September 21 Research Methods:

1
September 21 Research Methods

• Begin discussion of social science research
  methods
• Steps in scientific method
• Key concepts in research design
• Forms of data collection
• Tie data to Thursdays discussion of theory
• Ask you to draft a question youd like to ask
  your classmates (may do this on thurs).

2
Is your name on this slide?

• Stacy Clelland, Mark Davis, Kristopher Goetz,
  Ajiboye Laosebikan,
• If so, please see us after class about your group
  project.

3
Summary from last time Qs???

• Nature remains the main source of material wealth
  maintains the life-support functions of the
  ecosphere
• Rural areas comparatively rich in natural wealth
• What is natural capital and why is it important?
• Questions of natural capital
• Not all of nature is natural capital, based on
• Tech
• markets
• How are benefits distributed and who should
  decide?
• Are our models of natural capital sustainable?
• How do ALL of these questions speak to rural
  places?
• Natural resource and environmental sociology as
  the study of the interaction between individuals,
  social structure, and nature
• Multiple theoretical perspectives can be used to
  understand this interaction.

4
Switching Gears Research Methods

• We are bombarded with statistics and figures how
  do we separate the good stuff from the junk?
  What criteria?
• Polls a great example
• some show Bush with double digit lead, others
  show Kerry with almost as much of a lead
• Variation in support for Iraq war
• Being well grounded in soc sci research methods
  helps you to ask good questions about the numbers.

5
FundamentalTwo types of logic used in research

• Deductive start with known general principles,
  then see if these hold in the real world (theory
  testing)
• Inductive start with observations of the real
  world to develop general principles that explain
  these observations (theory building)
• In most research there is an interplay between
  these approaches.

6
An example the relationship between rural
residence and hunting participation

• Deductive start with the general logical
  principles
• Rural people live closer to places to hunt
  (access)
• Rural people more likely to have backgrounds
  /employment that use resources (including
  animals)
• Rural people have different attitudes toward
  nature
• Ask Do these principles also result in rural
  people being more likely to hunt? In what
  circumstances?
• Inductive start with observations of the real
  world
• rural residents are four times more likely to
  participate in licensed, recreational hunting
  than urban residents.
• Ask what are some possible explanations for
  this? How do we sort out between competing
  explanations?

7
The Scientific Method

• Define the problem
• Review the literature (general principles)
• Formulate a testable hypothesis
• Select (develop) an appropriate research design
• Collect and analyze the data
• Develop conclusions

New / changed (refined) research questions
8
Defining the Problem What is involved?

• Identify a problem, issue or question you want to
  examine
• How would you decide on a research question?
• How does value neutrality come into play in
  selecting a research question?
• The Example Understanding rural and urban
  participation in hunting

9
Defining the problemYour theoretical approach
affects the kind of questions you ask

• Functionalist
• Examples?
• Conflict
• Examples?
• Symbolic interactionist
• Examples?

10
Defining the problemYour theoretical approach
affects the kind of questions you ask

• Functionalist
• How are attitudes towards hunting passed down in
  the family?
• Conflict
• How are resources like access to land distributed
  throughout society?
• Symbolic interactionist
• What are the shared agreements about what hunting
  means?
• How are they created and communicated?

11
Reviewing the literature Its more important
than the text indicates

• Provides the background for the entire study
• Which theoretical perspectives applied
• How concepts have been operationalized in other
  studies
• Results from prior researchused to formulate
  hypotheses
• Identify past research designshow have others
  approached this type of question, avoid their
  mistakes
• What is already known? Has your question already
  been answered? (dont re-invent the wheel)

12
Important journals in rural sociology/ sociology

• Rural Sociology
• J. of Community Development Society
• American Journal of Sociology
• American Sociological Review
• Social Forces
• Plus specialty journals (according to topic) for
  this
• Environment and Behavior
• Society and Natural Resources
• Human Ecology Review
• Human Dimensions of Wildlife

13
Reviewing the literature (cont.)

• Identify theoretical approaches to hunting
  participation used in past (rural-urban)
• Socialization / culture
• Opportunity / access
• Symbolic interaction /
• Use as the basis for operational definitions of
  key concepts translate bigger concepts into
  measurable variables (not as simple as you might
  think!!!). How should we measure
• Hunting participation (?)
• Rural / urban (?)

14
Formulate a hypothesis

• What is a hypothesis?
• A testable statement about the relationship
  between two or more measurable factors or
  variables.
• Key concepts in a hypothesis
• Independent variable (a variable that causes a
  change in another variable)
• Dependent variable (a variable that is caused by
  the action of another variable)
• Control variables (a variable held constant to
  test the relative impact of an independent
  variable.
• The ability to assess cause (versus correlation)

15
Example
Independent Variable
Dependent Variable
Rural Residence
Hunting Participation
16
Example with control variable
Independent Variable
Dependent Variable
Rural Residence
Hunting Participation
Access to Land
17
Another Example with a different control variable
Independent Variable
Dependent Variable
Rural Residence
Hunting Participation
Family member hunt
18
Important characteristics of a testable
hypothesis

• Relatively simple
• Make sure the variables can be measured
• Indicate the expected direction of the
  relationship (what causes what?)
• Be clear about causation versus correlation
• causation An increase in X causes a decrease in
  Y.
• correlation A decrease in X is correlated with
  (is associated with) a decrease in Y.

19
More on Causation vs. Correlation

• Many correlations are meaningless from a causal
  standpoint!!
• Pos association between height and interest in
  PSU football.
• Pos association between incidence of forest fire
  and average age of local residents.
• (whats wrong with these)
• Sociologists seek to identify causal links
  between variablesthis is ideal, BUT difficult
  when dealing with humans
• studying the same people over time (time order is
  one way to establish causality)
• Random assignment to treatment / control groups
• Most often, sociologists have correlations but
  use theory to help identify cause

20
An example Correlation versus causation

• Relationship between mortality and income
• Correlation Poor people are more likely than
  wealthy people to die at younger ages
• So hypothesize High income decreases mortality
  risks
• (lots of theory to support this hypothesis)

21
An example Correlation versus causation (cont.)

• But, also argued that poor health or disability
  cause both earlier death and lower incomes.
• This is an alternative hypothesis.
• How do we test these (conflicting) hypothesis?
  By the way we decide to collect and analyze the
  datathe research design.

22
Research Designs Collecting and analyzing the
data

• Primary Data Collection (you do it yourself)
• Experiments
• Surveys
• Qualitative techniques observation, interviews
• Secondary data analysis (you use data someone
  else already has collected)
• Quantitative i.e., survey data
• Qualitative techniques review of secondary
  materials

23
Experimentstest hypotheses

• Ideal way to determine causal relationshipsbased
  in natural sciences model
• Use experimental and control groupspeople
  randomly assigned to groups
• A treatment is applied to the experimental group,
  but not the control group
• Both groups are examined and the outcome or
  dependent variable is measured for both
• The values of the dependent variable are compared
  across the experimental and control groups to
  test if they are different.

24
Experiments (cont.)

• Difficult to conduct experiments in social
  research
• Often not feasible cant do the manipulation
• Ethical issues potential risks to humans from
  receiving the treatment, or consequences if they
  dont receive the treatment (e.g., testing new
  drugs school-based interventions)

25
Surveystest hypotheses

• Used when experimental design is not feasible
• Allows large sample sizes
• Collect information to test multiple hypotheses
  at one time
• By sampling carefully and asking the right
  questions, differences across groups can be
  determined

26
Survey questions

• A series of questions respondents answer
• How satisfied are you with the quality of
  education you are getting at PSU? (1very
  dissatisfied, 5very satisfied)
• What was your high school GPA?
• A series of statements respondents agree or
  disagree with
• The profs at PSU assign too much reading
  (1agree, 5disagree)
• An open-ended question where respondents write a
  short answer
• What single PSU course have you found the most
  worthwhile?
• Or a longer answer
• How did you decide what to choose for your Major
  at PSU?

27
Survey methods

• Mail
• Phone
• Face to face
• Each has advantages and disadvantages

28
Potential problems with surveys

• Declining response rates
• Confusion with marketing surveys
• Respondent willingness to answer complex or
  personal questions truthfully
• Relationship between actual behavior and what
  people say they do
• Recall problems
• Social desirability bias

29
Qualitative Research

• Often used to generate hypotheses rather than
  test them (good when relatively little prior
  knowledge exists)
• Can go much deeper
• Some phenomena not easily reduced to 5 point
  scales
• Ethnography often through participant observation
• Case studies
• Participatory rapid appraisal

30
Qualitative Research

• Often used to generate hypotheses rather than
  test them (good when relatively little prior
  knowledge exists)
• Can go much deeper
• Some phenomena not easily reduced to 5 point
  scales
• Ethnography often through participant observation
• Case studies
• Participatory rapid appraisal

31
Potential issues with qualitative methods

• Generally, limited to specific locations,
  situations (harder to create general principles)
• Hard to replicate the study (time, researcher,
  research subjects vary)
• Presence of researcher changes what youre
  studying (Hawthorne effect)
• Lack of precise definitive answers not as
  scientificmore difficult to observe the
  working of the scientific method.
• Potentially harder to defend / justify results

32
Secondary data analysis

• Use existing data sources
• Surveys or interviews others have designed and
  administered (Censuses, Add Health, YBSS)
• Administrative data (from government) (Uniform
  Crime Reports, Vital Statistics, welfare data,
  school data)
• Other existing sources (newspapers, voting
  records, scientific records, books, transcripts,
  movies, commercials)

33
Potential issues with secondary data

• Validity issues do these data really get at your
  hypotheses? (cant ask things exactly the way you
  want them).
• Dont always know all the particulars about how
  the data was collected.
• May not apply to your specific population of
  interest.

34
Issues in Data Collection and Analysis

• Who do you want to find out aboutsample
  selection
• Entire populationcensus
• Random sample
• Stratified samplesubpopulation of special
  interest
• When is a sample representativeor representative
  of whom?

35
Reliability and validity

• Important for measurement of all variables
• Reliabilitythe extent to which a measure
  provides consistent results
• Validity the extent to which a measure truly
  reflects the phenomenon or concept under study

36
An illustrationof reliability and validity
Valid, not Reliable
Reliable, not Valid
Valid and Reliable
37
Class example hunting participation by urban /
rural residence

• What is the best way (research design) to test
  our hypothesis?
• How will we obtain our data?
• What is the population to which we want to make
  inference?
• If we do the study in PA, how much confidence do
  we have that similar result would occur in other
  places?
• Should we sample, if so how will we do that?

38
Class example hunting participation by urban /
rural residence

• Important variables to measure
• Operational definitions
• Dependent variable(s)
• Independent variable(s)
• Control variable(s)?
• Will we be able to determine causation or are we
  limited to correlation?

39
Survey question of your classmates

• Not open ended (have categories)
• Content
• Attitudinal
• Experiential
• Try to be as (a) specific, (b) unbiased as
  possible e.g., attitudes toward logging
• Are you a tree hugger who thinks rural families
  are unimportant?
• Do you care about the rural environment?
• I would support a policy that protected Old
  Growth forests even if it meant losing some rural
  logging and timber processing jobs.

40
Next class

• Continue examining scientific method and research
  design
• Talk about the ethics of social research
• Begin preparation of an in-class survey, so think
  about questions related to topics in this course,
  and control variables