Why has the bootstrap display been changed since Stata 16?

[Felix Bittmann]

Dear all,
ever since switching from Stata 16.1 to 18.0, I am so confused of why the display of the bootstrap (or simulation) progress display has been changed. For comparison:


Stata 16.1

. reg mpg weight foreign, vce(bootstrap, reps(500))
(running regress on estimation sample)

Bootstrap replications (500)
----+--- 1 ---+--- 2 ---+--- 3 ---+--- 4 ---+--- 5
..................................................    50
..................................................   100
..................................................   150
..................................................   200
..................................................   250
..................................................   300
..................................................   350
..................................................   400
..................................................   450
..................................................   500

Linear regression                               Number of obs     =         74
                                                Replications      =        500
                                                Wald chi2(2)      =     142.64
                                                Prob > chi2       =     0.0000
                                                R-squared         =     0.6627
                                                Adj R-squared     =     0.6532
                                                Root MSE          =     3.4071

------------------------------------------------------------------------------
             |   Observed   Bootstrap                         Normal-based
         mpg |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
      weight |  -.0065879   .0005595   -11.78   0.000    -.0076844   -.0054914
     foreign |  -1.650029   1.165938    -1.42   0.157    -3.935226    .6351676
       _cons |    41.6797   1.833757    22.73   0.000      38.0856     45.2738
------------------------------------------------------------------------------

Stata 18.0

. sysuse auto
(1978 automobile data)

. reg mpg weight foreign, vce(bootstrap, reps(500))
(running regress on estimation sample)

Bootstrap replications (500): .........10.........20.........30.........40.........50.........60.........70.........80.....
> ....90.........100.........110.........120.........130.........140.........150.........160.........170.........180.......
> ..190.........200.........210.........220.........230.........240.........250.........260.........270.........280........
> .290.........300.........310.........320.........330.........340.........350.........360.........370.........380.........
> 390.........400.........410.........420.........430.........440.........450.........460.........470.........480.........4
> 90.........500 done

Linear regression                                       Number of obs =     74
                                                        Replications  =    500
                                                        Wald chi2(2)  = 140.80
                                                        Prob > chi2   = 0.0000
                                                        R-squared     = 0.6627
                                                        Adj R-squared = 0.6532
                                                        Root MSE      = 3.4071

------------------------------------------------------------------------------
             |   Observed   Bootstrap                         Normal-based
         mpg | coefficient  std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
      weight |  -.0065879   .0005623   -11.72   0.000      -.00769   -.0054858
     foreign |  -1.650029   1.177672    -1.40   0.161    -3.958225    .6581663
       _cons |    41.6797   1.832716    22.74   0.000     38.08764    45.27176
------------------------------------------------------------------------------

The old display is clean, all ducks in a row. The new one is chaotic, the dots are not aligned at all and it is less clear overall. Line break in the middle of a number? Yikes. Maybe this is just an old man yelling at cloud moment for me, but is there any reason for this change? Just curious about the thought process. Do other users prefer the new display? Comments welcome!

[daniel klein]

The dots used to be displayed by non-documented _dots.ado; they are now displayed by undocumented Mata (class) _dots(). The messed-up display might well be an unintended side-effect of the transition.

[Sebastian Kripfganz]

I believe, the intention is to conserve vertical space in log files. I still have to encounter someone who actually prefers the new way how dots are displayed. It would be nice if the user was able to set a style for the dots.

[daniel klein]

If users wish to conserve (vertical) space, there has always been the dots(#) or, more radically, nodots options. If StataCorp wanted to conserve space by default, changing the default from dots(1) to dots(>1) would have been the better choice IMHO.