[new on SSC] ineqord: module to calculate indices of inequality and polarization for ordinal data

[Stephen Jenkins]

With thanks as ever to Kit Baum, ineqord is now available on SSC. Its functionality is described below. For references to literature, see the help file.

TITLE
'INEQORD': module to calculate indices of inequality and polarization for ordinal data

DESCRIPTION/AUTHOR(S)

ineqord calculates indices of inequality and polarization for
ordinal data recorded in the response variable: the
Allison-Foster index, the normalized Average Jump index, multiple
Apouey indices (parameters 0.5, 1, and 2), multiple Abul
Naga-Yalcin indices (parameters (a,b) = (1,1), (2,1), (1,2),
(4,1) and (1,4)), multiple Cowell-Flachaire indices (for
peer-inclusive downward and upward-looking status; parameter
alpha = 0, 0.25, 0.5, 0.75 and, optionally, another alpha value
between 0 and 1), the Jenkins index, and also the standard
deviation. Optionally, ineqord also derives estimates of
cumulative distribution functions, survivor functions, and
Generalized Lorenz curves. These can be used to describe ordinal
data distributions and to undertake dominance checks of
differences between distributions.

KW: inequality
KW: polarization
KW: indices
KW: survivor functions
KW: Generalized Lorenz curves

Requires: Stata version 14

Distribution-Date: 20191214

[Stephen Jenkins]

With thanks as ever to Kit Baum, a revised version of ineqord (and its help file) is available from SSC.

Thanks to Benoît-Paul Hébert who pointed out an issue with levelsof that affected users with a Stata version below 15.1. (ineqord is for version 14 or later.) That issue has now been addressed.

In addition, I've added options to calculate the ordinates of "H+" and "H-" curves. These may be used to check for inequality dominance consistent with Hammond transfers: see Gravel, Magdalou, and Moyes (2020), 'Ranking distributions of an ordinal variable', Economic Theory, online ahead of print. http:// https://link.springer.com/article/10.1007/s00199-019-01241-4

[Stephen Jenkins]

With thanks again to Kit Baum, a further revised version of ineqord (and its help file) is available from SSC. There are a few cosmetic changes to the ado- and help-file, one of which is intended to help version 14 users. An article based around the command is forthcoming in the Stata Journal.

[Soren Nielsen]

Thanks for a very useful program.
I am currently using it to evaluate the effect of eating disorders on educational attainment in an observational epidemiological register follow-up using record linkage. The design is 1:4 matching on sex at birth, age, and place of residence. N: ~ 20 000 in the ed-group and ~80 000 matched controls. I have had but little trouble in running ineqord and creating graphs analoguous to Fig. 1 in Jenkins (2020) your net sj 20-3 paper.
To your knowledge, has anybody else used ineqord to evaluate the distributions of ISCED 2011, (which are ordinal data)?
Preliminary findings show that there are 'too many' of the eating disorders cases below the median when compared to the controls.
I am using the ineqord procedures as preparation before xtologit procedures and model building etc

I hope that I am not unneccesarily disturbing your universe

Søren Nielsen

[Stephen Jenkins]

Soren Nielsen : I don't know of research using -ineqord- to summarize ISCED data. (Perhaps most researchers of those data are unaware of the program's capacity to easily draw graphs like Fig 1, i.e., discrete CDFs ... or perhaps they draw such graphs using a different approach ... or (worse) don't inspect the distributions of data that they are working with!) Good luck with your work

[Soren Nielsen]

I shall do my best
Thanks for the encouraging words