build reporting template

[maxheld83]

build a template for how q-results may be reported, including more accessible formatting in LaTeX / Knitr / RMarkdown.

This is maybe less of a technological hurdle, but one that may help people do their own (first) q study.

[aiorazabala]

How about this draft structure? (an old draft of mine)

"I suggest the following standard to report a Q methodology study in order provide all the necessary information to ensure its reproducibility and replicability:

1. Research design:
   • The process of constructing the concourse, the selection of statements and their structure (if any).
   • The number of statements.
   • The shape of the distribution grid, and whether respondents were forced to fit their sorting
     to this shape.
   • The number of participants and criteria for selection.
   • Administration: format (paper, online, etc.) and condition of instruction.
2. Analysis:
   • Method for extraction.
   • Number of factors extracted.
   • Method for rotation.
   • Number of Q-sorts flagged for each factor.
   • Number of Q-sorts confounding and/or not loading high on any factor.
3. Results:
   • The table of factor loadings (non-essential), indicating which Q-sorts are flagged.
   • The table of statement results, either z-scores or factor scores, and an indication of which
     statements are distinguishing and consensus.
4. Interpretation: semantic denomination and description of each of the perspectives."

[maxheld83]

That's great.

For the analysis I have nothing to add (except maybe the correlation coefficient used, but that's a bit nitpicky).

What do you think about adding some standard set of plots (as available in qmethod) to the reporting template?
I was thinking (in that order, part of 3,)

1. correlation matrix correlation matrix plot #92
2. factor space (loadings) (works only for 2-3 dims, maybe work together with manual rotation) loadings plot: visualization for factor space (loadings) (2D * x) #94
3. Factor correlation matrix (same as correlation matrix, except for cross-factor cors) make visualization out of factor loading matrix #93
4. factor arrays (the ideal-typical scores) make visualizations out of factor arrays #14
5. factor composition factor composition plot #98 (that's a very handy plot that I've so far only seen in Brown)

[maxheld83]

also, I agree that this should be kept quite minimal, and leave out the methodological parts – we can't get into that bag of worms.
Should just be a helper function (?) to give people easily everything they should probably be looking at (much like qmethod already does).

[maxheld83]

I'm a little hazy about where/how to best implement this.
I see 3 options:

1. expand the default output by qmethod() (probably not a good idea because it leads to bloat?)
2. make a reporting function qreport() that wraps a bunch of other functions. Could go two ways (at least):
   a) via knitr / rmarkdown and return a PDF and/or website (requires TeX on machine)
   b) via shiny
3. just make it a recommendation of things to run and how to produce them on the wiki (which, btw, is great!)

[aiorazabala]

Check out the function export.qm(), this exports a report as a text file, similar to the PQmethod LIS file