Skip to main content

Table 1 Summary of CI-based index properties and their health variable requirements

From: When the technical is also normative: a critical assessment of measuring health inequalities using the concentration index-based indices

 

Properties of the health variable (hi)

Properties of the index

Intervalb

Ratio

Unbounded

Bounded

Absolute

Relative

Mixed

Quasi-absolute

Mirror

Transfer

Weighting scheme

Index equationa

Standard CI

 

✓

✓

  

✓

   

✓

Fixed. Inequality aversion parameter (v) = 2

\(\frac{1}{n}\mathop \sum \limits_{i = 1}^{n} \left\{ {\frac{{h_{i} }}{{\overline{h}}}\left( {2R_{i} - 1} \right)} \right\}\)

Modified CI

✓

 

✓

  

✓

   

✓

Inequality aversion parameter (v) = 2

\(\frac{1}{n}\mathop \sum \limits_{i = 1}^{n} \left\{ {\frac{{h_{i} }}{{\overline{h} - a_{h} }}\left( {2R_{i} - 1} \right)} \right\}\)

Generalized CI

✓

✓

✓

 

✓

   

✓

✓

Inequality aversion parameter (v) = 2

\(\frac{1}{n}\mathop \sum \limits_{i = 1}^{n} \left\{ {h_{i} \left( {2R_{i} - 1} \right)} \right\}\)

Extended CI

 

✓

✓

  

✓

   

✓

Asymmetric Inequality aversion parameter (v) can be varied

\(\frac{1}{n}\mathop \sum \limits_{i = 1}^{n} \left[ {\frac{{h_{i} }}{{\overline{h}}} \left\{ {1 - v\left( {1 - R_{i} } \right)^{v - 1} } \right\}} \right]\)

Generalized extended CI

✓

✓

✓

✓

✓

   

✓

✓

Inequality aversion parameter (v) can be varied

\(\frac{1}{n}\mathop \sum \limits_{i = 1}^{n} \left[ {\left( {\frac{{v^{{\frac{v}{v - 1}}} }}{v - 1}} \right)\left( {\frac{{h_{i} - a^{h} }}{{b_{h} - a_{h} }}} \right)\left\{ {1 - v\left( {1 - R_{i} } \right)^{v - 1} } \right\}} \right]\)

Wagstaff index

✓

✓

 

✓

  

✓

 

✓

✓

Fixed

\(\frac{1}{n}\mathop \sum \limits_{i = 1}^{n} \left[ {h_{i} \left\{ {\frac{{h_{i} \left( {b_{h} - a_{h} } \right)}}{{\left( {b_{{\text{h}}} - \overline{h}} \right)\left( {\overline{h} - a_{{\text{h}}} } \right)}}} \right\}\left( {2R_{i} - 1} \right)} \right]\)

Erreygers index

✓

✓

 

✓b

   

✓

✓

✓

Fixed

\(\frac{1}{n}\mathop \sum \limits_{i = 1}^{n} \left\{ {4\frac{{h_{i} }}{{\left( {b_{h} - a_{{\text{h}}} } \right)}}\left( {2R_{i} - 1} \right)} \right\}\)

Symmetric index

 

✓

✓

  

✓

   

✓

Symmetric Inequality aversion parameter (β) can be varied β can be varied

\(\frac{1}{n}\mathop \sum \limits_{i = 1}^{n} \left( {\frac{{h_{i} }}{{\overline{h}}}} \right)\left[ {\beta 2^{\beta - 2} \left\{ {\left( {R_{i} - \frac{1}{2}} \right)^{2} } \right\}^{{\frac{\beta - 2}{2}}} \left( {R_{i} - \frac{1}{2}} \right)} \right]\)

Generalized symmetric index

✓

✓

✓

✓

✓

   

✓

✓

Symmetric Inequality aversion parameter (β) can be varied β can be varied

\(\frac{1}{n}\mathop \sum \limits_{i = 1}^{n} 4\left( {\frac{{h_{i} - a^{{\text{h}}} }}{{b_{{\text{h}}} - a_{{\text{h}}} }}} \right)\left[ {\beta 2^{\beta - 2} \left\{ {\left( {R_{i} - \frac{1}{2}} \right)^{2} } \right\}^{{\frac{\beta - 2}{2}}} \left( {R_{i} - \frac{1}{2}} \right)} \right]\)

  1. aRi is the rank of the ith person in the SES distribution and \(\overline{h}\) represents the sample mean. bExcept for Erreygers index where measurement scale does not matter, hi should be measured in same unit to ensure differences in estimates are not reflective of arbitrary differences in measurement unit, i.e., weight measured in same units (kg) for all observations