new coding form

[rubenarslan]

About the new coding form I like:

• it's clean
• there's lot of unobtrusive hints and help
• progress bars for sub-sections
• menu to the left shows which study one is currently working on
• deletion buttons are much clearer and better positioned
• it's clear how to change the study and test names
• I like that you scrapped the "tagging" system I used for methodological codes, IVs and DVs, it was dumb.

About the old form I liked:

• grouping, on a line, of input fields that belong together, e.g effect size statistic and reported effect size, hypothesis-related items or the optional replication items. Currently this organisation is sort of lost.
• it was a little tighter-spaced. I know there's a general trend to avoid this in current web design, but I think few people will fill out the form on a mobile phone, so maybe looking nice on mobile shouldn't come first for once.
• because I used a greater variety of different input fields, there was something of a visual indicator of where you currently are in the form. I think you over-relied on dropdowns, that way the fact that the methods questions for ...hypothesis, ... statistics etc. all have the same options is lost. Also, dropdowns require two clicks or to get the right choice, radio buttons just one. We might have "power coders" one day, after all.
• the help text in the new form should also have a note regarding what constitutes "other variables" (mainly covariates).
• Significance reported (and also effect size statistic and test statistic) both allowed you to enter text if the necessary choice was missing (if a precise significance level was reported such as p=0.04, but we also provided various common representations like ***). This ability is lost in the new form, but we do need that information. It may be that the combined text input/select wasn't clear (probably you didn't notice it could do that), so maybe we need a different option, but maybe a hint would suffice.

I don' t have a really good idea of how the two layout things I like about both forms (help bubbles to the right and progress report to the left in the new form and tighter layout and grouping on a line in the old form) can be combined.
Maybe, we could emphasise grouping through decreased spacing in some places (e.g. hypothesis-related items) and sometimes just squeeze it in one line, by putting the label on top and making the input fields for numbers narrow (like Test stat. and Effect size in the old form).

We also have this "computed significance" field, which I see a bit as a remnant of when we still wanted students to do analysis. Is this something to scrap? Maybe Roger knows.

[lyndsysimon]

I was going for function over style, at least in the beginning.

About the old form I liked:

• grouping, on a line, of input fields that belong together, e.g effect size statistic and reported effect size, hypothesis-related items or the optional replication items. Currently this organisation is sort of lost.

I've begun to address your comments one-by-one, and just rolled out an update that groups the "N total", "N Exclusions", and "N Used" fields on a single line, and reinstates the autocalculation piece as well. I agree that there should be some additional work here in terms of organization and would welcome input on specifically what needs to go where.

I've purposefully not used fieldsets to this point, so I'll have a semantic element to use to group related inputs together in the future.

• it was a little tighter-spaced. I know there's a general trend to avoid this in current web design, but I think few people will fill out the form on a mobile phone, so maybe looking nice on mobile shouldn't come first for once.

I wasn't really thinking about mobile, but I know what you mean. The form feels "long" right now. I think this is a consequence of serializing the fields more than the spacing between them - hopefully, this concern will be addressed as we group related fields.

• because I used a greater variety of different input fields, there was something of a visual indicator of where you currently are in the form. I think you over-relied on dropdowns, that way the fact that the methods questions for ...hypothesis, ... statistics etc. all have the same options is lost. Also, dropdowns require two clicks or to get the right choice, radio buttons just one. We might have "power coders" one day, after all.

I defaulted to dropdowns for all "select one" fields and checkboxes for all "select multiple" fields. I agree that radios would be more appropriate for the questions about the coder's confidence, and will change them.

• the help text in the new form should also have a note regarding what constitutes "other variables" (mainly covariates).

I'm not comfortable writing this sort of content yet - @jlcohoon, perhaps you could contribute this?

• Significance reported (and also effect size statistic and test statistic) both allowed you to enter text if the necessary choice was missing (if a precise significance level was reported such as p=0.04, but we also provided various common representations like ***). This ability is lost in the new form, but we do need that information. It may be that the combined text input/select wasn't clear (probably you didn't notice it could do that), so maybe we need a different option, but maybe a hint would suffice.

Yes, I did miss that. Comboboxes are glarlingly missing from HTML prior to HTML5, and they're not well supported yet at all. I assume you have Javascript to fill that gap, and I'll do the same.

Maybe, we could emphasise grouping through decreased spacing in some places (e.g. hypothesis-related items) and sometimes just squeeze it in one line, by putting the label on top and making the input fields for numbers narrow (like Test stat. and Effect size in the old form).

I'd rather not decrease vertical space between inputs if I can help it. I'd much rather use the <fieldset> element to mark related sections semantically and display that however seems best. Ther way I wrote the progress bars plugin will also allow each of these areas to have their own progress bar built in, even through I don't think we'll be showing it in the left bar.

Collapsing each section as they are completed is an option, as is allowing multiple sections to appear side-by-side on the page. Right now much of the space on the right is wasted, and that would recoup a lot of it.

We also have this "computed significance" field, which I see a bit as a remnant of when we still wanted students to do analysis. Is this something to scrap? Maybe Roger knows.

Let me know - trimming it off the form shouldn't be a problem.

[rubenarslan]

Thanks for the reply – I think going for function over form is definitely worthwhile, so reinstating the combo-boxes should be first priority. I used select2 for that, it has a fair browser coverage and our audience is likely to use one of those it covers. It degrades gracefully too. I've also webshims for polyfilling number inputs and so on on Firefox which is really lagging behind on HTML5 forms. I also remembered why I was using this tagging system for methodology codes - it's a combobox too, because there will me methodology codes we missed.

I think Effect size statistic and effect size belong on a line. So do Test stat., degrees of freedom and and test stat value. The df field can be very narrow. Reported significance and computed - should wait for Roger's go-ahead to axe the computed field. For the agreement stuff, I think you could just go back to my version, you also missed one item (... hypothesis).

The conditional hiding part for the hypothesis should be reinstated, though I don't think they can or should be on one line. Maybe you can style fieldsets like bootstrap "well" if you decide to use it here.

I didn't check if you went with my code for the progress calculation, but I only calculated progress on visible fields, so the conditional hiding comes handy here.

The idea with collapsing studies and test optionally is good. People could use a button to collapse them, which would reduce them to their structure and a progress indicator (maybe preferably just text there). I don't like side-by-side and I don't think the space to the right is wasted, if anything the space to the left leaves something to wish for. Maybe the abstract and the link to the full paper could go below the progress bars?