Background for research consists of everything a researcher knows about a topic:
o Informal-personal experiences, and everything you pick up in life
o Formal-education and training in your particular professional area
Two factors influence the choice of a research topic: the researcher's interest in the topic and the feasibility of carrying out research on the topic
Interest: Choice of a topic because the researcher is inherently interested in something. Choosing a topic of personal interest to the researcher generally leads to better quality research as the researcher is generally more invested in the topic than if it were a topic of less interest.
Feasibility: Has to do with whether the researcher will actually be able to conduct the research; resources and access to the population are of particular importance
Ask yourself: "Do I want to do research on this topic?" AND "Can I do the research on this topic?"
There are three aspects of question formulation: refining a broad topic into a specific, researchable question; the characteristics of a good research question; and sources of ideas for research questions.
Refining a Topic into a Question: narrowing down a topic into a more specific research question for which data can be collected
You've refined your topic into a researchable question when you can phrase the question in terms of the relationship between two operationally defined variables (we'll discuss operationalization further)
Characteristics of a Good Research Question: has the potential to expand our knowledge base (3 characteristics affect a research question's potential for increasing knowledge)
1) How well grounded the question is in the current knowledge base (the problem must have a basis in theory, research, or practice (we need to know what is already known so that we can judge how much it can add to the knowledge base; gives us an anchor)
2) How researchable it is (how easy it is to formulate clear operational definitions of the variables involved and clear hypotheses about the relationships between the variables)
3) Importance: the more information the answer to a research question provides, the more important it is
Theory-confirmation, refutation, comparison, merger
Practical Problems-problem definition, solution seeking, validating
Practitioners' assumptions
Prior Research-case studies, conflicting findings, overlooked variables, setting
and expanding boundaries, testing alternative explanations
Logical Analysis-analogy, looking at things backwards
Everyday Experience
Purposes of the Literature Review: (1) to provide a scientific context for the research and to validate it against the three criteria for a good research question; (2) avoid duplication of effort (if a question has been addressed in numerous ways and the answer is always similar then it might not be worth pursuing further; or, you might want to try to examine the question in a new way (new setting, new population, etc); (3) identify potential problems in conducting the research (knowing in advance the potential problems that can arise in the research can help you to avoid them)
Types of Information: (1) look for relevant theories (be sure you know all relevant theories than can explain a phenomenon); (2) look for information on what has been previously done on your research question (you want to know what has been done, what has not been done, and what still needs to be done); (3) look form information concerning methodology (can borrow from methods previously used and can make changes in your design based upon what did not work well in the past); (4) look for information on data analysis (need to know how you are going to analyze your data to be sure that you can actually answer your research question-analytic techniques must match the data that you collect)
Primary vs. Secondary Sources: Primary sources are original research reports whereas a secondary source summarizes primary sources. Do NOT do secondary citation! Go to primary sources and see what the authors actually found (even if this has been summarized somewhere else). Looking at reference lists on primary sources is extremely useful.
Where to Find Information/ Evaluating Information -read section on pages 99-104 about finding information, using library research tools, and evaluating research. Be CRITICAL when evaluating research (see, especially, Box 4-2).
Need to answer the following five questions:
(1) How will the study be conducted?
a. Choose a research strategy and a specific design within the chosen strategy
(2) What will be studied?
a. Choice of operational definition for the hypothetical constructs you're studying
(3) Where will the study be conducted?
a. Lab or field setting? Location?
(4) Who will be studied?
a. What population and what sampling technique?
(5) When will the study be conducted?
a. Time factors? Cross-sectional or longitudinal?
Each hypothesis should take two forms-a research hypothesis and a statistical hypothesis
Research Hypotheses: states an expectation about the relationship between two variables; this expectation derives from and answers the research question, and so is grounded in prior theory and research on the question
Statistical Hypotheses: transforms the research hypothesis into a statement about the expected result of a statistical test (directional); must accurately represent the research hypothesis
The conclusions about a research hypothesis are correct if and only if the statistical hypothesis is congruent with the research hypothesis. That is, the validity of all tests of theory depends on the congruence between the research hypothesis and the statistical hypothesis. The more specific the research hypothesis, the easier it is to formulate a congruent statistical hypothesis.
Should lay out the answers to each of these questions and includes an introduction and a methods section. (See handout).
What is operationalization? Operationalization occurs when we take a hypothesis, e.g. violence causes further violence, and develop a procedure, or operation, for identifying instances of the critical terms, here, violence. Our operation should give us answers to questions like:
1. How can we recognize violence?
2. What is or isn't a case of violence?
3. How will we determine if violence has increased or decreased?
Consider the hypothesis, "Watching depictions of violence on TV makes kids more violent." Trying to operationalize the critical terms here might bring us to ask:
1. What is watching TV? Need a child be paying close attention to it, or would just having it on in the background count? How do we determine how much TV a child is watching?
2. What counts as violence? Football? Mighty Mouse? A depiction of an assault? Documentary footage from a war?
3. How are we to determine if kids have become more or less violent? From their play-acting? From their actual fighting? From their arguments or threats?
In its strongest sense, operationalization occurs when we define variables so as to make
them measurable.
Operationalization has both advantages and disadvantages. One advantage is specificity. After operationalization, we should be able to determine whether there is evidence for or against a given hypothesis.
A disadvantage is that operationalization necessary involves interpretation and a narrowing down from broad, though less clear, concepts, to sharper and often less generally agreed to specifications.
It is possible to arrive at competing operationalizations of the same term and thus provoke disagreement about which is "best". The greater the specificity, the more likely a complaint that the defined variable is "too narrowly interpreted." Such disagreements are not settled by further operationalization, but by philosophical, moral, political or pragmatic argument.
a. Identify/specify the hypothesis
b. Identify/specify the variables
c. Specify the identity criteria for each variable (what components make up the particular variable; need to be mutually exclusive and exhaustive)
d. Specify a measurement procedure for each variable (how will you measure or quantify this particular variable)
e. Indicate what would count as evidence for or against the hypothesis (if your hypothesis were to be confirmed (or disconfirmed) what would you expect to find in terms of each variable)
These are the simplest form
Do not go beyond being mutually exclusive and exhaustive
Examples: Sex, ethnicity, religion, birthplace, college major
Simply the case that everyone in your sample can fit into only one category
These categories can be coded for analysis purposes (Male 1, Female 2); yet, these numbers do not refer to any interval (2 is not more of something than 1)
Consequently, little can be done, statistically, with this information beyond frequencies
These are variables with can be logically be rank ordered
Examples: social class, broad categorizations of level of education, levels of satisfaction, opinions and attitudes
Note that many social science variables have this feature
While we know that one group has more of the variable than another, we can say nothing about the amount more they have
Again, these can be coded, but a higher number does not correspond with the amount
Distance between attributes has some meaning, but they lack an absolute zero
More useful in hard sciences (Temp.) differences between points on these variables are equivalent but not in ratio format (i.e., 40-50 degrees is the same difference as 70-80 degrees BUT 80 degrees is not twice as hot as 40 degrees)
Common social science example is IQ-scores are interval-based; that is, a difference of 110 ad 120 is that same as the difference between 70-80 BUT there is no absolute zero as cannot have ZERO intelligence
We have few of these variables to work with
Those that we do have are those we construct (i.e., IQ)
These have all the characteristics of the above and have an absolute zero
Most of the variables that we use that meet interval requirements also meet ratio
Examples: Age, education, income, years of services, days of hospitalization, etc.
These are also coded and we can thus say that one 2 is not only more than 1 but is twice as large
Most of the implications occur in the data analysis phase
Thus, these needs should be anticipated in order to meet your demands later
The level of measurement you use will determine what you can say about you findings later
For example, you will be able to report a mean age for your population, but should not plan on reporting a mean religious affiliation
Mean requires interval or ratio-level variable
Nominal variables allow you to report the mode
It is important to note that some variables may be treated as different levels of measurement
That is, ratio can be treated as any lesser type, but not the other way around
Be sure to note the hierarchy of these levels
For example: age can be treated in a variety of ways if you define certain cut off points
The level of measurement for a variable that you should seek will depend on the analysis that you have planned for it
Keep in mind that some variables are inherently limited
If you are to use a variable in a variety of ways, you should seek the highest level possible or necessary
You need not measure variables at their highest level if you are sure that you have no need for the information later
But it is advisable to choose the highest because ratio can be reduced, but nominal or ordinal can not be expanded after collection