Elements of Research: Units of Analysis, Variables, Hypotheses Research Question A question that stimulates a response in the form of a structured scientific inquiry Excludes basic assumptions of science As they are not falsifiable Excludes subjective (normative) questions Such as beliefs, values, tastes Also because they are non-falsifiable Research Question Concentrate on behavioral outcomes Must be clearly defined And highly specific Research Question Wrong What incentives promote energy conservation? Too general Better: Which economic incentives best promote residential energy conservation? Also need to specify which economic incentives are to be investigated. Units of Analysis Individual - most common level Groups - also common Organizations Social Artifacts - [products of social beings or their behavior] Units of Analysis Individual Level: Most typical unit of analysis Units of Analysis Groups: Also common Units of Analysis Organizations: Also popular Units of Analysis Social Artifacts = any product of social beings or their behavior Units of Analysis: Ecological Fallacy Keep analyses at the same level of unit of analysis Do not generalize from a more complex to a simpler unit of analysis Nor the opposite Otherwise you commit an ecological fallacy Units of Analysis: Ecological Fallacy When relationships are estimated at one level of analysis (e.g., groups) and then extrapolated to another level (e.g., individuals) distortions are likely to result. Most commonly: Falsely assuming study of group or organization- level tells you about individuals Example: assuming prejudice of an individual because group is thought to be prejudiced Units of Analysis: Ecological Fallacy Robinson's (1950) contradictory findings concerning literacy & immigration Region unit of analysis Higher rates of literacy in regions with higher percentages of immigrants Individual unit of analysis Within the same region higher rates of literacy among "native-born" individuals Units of Analysis: Ecological Fallacy Robinson's (1950) contradictory findings concerning literacy & immigration What explains the reversal of the findings? Apparently, there were dramatic regional differences in the quality of public education and the tendency of immigrants to initially settle in regions that happened to have better public education that account for the findings at the regional level of analysis. Units of Analysis: Individualistic Fallacy Also known as reductionism We cannot test hypotheses about groups when we have only individual-level data. Reductionism is the tendency to reduce complex social phenomena to a single cause. Explaining social phenomena solely in terms of individual psychological characteristics is a kind of psychological reductionism. Economic reductionism is the tendency to explain social phenomena purely in economic terms. Units of Analysis: Individualistic Fallacy Reductionism might result if inferences about groups, societies, or organizations are drawn from individuals To count the percentage of individuals who agree with particular statements on democracy and to take this as an indicator of the degree to which a political system is democratic is to commit the individualistic fallacy. A political system can have an authoritarian regime even if most citizens share democratic values. Variables While the researcher observes units of analysis in the process of research, it is relationships among characteristics of units that are of primary interest.  Characteristics of units that vary, taking on different values, categories, or attributes for different observations, are called variables.  Variables may vary over cases, over time, or over both cases and time. By definition they must vary! Variables It is not unusual to see some confusion between variables and their attributes or categories. "Gender" or "Sex" is a variable consisting of the categories "male" and "female"; but "male" and "female" by themselves are not variables. To keep this distinction clear, note that any term your would use to describe yourself (junior, PA major) or is an attribute or category of a variable (academic class, major) This same principle applies to other units of analysis to distinguish variables from attributes. Dependent Variable The variable we are trying to explain The effect Main subject of our research project It's variation is said to be "dependent" on other variables Independent Variable The variable we claim contributes to or explains the dependent variable The cause or predictor There may be more than one Control variable A variable held constant during the course of our analysis It is an independent variable, but not one of interest to our research question Often (usually) included in analyses to reduce risk of unwarranted conclusions Causal Relationships 4 formal criteria to show causation Time order Covariation Nonspuriousness Theoretical justification Formal criteria for causation Time order Cause must precede the effect Studying " grades Job tenure " job satisfaction Formal criteria for causation Covariation When the independent variable changes so does the dependent variable This may mean the independent variable is the cause of change in the dependent variable Increase in studying " increase in grade Increase in salary " increase in production Decrease in employment " increase in crime Formal criteria for causation Nonspuriousness Change in dependent variable is not caused by some 3rd unknown variable Spuriousness means the changes we observed in the dependent and independent variables are actually caused by some third variable Ice cream sales " juvenile crime ? Marijuana use " high grades? Formal criteria for causation Theoretical justification Basically, does the hypothesized causal relationship make sense? For the time being, rely primarily on Theory from the4 scholarly literature rather than so called "common sense" Fluoridated water " AIDS? Also, note that most social phenomena are complex-- meaning there will be several independent variables Hypotheses Basic statement in research This is what research tests A tentative answer to your research question Hypotheses A formal statement that proposes a relationship between two variables - a cause and a result A formal statement of the expected relationship between the independent variable(s) and the dependent variable Hypotheses Purpose? To account for changes in the dependent variable by linking causation between the independent and dependent variables Hypotheses Requirements concepts and variables must allow measurement and empirical testing Must be stated clearly A clear and precise statement of the expected relationship What change is expected & Direction of the change Hypotheses Requirements Must be specific - especially the direction Must be value free No ought to, no should, no "it is only right," no "justice demands" Hypotheses Too vague Age effects voting Economy effects voting Race effects sentencing Gender effects voting Hypotheses Better Older people tend to vote Republican or . . . Age is positively related to voting Republican A good economy causes people to vote for incumbents or . . . The state of the economy is positively related to incumbent success in elections Hypotheses Better Minority convicted persons receive harsher sentences than Anglos Women tend to vote Democratic Hypotheses Most common errors Must relate two or more variables Unclear relationship Lacking generality Value judgment Null Hypothesis Empirical research actually tests the null hypothesis This is a more stringent test Test attempts to demonstrate that the null hypothesis is true (really likely to be true) Simplest way to say the research hypothesis is untrue Null Hypothesis Examples H1 Age is negatively related to support for women's liberation. H0 Age makes no difference on support for women's liberation. Symbol H0 means null hypothesis Null Hypothesis H1 The greater the level of education the higher voter turnout What is H0 ? Null Hypothesis H1 The higher private hand-gun ownership the lower the violent crime What is H0 ? Sample Research Questions Who makes public policy in the U.S.? What causes inflation? Does bureaucracy threaten democracy? Are affirmative action programs achieving their objectives? Does school integration enhance educational attainment? Sample Research Questions & Hypotheses What factors determine urbanization? Hypotheses derived from this RQ: The degree of urbanization in a society varies directly with the dispersion (geographic)of objects of consumption. The degree of urbanization in a society varies directly with the division of labor. The degree of urbanization in a society varies directly with technological development. Sample Research Questions & Hypotheses What causes political violence? Hypotheses derived: The potential for political violence varies strongly with the intensity and scope of normative justifications for political violence among members of the group. The potential for political violence in any specific case varies strongly with the potential for group violence generally. Time Dimension Cross-sectional (most research) Longitudinal = over time Trend studies Cohort studies Panel studies Trend Study Examines change in a population over time But not the same people, group, or artifact-- each time a new sample US Census Most often a series of surveys administered over several years Cohort Study Study a specific subpopulation (cohort) over time That is, by birth year A new sample each time, but from or of the same cohort Panel Study Theoretically the best, but very expensive and hard to accomplish Study over time the exact same respondents each time Research Report Introduction - importance, research question Literature Review Subjects for study [and hypotheses] Measurement and Data Analysis & Findings Conclusions-- meaning of project findings, problem areas, suggestions for future research References Tables Figures