Research Methodologies: Descriptive Research

1
Research MethodologiesDescriptive Research

• JJ Sandvig
• DePaul CTI

2
Topics

• What is descriptive research?
• Types of descriptive studies
• Guidelines for interviews/questionnaires
• Population sampling

3
Descriptive Research

• Quantitative methodology
• Examines situation as it is
• Identify characteristics of observed phenomenon
• Explore possible correlations among phenomenon
• NOT to determine cause-and-effect relationships

4
Descriptive Research

• Observation studies
• Developmental designs
• Correlation studies
• Survey research

5
Observation Study

• Focus on particular aspect of behavior
• Must be quantifiable
• Maintain objectivity
• Define behavior in precise manner
• Segmented observation periods
• Rating scale to evaluate behavior
• Two or more independent raters
• Raters follow specific criteria

6
Developmental Designs

• How do particular characteristics change with
  age?
• Cross-sectional study
• Sample of different age groups
• Different environmental conditions
• Cannot compute correlation at different age
  levels
• Longitudinal study
• Single group followed over time
• Responses likely to improve because of practice

7
Correlation Study
8
Correlation Study

• Examine the relationship between multiple
  characteristics or variables
• Correlation
• When one variable increases, another variable
  increases or decreases consistently
• Regression Line predict perfect correlation

9
Correlation Study

• What can we describe?
• Homogeneity/heterogeneity of variables
• Degree of correlation between variables
• Correlation Coefficient
• Meaning and Interpretation
• Income improves with experience
• Experience can help estimate income

10
Correlation Study

• Watch out!!
• Children from lower socioeconomic groups have
  lower GPAs than children from higher
  socioeconomic groups
• Does this mean socioeconomic status affects
  academic achievement?
• What about other factors?
• Parents education levels, racial/ethnic
  discrimination, etc.
• Correlation does not indicate causation!

11
Survey Research

• A.k.a. descriptive survey, normative survey
• Learn about a large population by surveying a
  sample of that population
• Moment in time
• Extrapolate over a longer period
• Rely on self-report data
• Truth as respondent believes it
• Tailor response to what researcher wants to hear

12
Interview Guidelines

• Interviewees representative of group
• Suitable location
• Written permission
• Establish rapport
• Focus on actual, not hypothetical

13
Interview Guidelines (cont.)

• Dont put words in peoples mouths
• Record responses verbatim
• Keep reactions to yourself
• Not necessarily getting the facts!!
• Truth as they believe it
• Saying what you want to hear

14
Interview Guidelines (cont.)

• Write questions with quantifiable responses
• Consider eliciting limited qualitative info
• Pilot-test the questionnaire
• Restrict each question to single idea
• Save controversial until the end
• Seek clarifying information for vague responses

15
Questionnaires

• Facilitate both evaluation and quantification
• Checklist
• List of characteristics or behaviors
• Check items that are observed or present
• Rating scale (Likert scale)
• Evaluate attitude or behavior on a continuum
• inadequate to excellent
• never to always
• strongly disapprove to strongly approve
• Mixed opinions about neutral responses

16
Questionnaires

• Keep it short
• Keep it simple
• Precise, unambiguous language
• Check your assumptions
• how many cigarettes do you smoke each day?
• Does this assume the person is a smoker?
• Does this assume the person smokes the same
  amount every day?

17
Questionnaires (cont.)

• Do not give clues to desired response
• Check for consistency
• Determine in advance how to code responses
• Should look attractive and professional
• Every question should address research problem

18
Sampling

• Many situations where not possible to study the
  entire population
• Subset of population can be used to make
  generalizations
• Must be representative of total population
• See all characteristics of total population in
  same relationship
• Microcosm of total population

19
Probability Sampling

• Specify in advance that each segment of
  population will be represented
• Random selection
• Is a random number generator truly random?
• What is a seed? What good is it?
• Are the following strings random?
• 1-5-2-6-3-7-4
• 3-8-6-6-1-9-2
• 4-5-8-3-7-9-1

20
Probability Sampling

• Simple random sampling
• Stratified random sampling
• Proportional stratified sampling
• Cluster sampling
• Systematic sampling

21
Stratified Random Sampling

• Layers (strata) of distinctly different types
• Sample equally from each strata
• Guarantee equal representation of each strata
• Appropriate when strata equal in size

22
Proportional Stratified Sampling

• Proportions of each strata are different
• Random sample from each strata
• Proportion from each strata same as total
  population

23
Cluster Sampling

• Subdivide total population into clusters
• Clusters must be equally heterogeneous
• Choose random subset of clusters

24
Systematic Sampling

• Select individuals according to predetermined
  sequence
• Originate by chance
• ex. every 10 units

25
Non-probability Sampling

• Convenience sampling
• No identification of representative subsets
• Quota sampling
• Same proportions as total population
• Nonrandom
• Purposive sampling
• ex. typical representatives of group
• ex. representatives with diverse perspectives

26
Sample Size

• The larger the sample, the better
• Guidelines
• lt 100 survey entire population
• 500 survey 50 of population
• 1,500 survey 20 of population
• gt 5,000 survey 400
• Larger sample may be necessary for heterogeneous
  population

27
Sample Bias

• Influence disturbing randomness of sample
  population
• Sample selection
• Response rate

28
Descriptive Research

• Questions?