MAIN POINTS

All research begins with a problem, and such problems must be empirically grounded and clearly and specifically articulated.

Units of analysis are entities to which scientific concepts pertain and which influence subsequent research design, data collection, and data analysis decisions. Units of observation, on the other hand, refer to the unit that is afforded by a particular data set.

The ecological fallacy is a kind of distortion that occurs when relationships are estimated at one level of analysis and then extrapolated to another. The individualistic or reductionist fallacy occurs when scientists draw conclusions about groups, societies, or nations based on the observation of individuals.

Research problems are conveyed with sets of concepts, which are abstractions representing empirical phenomena. In order to move from the conceptual to the empirical level, concepts are converted into variables.

The variable that the researcher wishes to explain is viewed as the dependent or criterion variable. The variable expected to explain change in the dependent variable is referred to as the independent or predictor variable.

Control variables are used to determine whether an observed association between independent and dependent variables is a spurious relationship (one that is explained by other variables). A variable is continuous if it does not have a minimum size unit. Discrete variables do have a minimum size unit.

A continuous variable does not have a minimum sized unit by definition. Length and time are examples of these. An object can be measured at 10 feet, at 10 feet and 3 inches, at 10.251 feet, or an even smaller denomination, and all answers would be correct. Any divisions of length or time are imposed using tools we developed. Discrete variables do have a minimum sized unit, and examples of this include monetary currency (the penny being the smallest denomination) and the total number of people in a household, community, etc. (a single person being the smallest denomination).

A relation in research always means a relation between two or more variables. When two or more variables are related, changes in one variable are systematically related to changes in another. This link is known as covariation. The direction of this link refers to relations between variables being either positive or negative. A positive relation means that as values of one variable increase, values of the other also increase. A negative relation indicates that as values of one variable increase, values of the other decrease. Relations between variables are characterized not only by direction, but also by magnitude. The magnitude of a relation is the extent to which variables covary positively or negatively. The highest magnitude of relation is a perfect relation, in which knowledge of the value of one or more independent variables determines exactly the value of the dependent variable. At the other extreme is the lowest magnitude of relation, the zero relation, where no systematic covariation between the values of an independent variable and a dependent variable can be discerned. If a relation has no magnitude, it also has no direction.

Research hypotheses are tentative answers to research problems; they are tentative answers because they can be verified only after they have been tested empirically. Research hypotheses share four common characteristics; they are clear, value-free, specific, and amenable to empirical testing.

Katherine Mason's study of income disparities between severely obese and non-severely obese individuals shows a series of hypotheses developed for each potential conclusion, which consist of these: 1) there will be no income disparities between severely obese and non-severely obese people after education and intelligence are controlled. If so, this finding would indicate the presence of meritocratic discrimination, 2) income disparities will persist after other factors like education and intelligence are controlled, but will narrow as severely obese individuals gain experience, indicating statistical discrimination, 3) income disparities between severely obese and non-severely obese individuals will again persist, but the gap will not narrow with experience, indicating prejudicial discrimination, and 4) relative to non-severely obese women and men, severely obese women and men experience different types of discrimination, with anti-obesity discrimination more damaging to women's incomes than to men's incomes.