Qualitative analysis consists of three activities: data reduction, data display and drawing conclusions, according to Matthew B. Miles and A. Michael Huberman, authors of "Qualitative Data Analysis." Data reduction refers to simplifying and summarizing raw data -- from materials like field notes and interview transcripts -- into a more manageable form. This implies subjective decisions by the researchers about what to include in the summaries or memos they prepare. This stage of analysis occurs during data collection. The process of data reduction often leads to additional questions, requiring further fieldwork and data collection.
Data display involves the organized display of reduced data from which researchers can highlight findings and draw conclusions. Common displays in qualitative research include charts and matrices, with data organized in rows and columns. Designing a data display requires more subjective decisions by the researchers, involving what data to display and how to display it. Miles and Huberman write that designing charts and matrices has data reduction implications, as the design of these display instruments help structure the data-reduction process.
Conclusion drawing involves noting regularities and themes in the data. Qualitative researchers, especially in anthropology, are interested in the regularities of life and behavior. As they analyze qualitative data, researchers look for these regularities, which they call themes, patterns and rituals. After data reduction and display, qualitative researchers draw conclusions, implying an overall examination of the reduced data being displayed, followed by a description of what the rituals and patterns reveal.
Miles and Huberman point out that these three stages of qualitative analysis -- reduction, display and conclusion drawing -- imply a cyclical process. Data reduction, for example, leads to new ideas on what to display and how. Displaying the data requires further reduction. The display enables researchers to draw preliminary conclusions, after which they may expand the chart or matrix to test that conclusion.