Retirement of SocialSciences task view

[zeileis]

The existing Social Sciences task view is one of the old existing task views that was useful when it was introduced but turned out to be too broad to be useful. John Fox @john-d-fox and I both agree on this and it would probably be better to establish separate task views for topics like: mixed-effects models, social network analysis, demography, and possibly causal analysis or matching/propensity scores. Unless there are any objects I therefore suggest:

• We retire the SocialSciences task view immediately (still in the SVN on R-Forge).
• When we announce the possibility of proposing new task views, we reach out to some people who might be willing to put together task views on the topics above.

[tuxette]

I agree with a few comments:

• I suppose that you are aware that some people have already been working on a task view on causal analysis, which is (as far as I know) almost ready to be submitted?
• I wonder if the task view "social network analysis" is not a bit overlapping "graphical models"?

[zeileis]

Thanks, Nathalie @tuxette !

Causal inference: Julie Josse once mentioned to me that she might be interested in putting together a task view on that topic but that was 2.5 years ago and I hadn't heard back since.

Social network analysis: I'm not sure how much overlap there really is but I'll mention it to Søren in a separate issue.

[john-d-fox]

HI,   I can’t speak to existing plans for a task view on causal inference (though that would be good news), but I did take a look at the graphical models task view, and there is overlap with network analysis. Some of the overlap makes sense, such as packages for drawing graphs, but I find some of it a bit mystifying, such as the inclusion of the ergm package (for exponential random graph models), which, unless I’m missing something, is unrelated to graphical models. I think that there’s still probably room for a task view that focuses on social network analysis, possibly coordinating with the content of the graphical models task view.   Best, John  

[zeileis]

I personally also lean more towards a separate task view but wanted to get Søren's opinion on this.

[john-d-fox]

I agree that makes sense, and it would be desirable to coordinate the content of the two task views if a social networks task view materializes.

[eddelbuettel]

@john-d-fox You email setup makes replies borderline readable. Can you consider another email client, or answers here? Editing your second answer down to something more useful.

[john-d-fox]

Dirk,

Until recently, I used the Thunderbird email client, which was easily configurable, but my university IT department now requires use of either the Outlook or Apple Mail clients, or Outlook web mail. I'm using the Outlook email client, and have found it frustrating.

OTOH, I just looked at my messages in this thread on GitHub, and, apart from a bad line break and a typo in my first message, they seem perfectly intelligible to me.

My second message concerns coordinating a new social networks task view, should one materialize, with the existing graphical models task view. That, of course, would be up to the maintainers of the two task views. Is that sufficiently clear?

John

[eddelbuettel]

Well yes, after I edited them. Mail replies often leave the quoted trail, and GitHub is clever enough to not show that for the forms it recognized. But that failed here. But what was a first I had never seen in all those years is that the mailer you used (possibly in the way IT configured it) also spammed in thirty or so lines of CSS. Which is fine and well if done in a way that GitHub can filter it which it didn't here. It's quick enough to edit out as a one-off (via the the '...' menu entry) but can get repetitive. The much better news is that this reply of your has neither shortcoming. So thank you!

[john-d-fox]

Now I understand -- the message was garbled before you edited it. Thanks for that (and I've now fixed the typo and bad line break).

I managed a while ago to configure Outlook so that it sends plain-text emails by default, but I can't get it to reply in plain text to HTML emails. As I said in my previous message, I wish that I could use a better email client, but I can't.

I'll simply continue to avoid using email to reply to GitHub messages.

[robjhyndman]

Maybe @timriffe would be interested in doing a Demography task view?

[timriffe]

@robjhyndman sure but I couldn't dedicate much time until May/June. But it wouldn't be too hard to rally support, would it @MJAlexander ?

[zeileis]

Thanks for the suggestion Rob @robjhyndman and for the follow-up Tim @timriffe , this is very much appreciated. Time is not crucial here. If you and a couple of potential co-maintainers like @MJAlexander would put together a first proposal, then we (the CRAN Task View Editors) can have a look and you will still have enough time to come up with the actual first version of the full task view. The guidelines for proposing a new task view are described at: https://github.com/cran-task-views/ctv/blob/main/Proposal.md. Let us know if you need more information or feedback etc.

[zeileis]

OK, thanks everyone for the discussion so far. I will close this issue now and just note some ideas and concrete plans:

• A CausalInference task view has already been proposed in CRAN task view proposal: CausalInference #15
• A Demography task view might be proposed by @timriffe and potentially @MJAlexander
• A SocialNetworkAnalysis task view would also be a good idea but no potential maintainers have been encouraged to submit a proposal.

Further "spin-offs" might be considered in the future as well.

[john-d-fox]

One further suggestion: The people behind statnet <https://statnet.org/attribution> (I don't know them) might be interested in a social-networks taskview.

…

On 2022-03-06 6:07 a.m., Achim Zeileis wrote: OK, thanks everyone for the discussion so far. I will close this issue now and just note some ideas and concrete plans: * A |CausalInference| task view has already been proposed in #15 <#15> * A |Demography| task view might be proposed by @timriffe <https://github.com/timriffe> and potentially @MJAlexander <https://github.com/MJAlexander> * A |SocialNetworkAnalysis| task view would also be a good idea but no potential maintainers have been encouraged to submit a proposal. Further "spin-offs" might be considered in the future as well. — Reply to this email directly, view it on GitHub <#8 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADLSAQQLUMGYJZCFWB6WA23U6SGYLANCNFSM5KKCUBOA>. Triage notifications on the go with GitHub Mobile for iOS <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android <https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub>. You are receiving this because you were mentioned.Message ID: ***@***.***>

[zeileis]

Thanks, I agree. They would be a natural group to ask.

[john-d-fox]

Yet another suggestion: I originally mentioned a mixed-effects models taskview as one possible partial successor to the social-sciences taskview. A natural person to approach about that would be Ben Bolker. I do know him and could approach him, if that would be helpful. (And is someone planning to follow up with the statnet folks about a social networks taskview?)

[zeileis]

We're currently working towards the first official release of the task views based on the new infrastructure. So I would like to get this out first before starting to invite new task view proposals.

Also, in case of the mixed-effects topic I'm not completely sure that this would be the best way to encapsulate a new topic. There is a lot of overlap with existing task views and it would exclude non-mixed-effects methods for the same kind of data (longitudinal, panel, multi-level), e.g., OLS variants or GEE.