Académique Documents
Professionnel Documents
Culture Documents
Table of Contents
0. Introduction
1. Definitions
1.1. The qualitative approach
1.2. The quantitative approach
1.2.1. Aggregated
1.2.2. Reproducible
1.2.3. Others characteristics of measurements
1.3. Overlaps
2. Advantages and disadvantages of either approach
2.1. The qualitative approach
2.1.1. Advantages
2.1.2. Disadvantages
2.2. The quantitative approach
2.2.1. Advantages
2.2.1. Disadvantages
3. Why should human rights violations be measured?
3.1. Measuring and promoting success
3.2. Identifying causes
4. How should human rights volations be measured?
4.1. Transparency and comparability
4.2. Statistical soundness
4.3. Cultural sensitivity
4.4. Simple or composite measures
4.5. Useful output
5. What should be measured?
5.1. Types of measurements
5.2. Cross-measurements
5.3. Meta-measurements
5.4. Stylized example
5.5. Other examples
6. What are the problems with measurements?
6.1. Statistical problems
6.2. Political problems
6.3. How to improve measurements?
7. Conclusions
0. Introduction
There are two ways of describing or documenting the phenomenon of human rights
violations. There's the qualitative approach, and there's the quantitative approach.
The aim of both approaches is the same: to describe or document human rights vi-
olations, and thereby to contribute to the realization of a higher level of respect for
human rights. They only differ in the preferred descriptive methods, not in the ob-
ject of description or the purpose that is served by the description.
1. Definitions
A particularly interesting data source in the quantitative approach are so-called ex-
pert-group data. These result from interviews with expert, independent researchers,
Page | 4 field workers, forensic specialists, psychologists, government officials etc.
Broadly speaking, whereas the qualitative approach uses words to describe rights
violations, the quantitative approach uses numbers, scores and scales.2 The quantit-
ative study of human rights aims:
Let me focus for a moment on two core elements of this definition, aggregation
and reproduction. I will return to the definition of measurement, indicator or score
(henceforth MIS) in paragraph 4, "How should human rights violations be meas-
ured".
1.2.1. Aggregated
The MISs that make up the quantitative approach are aggregated in the sense that
they transcend specific events, cases, persons, perpetrators or victims. The evidence
that is aggregated into an MIS must come from a large and statistically representa-
tive (and random) sample or even the whole population of relevant cases of rights
violations.3
Case studies as such, even a few of them, are insufficient sources for MISs. The
usual statistical and scientific methods regarding representativity, randomness of
the samples, quality control, coherent sources etc. must be respected. Data sources
for MISs are typically household surveys, opinion surveys, national accounts data,
1
Of course, in many cases, violations do not reach the court, and the only "verdict" is the one that is issued
by public opinion and "history". However, this makes the quasi-judicial role of the qualitative approach all
the more important.
2
See The Metagora Project, http://www.metagora.org/html/index.html. I consider ordinal scales (such as
"good, medium, bad") to numeric in nature.
3
The Quantitative Study of Human Rights Violations, Cingarelli & Richards (2007), p. 1.
The quantitative approach is therefore more than the sum of results of the qualita-
tive approach, although it's not impossible to conceive of cases in which qualitative
Page | 5 results can be a useful input in quantitative measurements, especially if the events
included in qualitative results are systematically described, in a standardized fa-
shion using common definitions and attributes that make it possible to aggregate
the descriptions. For example, registration of the sex, age, race, ethnicity of the vic-
tims and perpetrators.4 Qualitative data are then turned into numbers.
However, often this is not the case, either because qualitative descriptions of human
rights violations are not done in this way, or because it is impossible to do it in this
way. And even if they are, it is almost inevitable that the events will be recorded in
a highly selective way, even if each individual event is recorded systematically and
completely. As a result, generalizations, country comparisons and the creation of
global or even regional progress indicators are impossible.
Hence, events described in the qualitative approach will not often be useful data
sources for the quantitative approach. Which doesn't mean their recording is use-
less. They are part of the qualitative approach and offer all the advantages of that
approach (see paragraph 2.1. below). They may not be the source of MISs or statis-
tics, but they are the source of history and they are especially useful in court proce-
dures.
Ideally, quantitative human rights data should also provide explanations of the
causes of rights violations, but this paper focuses only on the descriptive power of
these data, and their ability to offer measurements of variations in human rights vi-
olations (variations over space and time) and to offer the possibility to evaluate and
adjust policies and human rights projects.
1.2.2. Reproducible
The MISs need to be reproducible, both over space and time. It must be possible to
calculate the same score in different places (e.g. countries) and at different times.
This will allow comparisons over space and time, an important aspect of the quan-
titative human rights approach. Different countries, for example, can be compared
by way of their different scores. And progress or retreat over the course of time can
also be measured.5
This is important because it is widely accepted and legally codified that some hu-
man rights, and in particular some economic human rights, can be realized only
4
Quantitative Human Rights Indicators, A Survey of Major Initiatives, Malhotra & Fasel (2005), p. 6. The
organization http://www.huridocs.org has made a particular effort in this field.
5
ibidem, p. 3.
Page | 6
1.2.3. Others characteristics of measurements
MISs should not only be aggregated and reproducible. As pointed out by Sano,8
they should also be:
1.3. Overlaps
The distinction between the two approaches should not become a fetish. Quantita-
tive measures always imply qualitative statements because they always express a
judgment on the quality (or lack thereof) of human rights protection.9 On the other
hand, purely quantitative evaluations without examples and telling cases, are often
meaningless because they reduce real suffering, deprivation and tyranny into a set
of abstract numbers.
6
For example, article 11 of the International Covenant on Economic, Social and Cultural Rights: The States
Parties to the present Covenant recognize the right of everyone to an adequate standard of living for himself
and his family, including adequate food, clothing and housing, and to the continuous improvement of living
conditions. The States Parties will take appropriate steps to ensure the realization of this right. (My empha-
sis).
7
Quantitative Human Rights Indicators, A Survey of Major Initiatives, Malhotra & Fasel (2005), p. 27.
8
Human Rights Indicators, Purpose and Validity, Sano (2005).
9
ibidem, p. 3. See The Metagora Project, http://www.metagora.org/html/index.html.
Qualitative approaches are mostly anecdotal. We can find them, for example, in AI
or HRW reports. They have the advantage of showing the human side of stories.
They engender empathy and motivate to take action. Their often very detailed narr-
atives of events make them the ideal sources of history as well as invaluable contri-
butions in the administration of (criminal) justice in those cases in which rights vi-
olations reach the courts. But even when they don't, it may be possible that viola-
tions stop because of the mobilizing power of the qualitative approach. This ap-
proach also allows researchers to uncover the motivation(s) behind rights viola-
tions.
2.1.2. Disadvantages
Qualitative approaches may also fail to show the larger perspective of the human
rights situation in a particular place. Magnitudes and trends of rights violations can
only be offered by the quantitative approach.10
The main advantage of the quantitative approach is that it can measure a general
state of affairs, not limited to specific cases or events. It provides a more general
understanding of the level of respect for (a) human right(s) in a particular country
or region, and it uncovers trends that are not immediately detectable on the micro-
scopic level of individual events.11
Rather than reporting individual events or incidents, the quantitative approach tries
to calculate a score, using surveys or other available quantitative information.
When these calculations and scores are repeated over time, it becomes possible to
measure progress and success (or lack of it). Government actions or policies, NGO
activity and intervention by international agencies can then be evaluated and redi-
rected if necessary.
10
See The Metagora Project, http://www.metagora.org/html/index.html.
11
Human Rights Indicators, Purpose and Validity, Sano (2005), p. 3.
Page | 8 This also holds for judicial action which, for example in the framework of the ICC,
can benefit, not only from highly accurate accounts of events (provided by the qua-
litative approach), but also from accurate, objective and general statistics.13 These
statistics can also assist charities seeking to justify demands for funds. And NGOs
may use them in order to focus or target their more qualitative efforts. Governments
and international institutions may find it necessary to justify investment in human
rights projects, and can do if they have indicators that show the success of these
projects.14
2.2.1. Disadvantages
The quantitative approach makes it all somewhat more impersonal because it fo-
cuses on numbers and rankings rather than stories, and because it takes the general
rather than the anecdotal view (often on the relatively high and abstract level of a
country, a region or even the world). It's purpose is not mobilization or empathy,
but evaluation and prioritization.
In some countries or some domains, statistical data are lacking or of inferior quali-
ty, making it hard to calculate correct MISs and country comparisons. These diffi-
culties are often compounded by definition problems, survey population issues,
question formulation issues, breaks in times series or insufficiently long time series.
12
See The Metagora Project, http://www.metagora.org/html/index.html.
13
See for example the work of Patrick Ball in Kosovo, http://www.csmonitor.com/2008/0207/p20s01-
wogn.html.
14
What’s the Goal? What’s the Purpose? Observations on Human Rights Impact Assessment, Andreassen &
Sano (2004), p. 5.
15
See The Metagora Project, http://www.metagora.org/html/index.html.
There are other disadvantages to the quantitative approach. It is often very costly to
gather the required information. And the use of structured surveys makes it hard to
understand the context, background and motivation of rights violations.18
A clear picture of the evolution of violations over time does not only bring out the
force of shame and peer-pressure. It also helps to evaluate of the success of human
rights policies of international institutions and NGOs.
16
The Quantitative Study of Human Rights Violations, Cingarelli & Richards (2007), p. 4, and Human
Rights Indicators, Purpose and Validity, Sano (2005), p. 2.
17
Human Rights Indicators, Purpose and Validity, Sano (2005), p. 2.
18
See The Metagora Project, http://www.metagora.org/html/index.html.
On the output side, one has to reflect on the type of result one expects from an MIS.
Does one want an ordinal scales,21 and, if so, of what type? In its simplest form, an
MIS can rank for example a country according to the level of respect for a particu-
lar right during a particular period (the ranking may be for example "good", "me-
dium", "bad"). Of course, many MSIs will go way beyond this simple type of
measure.
19
The Scope of Human Rights: From Background Concepts to Indicators, Todd Landman (2005), p. 13.
20
See The Metagora Project, http://www.metagora.org/html/index.html.
21
See here: http://en.wikipedia.org/wiki/Level_of_measurement
22
Human Rights Indicators, Purpose and Validity, Sano (2005), p. 6.
23
World Bank, http://www.worldbank.org/poverty/
24
Human Rights Indicators, Purpose and Validity, Sano (2005), p. 2.
It is also important to avoid measuring human rights independently from each oth-
er. One should compare different indicators and try to establish correlations be-
tween different measurements for different human rights. Such correlations are ex-
Page | 12 amples of the power of statistical analysis to uncover what other types of analysis
cannot (although one should warn about the difference between correlation and
causation). At the same time, they show how different types of rights are interde-
pendent. A useful reminder for some governments tempted to sacrifice some rights
for the sake of others.
5.3. Meta-measurements
Another important distinction is between, on the one hand, measures of actual re-
spect or violations of human rights (de facto realization of rights), and, on the other
hand, what we could call meta-measures:
One can safely assume that bad score in these measurements are likely to result
sooner or later in bad scores for actual respect for human rights.
It's clear that not all indicators are directly linked to human rights, or, in other
words, try to measure human rights directly. Other aspects, which have an impact
on human rights, should also be measured: e.g. government efficiency, corruption,
GDP, income inequality, etc. The quantitative approach to human rights will there-
fore overlap somewhat with other fields of investigation, such as development, go-
vernance, crime etc.25 This is inevitable given the fact that many human rights vi-
olations are of an economic, criminal or political nature.
25
Human Rights Indicators, Purpose and Validity, Sano (2005), p. 8.
Take the example of the right to a certain standard of living or the right not to suf-
fer poverty. We can create a time series of the number of people (or the proportion
of a population) having a level of income which we deem to be sufficient (e.g. an
Page | 13 income more than $1 PPP a day), and evaluate if there is progress or not, and if
some countries (or regions within a country) progress faster than others (e.g. in the
form a colored map of a country or of the world). We can estimate, on the basis of
the time series, when a certain proportion of a given population will achieve a level
of income which we deem to be sufficient.
Subsequently, we can try to identify which other indicators seem to correlate with
low or high income (for example levels of corruption). These correlations can then
point towards a possible causal link, suggesting further investigation of the causes
of poverty. If this further investigation is successful in identifying a causal link,
then governments and international institutions can focus their anti-poverty policies
more successfully.
Apart from measuring poverty directly, we can measure to which extent a country
or countries have committed themselves to the fight against poverty, have created
and funded efficient institutions and policies etc. And we can measure popular sup-
port for this type of government action.
As stated before, we are dealing here with a highly immature field of research,
which is evident from the wide range of problems that we face. Some of those
problems have already been mentioned.
Data from different countries may also be incomparable due to differences in me-
thodologies or concepts and definitions (representativity of survey populations26,
question formulation issues etc.).
Some of the indicators which are doing the rounds are not transparent. It's not clear
how they are established.
Some ordinal scales show "variance truncation": due to a low number of gradations,
countries that are vastly different find themselves in the same score.
Sampling problems in the creation of survey data are quite common. Samples may
not be fully random. The Metagora project27 gives the following example. Suppose
we make a random sample of households in order to measure income and poverty,
but we omit to include homeless people in the sample because we do not have in-
formation about them (addresses etc.). Our results will be biased.
26
Report of the Eurostat Conference on Statistics and Human Rights (2002), p. 14.
27
The Metagora Project, http://www.metagora.org/html/index.html.
Statistics are often produced by government agencies, and governments are the
most common rights violators.28 Neutrality and objectivity may therefore be lack-
ing, and some data may be withhold or manipulated for political reasons, or simply
Page | 15 not collected.
Surveys may be biased because they tend to become less random when repression
or violence is more common. People de-select themselves from the survey because
they are afraid. Another reason for biased surveys is the possibility that questions
are understood differently in different cultures or countries, creating problems for
country comparisons.
There's also the "Paradox of human rights statistics": less information on rights vi-
olations may imply the existence of more violations. The difficulty of gathering in-
formation on rights violations is in itself an indicator of rights violations. There is
more information available about human rights violations in open societies because
they have a free press, freedom of entry and movement for NGOs and other moni-
toring bodies, etc.29 This imbalance in the availability of information can bias mea-
surements and country comparisons. Differences in scores may reflect differences
in information on rights violations rather than differences in rights violations as
such.
Measurements can be misused in order to prioritize and select certain human rights
policies, for example on the basis of the numbers of certain measurements. This
will ultimately fail because of the interdependence of human rights.
Given the importance of statistical and survey data, governments should invest in
their statistical apparatus, in data collection efforts (improving existing efforts and
collecting new data if necessary) and in codes of conduct and ethical standards for
statistical work. International institutions such as the UN, the IMF and others have
to assist them in this, and have a special responsibility regarding harmonization of
definitions and statistical procedures.
Governments, but also NGOs engaged in rights measurement, should strive towards
transparency and methodological accountability for all indicators, including trans-
parency about possible shortcomings in the data so as to avoid misinterpretation.
28
Report of the Eurostat Conference on Statistics and Human Rights (2002), p. 18.
29
The Quantitative Study of Human Rights Violations, Cingarelli & Richards (2007), p. 4.
This paper started from the premise that the quantitative approach to human rights
violations is as important as it is immature. The problems it faces are gigantic, as
Page | 16 are the investments necessary to improve and expand the available data. However,
these investments will pay themselves back many times over. I tried to offer a defi-
nition and a defense of the approach, citing its numerous theoretical advantages. At
the same time, I described the disadvantages and the existing shortcomings, as well
as the way to overcome them.
A final remark could be that it is not enough to measure human rights. Rights viola-
tions will not disappear when we know their magnitude. The measurements should
be used and should drive policies.
References
Andreassen & Sano (2004), What’s the Goal? What’s the Purpose? Observations
on Human Rights Impact Assessment,
http://www.humanrights.uio.no/forskning/publikasjoner/arkiv/rn/2004/0204.pdf
Asher, Banks & Scheuren (2007), Statistical Methods for Human Rights, Springer
Ball & Spirer, Making the Case, Investigating Large Scale Human Rights Viola-
tions Using Information Systems and Data Analysis, http://shr.aaas.org/mtc/
Benetech's Human Rights Data Analysis Group (HRDAG), http://www.hrdag.org/
Cingarelli & Richards (2007), The Quantitative Study of Human Rights Violations,
https://umdrive.memphis.edu/drich1/public/POLS%203320%20Spring%202008/Ci
ngranelli_%20Richards_Quantitative_Study_of_Human_Rights_Violations.pdf
CIRI Human Rights Data Project, http://ciri.binghamton.edu
Hafner-Burton & Ron (2009), Seeing Double: Human Rights Impact Through Qua-
litative and Quantitative Eyes, World Politics 61, no. 2,
http://papers.ssrn.com/sol3/Delivery.cfm/SSRN_ID1292431_code1003168.pdf?abs
tractid=1287858&mirid=1
Jabine & Claude (1992), Human Rights and Statistics, University of Pennsylvania
Press
Landman, Todd (2003), Map-Making and Analysis of the Main International Initia-
tives on Developing Indicators on Democracy and Good Governance,
http://web.abo.fi/instut/imr/research/seminars/indicators/final-report.pdf
Landman, Todd (2005), The Scope of Human Rights: From Background Concepts
to Indicators,
http://web.abo.fi/instut/imr/research/seminars/indicators/Background.pdf
Malhotra & Fasel (2005), Quantitative Human Rights Indicators, A Survey of Ma-
jor Initiatives, http://www.jus.uio.no/forskning/grupper/humrdev/Project-
Indicators/Workshop06/Background/Malhotra&Fasel.pdf
The Metagora Project, http://www.metagora.org/html/index.html
Report of the Eurostat Conference on Statistics and Human Rights in Brussels,
Belgium (2002),