Wishlist for Stata 18

[Andrew Musau]

Now that Stata 17 is out, use this thread to request features that you desire for Stata 18.
sladmin

[Dario Maimone Ansaldo Patti]

I saw a large improvement in the bayesian analysis. I would be very glad if in the new release even bayesian spatial analysis could be performed.

[Clyde Schechter]

Wasn't expecting to see Stata 17 come out today, but it looks like it has a lot of good new stuff.

Anyway, here's what I was going to add to my wishlist for Stata 18--I assume this hasn't been implemented in 17 (which I've ordered, but don't have my download yet.): let's have the ability to open a separate browser for each frame.

[I-Chin Wu]

I hope Stata 18 could provide official support for non-parametric cumulative incidence estimation and state occupation probability in the presence of competing risks. Currently, these could be estimated by stcompet and msaj. Thank you.

[River Huang]

I wish to see commands on nonlinear panel (both static and dynamic) quantile and threshold models!

[Sergiy Radyakin]

I'd like some more functionality for the recently added frames :

• the append command to work across frames (without the need to save data to even a temporary file at all);
• frlink+frget to work like a true link, not copying the data, but just linking to it (as a possibility in addition to the current behavior, which is also useful).
• this link to be bi-directional (so that when the linked info is changed in one frame it is changed in all linked frames).
• saving multiple frames as a single dataset (and loading all the saved frames or selectively, by frame name).
• for links to work within the same data frame:

clear all

input id fid strL name
1 . "John"
2 1 "Peter"
3 2 "Victoria"
4 2 "Simon"
5 4 "Gretta"
6 4 "Anna"
7 4 "Maria"
end

// using the same frame does not produce the desired result:
frlink m:1 fid, frame(default id)
frget fname=name , from(default)
list id fid name fname, sepby(fid)
drop default fname
// assert fname=="John" in 2 // fails

// using a copy in a different frame does produce the desired result:
frame copy default reference
frlink m:1 fid, frame(reference id)
frget fname=name , from(reference)
list id fid name fname, sepby(fid)
assert fname=="John" in 2

Thank you, and looking forwards for Stata 18 already!

[Richard Williams]

Here are a few things I would like, if they are statistically justifiable. I have seen that there is work being done in these areas, but I don't know how good it is.

Add an fe (fixed effects) option to xtologit. The new xtmlogit has an fe option, which is nice.

Alternatives to BIC and AIC that can be used with svyset data. For that matter, alternatives for a lot of things when data are svyset, e.g. various diagnostic tests.

Add an memlogit command. Many xt commands have an me counterpart, so I am mildly surprised xtmlogit doesn't.

Whenever Stata does not have something I would like or expect, I figure it is usually because (a) it is statistically inappropriate, or at least an appropriate means has not been worked out yet, or (b) nobody has gotten around to programming it yet. xtmlogit seems to have been in the latter category until yesterday, so perhaps some of these other things do too.

[Roy Berghaus]

I would appreciate the ability to perform Tukey pairwise comparisons after running a linear mixed model.

[Rich Goldstein]

I think that the "teffects" modules need a little work; at the least, I would like to see (1) the ability to get balance statistics on what the manual calls the raw data before any balancing (helps to see if one needs to go through the balancing steps); (2) for the various models under, e.g., teffects ra, I see no way to diagnose the model (i.e., tell whether I have a good model) as exists for regress, logit, etc. and I would like to be able to do those

[daniel klein]

for the various models under, e.g., teffects ra, I see no way to diagnose the model (i.e., tell whether I have a good model) as exists for regress, logit, etc. and I would like to be able to do those

Most of the models in teffects combine multiple models (outcome and treatment) for multiple groups (at least treatment vs. control). How would one model fit look like? Seriously interested; is there any literature on how to assess the fit for such combined models?

[Rich Goldstein]

daniel klein - thanks for the question which allows me to clarify at least a little - I was thinking primarily of the outcome model but diagnostics for each model should be possible; I do not know of diagnostics for the combined model

[Marc Kaulisch]

I wonder whether a next good step will be to use the collections for graphing as well. The reasoning to use to collect instead of table - reduce computing time in large data environments - would be valid for graphs as well, or? And the opportunity to change labels easier would be nice for graphs too.

[Sergiy Radyakin]

I'd like some more functionality for the recently added frames :[LIST]
...[*]this link to be bi-directional (so that when the linked info is changed in one frame it is changed in all linked frames).
...

To clarify the above point, here is a test example that I wish could pass (currently does not):

clear all
input long interview__id strL address
1 "123 Main St"
2 "456 Broadway St"
3 "789 Winter Hwy"
9 "101 Church St"
end
list

frame create MEMBERS
frame change MEMBERS
input long interview__id byte mem__id strL mem_name int mem_age
1 1 "John SMITH" 39
1 2 "Mary SMITH" 32
1 3 "Peter SMITH" 7
2 1 "Jennet JONES" 19
2 2 "Kenneth JONES" 27
3 1 "Ted RYAN" 69
3 2 "Clara RYAN" 65
end
list, sepby(interview__id)


frlink m:1 interview__id, frame(default)
frget address, from(default)
list

frame change default
replace address="TEST" in 1
frame change MEMBERS
list, sepby(interview__id)
assert address=="TEST" in 3
assert address=="TEST" in 2
assert address=="TEST" in 1

Clearly frlink could be frlink2 or some mode switch to establish the live link.

Thank you, Sergiy Radyakin

[Tiaga Falcao]

Network Analytics (Econometrics and Visualization)

[JanDitzen]

Tiaga Falcao are you thinking of estimations of specific models?