Abstract
There are many Computer Assisted Qualitative Data AnalysiS (CAQDAS) packages. Most are powerful tools supporting the range of qualitative analytic techniques. QDA Miner (http://www.provalisresearch.com/) is one of a suite designed by Provalis Research for textual and image analysis. In its own right QDA Miner is a sophisticated package for conducting robust code-based analysis; when integrated with WordStat and SimStat, possibilities for conducting mixed analyses of mixed data are greatly extended. Provalis Research’s tools in most respects equal others’ in supporting code-based qualitative analysis but outstrip those for quantitative, mixed, and statistical analysis. This review highlights QDA Miner’s distinguishing features and summarizes functions of its concurrent use with WordStat and SimStat that make it a particularly good choice for mixed methods analysis.
Inherent to QDA Miner’s developmental philosophy is the facilitation of analysis from different epistemological standpoints and the provision of tools that visualize coding patterns and trends, explore relationships in coding applications, and test hypotheses without the need to export to a separate statistical program. Sharing the architectural format of WordStat (text mining) and SimStat (statistical analysis), large mixed corpora can be stored within one file. Thus, text mining, quantitative content, or statistical analytic techniques can be conducted concurrently—stretching the boundaries of what is possible in mixed methods.
Tools for undertaking user-driven qualitative coding are well-developed in QDA Miner and equivalent to those provided by other CAQDAS packages. Alongside are hyperlinking, memo-ing, and querying tools. Particularly distinctive is the “geo-tagging” facility that allows data to be linked to geographical locations and corresponding temporal dimensions. QDA Miner particularly stands out, however, with respect to assistance with coding large volumes of textual material, the range of retrieval options available without recourse to manual creation of complex queries, and the creation and visualization of joint qualitative/quantitative displays.
QDA Miner’s text-based search tools providing coding assistance include keyword in context (KWIC) searching, section retrieval, query by example searching, and a cluster extraction and coding tool. Utilizing artificial intelligence technology, these tools increase the speed and reliability with which textual data can be identified, thus contributing to ensuring coding work is consistent, equivalent, and thorough. In addition, its interrater reliability tool enables contributions of multiple coders’ to be assessed. The user remains in control of coding work; suggestions made by the software can be rejected as irrelevant to the context of a specific requirement, and the tool “learns” from such assessments and refines its suggestions as a result. This ensures that the benefits of computer assistance are, rightly, mediated by the researcher, who remains ultimately in control of coding work.
Where WordStat is used in combination with QDA Miner, the possibilities for conducting quantitative content analysis and text mining are significantly extended, because WordStat includes additional tools for identifying textual material for coding. It comes with standardized dictionaries or users can create their own; composed of words, word patterns, phrases, and proximity rules (e.g., NEAR, AFTER, BEFORE) to generate concepts and categorization dictionaries relevant to the specific field of study. Textual data extracted via WordStat can subsequently be coded via seamless integration with QDA Miner.
QDA Miner includes many tools for retrieval based on earlier coding work. They are easily accessible and maniputable without recourse to complex query dialogs. As well as the standard retrieval tools routinely provided by CAQDAS packages (e.g., in-context retrieval, code co-occurrences, code frequency information), QDA Miner includes tools for identifying coding sequences and assessing relationships between coding and numerical or categorical properties. Retrieval is usually displayed in tabular format, and various options exist for varying the display according to alternative measurement criteria. QDA Miner’s integrated statistical and visualization tools offer additional opportunities for displaying, exploring, and interpreting results alternatively. These include clustering, multidimensional scaling, heatmaps, correspondence analysis, and sequence analysis. Such tools offer means of displaying qualitative and quantitative data and the results of interrogations jointly, allowing, for example, identification of patterns, trends and anomalies; comparison across cases and according to the presence of variables; the testing of hypotheses; and the building of theory.
Many of the retrieval and visualization tools outlined here provide innovative mixed methods analytic possibilities. These can be vastly extended, however, by the concurrent use of SimStat. In its own right a powerful analytic tool, equal to its more well-known rivals, SimStat provides a number of unique output features. When used in combination with QDA Miner and/or WordStat, however, options for conducting various types of mixed analyses on qualitative or mixed data are considerable. Having numerical, categorical, and textual data contained within the same data file and analyzable concurrently, enables possibilities of exploring relationships between variables of different types (e.g., numeric and categorical) and comparisons of qualitative codings or content-based categories within and between complex subsets of data, individuals, groups, or settings. These options are unique among the range of packages designed for qualitative and mixed methods analysis.
QDA Miner, WordStat, and SimStat offer in their own right sophisticated solutions for qualitative analysis, quantitative content analysis/text mining, and statistical analysis, respectively. Used together they offer unparalleled options for mixed methods analyses; opening up doors for handling heterogeneous data in native formats, without the need for manual conversion or the use of multiple propriety products.
Their use may not, however, be suitable in all circumstances; individual products have their pros and cons and other CAQDAS packages also offer sophisticated options. 1 However, in using Provalis Research’s products, researchers undertaking mixed methods analyses can pick and choose from an array of tools, to suit the particular needs of the research design and analytic strategy. Those engaged in customary forms of qualitative data analysis may not require the types or range of more quantitatively oriented tools. However, researchers undertaking mixed methods studies seeking to integrate qualitative and quantitative analytic strategies in complex ways are well advised to consider Provalis Research’s tools.