Back at it.... :-(

Ch4 Summary:

3 goals of research:

1. Exploration - familiarization, focus groups. Shortcomings: seldom satisfactory answers, not representative. 3 reasons for use:

   • curiosity

   • pre-study for larger study

   • dev methods for larger study

2. Description - describe accurately and precisely chars of a population (Ex: US Census)

3. Explanation - Answer Why? (Ex: ID vars to explain why City A higher crime rate than City B)

   • Nomothetic Explanation - few factors that lead to 'most' changes in results

   • Idiographic Explanation - all the reasons, all the time

Criteria for Nomothetic Causality; 3:

1. Variables correlated (established relationship)

2. cause before effect (time order)

3. vars non spurious (no 3rd var)

Note what is NOT an interest of Nomothetic causality:

• Complete explanation of causes,

• Exceptions don't disprove the rule

• Rel can be true even if it only applies to minority of cases

Defs:
Necessary Cause - Must be present for result (take classes to get degree)
Sufficient Cause - Condition that guarantees effect, not the only way to get effect (take right classes, get good GPA, get degree)

Idiographic causes are sufficient but NOT necessary. (Anyone with your details of life would have attended college, however, other people attend college through different means).


Units of Analysis

Units of analysis (thing seeking to describe) are generally also units of observation, individuals being most typical. Assertions about one unit of analysis need to be based on exam of that same unit of analysis.

Ecological Fallacy - applying findings from group to individuals (Ex: crime rates in large cities, large African American pop, can't say that the AA are making the crime.)

• Individuals

• Groups

• Organizations

• Social Artifacts (are you studying marriage or marriage partners?)

Individualistic fallacy - probabilistic statements not invalidated by individual exceptions

Reductionism - reducing complex phenomena to simple explanation (Ex: answering What caused the Am revolution? - with a single factor).


Other types of Studies/Time-based:

Cross-Sectional Study - observation of sample at _one_ point in time (single US Census)
Longitudinal Study - sampling over time

• Trend Study - re sampling of same population more than once, over time

• Cohort Study - following an age group and re sampling that age group, over time

• Panel Study - follow the same group of people over time (special problems with panel attrition and possibility that dropouts will be atypical)

Triangulation - use of several different research methods to test same finding