Empirical Strategies Sources Wohlin et al' 2000

1
Empirical StrategiesSources Wohlin et al.
2000

• Orest Pilskalns
• CS 521
• Washington State University Vancouver

2
Two types of Research Paradigms

• note subject is person taking part in the
  experiment to evaluate an object
• Quantitative quantifying a relationship or to
  compare two or more groups
• Testing the effects of a treatment (controlled
  experiment)
• Looking for cause-effect relationship
• Promotes comparison and statistical analysis
• Qualitative concerned with studying objects in
  their natural setting.
• Find out why the results from a quantitative
  investigation are as they are
• Discovering causes noticed by subjects in the
  study

3
Example

• Possible for qualitative and quantitative
  research to investigate same topic but address
  different questions
• Quantitative investigate how much a new
  inspection method decreases the number of faults
  found in test
• Qualitative find the source of variations
  between different inspection groups

4
Empirical Strategies Survey

• Survey investigation performed in retrospect
• Gather qualitative or quantitative data by
  interview or questionnaires
• Taking a sample which is representative of a
  population to be studied

5
Empirical Strategies Case Study

• Case Study use for monitoring projects,
  activities, or assignments
• Data is collected for specific purpose (tracking
  a specific attribute) use multivariate
  statistical analysis (linear regression and
  principal component analysis)
• Level of Control lower in Case Study than an
  Experiment
• Case Study observational study
• Experiment controlled study

6
Empirical Strategies Experiment

• Experiments objective is to manipulate one or
  more variables and control all the other
  variables at fixed levels, the effect of the
  manipulation is measured (can apply statistical
  analysis)
• High Level of control
• Quasi-Experiment Impossible to perform random
  assignment of the subject to the different
  treatments

7
Example

• Compare two different methods of inspection
• Statistical inference are applied with the
  purpose of showing with statistical significance
  that once method is better than the other.

8
Qualitative vs. Quantitative
9
Survey

• Used when technique or tool already has taken
  place or before it is introduced
• Aimed at developing generalized suggestions
• Select variables ahead of time to study
• Types
• Descriptive enable assertion about some
  population
• Explanatory make explanatory claims about a
  population
• Explorative pre-study to a more thorough
  investigation to assure that important issues are
  not foreseen
• Data Collection
• Questionnaires and interviews (paper, email, www)

10
Case Study

• Conducted to investigate a single entity with in
  a specific time space
• More suitable of industrial evaluation
• Samples over variables being sampled, but cannot
  control
• Harder to interpret
• Strategies
• Comparison between a new method and Baseline
• Sister projects (both should have same
  characteristics)
• When studying one project or component, randomly
  pick another to study

11
Confounding Factors

• Makes it impossible to distinguish the effect
  from two factors from each other

12
Experiments

• Control over environment and want to manipulate
  behavior directly.
• Involves more than one treatment to compare the
  outcomes
• Characteristics
• Confirm theories
• Confirm conventional wisdom
• Explore relationships
• Evaluate the accuracy of a model
• Validate Measures

13
Experiment process

• Definition
• Planning
• Operation
• Analysis
• Presentation

14
Research Strategy Factors