Research Design

1
Chapter 4

• Research Design

2
Many Ways to the Same Thing

• Asking good theoretical questions is more
  difficult than answering them
• What do you want to find out?
• No single best way, but we strive to choose the
  best method (tool) for the task
• A topic isnt enough, you must be specific about
  your concepts (variables)
• Who or what are you studying?
• What is your purpose?

3
Three Main Purposes of Research

• Exploration
• Usually used to break new ground
• Description
• Accurate and precise description of events,
  persons or situations
• Answers the what, where, when and how questions
• Explanation
• The why
• These are not mutually exclusive

4
Purpose to Explore

• To go where no one has gone before
• Satisfy the researchers curiosity and desire for
  better understanding
• Rarely conclusive they test the feasibility of
  undertaking a more extensive study
• They help develop the methods to be employed in
  any subsequent study
• They help identify the variables
• Often inductive, rarely scientific (although they
  often lead to a more scientific approach)

5
Purpose to Describe

• Careful and deliberate scientific descriptive
  studies strive for precision and accuracy
• While most studies have descriptive elements
  some are purely descriptive
• Census is an example
• Demographic profile of aggregate group
  (population)

6
Purpose to Explain

• Answers questions like Why do people? and how
  do people? and when do people? etc.
• Usually more complex
• Usually more ambiguous
• Usually more experimental and scientific in design

7
Nomothetic Explanation

• To what extent (if any) do the independent
  variable(s) account for variations in the
  dependent variable(s)?
• Since it applies to aggregates, and is more
  naturally probabilistic, it is more likely to
  lead to misinterpretation and misunderstanding
• For example what caused Joe to support liberal
  views? is less difficult than what causes
  people to be more liberal?

8
Criteria for Nomothetic Causality

• Three things required to show causation
• A statistical relationship between the two
  variables
• One variable changes consistently with another
  (either direction). Alone, this is NOT enough to
  claim causation
• Time order the cause must takes place before the
  effect
• Relationships must not be spurious
• I.e. there are no other variables that can
  (partly or entirely) explain the observed
  relationship

9
Problems with Nomothetic Causality

• Example research has found that there is a
  positive relationship between the number of
  firefighters at a fire and the damage that
  results
• Therefore, firefighters are the cause of some of
  the damage?
• Does this fulfill the requirements of causal
  logic?

10
Necessary vs. Sufficient Causes

• Necessary cause something that MUST be present
  for the effect to occur
• If youre determining the causes of classroom
  disruptions, a necessary cause, but not
  sufficient, is that the subject be a student
• Sufficient cause by itself guarantees the effect
• Not necessarily the only cause
• Causes that are necessary and sufficient are the
  most satisfying outcome in research. However,
  single causes rarely exist

11
Units of Analysis

• What or whom is to be studied
• We must be careful to make sure our units of
  analysis and the aggregates we generalize about
  concur
• E.g. Gangs vs. gang members
• Units of analysis include
• Individuals
• Groups/organizations
• Social interactions
• Social artifacts

12
Units of Analysis

• Individuals
• Most common, even in studies of groups.
  Generalize about the behavior of the individual,
  not the group
• Groups and organizations
• Generalize about the behavior of the group or
  organization as single social entities
• Social artifacts
• Any product of social beings, usually a means of
  communication
• Books, movies, newspapers, ceremonies, riots,
  etc.
• Generalize about the artifact, not the people
  involved

13
Tip

• Three ways to determine unit of analysis versus
  unit of observation (who or what unit you
  sampled to collect data)
• Unit of analysis
• Determine who or what does the study intends to
  generalize about. Groups? Individuals? Social
  artifacts?
• Determine the category of the independent
  variable. Groups? Individuals? Social artifacts?
• Create a sample conclusion. Does it involve
  groups, individuals, social artifacts?

14
Example

• A researcher collects data regarding the effects
  of five different programs to determine which is
  most effective
• Unit of analysis?
• Unit of observation?

15
Example

• A researcher collects data from software
  engineers about their perceptions of what
  constitutes best practices with regard to
  program design. She wants to see which software
  products (from 20 different companies) are most
  reliable.
• Unit of analysis?
• Unit of observation?

16
Example

• A researcher collects interview data from a
  random sample of HR directors about hiring
  practices in order to determine which
  characteristics of potential job candidates were
  most often used in the selection process
• Unit of analysis?
• Unit of observation?

17
Fallacies Caused by Incorrect Identification of
Units of Analysis

• Ecological fallacy assuming something learned
  about an ecological unit (group) says something
  about the individuals in the unit
• Be careful of the reverse as well
• Reductionism reducing something to a simple
  explanation when in reality it is complex
• Distinct disciplines often filter their
  perceptions in terms of that discipline eg.
  Economic reductionism
• Likely when we search for a single cause for
  anything

18
Time and Research

• Do you collect data at one time, or over time?
• Causal studies cannot effectively use a
  snapshot data collection method they require
  multiple data collections over time
• Two broad types of studies as relates to time
• Cross-sectional (one-time, or snapshot)
• Longitudinal three types
• Trend
• Cohort
• Panel

19
Time-Related Studies

• Cross-sectional studies
• One time measurement of a sample population or
  phenomena
• Common in exploratory and descriptive studies
• Not as useful for explanatory studies (causation)
• However, time effects can be logically drawn from
  a single measurement at times
• Longitudinal studies
• Observations over an extended period
• Difficult for quantitative studies (threats)

20
Time-Related Studies

• Longitudinal studies
• Trend studies changes within a population over
  time (Census). Each measurement may be different
  individuals
• Cohort studies measures a specific subpopulation
  (e.g. U.S. males between 20-25). Each
  measurement may be different individuals
• Panel studies measures the same individuals over
  time
• Each has unique strengths and weaknesses

21
Example

• Cross-sectional studies indicate that IQ declines
  with age. Longitudinal studies claim that there
  is either no decline or less of a decline with
  age than cross-sectional studies indicate
• What might cause this difference in results?

22
Identify the Problem

• Professor Root was studying the arrest rates for
  drunken driving in urban and rural areas of Ohio.
  It was found that the arrest rate was higher in
  the rural areas. Professor Root concluded that
  people who live in rural areas are more likely to
  drive while intoxicated than are people who live
  in urban areas

23
How to Design a Research Project

• Define the purpose of your project
• Explore, describe or explain? What outcomes do
  you want?
• Specify exact meanings for the concepts you want
  to study (conceptualize)
• Example excessive stress visiting a mental
  health practitioner
• Choose a research method
• Survey, content analysis, secondary data analysis
  etc.
• Decide how to measure the results
  (operationalization)
• Example recording the number of times an
  individual visits a mental health practitioner.
  Also, the wording of questionnaires

24
How to Design a Research Project

• Decide whom or what to study
• What is the unit of analysis
• Define the population or sample specifically
• Collect empirical data
• Make your observations, distribute and collect
  your surveys etc.
• Process the data
• For qualitative or quantitative analysis
• Analyze the data
• Interpret and draw conclusions
• Report your findings
• Use a standard method, report implications,
  problems, future research