@Mark – if you are interested in assessing the reliability of symptom ratings, then you need ICCs for each item. If you are interested in assessing inter-rater reliability in general (. the reliability relevant to any future statistics computed using a “borderline score”), you’ll want to compute the scale means for each rater and compute ICC for those means. You should not take an average of ICCs, as that’s not an interpretable value – if you’re interested in the ICC of overall scales, that’s what you should compute the ICC on.
Good article, especially the notes that say when these tools are most useful.
I came across your page because I’m deluged with adverts for Grammarly, and when I see so many adverts, I have to do some research. I conclude from this that Grammarly is slightly better than Word’s own spelling/ grammar/ Fleisch score, and I’m not going to pay for that small an improvement.
I’ve used StyleWriter for years. It seems to address most of the above concerns (in the comments as well), as
1) it identifies the obvious process errors (spelling, grammar), but
2) before it tries to tell you how to improve, it lets you define your audience and then tells you how well your current piece matches what it thinks your audience will find most persuasive. Examples: How long and sophisticated should the sentences and words be (the reports of a professional association should be a little more demanding than a book for children, or the readers will feel insulted)? Are there Sticky words that slow down your reader or interrupt their train of thought?
What I particularly like is the graphical report. It can display my whole document in terms of sentence length and sentence complexity, so I can spot the super long and super short sentences, and by clicking on the graphical representation, get taken straight to them to make whatever corrections.
Of course it isn’t live – once I’ve made a series of corrections/ improvements then I have to rerun the report. And it doesn’t correct for me – I have to make the corrections. But my readership seems to find the results impressive.
By the way, none of the other tools are live either.