Location: Glasgow Caledonian University
Facilitators: Noori Akhtar-Danesh (McMaster University), Maximilian Held (University of Hohenheim), Byung S. Lee (Elon University)
The choice and mastery of Q-specific software can be an important factor for a successful Q study.
Q methodology includes some mathematic operations not easily accomplished with generic statistical software, and Q practitioners require custom interfaces to guide their analyses and interpretation.
Happily, there has recently been a lot of new software released to the Q community, often generously as free and open source code.
Three of those new tools will be featured in this workshop:
QPackage, a user-written Q analysis package for the commercial Stata software, developed by Noori Akhtar-Danesh at McMaster University (Hamilton, ON, Canada).
pensieve, a Q analysis package for the free and open source R Project for Statistical Computing, developed by Maximilian Held at the Department of Sociology at the University of Hohenheim (Stuttgart, Germany).
PQMethod (Julia wrapper), a tool that calls the Fortran-based PQMethod from within Julia, offering easier access. Developed by Byung S. Lee at Elon University (Elon, NC, United States).
This workshop is for those relative newcomers, but also for experienced Q methodologists, who are curious about new tools and developments or are looking for an introduction to new software.
We will proceed in a single–track fashion, so all attendees can learn about all three software packages.
Overview of Other Existing Q Software
|10:15 am||QPackage (Stata)||Noori Akhtar-Danesh|
|11:45 am||Accio / Pensieve (R Project for Statistical Computing)||Max Held|
|1:00 pm||Lunch Break|
|1:45 pm||PQMethod (Julia wrapper)||Byung S. Lee|
|3:00 pm||End of Workshop|
|3:00 pm||1-on-1 Support||(by arrangement)|
|4:30 pm||End of Session|
During each session, the developer will introduce his program, demonstrate its features and each participants how to obtain, install and use the program.
In addition, developers may offer one-on-one support to assist participants with their setup and analysis. To receive such individual support, contact the developer directly to arrange for a time during the conference.
QPackage (Stata Programs)
There are only a limited number of programs that offer Q-methodology analysis. In this workshop, I will introduce two new programs, qconvert and qfactor written in Stata.
qconvert reads statement numbers from each Q-sort into Stata and replaces statements’ numbers with their rankings.
qfactor is the main program. It performs factor analysis on Q-sorts. It extracts factors based on principal axis factor, iterated principal factor, principal-component factor, and maximum-likelihood factor methods. It rotates factors based on the rotation techniques available in Stata including varimax, quartimax, equamax, promax, and oblimin. qfactor displays the eigenvalues of the extracted factors, rotated and unrotated factor loadings, and communalities of each Q-sort. It also displays z-scores and composite ranks of statements for each factor, distinguishing statements, consensus statements, and number of Q-sorts loaded on each factor. In addition, qfactor saves the following files:
- FactorLoadings which unrotated factor loadings, uniqueness and communality of each Q-sort, and indicates which Q-sort was loaded on what factor. This file can be used for subsequent analysis, e.g. producing graphs for loadings.
- FactorScores includes statements, composite z-scores and composite rankings of statements for each factor.
Accio / Pensieve (R Package)
Pensieve is a new open-source R package to support researchers in their study of human subjectivity, supplemented by a web frontend for easy access.
As a feature-rich, one-stop-shop, the software supports a Q study all the way from conception to publishing, including:
- Custom Q data objects, extending the R S3 object-oriented system, including robust data validation.
- Helpers to construct and sample a P and Q-Set.
- Tools to administer Q sorts, online and on paper.
- Factor extraction, including PCA and other algorithms.
- Factor retention, including a Q-specific, modified parallel analysis.
- Factor rotation, including both automatic, as well as by-hand manual methods.
- Factor scoring, including Q-orthodox and alternative methods.
- Interactive plots, aiding in the presentation and interpretation of Q methodological results.
The package adheres to best practices in (R) software development and reproducible research, including source control and dependency management, unit and integration testing as well as an extensive documentation.
Because setting up a computer and data for analysis can have a bit of a learning curve, I am also offering individual support during the conference. Just get in touch (
firstname.lastname@example.org) and we’ll fix a time and place; I’ll be in Glasgow through the end of the conference.
If you want to use accio / pensieve, the following free downloads and preparations are recommended:
- The current version of R for your platform
- The free and open source variant of RStudio Desktop, a helpful integrated development environment (IDE) for R.
- Frequently used R packages (here’s a list); you can install each of them by calling `install.packages(“package_name”)` on the console, or clicking the requisite button in RStudio’s packages pane.
pensieveoff of CRAN (to be released) or GitHub.
- Uh, and don’t forget your computer. You’ll need to have full administrative privileges.
If you get stuck installing any of this, don’t fret, I’ll be happy to help in Glasgow.
PQMethod (via Julia wrapper)
Byung S. Lee will demonstrate in the workshop how a new program using Julia, a free programming language, has improved PQMethod. He will also discuss his ongoing improvement plan, part of which, he hopes, could be further implemented before the workshop.
In the case of Mac users now, they have to run PQMethod on DosBox while PC users can directly run PQMethod. Using the new program running on Julia, Q researchers enjoy a few advantages. 1) Even when using Mac computers, they don’t have to rely on DosBox; 2) They also can enjoy more crisp fonts and white background; 3) They can copy text on the screen and paste it on other programs; and 4) They also can scroll up and down to view the entire results that are displayed on the screen, while PQMethod users can only view one screen at a time. (Please compare the screenshots of PQMethod and a new Julia version.)
PQMethod, which Peter Schmolck has maintained, has served Q scholars over many years. It has rich features, such as entering Q-Sorts manually or as a fixed format file. It computes correlations among Q-Sorts, offers the Centroid or PCA method, and allows the statistical varimax method or theoretical, judgmental manual rotation on two-dimensional plots.
PQMethod, written in Fortran 77, has three components: 1) data entry, 2) numerical analysis and calculations, and 3) production of reports. The 2) portion works well, but 1) and 3) portions have room for improvements using a well-designed GUI (Graphic User Interface). The Julia Language can easily interface with other languages, such as C, Fortran, Python, R, graphical libraries, and others. The author will take an eclectic approach to integrate advanced parts from these external programs and further improve the new program.