional quality of a true zero point (i.e., amounts of some unit of measure that can be completely depleted).