Research Design

1
Research Design
Andrew Austin 2009
2
Three Purposes of Research Design Exploration Fe
asibility and pilot studies Description U.S.
Census Uniform Crime Report (UCR) Explanation
3

• Units of Analysis (UA)
• UA may be
• The individual (most common) gender, age,
  attitudes, etc.
• Group (family, peers) size, structure,
  composition, etc.
• Organization (corporation, university)
• Category (class, gender, race)
• Artifacts--any product of social beings or their
  behaviors
• (music, art, architecture, myths)
• Relations (class, race, interactions)
• Institution (religion, education, family)
• Society (nation, regions)
• UA may be either
• Descriptivee.g. the population of individuals
• Explanatorye.g. the dynamics within that
  population

Unit of analysisthe entity analyzed in the
study artifacts, groups, individuals, and
organizations. Unit of observationthe unit from
where data is collected, such as an individual in
a household.
4
Faulty Reasoning Reductionismreducing complex
reality to a simple explanation. Examples Human
natureall human behavior is to be explained in
terms of seeking pleasure and avoiding
pain. Methodological individualismsocial
behavior is to be explained in terms of
individual psychological principles. The
Ecological Fallacyoccurs because of confusion
about units of analysis. Making assertions about
individuals as the unit of analysis based on the
examination of groups or other aggregates. That
is, generalizing from the group to the
individual.
5
Clifford Shaw
Social Disorganization Theory
6
(No Transcript)
7
In their social disorganization theory, Shaw
and McKay predicted that persons living in a
particular neighborhood around groups associated
with high crime rate would be more likely to
become delinquent. Although aggregate studies of
delinquency explain variation in delinquency
rates in the social unit studied, they say
nothing about individual behavior or involvement.
8
The Question of Time Cross-sectionalobservation
of a sample (cross section) of a population or
phenomenon that is made at one point in time. For
example, U.S. Census. A snapshot. Problems
Difficult to determine causation (time
order) The peril of retrospective causal
analysis. Longitudinala study that extends over
time, using a group of individuals or samples
from a population. Stresses the importance of
analysis of within-subject changes. Prospectivebe
gin with young individuals and follow them
through time. (Desirable but expensive) Retrospect
ivebegin with adults and trace their histories.
(Problematic but inexpensive)
9
Types of Longitudinal Studies Trend
Studiessuccessive cross-sectional studies. For
example, comparing the annual UCR reports for
changes in official rates of criminal
offending. Cohort studiesfollowing a
subpopulation through time. For example, 15-20
year olds in 1960, 35-40 year olds in 1980, 55-60
year olds in 2000. Panel studiesfollowing the
same set of people through time. For example,
selecting 500 juvenile delinquents and 500
non-delinquents and then reassessing the two
groups every 5-10 years to determine involvement
in adult criminal activity.
What are the benefits of longitudinal
studies? For example, in the study of criminal
offending one can clarify notion of careers and
help resolve criminal specialization controversy.
determine the incidence and prevalence of
delinquency at various ages. understand age-crime
relation, e.g., relation between JD and adult
crime. forecast onset, recidivism, and
desistance. evaluate treatment or rehabilitation
regimes. determine the effects of social bonds,
such as marriage and work. determine the degree
of intergenerational transmission of crime.
10
The Standard Research Process