1. Review -- research General types of research Descriptive (“what”)
Exploratory (find out enough to ask “why”)
Explanatory (“why”)
Unit of analysis: “object, entity or process” under study
Contains the variables being measured
Case: A single instance of a unit of analysis

2. Review - variable
Any characteristic of a unit of analysis that is not fixed, meaning it can differ or change
How were officers accidentally killed
During which period (pre- or post-Ceasefire) were Boston youths shot dead
Any concept that can be divided into subcategories or values
Only limitation is that a variable must be able to have different values, scores or levels
Must be conceptually free to change
Number of officers accidentally killed each year
How officers were killed
Mean number of Boston youth shot dead each month
Period when Boston youth were shot dead (pre- or post-Ceasefire)
Coding
Assigning a measurement to a variable

3. Review - association and causation
Association means that the values of two or more variables change together
In Boston, the number of youth shot dead appears to be associated with the study period
After invoking Ceasefire the mean number of youths slain by gunfire drops
Causation means that changes in one variable cause corresponding changes in another variable.
The causal variable is called the “independent” variable (here it’s the time period)
The effect variable is called the “dependent” variable (here it’s the mean number of monthly deaths)
So, did Ceasefire cause the reduction?

4. Research design: Non-experimental
Poverty  Crime
Collect data on the independent and dependent variables at a certain point in time
Personal observation
Surveys
Existing data
Cross-sectional
Refine conclusions by creating subgroups for the independent variable
Subgroup 1 - male; Subgroup 2 - female
Look for significant differences between subgroup scores on the dependent variable
Panel designs
Longitudinal: take repeated measurements over time
Time1 (1980) ...T2 (1990) ...T3 (2000) ...T4 (2010)
Multiple group trend: measure subset of the group on each occasion
Group T1
Group T2
Group T3

5. Non-experimental design (observational study)
Research question: do police officers take harsher legal measures if youths display a bad attitude?
Hypothesis: worse demeanor  harsher disposition
Researchers rode along with cops to observe their interactions with youths
Researchers did NOT intervene -- they let things be
Researchers coded...
Independent variable: youth demeanor
Dependent variable officer disposition
At a later time they used statistical techniques to determine if there is an association between the variables
Depending on the strength of the association they might conclude cause-and-effect

6. Non-experimental design (survey)
Research question?
Possible hypothesis?
Data source?
DV?
IV?

7. Hypothesis
Frequency & amount of alcohol consumption  Greater risk of physical & sexual assault
Independent variables
Frequency of alcohol consumption
Amount of alcohol consumption
Dependent variables
Physical assault
Sexual assault
Unit of analysis = person (1 person = 1 case)
Independent & dependent variables embedded in survey items
Effectively “coded” by respondents who filled out the surveys
Retrospective survey -- asks about past experiences
Researchers stand back -- do not adjust the independent variables
Researchers use statistical tests to estimate the relationship between the independent & dependent variables
If there is a significant association in the predicted direction, it may be interpreted as cause-and-effect

8. Non-experimental design (existing data)

9. Hypothesis
Better academic performance  Less delinquency
Independent variables
Primary
Grades in 10th. grade
Control
Social bonds
Self-control
Dependent variable
Delinquency
Unit of analysis = person (1 person = 1 case)
Source was a national survey that was repeated on three years, 1988, 1990 and 1992
Grades came from official records
Social bonds, self-control and delinquency came from surveys filled out by students
Researchers use statistical tests to estimate the relationship between the independent & dependent variables
At first there was a significant association in the predicted direction between grades and delinquency
But when independent (“control”) variables (social bonds and self-control) were included, the relationship betwee grades & delinquency disappeared

10. Issues with non-experimental designs
Poverty  Crime
Causal order: Did the change in the independent variable precede the change in the dependent variable? Poverty  Crime ?
Crime  Poverty ?
Intervening variables: Could lack of education or living in a violent area be the true cause of crime?
Poverty  poor education  crime
Spurious relationships
Chief Bratton
Lunar cycles and homicide

11. Research design: Experimental
Purposes
Eliminate other possible “causes” (e.g., education)
Assure causal order (e.g., poverty really does precede crime)
Method
Randomly assign cases to two or more groups
Pre-measure independent and dependent variables
Designate one or more groups as “experimental” and one or more as “control”
Intervene (e.g., introduce independent variable or adjust level of existing independent variable in the hypothesized direction)
Post-measure dependent variable. Any substantial difference between the experimental and control groups can be attributed to the intervention
Simple experiment
( X ) DVt1…IV….DVt2 (intervention)
( C ) DVt1…..……DVt2 (no intervention)

13. Hypothesis: SOCP will reduce recidivism
SOCP  recidivism
Independent (causal) variable: SOCP (yes/no) (categorical/nominal)
Dependent (effect) variable: recidivism (rearrest rate, continuous)
Randomly assign youths being released to either X or C
Random assignment makes them about equal overall for background factors such as age, criminal record, disciplinary history, etc.
Give X (experimental) group special intensive supervision
This is called an “intervention”
Give C (control) group regular supervision
Wait two years
Compare recidivism (rearrest) rates for both groups
Does the X group have a significantly lower rate?

14. 1973 Kansas City Patrol Experiment
Does routine patrol deter crime?
Area randomly divided into 15 beats
Five C beats (Control - same patrol as before)
Five R beats (Reactive - no random patrol)
Five P beats (Proactive - more patrol)

15. Hypothesis: Random patrol reduces crime
Patrol  crime
Independent (causal) variable: Patrol (categorical/three values: same amount of patrol as before/less/more)
Dependent (effect) variable: crime rate(continuous)
Randomly divide an area into 15 beats
Randomly assign each a different value of the independent variable (same amount of patrol as before/less/more)
Can be interpreted as ten X groups and five C groups
Five C beats: same patrol as before
Five X 1 beats: More patrol
Five X 2 beats: Less patrol
After one year compare crime rates

16. Experimental designs (cont’d)
Solomon four-group design (checks for effects of premeasure) X1 DVt1 … IV … DVt2 C1 DVt1 …………. DVt2 X2 ……….... IV … DVt2 C2 ……………… DVt2
Issues with experimental designs
Practical constraints
How can we test poverty  crime experimentally?
Ethical constraints
Should we test poverty  crime experimentally?

17. Research design: Quasi-experimental
Experiment that lacks random assignment to groups
Groups might differ in an important respect (“matching” often used to try to make up for this)
Experiment without a control group
An extraneous event might be the true cause of the change in the dependent variable
A non-experimental design that mimics an experiment
A known intervention did take place (e.g., it’s known that the level of the independent variable did change at a certain time)
Measures of the dependent variable are available for the periods before and after the intervention

18. Class assignment Research question: can harsh sentencing reduce crime?
How would you test the hypothesis that it does experimentally?
Discuss it quietly so that other teams don’t overhear
Write down a brief, non-technical answer -- you’ll get a chance to explain it in more detail in a few minutes (one slip of paper per team)
Remember you must . . .
Randomly assign cases to two or more groups
Pre-measure independent and dependent variables
Designate one or more groups as “experimental” and one or more as “control”
Intervene (e.g., introduce independent variable or adjust level of existing independent variable in the hypothesized direction)
Post-measure dependent variable. Any substantial difference between the experimental and control groups can be attributed to the intervention