Fielding and Lee, Computer Analysis and Qualitative Research (1998)

· Systematic coding and sorting of qualitative data is more desirable and defensible than ad hoc collections of anecdotes

· Coding qualitative data (inclusive categories) differs in important ways from coding quantitative data (exclusive categories)

· Content analysis focused on counting differs in important ways from analysis focused on judgement and interpretation

· The goal ultimately is the creation of transparent methods and assumptions

Using Codes

· Codes are labels with theoretical properties and analytical purchase

· Codes enable an revisable comparative process in which the interrelationships between the codes and concepts become important to the analysis

· As coding proceeds, concepts are clarified, refined, solidified, modified or else they are discarded (notes about how this happens should appear in memos)

· Collection, coding & analysis of data not necessarily sequential, in fact, may be preferable in some cases to do all three simultaneously

· Successive reformulation of the research questions and codes is a gradual reductive process that seeks abstraction and generalization which is tough or impossible to undermine with the available data

· This amounts to a movement toward parsimony without parsimony as an end in and of itself

· The nature of the social world and analysis of that world is rife with uncertainty

· All theory is provisional, particularly causal analysis

Types of Codes

1. Code notes (initial categories)

2. Theory notes (emerging ideas)

3. Operational notes (what to do next or in the future)

Methods

· Given the difficulty of standardizing the qualitative research methodology, it is valid to seek to customize your method to fit your project

· Adapt your method to fit the questions you want to ask and the available data

· Use inductive logic to discover potential hypotheses

· Use deductive strategies to attempt to verify or disprove hypotheses

Three Important Steps (should be visible, explicit, transparent processes)

1. Data reduction (through summaries and codes)

2. Data display (maps, matrices, reports, flow charts, hierarchies)

3. Conclusion drawing and verification

· Always look for ways to allow others to assess the validity of your work

· Force yourself to construct and assess rival hypotheses

King, Keohane & Verba, Designing Social Inquiry (1994)

· Some very worthy goals:

· Make valid inferences under a unified logic of science

· Find common ground that qualitative and quantitative researchers can agree on

· This is a particularly worthy goal for political science, which is sharply and at times antagonistically divided along methodological lines

· A key point: employ well-established procedures for inquiry

· The ideal is unattainable: perfect experiments and research with certain conclusions will not be possible

· It is preferable to work on the best available models, data & theories to seek uncertain conclusions with carefully reported assessments of where the uncertainty lies, and to assess, as best as possible, the extent of your uncertainty

Four Basic Goals Restated

1. Seek inferences beyond the immediate data

2. Use explicit, codified and public methods

3. Reach uncertain conclusions

4. Acknowledge science is essentially methods and rules, as well as a social enterprise where criticism constitutes a contribution to the process

Research Design

· Inquiry ought to proceed within the precepts of scientific method, but also with a flexible structure

· New questions and concepts emerge within a disciplined technique, but not when the discipline is such that it stifles innovation and creativity

· The best time to struggle with creative research design problems may be prior to the field work or data collection

· At the early stage of research question formulation (discovery) the rules are much less formal, and more conducive to creative, intuitive and personal influences

· It is not that important why we choose to study something

· It is important to be able to convincingly demonstrate the results of our efforts

· Two points on the relevancy of research questions:

1. Questions should be connected to the real world

2. Questions should fit within the scholarly discourse

Building Theory or "Precise Speculation"

1. Choose theories that can be wrong and be sure to know what evidence you could find that would prove the theory is wrong

2. Falsifiable theories with as many observable implications as possible are the best bet for making scientific inferences

3. Be concrete without seeking parsimony as an end goal

4. Consider using pilot projects to test the viability of concepts and theories

Demonstrate Quality in Data Gathering

· Record and then report the process used to gather data to ensure transparency and allow outside assessment of the validity of your inferences

· Keep a paper or electronic trail in memos that describes how you moved from one step to the next in the process

· The final product or conclusions will be more compelling if they do not require the reader to make leaps of faith about how you got from point A to point Z

· Collect data on as many observable implications of a theory as possible

· Use valid measurement techniques

· Use reliable and replicable collection techniques