The Strategic Plan hypothesized how the operational factors it chose to form performance indicators may affect the strategic factors, including perception.10 The Plan stated, “Strategic [factors] measure whether the Office is achieving its intended mandate under the Rome Statute. Operational [factors] measure whether the Office is implementing its strategy which in turn is assumed to impact positively on the strategic [factors].”11 The OTP seems to be implying with this statement that if the OTP positively improves its operations then perceptions of the OTP will also positively improve. While it does align with common sense that operating well means that others will perceive the OTP more positively, this assumption glossing over the realization made earlier in the Plan that strategic [factors] are not controlled by the OTP. External forces that might affect this relationship must be measures to determine the validity of that assumption.

This comment aims to tackle that issue by measuring international media bias in relation to the OTP and ICC. Media bias is an external force not under the OTP’s control that may affect perceptions regarding the OTP and the ICC more generally.12,13,14,15 It would therefore be useful to determine what, if any, media biases regarding the ICC exist and then try to determine how that affects perceptions. This task for one comment, however, must be constrained for pragmatic reasons. Therefore, this comment hopes to serve a few key purposes.

First, this comment proposes an ideal method that could be used in the future to quantify and measure international media bias. Second, this comment then shows how this proposed measure can be applied by using the international reaction to the ICC’s latest guilty verdict in The Prosecutor v. Ahmad Al Faqi Al Mahdi as a case study to determine whether international media biases exist and test hypotheses about how the case’s timeline and countries’ characteristics may affect media bias.

The hopes of this comment are that the results of these hypotheses and the numbers produced from the Al Mahdi case study may then serve as the first of two parts necessary to form a baseline measure of how operational factors interact with strategic factors including perception. If a future study goes back to the records of the Al Mahdi case to determine the operational factor measures and then combines that information with the information in this comment, then a baseline could be formed to compare with all future cases to determine, for example, whether an increase in operational factors does increase positive perceptions of the ICC.

II. Background: The Prosecutor v. Ahmad Al Faqi Al Mahdi

A. The Malian Conflict and Destruction of Mausoleums in Timbuktu

Mali became a state party of the ICC on August 16, 2000.16 Then on March 21, 2012, the Malian army overthrew the government because of its inability to deal with Tuareg rebels.17 Two Islamic insurgency groups, Ansar Dine and Al-Qaeda in the Islamic Maghreb (“AQIM”), took advantage of the situation.18 They managed to gain partial control over northern Mali and started implementing their version of sharia law in the area.19 The Malian government referred the situation to the ICC on July 13, 2012.20

From then until January 2013, Ansar Dine and AQIM imposed their religious and political edicts on the territory of Timbuktu and its people.21 They did so through a local government, which included an Islamic tribunal, an Islamic police force, a media commission, and a “morality brigade” called the Hesbah.22 Ahmad Al Faqi Al Mahdi was very active in aspects of the Ansar Dine and AQIM administration.23 He led the Hesbah from its creation in April 2012 until September 2012 and wrote a document on its role and objectives.24

In late June 2012, the leader of Ansar Dine—Iyad Ag Ghaly—made the decision to destroy ancient mausoleums and mosques that were historically, culturally, and religiously significant to the people of Timbuktu.25 Al Mahdi, in his capacity as the chief of the Hesbah, was given the instructions to fulfill the destruction of the mausoleums.26 Despite his initial reservations, Al Mahdi agreed to conduct the attack without hesitation on receipt of the instruction.27 Al Mahdi wrote a sermon dedicated to the destruction of the mausoleums, which was read at the Friday prayer at the launch of the attack.28 Al Mahdi then made the necessary logistical arrangements and supervised the execution of the operations by using Hesbah’s members and determining the sequence in which the buildings would be destroyed.29 He also justified the attack to the broader world through media interviews.30 Al Mahdi was present at all of the attack sites, giving instructions and moral support, and he personally participated in the attacks that led to the destruction of at least five sites.31

B. International Intervention and Al Madhi’s Arrest

In response to an official request by the Malian interim government for French military assistance and the United Nations Security Council Resolution 2085, the French military started Operation Serval on January 11, 2013.32 The aim of the operation was to oust Islamic militants including Ansar Dine and AQIM in the north of Mali, who had begun a push into the center of Mali.33 Three days later, on January 16, 2013, the OTP opened its investigation into war crimes in Mali.34 Al Mahdi attempted to flee to Niger most likely due to Operation Serval.35 However, Al Mahdi was later arrested in Niger in 2014 when French troops intercepted an arms-smuggling convoy.36,37

C. The Initial Appearance and Charges Confirmation

Multiple important events in the Al Mahdi case then happened in quick succession in late September 2015. The ICC filed an arrest warrant for Al Mahdi on September 18, 2015.38,39 A week later, on September 26, 2015, Al Mahdi was transferred to ICC custody by Niger authorities.40 Al Mahdi then made his first court appearance on September 30, 2015 for the initial appearance hearing.41 At the initial appearance hearing, a judge verified the identity of Al Mahdi and determined that Al Mahdi would be able to follow along with the court proceedings in Arabic.42

The confirmation of charges took place on March 1, 2016.43 Al Mahdi was accused, under Article 8(2)(e)(iv) of the Rome Statute, of:

Intentionally directing attacks against buildings dedicated to religion, education, art, science or charitable purposes, historic monuments, hospitals and places where the sick and wounded are collected, provided they are not military objectives.”44

Nine out of the ten sites that were destroyed in Timbuktu were designated as UNESCO World Heritage sites.45

On March 24, 2016, ICC Pre-Trial Chamber I confirmed against Al Mahdi the war crime charge regarding the destruction of historical and religious monuments in Timbuktu and committed Al Mahdi to trial before a Trial Chamber.46 This would be the first time an Islamic jihadist would be tried in front of the ICC.47

D. The Trial and Guilty Plea

The trial took place on August 22–24, 2016.48 At the opening of the trial, Al Mahdi admitted guilt as to the destruction of historical and religious monuments.49 This was the first time a defendant pleaded guilty at the ICC.50 The judges made sure that Al Mahdi realized he was giving away most of his rights by pleading guilty.51 The judges also emphasized that they had the right to decide his sentence, which could have resulted in a prison sentence of up to 30 years, despite the plea deal between the Prosecution and Defense which asked for nine to eleven years.52

After establishing that Al Mahdi understood these facts and made a knowing plea after consultation with his legal representatives, the Prosecution presented its evidence and called three witnesses.53,54 The legal representative of the victims and the Defense presented their remarks before the judges on August 24, 2016.55

E. Verdict and Sentence

On September 27, 2016, the ICC Trial Chamber declared Ahmad Al Faqi Al Mahdi guilty of the war crime of attacking historic and religious buildings in Timbuktu.56 The Chamber sentenced him to nine years in prison.57

To determine the nine-year sentence, the Chamber took into consideration the gravity of the crime, Al Mahdi’s culpable conduct, and his individual circumstances.58 The Chamber also considered five mitigating circumstances including Al Mahdi’s admission of guilt, his cooperation with the Prosecution despite potential security implications for his family in Mali, and his initial reluctance to commit the crime.59 The Chamber also noted that, even if inherently grave, crimes against property are generally of less gravity than crimes against persons.60

III. Methodology

A. Selection of Al Mahdi Case

This comment uses The Prosecutor v. Ahmad Al Faqi Al Mahdi as a case study to measure international media bias towards the OTP and ICC. The Al Mahdi case was selected for various reasons. The case is the most recent guilty verdict and successful prosecution of a war crime by the ICC. Therefore, it was more likely that material would available online for this case study. The Al Mahdi case was also a prosecution that did not involve an extensive trial. This would minimize the amount of variables that could affect the publications regarding the case. The articles written about the Al Mahdi case would be more focused on the ICC’s processes and the subject of the prosecution, rather than on the fairness or procedure of a trial.

B. Conception Framework for Measuring Media Bias

The conception framework for measuring media bias in this case study was developed from a combination of factors that were derived from the literature review and the OTP’s 2016–2018 Strategic Plan. From the initial literature review of news articles relating to the Al Mahdi case, it was clear that the three major themes most writers covered in their articles was the deterrent effect of the case, the proper priority of this case in relation to other court cases, and whether justice was truly delivered in this case. The last point fit neatly into a category the OTP recognized in its 2016–2018 Strategy Plan as important in shaping perceptions regarding the court: “Whether justice is seen to be done.”

The OTP also hypothesized that people’s views of the OTP’s professionalism, independence, impartiality, and objectiveness, would heavily impact their general view of the OTP. The OTP’s strategic factors and the literature review themes were combined to create a total of six “perception factors” that likely impact perception of the OTP: (1) deterrence; (2) priority; (3) justice; (4) professionalism; (5) independence; and (6) impartiality/objectivity. Whether the world news media included phrases relating to one of these six perception factors in a positive or negative context, or simply ignored it completely, was the basis for measuring media bias in this case study.

C. Units of Measurements

1. Dependent Variable: “News Presentation”

There were three potential units of measurement to measure media bias: numbers of news articles published, numbers of sentences published, and “news presentation.”

i. Reasons for rejecting numbers of news articles as the unit of measurement for media bias

Media bias could be measured by counting and classifying the number of news articles published into categories like positive, neutral, and negative. The issue that comes up with using this unit of measurement is that many news articles contain both positive and negative sentences. This means that news articles as a unit of measure would also have to include a fourth category, like mixed news articles, one news article might be counted twice, as both a positive and negative news article, or news articles would have to be counted as positive or negative through a determination of whether the news article was mostly positive or negative.

The problem with the creation of a fourth category like mixed news articles is that it does not help provide information regarding the actual state of media bias. Although it could be inferred that a mixed article means that the country has a balanced presentation of different viewpoints regarding the ICC, that is not necessarily true because a mixed article could theoretically contain one negative sentence and the remainder could be positive. Also, one of the goals of this comment is to help show media bias because of the common sense link between media bias and people’s actual perceptions.61 Categories like positive, neutral, and negative would help facilitate a better understanding how media bias affects people’s perceptions because the basic hypothesis would be that people would tend to follow the skew of their media. A mixed category would not as easily help determine how media bias may affect people’s perceptions because it is not on the logical sliding scale as the other three categories. Mixed media bias would require a closer look at the balance between the positive and negative statements in the articles to get a better sense of how it affects the public, and in even doing that, it may not be clear how the interaction of positive and negative statements in one article may affect other’s views.

The alternative to avoid the issues with a fourth category of mixed media bias is to count one mixed news article as both positive and negative. The first problem with this approach is that then it would be a misnomer to state that the unit of measurement is number of published news articles. The second problem with this approach is that then solely positive or negative articles are under counted. A mixed article would count twice, while positive and negative articles would only be counted once. The third problem with this approach appears when trying to show how media affects people’s perceptions. Solely positive and negative news article could have a clear affect on people’s perceptions. To include in mixed media as equivalent as each news article would assume that the mix of positive and negative information within one article has no affect.

A second alternative then would be to count mixed news articles as either only positive or negative through a determination of whether the news article was mostly positive or negative. This, however, brings up bigger issues with regards as to how to categorize an articles as mostly positive or negative. It could be that each sentence would be categorized as positive, neutral, or negative, and then the totals for each news article are added up to determine a majority skew. The problem with coding each sentence is that it would be very time consuming and also the result would likely be that every news article would end up being majority neutral. Also, imagine if one sentence in the news article was very negative towards the ICC, but then the news article contained two barely positive sentences. Under this approach the news article would be categorized as positive, despite the likely result that people would start forming a negative perception of the ICC after reading the article.

Due to the issues relating to counting mixed media articles, the numbers of articles published was rejected as the unit of measurement for media bias.

ii. Reasons for rejecting numbers of sentences published as the unit of measurement for media bias

The second potential unit of measurement for media bias was counting and categorizing the sentences published in each news article. This unit was rejected because it would be very time consuming. The other issue with this unit of measurement is that sometimes sentences could convey both a positive and negative sentiment. The issues arising with how to categorize these mixed sentences would be the same as the issue arising with how to categorized mixed articles if the unit of measurement was number of published news articles.

iii. Reasons for using news presentation as the unit of measurement for media bias

The unit of measurement this comment uses to measure media bias is called “news presentation.” This comment uses news presentation as a technical term. News presentation is shorthand for the counting and categorization of the presentation of ideas within a news article. News presentation is a unit of measurement where the amounts of times a news article introduced an idea relating to the six perception factors would be counted and categorized as either positive or negative. For example, if a news article mentioned that the court had done justice in this case but that it was not a priority, that article would count as one positive news presentation based on the justice factor and one negative news presentation based on the priority factor. It is important to note that the measure was the introduction of ideas, and not how many times that idea was mentioned throughout the article.62 Otherwise, this would be similar to simply codifying each sentence. Under this unit of measurement, it was theoretically possible for one news article to count as six positive “news presentations” because of six positive perception factor mentions and six negative “news presentations” for six negative perception factor mentions.

If a news article did not introduce any positive or negative ideas regarding the six perception factors, then that article was counted as one neutral “news presentation.” This would mean that positive and negative news articles could have a higher value compared to neutral news articles. However, this is not a big concern because by using news presentation as a measure, it gets to a more core measure of how articles might affect people’s perceptions. For example, if one article has multiple positive news presentations, then it is more likely that a person reading this article will start forming a stronger positive perception of the ICC, than compared to reading a news article with only one positive presentation.

2. Independent Variables

Multiple independent variables were proposed as potential explanations for positive, neutral, or negative news presentations in the international media. This study was able to measure how the following independent variables could affect or correlate with the presentation of information in the world’s news articles:

i. Case Timeline

The first variable that was established was the timeline of the case. The descriptions regarding the Al Mahdi case could change as the case unfolded. Thus, this study categorized news articles into three separate timespans based on significant developments in the Al Mahdi case. The first timespan (“Time 1”) starts on September 18, 2015—the day the ICC issued an arrest warrant for Al Mahdi. Time 1 includes the events that occurred in quick succession during September 2015, including the transfer of Al Mahdi to ICC custody and his first court appearance. The second timespan (“Time 2”) starts on March 24, 2016—the day the ICC had its preliminary hearing for the case and when it was first reported that Al Mahdi might plead guilty to the charges. Time 2 also includes the week-long trial that happened on August 22–24, 2016, where Al Mahdi did plead guilty to the charges. The third timespan (“Time 3”) starts on September 27, 2016—the day the verdict and nine-year sentence were announced by Chambers. Time 3 spans all the way to the present of when data collection began on November 22, 2016.

ii. Government Structures

The second variable was the government structure of the different countries. A 2015 study by Matthew A. Baum of Harvard Kennedy School and Yuri M. Zhukov of the University of Michigan found that the difference between democratic and authoritarian governments affected the way the media in that country reported on international armed conflicts.63,64 Similarly, a country’s government may affect the way the media in that country reports about the ICC and OTP. This study used the Democracy Index compiled by the UK based Economist Intelligence Unit as a way to categorize the government structure of different countries.65 The Democracy Index categorizes countries as being a (1) full democracy; (2) flawed democracy; (3) hybrid regimes; or (4) authoritarian.66

iii. Muslim-Majority Nations

The third variable was whether a country had a majority Muslim population. The Al Mahdi case represented the first Muslim person to be prosecuted by the OTP. A country with a Muslim majority population could potentially portray the case differently than a Muslim minority country. This study, therefore, separated the two countries on a simple majority. If a country had a Muslim population of 50% of more, then it was a Muslim majority country, and vice-versa. The percentage of Muslim population in one country was based on continually updated World Factbook compiled by the United States Central Intelligence Agency.67

iv. ICC States Parties

The fourth variable was whether a country was a state party to the Rome Statute. Countries that were states parties at the time could have reacted differently to the countries that were not states parties, especially in terms of the deterrent effect of the case. Thus, this study separated countries into states parties and non-states parties at the time of the verdict announcement on September 27, 2016.68 Countries that were signatories, but had not ratified the Rome Statute were considered to be non-states parties.

v. Africa Bias

The fifth variable was whether a country was from Africa or not. Since the ICC’s inception, it has been accused of being biased against African countries.69 The issue has become more prominent since Burundi, South Africa, and Gambia announced their intentions to withdraw from the ICC.70 African countries could therefore have reported more negatively on the Al Mahdi case to reinforce negative views it had regarding the ICC. The Al Mahdi case, however, could also serve as a counter point to the perception of ICC’s Africa bias because the cooperation of two African nations, Mali and Niger, was necessary to prosecute Al Mahdi. Thus, the articles were also separated by geographic lines between African countries and non-African countries.

D. Coding Lexis Newspaper Database

In order to collect data on media bias in relation to the Al Mahdi case, news articles from Lexis’ Newspaper Database71 were coded to transform qualitative information into quantitative information. Coding is the process of counting the amount of times qualitative information falls into a distinct category. Prior studies have proposed coding as a method of measuring media bias.72

In order to retrieve the relevant news articles, each country was individually searched using a broad terms-and-connectors script.73 The broad, initial script ensured that news articles in English and French would appear in the results. At the same time, the initial script excluded irrelevant news articles that appeared often in the initial literature review like stories relating to Sudanese president Sadiq Al-Mahdi.

Each news article that resulted from the initial search was read and whether an article introduced a positive or negative idea relating to the perception factors was recorded. If an article included more than one positive or negative phrase relating to the same perception factor, then those additional mentions were not counted. If an article did not mention anything positive or negative, then that article would be categorized as neutral.

A news article relating to the Al Mahdi case would introduce a positive idea about the case if it included a sentence or phrase that talked about one of the six perception factors in a positive light. For example, an article with the phrase, “The first ICC prosecution solely for tearing down monuments will deter other wreckers, such as the Islamic State of Iraq and the Levant (ISIL) group,” would be categorized as a positive mention in the article because it fits under the first perception factor of deterrence. Phrases in articles with synonymous meanings were also counted as a positive inclusion. For example, and article with the phrase, “Archaeologists hope, in light of the prevalent war on art, that the trial will send a hard-hitting message to groups that wrecking culturally cherished artefacts will not go unpunished,” would also be categorized as one positive mention for the same reason. For countries with large amounts of news articles, a second, more narrow script was created that could be used to “focus” search the articles in order to highlight key terms relating to the perception factors. These terms are included in Amendment 1 and demonstrate the different synonyms that were accepted as indicators of the six perception factors.

A news article relating to the Al Mahdi case would include a negative phrase if it mentioned the case in a negative context under one of the six perception factors. For example, an article that introduced the idea that the Al Mahdi case would have little deterrent effect would count as one negative news presentation because it was a negative idea relating to the deterrent perception factor. Another example would be an article that argued Al Mahdi should have been charged more harshly would be categorized as one negative mention in an article under the justice perception factor.

An article that did not include any phrase relating to the six perception factors was categorized as a neutral article. Usually, neutral articles would be shorter articles with little detail about the case.

E. Regression Analysis

1. Hypothesis developed for regression

The aim of this comment was to measure the media bias of each country and try to determine how different aspects of each country might affect or correlate that country’s media bias regarding the reports published about the Al Mahdi case. As explained in the methodology section, the news presentations of each country was categorized as positive, neutral, or negative. Before using regression to compare the different country’s characteristics, a hypothesis for how independent variable might affect the news presentations was developed.

My first hypothesis is that as time progressed, the reports about the case would become more polarized, meaning the reports would become more positive and more negative and less neutral. The reason for this hypothesis was that countries would likely form stronger opinions regarding the effects and outcome of the case as it progressed. For example, some countries would be positive that a guilty verdict was reached, while other countries may be more negative after it was announced that the sentence would be nine years instead of the maximum of 30 years.

My second hypothesis is that democratic government would be more positive or negative than authoritarian governments. In other words, a full democracy would be less neutral than an authoritarian government. The reason for this hypothesis was that under democratic governments, media corporations tend to have more editorial freedom.74 This would mean that news articles could contain more diverse opinions about the ICC and the Al Mahdi case compared to the news outlets in authoritarian regimes.

My third hypothesis is that Muslim countries would report the case more negatively or neutrally than non-Muslim countries. The Al Mahdi case represented the first time a Muslim man was prosecuted by the ICC. Muslim majority nations, similarly to African nations, could react more negatively to this prosecution by claiming this was a form of western suppression.

My fourth hypothesis is that ICC state parties would report the case more positively than non-state parties. The simple logic behind this hypothesis is that the very fact countries are states parties means they are supportive of the ICC’s efforts. Therefore, states parties are more likely to see the Al Mahdi case positively because it resulted in a rare guilty verdict.

My last hypothesis is that African countries would be more neutral or negative in their reports of the case, compared to Non-African countries. The Africa bias issue that has plagued the ICC was the reason for this hypothesis. Many African countries are supportive of the ICC. For example, Niger in the Al Mahdi case cooperated with the ICC by transferring Al Mahdi to ICC custody. However, the Africa bias issue has always caught constant negative criticism in media and therefore the hypothesis is that this may have permeated onto other coverage regarding the ICC.

2. Regressions

For the first two hypotheses, a logistical regression or logit model was used to determine how time and governance structure correlated with the polarization of media bias (moving away from neutrality and towards a more positive and negative news presentation). The logit model was selected because the first two hypotheses dealt with dichotomous outcome (whether the news presentations were more or less neutral). In the logit model, the log odds of the outcome were modeled as a linear combination of the predictor variables. After time and government structure were processed individually, the two variables were compiled into one logit model to ensure that the results would remain consistent and determine whether the statistical significance would vary at all.

For the three remaining hypotheses, an ordered logistical regression or ordered logit model was used to determine how much explanatory power the country’s characteristics as being Muslim, an ICC states party, and African, had on the positive, neutral, or negative news presentation. The ordered logit model was selected because these hypotheses dealt with ordered discrete categorical variables. Our response variable, news presentation, was treated as ordinal under the assumption that the levels of news presentation status have a natural ordering (high to low), but the distances between adjacent levels are unknown. One important thing to note is that for the ordered logit model the values for the news presentations were positive = 1, neutral = 2, and negative = 3. Therefore, the higher the coefficient in the ordered logit model, the more media bias was skewed towards negative. After the variables of being Muslim, states parties, and African were processed individually, the three variables were inputted into one ordered logit model to ensure that the results would remain consistent and determine whether the statistical significance would vary at all.

F. Methodology Challenges

The single biggest challenge to this case study is the language barrier. News articles were coded in English and French, the national languages of the ICC. However, other languages were excluded. For example, Germany’s news articles were mostly excluded because they were in German. The language barrier likely means that Western media influence is more heavily represented than other nations’ media. This is an unfortunate yet unavoidable issue.

Western media may be over represented in the study also because the Lexis database may not contain the widest array of international newspaper. Although the Lexis database contains 119 countries, many of these countries consist of few newspapers that are archived by the database. In some cases, like Estonia, the single newspaper for that country is a European wide circulation of the BBC. These types of wide ranging new sources were excluded if it did not have a specific, targeted message to one country, because then the measure of a single country’s media bias would be distorted by media from other countries and publishers.

The other issue with this study is the lack of controlling variables. There are many variables besides the ones identified in the hypotheses, like a country’s media norms, the effect of libel laws on news publications, and incentive to write eye-catching stories to name a few, that were not measured in this case. These variables were not controlled for in the regression analysis.

IV. Findings

A. News Publications by Country

A total of 40 countries, out of a potential 119 countries in the Lexis Newspaper Database, had at least one news article about the Al Mahdi case. A majority of the articles found came from large western-influenced countries like Australia, Canada, France, United Kingdom, and United States of America. The two notable examples of non-western countries with large amounts of news coverage of the Al Mahdi case are Pakistan and India.

Table 1. National News Presentations

Countries	Positive	Neutral	Negative
Australia	24	20	1
Bangladesh	0	3	0
Canada	45	6	10
China	9	5	2
Czech Republic	0	1	0
Denmark	1	0	0
Egypt	0	1	0
France	48	73	41
Greece	0	1	0
Hong Kong	10	1	2
India	25	20	3
Iran	1	0	0
Ireland	5	8	2
Israel	0	3	1
Italy	0	1	2
Jordan	2	7	4
Kenya	2	3	1
Lebanon	1	1	0
Malaysia	0	2	1
Mexico	0	0	1
Morocco	0	1	0
Netherlands	8	14	3
New Zealand	0	5	3
Nigeria	1	0	0
Oman	1	0	0
Pakistan	44	26	5
Philippines	2	1	1
Portugal	0	1	0
Russia	2	1	0
Saudi Arabia	2	1	0
Slovakia	0	1	0
South Africa	10	13	6
South Korea	1	0	1
Spain	0	7	0
Switzerland	3	40	3
Thailand	3	3	2
Tunisia	0	1	0
United Arab Emirates	1	0	0
United Kingdom	57	164	15
United States of America	49	48	11
Total	357	483	121

B. Media Bias Relating to The Perception Factors

50.26% of the total news articles were neutral.

There was a total of 357 positive news presentations relating to the perception factors within the news articles. A majority of the positive mentions in news articles related to the potential deterrent effect of the Al Mahdi case. 228 news presentations, or 68.87% of total positive news presentations, mentioned that the Al Mahdi case would have a positive deterrent effect against similar actors like ISIL. The second most common theme in positive news presentations was the justice factor. 102 news presentations, or 28.57% of total positive news presentations, mentioned that justice was done or that the Al Mahdi case was a positive step towards establishing accountability for the actions in Mali. The other four factors accounted for the remaining 2.56% of positive presentations in articles.

A total of 121 negative news presentations relating to the perception factors were found within the news articles. The biggest negative theme in news presentations related to the justice factor. 78 news presentations, or 64.46% of total negative news presentations, mentioned that the Al Mahdi case was a disappointment because the case did not provide complete justice to the people of Mali. These articles argue that Al Mahdi should have been charged for committing and facilitating murder, enforced disappearances, rape and other sexual violence committed by Ansar Dine and AQIM while he was in charge of the Hesbah. 22 news presentations, or 18.18% of total negative news presentations, mentioned that the Al Mahdi case would not have a deterrent effect. These articles focused mainly on the fact that Syria and Iraq were not states parties to the Rome Statute and therefore the ability of the OTP to prosecute ISIL for the destruction of cultural sites in Aleppo and Palmyra was severely limited. 20 news presentations, or 16.53% of total negative news presentations, mentioned that the OTP was not exercising proper judgment in prioritizing this case over other issues facing the ICC. The negative priority news presentations also argued that prosecuting the Al Mahdi case weakened the importance of the ICC as a center for prosecuting the most heinous of crimes against humanity. There was only one other negative news presentation, from Malaysia, that cast a negative light on the professionalism of the ICC by stating, “Cultural destruction, unfortunately, was not clearly defined in the court documents in The Hague.”

C. News Presentations Over Time

As time went on, the articles became relatively more positive and less neutral when looking at a simple bivariate relationship. The percentage of negative articles remained relatively the same over time.

Table 2. Difference in News Presentation Over Time

Case Timeline	Positive	Neutral	Negative	Total
Time 1	46	113	30	189
Time 2	139	261	45	445
Time 3	172	109	46	327

A total of 189 articles were published in Time 1; the time between the ICC’s arrest warrant of Al Mahdi was announced and the day before Al Mahdi was reportedly going to plead guilty. Time 1 includes the events that occurred in quick succession during September 2015, including the transfer of Al Mahdi to ICC custody and his first court appearance. During Time 1, a majority of news articles were neutral. 59.79% of news presentations were neutral, 24.34% of news presentations were positive, and 15.87% of news presentations were negative.

A total of 445 articles were published in Time 2; the time between the first preliminary hearing for the case when it was reported that Al Mahdi might plead guilty to the charges and the day before the nine-year prison sentence was announced. Time 2 also includes the week-long trial that happened on August 22–24, 2016, where Al Mahdi did plead guilty to the charges. During Time 2, a majority of news presentations were also neutral. 58.65% of news presentations were neutral, 31.24% of news presentations were positive, and 10.11% of news presentations were negative.

A total of 327 articles were published in Time 3; the time between the announcement of the verdict and nine-year sentence and the day data collection began on November 22, 2016. During Time 3, a majority of news presentations were positive. 52.60% of news presentations were positive, 33.33% of news presentations were neutral, and 14.07% of news presentations were negative.

The results from the logit model taking time as a variable further confirmed the trend that countries became more polarized as time went on. In order to first run the logit regression, the news presentation had to be divided between neutral (value = 1) and non-neutral (value = 0) news presentations.

Table 3. Separation of News Presentation to Neutral and Non-Neutral

News Presentation	Frequency	Percentage
Non-Neutral (0)	478	49.74
Neutral (1)	483	50.26
Total	961	100.00

The logit model then produced the following results:

Table 4. Logit Model for Correlation Between Time and News Presentation Polarization [75]

Neutrality	Coef.	Std. Err.	z	P>|z|	[95% Conf. Interval]
Time 1	—	—	—	—	—	—
Time 2	−.0339977	.1763598	−0.19	0.847	−.3796566	.3116611
Time 3	−1.072132	.1887304	−5.68	0.000	−1.442037	−.702227
Constant	.3835824	.147772	2.60	0.009	.0939546	.6732102

The regression demonstrates that there was a general trend towards decreasing neutrality. The constant demonstrates the expected log-odds of neutrality if time was not used as a predictor variable. Therefore, normally a neutral news presentation would be expected (with a positive value of .384). But, when time is taking into account, the expected log-odds of neutrality decreases by .034 between Time 1 and Time 2; and further decreases by 1.07 between Time 1 and Time 3. This means there was a large spike in polarization after the sentence and verdict were announced in Time 3. The decreasing amount of neutrality between Time 1 and Time 3 is statistically significant (at P>|z| = 0.000). The high statistical significant in polarization between Time 1 and Time 3 serves as strong evidence to support the first hypothesis that countries would become less neutral as the Al Mahdi case progressed.

D. News Presentation Between Different Country Governance Structures

When looking at bivariate relationships, three trends were observed if the news presentations of the Al Mahdi case were analyzed through the lens of country governance structures. The first trend, unsurprisingly, is that more media was available online for democratic countries. The second trend, less intuitively, is that the media bias becomes more positive as the country analyzed is scored as less democratic. The third trend is that general media bias becomes more prominent, meaning the news presentation is less neutral, as countries are scored as less democratic.

Table 5. Difference in News Presentation Between Different Country Governance Structures

Governance Structures	Positive	Neutral	Negative	Total
Full Democracy	192	312	48	552
Flawed Democracy	96	119	59	274
Hybrid Regime	51	37	8	96
Authoritarian	18	15	6	39

552 articles were published by countries categorized as full democracies. 34.78% of the news presentations were positive, 56.52% of the news presentations were neutral, and 8.70% of the news presentations were negative.

274 articles were published by flawed democracies, less than half as many articles as full democracies. 35.04% of the news presentations were positive. The transition between full democracies and flawed democracies hinted at the trend of increasingly positive news presentations. 43.43% of the news presentations were neutral. Neutrality went down by 13.09% between full democracies and flawed democracies. This was the biggest reduction in neutrality between two groups. 21.53% of the news presentations were negative.

96 articles were published by hybrid regimes, around a third as much as flawed democracies. The positive news presentations jumped up to 53.13% and became a majority of the articles published. 38.54% of the news presentations were neutral, and 8.33% of the news presentations were negative.

Authoritarian governments published only 39 articles, around a third as many articles as hybrid regimes. The positive news presentations dipped to 46.15%, still higher than both full or flawed democracies. Neutrality continued to fall, reaching its lowest levels at 38.46% and negative news presentations continued to fluctuate back up to 15.38%.

The logit model for the correlation between neutrality and governance structure also confirmed the patterns observed through the bivariate relationships.

Table 6. Logit Model for Correlation Between Governance Structure and Polarization [76]

Neutrality	Coef.	Std. Err.	z	P>|z|	[95% Conf. Interval]
Full Democracy	—	—	—	—	—	—
Flawed Democracy	−.5266659	.1490864	−3.53	0.000	−.81887	−.234462
Hybrid Regime	−.7289838	.2266009	−3.22	0.001	−1.173113	−.284854
Authoritarian	−.7323679	.3401546	−2.15	0.031	−1.399059	−.065677
Constant	.2623643	.0858592	3.06	0.002	.0940834	.4306451

As countries were scored as being less democratic, the countries’ news presentations became less neutral. Flawed democracies were .527 less neutral than full democracies. Hybrid Regimes and Authoritarian governments were .729 and .732 less neutral than full democracies respectively. This increase in polarization between full democracies and every other governance structure was statistically significant.

This is strong evidence against the hypothesis that democratic countries would be less neutral and authoritarian countries would be more neutral. It may be that since authoritarian governments have more control over media production, these countries may be able to better coordinate a single narrative. This evidence shows that this narrative is not neutral however, and may be positive or negative depending on government’s views of the Al Mahdi case and the ICC.

E. Logit Model with Time and Governance Structure

In order to double check the values of the two individual time and governance structure logit models, the two variables were used in the same model to see if any statistical significance changed and how the two variables might interact.

Table 7. Logit Model for Correlation Between Polarization and Time and Governance Structure [77]

Neutrality	Coef.	Std. Err.	z	P>|z|	[95% Conf. Interval]
Time 1	—	—	—	—	—	—
Time 2	−.0927473	.1787894	−0.52	0.604	−.4431682	.2576736
Time 3	−1.058905	.1904655	−5.56	0.000	−1.432211	−.6856995
Full Democracy	—	—	—	—	—	—
Flawed Democracy	−.4555369	.1534172	−2.97	0.003	−.7562291	−.1548447
Hybrid Regime	−.5621364	.2351827	−2.39	0.017	−1.023086	−.1011868
Authoritarian	−.70866	.349579	−2.03	0.043	−1.393822	−.0234977
Constant	.6204293	.1624221	3.82	0.000	.3020879	.9387708

Everything that was statistically significant in the two prior models remained statistically significant in the combined model. The time coefficients remained relatively the same. The governance coefficients shifted upwards very slightly. It seems that the timeline of the case explains somewhat better the reduction in news presentation neutrality because the three government coefficients shows that there was now less neutrality reduction between full democracies and the other types of governments.

F. News Presentation Between Muslim Majority and Muslim Minority Countries

The Al Mahdi case was the first time a Muslim was tried in the ICC. At first impression from the bivariate relationships, the Al Mahdi case did not result in any clearly visible division in news presentations between countries with a majority Muslim population and countries with a minority Muslim population.

Table 8. Difference in News Presentation Between Muslim Majority and Minority

Muslim Population	Positive	Neutral	Negative	Total
Muslim Majority	53	43	10	106
Muslim Minority	304	440	111	855

The lack of English or French publications from Muslim majority countries likely limited the useful data for this independent variable. From the 106 news presentations in Muslim majority countries, 50% were positive, 40.57% were neutral, and 9.43% were negative. From the 855 news presentations in Muslim minority countries, 35.56% were positive, 51.64% were neutral, and 12.98% were negative.

To better understand the relationship between Muslim population and news presentation, the information from Table 8 was used to make an ordered logit model.

Table 9. Ordered Logit Model for Correlation Between News Presentation and Muslim Majority Countries [78]

News Presentation	Coef.	Std. Err.	z	P>|z|	[95% Conf. Interval]
Muslim Majority	.5571664	.2007754	2.78	0.006	.1636539	.9506788
Cut 1	−.0320557	.1898912	—	—	−.4042357	.3401242
Cut 2	2.444726	.208414	—	—	2.036243	2.85321

As mentioned before in the methodology section, the response variable, news presentation, is treated as ordinal under the assumption that the levels of news presentation status have a natural ordering (positive to negative), but the distances between adjacent levels are unknown. Cut 1 and Cut 2 serves as ancillary parameters for this model, meaning that Cut 1 and Cut 2 are thresholds used to differentiate the adjacent levels of news presentation. For the purpose of this model, news presentation is clearly positive if it is-.03 or lower, and negative if it is 2.44 and above. This provides a compass for interpreting this table, although it is a bit counter-intuitive. If the coefficient is a negative number, then it means that Muslim majority countries have a positive news presentation media bias. If the coefficient is a positive number, then it means that Muslim majority countries have a negative media bias.

Standard interpretation of the ordered logit coefficient is that for a one-unit increase in the predictor (in this case whether the country is Muslim), the response variable level (news presentation) is expected to change by its respective regression coefficient in the ordered log-odds scale while the other variables in the model are held constant. This means that news presentations grows increasingly negative by .56 when a country has a majority Muslim population. This increase in negative news presentation when a country is majority Muslim is statistically significant. This, then, supports the hypothesis that Muslim majority countries would report about the Al Mahdi case and, potentially in extension, the ICC more negatively.

G. News Presentation Between ICC States Parties and Non-States Parties

First impression from the bivariate relationships showed that there were no clear differences between ICC states parties and non-states parties. Out of the 626 news presentations published by states parties, the news presentations were 32.91% positive, 52.56% neutral, and 14.54% negative. Similarly, out of the 335 presentations published by non-states parties, the news presentations were 45.07% positive, 45.95% neutral, and 8.95% negative.

Table 10. Difference in News Presentation Between ICC States Parties and Non-States Parties

ICC States Parties	Positive	Neutral	Negative	Total
States Parties	206	329	91	626
Non-States Parties	151	154	30	335

The ordered logit model provides more useful information regarding the relationship between ICC states parties and news presentation.

Table 11. Ordered Logit Model for Correlation Between News Presentation and States Parties [79]

News Presentation	Coef.	Std. Err.	z	P>|z|	[95% Conf. Interval]
ICC States Parties	−.5215395	.1307544	−3.99	0.000	−.7778135	−.2652656
Cut 1	−.7152676	.0827983	—	—	−.8775493	−.5529858
Cut 2	1.777959	.1043912	—	—	1.573356	1.982562

The ICC states parties coefficient demonstrates that countries that are ICC states parties have a positive news presentation media bias in how they reported on the Al Mahdi case. This trend of having a positive news media bias was statistically significant. Thus, supporting the hypothesis that countries that are ICC states parties will report more positively on ICC cases because they intrinsically support the mission of the Court.

H. News Presentation Between African and Non-African Countries

There were no clear differences between African and Non-African countries. Out of the 39 news presentation by African countries, the presentations were 33.33% positive, 48.71% neutral, and 17.95% negative. Similarly, out of the 922 presentations published by non-states parties, the news presentations were 37.31% positive, 50.32% neutral, and 12.36% negative. There was no more than a 5.59% difference in the media bias measures between African and Non-African countries.

Table 12. Difference in News Presentation Between African and Non-African Countries

News Presentation	Positive	Neutral	Negative	Total
African	13	19	7	39
Non-African	344	464	114	922

Unfortunately, the ordered logit model for African countries does not provide further insights as well.

Table 13. Ordered Logit Model for Correlation Between News Presentation and African Countries [80]

News Presentation	Coef.	Std. Err.	z	P>|z|	[95% Conf. Interval]
African Countries	−.2632453	.3175755	−0.83	0.407	−.8856818	.3591913
Cut 1	−.7790708	.3129516	—	—	−1.392445	−.1656969
Cut 2	1.685773	.3183466	—	—	1.061825	2.309721

The coefficient would seem to suggest that African countries have a positive news presentation media bias. However, this coefficient is not statistically significant. The small sample size of African countries’ news presentations likely created this result. Therefore, the hypothesis as to whether African countries would have a negative media bias against the court is inclusive, and if anything rebutted by the ordered logit model.

I. Ordered Logit Model with Muslim, ICC States Parties, and African Variables Combined

The ordered discrete categorical variables were combined into one ordered logit model in order to double check the validity of the previous models to see if any statistical significance changed and how the two variables might interact.

Table 14. Ordered Logit Model for Correlation Between News Presentation and Muslim Countries, ICC States Parties, and African Countries [81]

News Presentation	Coef.	Std. Err.	z	P>|z|	[95% Conf. Interval]
Muslim Majority	.3163451	.2148801	1.47	0.141	−.1048121	.7375023
ICC States Parties	−.4413296	.1407542	−3.14	0.002	−.7172027	−.1654565
African Countries	−.1552474	.3217726	−0.48	0.629	−.7859102	.4754154
Cut 1	−.5552214	.3698602	—	—	−1.280134	.1696913
Cut 2	1.941364	.3757532	—	—	1.204901	2.677827

The combined ordered logit model shows that the Muslim majority coefficient, that was previously statistically significant, is no longer statistically significant. It is possible that the interaction between the ICC states parties variable and the Muslim variable might have led to this result. It is unclear how the interaction between these two variables lead to this result. Future research may be able to shed light on this.

V. Conclusions

This comment proposes coding news presentations as an ideal method that could be used in the future to quantify and measure international media bias. Although news presentations do not escape for some issues that plague other units of measurement, news presentations best reduce the issues associated with coding news articles while still being practical unlike coding sentences published. News presentations also serves as a stronger core measure of how media bias may affect people’s perceptions by focusing on the number of ideas introduced as the central measure of media bias.

This comment then shows how news presentations can be measured and applied by using the international reaction to the ICC’s latest guilty verdict in The Prosecutor v. Ahmad Al Faqi Al Mahdi as a case study to determine whether international media biases exist and test hypotheses about how the case’s timeline and countries’ characteristics may affect media bias. Looking at the raw numbers of news presentations, each country seemed to exhibit some type of media bias. A few countries had entirely neutral news presentations, but these few countries only had one news presentation. Thus, in order to determine how the case’s timeline and countries’ characteristics may affect media bias, this comment looked at bivariate relationships and created logit and ordered logit models for five independent variables.

The hypothesis that as time progressed news presentations would become more polarized, meaning more positive and negative, and less neutral, was confirmed by the bivariate relationship and the logit model. The decrease in neutrality between Time 1 and Time 3 was statistically significant in both the single variable and combined variable logit model. The second hypothesis that democratic countries would be less neutral was debunked. Instead, the bivariate relationship and logit model showed that authoritarian countries would be less neutral. The decrease in neutrality between full democracies and every other type of governance structure was statistically significant in both the single variable and combined variable logit models.

The third hypothesis that Muslim majority countries would report more neutrally or negatively about the Al Mahdi case were inconclusive based on the bivariate relationship and ordered logit models. For the single variable ordered logit model, Muslim majority countries had an overall negative media bias that was statistically significant. However, this statistically significant result disappeared when other variables were introduced into an ordered logit model. The fourth hypothesis that ICC states parties would report the Al Mahdi case more positively was confirmed by the ordered logit models. Both the single variable and combined variables models showed that ICC states parties had an overall positive media bias that was statistically significant. The last hypothesis that African countries would report about the Al Mahdi case more negatively was also inconclusive since no clear trends appeared on the bivariate relationship chart and the ordered logit model was not statistically significant.

These results show that the OTP in the Strategic Plan was right that perceptions are not controlled by the OTP, but are likely affected by outside forces like media bias. The hope now is that the effects for time, governance structure, and being an ICC states party can be controlled for in future studies to establish a better understanding of how the ICC’s operations affect people’s perceptions. In addition, future studies may be able to complete the measures of performance indicators laid out in the Strategic Plan for the Al Mahdi case and pair those results with the results of this case study to then compare with future cases to see if improvements in performance indicators correlate with an increase in positive news presentations.

Table 15. Focus Search Terms & Connectors

Perception Factors	Positive	Neutral	Negative
General	(“welcome” or “welcomes” or “commends” or “commend” or “positive” or “ground-breaking” or “landmark”)
Deterrence	(“deter” or “deterrence” or “deterrent” or “discourage” or “dissuade” or “inhibit” or “repel” or “chill” or “hinder” or “hindrance” or “holdback” or “restrain” or “restraint” or “stop” or “obstruct” or “fetter” or “rein” or “saving the past” or “new step” or “message” or “ISIS” or “ISIL” or “precedent”) and no ((“not” or “no” or “little” or “minimal” or “few” or “ineffective” or “useless” or “nominal” or “negligible” or “unsuccessful” or “fruitless” or “unproductive” or “futile” or “purposeless” or “inefficient” or “inefficacious” or “inadequate” or “feeble” or “inept” or “impotent” or “weak” or “poor” or “scant” or “meager” or “thin” or “small” or “negative” or “oppose” or “dissent” or “harmful” or “damaging” or “unfavorable” or “disadvantageous” or “bad” or “adverse” or “detriment” or “detrimental” or “dismissive”) w/3 (“deter” or “deterrence” or “deterrent” or “threaten” or “discourage” or “dissuade” or “inhibit” or “repel” or “chill” or “hinder” or “hindrance” or “holdback” or “restrain” or “restraint” or “stop” or “obstruct” or “fetter” or “rein” or “saving the past” or “new step” or “message”))	No (“deter” or “deterrence” or “deterrent” or “threaten” or “discourage” or “dissuade” or “inhibit” or “repel” or “chill” or “hinder” or “hindrance” or “holdback” or “restrain” or “restraint” or “stop” or “obstruct” or “fetter” or “rein” or “saving the past” or “new step” or “message”)	((“not” or “no” or “little” or “minimal” or “few” or “ineffective” or “useless” or “nominal” or “negligible” or “unsuccessful” or “fruitless” or “unproductive” or “futile” or “purposeless” or “inefficient” or “inefficacious” or “inadequate” or “feeble” or “inept” or “impotent” or “weak” or “poor” or “scant” or “meager” or “thin” or “small” or “negative” or “oppose” or “dissent” or “harmful” or “damaging” or “unfavorable” or “disadvantageous” or “bad” or “adverse” or “detriment” or “detrimental” or “dismissive”) w/3 (“deter” or “deterrence” or “deterrent” or “threaten” or “discourage” or “dissuade” or “inhibit” or “repel” or “chill” or “hinder” or “hindrance” or “holdback” or “restrain” or “restraint” or “stop” or “obstruct” or “fetter” or “rein” or “saving the past” or “new step” or “message”))
Priority	OR (“inseparable” or “priority” or “arrangement” or “preference” or “antecedence” or “order” or “preeminence” or “prerogative” or “important”) OR ((“correct” or “right” or “appropriate” or “justified” or “right” or “spot-on” or “valid” or “legitimate” or “perfect” or “infallible” or “sound” or “apt” or “suitable” or “acceptable” or “fitting”) w/3 (“priority” or “arrangement” or “preference” or “antecedence” or “order” or “preeminence” or “prerogative” or “important”)) AND NO ((“not” or “no” or “little” or “minimal” or “few” or “ineffective” or “useless” or “nominal” or “negligible” or “unsuccessful” or “fruitless” or “unproductive” or “futile” or “purposeless” or “inefficient” or “inefficacious” or “inadequate” or “feeble” or “inept” or “impotent” or “weak” or “poor” or “scant” or “meager” or “thin” or “small” or “negative” or “oppose” or “dissent” or “harmful” or “damaging” or “unfavorable” or “disadvantageous” or “bad” or “adverse” or “detriment” or “detrimental” or “dismissive”) w/3 (“priority” or “arrangement” or “preference” or “antecedence” or “order” or “preeminence” or “prerogative” or “important”))	And no (“priority” or “arrangement” or “preference” or “antecedence” or “order” or “preeminence” or “prerogative” or “important”)	((“not” or “no” or “little” or “minimal” or “few” or “ineffective” or “useless” or “nominal” or “negligible” or “unsuccessful” or “fruitless” or “unproductive” or “futile” or “purposeless” or “inefficient” or “inefficacious” or “inadequate” or “feeble” or “inept” or “impotent” or “weak” or “poor” or “scant” or “meager” or “thin” or “small” or “negative” or “oppose” or “dissent” or “harmful” or “damaging” or “unfavorable” or “disadvantageous” or “bad” or “adverse” or “detriment” or “detrimental” or “dismissive”) w/3 (“priority” or “arrangement” or “preference” or “antecedence” or “order” or “preeminence” or “prerogative” or “important”))
Professional	OR (“professional” or “professionalism” or “skilled” or “proficient” or “proficiently” or “competent” or “competently” or “able” or “abled” or “top-notch” or “expertly” or “skillful” or “skillfully”) AND NO ((“not” or “no” or “little” or “minimal” or “few” or “ineffective” or “useless” or “nominal” or “negligible” or “unsuccessful” or “fruitless” or “unproductive” or “futile” or “purposeless” or “inefficient” or “inefficacious” or “inadequate” or “feeble” or “inept” or “impotent” or “weak” or “poor” or “scant” or “meager” or “thin” or “small” or “negative” or “oppose” or “dissent” or “harmful” or “damaging” or “unfavorable” or “disadvantageous” or “bad” or “adverse” or “detriment” or “detrimental” or “dismissive”) w/3 (“professional” or “professionalism” or “skilled” or “proficient” or “proficiently” or “competent” or “competently” or “able” or “abled” or “top-notch” or “expertly” or “skillful” or “skillfully”))	And no (“professional” or “professionalism” or “skilled” or “proficient” or “proficiently” or “competent” or “competently” or “able” or “abled” or “top-notch” or “expertly” or “skillful” or “skillfully”)	((“not” or “no” or “little” or “minimal” or “few” or “ineffective” or “useless” or “nominal” or “negligible” or “unsuccessful” or “fruitless” or “unproductive” or “futile” or “purposeless” or “inefficient” or “inefficacious” or “inadequate” or “feeble” or “inept” or “impotent” or “weak” or “poor” or “scant” or “meager” or “thin” or “small” or “negative” or “oppose” or “dissent” or “harmful” or “damaging” or “unfavorable” or “disadvantageous” or “bad” or “adverse” or “detriment” or “detrimental” or “dismissive”) w/3 (“professional” or “professionalism” or “skilled” or “proficient” or “proficiently” or “competent” or “competently” or “able” or “abled” or “top-notch” or “expertly” or “skillful” or “skillfully”))
Independence	OR (“Independence” or “independent” or “autonomous” or “free” or “self-determining” or “freestanding” or “self-governed” or “self-governing” or “self-ruling” or “unsupervised” or “nonaligned” or “nonpartisan”) AND NO ((“not” or “no” or “little” or “minimal” or “few” or “ineffective” or “useless” or “nominal” or “negligible” or “unsuccessful” or “fruitless” or “unproductive” or “futile” or “purposeless” or “inefficient” or “inefficacious” or “inadequate” or “feeble” or “inept” or “impotent” or “weak” or “poor” or “scant” or “meager” or “thin” or “small” or “negative” or “oppose” or “dissent” or “harmful” or “damaging” or “unfavorable” or “disadvantageous” or “bad” or “adverse” or “detriment” or “detrimental” or “dismissive”) w/3 (“Independence” or “independent” or “autonomous” or “free” or “self-determining” or “freestanding” or “self-governed” or “self-governing” or “self-ruling” or “unsupervised” or “nonaligned” or “nonpartisan”))	And no (“Independence” or “independent” or “autonomous” or “free” or “self-determining” or “freestanding” or “self-governed” or “self-governing” or “self-ruling” or “unsupervised” or “nonaligned” or “nonpartisan”)	((“not” or “no” or “little” or “minimal” or “few” or “ineffective” or “useless” or “nominal” or “negligible” or “unsuccessful” or “fruitless” or “unproductive” or “futile” or “purposeless” or “inefficient” or “inefficacious” or “inadequate” or “feeble” or “inept” or “impotent” or “weak” or “poor” or “scant” or “meager” or “thin” or “small” or “negative” or “oppose” or “dissent” or “harmful” or “damaging” or “unfavorable” or “disadvantageous” or “bad” or “adverse” or “detriment” or “detrimental” or “dismissive”) w/3 (“Independence” or “independent” or “autonomous” or “free” or “self-determining” or “freestanding” or “self-governed” or “self-governing” or “self-ruling” or “unsupervised” or “nonaligned” or “nonpartisan”))
Impartiality / Objectivity	OR (“impartial” or “equal” or “equitable” or “evenhanded” or “fair” “unbiased” or “unprejudiced” or “neutral” or “square” or “disinterested” or “uninvolved” or “uncommitted” or “detached” or “dispassionate” or “objective” or “objectively” or “objectivity” or “objectiveness” or “nondiscriminatory” or “indisputable” or “undeniable” or “demonstrable” or “provable” or “verifiable” or “neutralism” or “neutrality”) AND NO ((“not” or “no” or “little” or “minimal” or “few” or “ineffective” or “useless” or “nominal” or “negligible” or “unsuccessful” or “fruitless” or “unproductive” or “futile” or “purposeless” or “inefficient” or “inefficacious” or “inadequate” or “feeble” or “inept” or “impotent” or “weak” or “poor” or “scant” or “meager” or “thin” or “small” or “negative” or “oppose” or “dissent” or “harmful” or “damaging” or “unfavorable” or “disadvantageous” or “bad” or “adverse” or “detriment” or “detrimental” or “dismissive”) w/3 (“impartial” or “equal” or “equitable” or “evenhanded” or “fair” “unbiased” or “unprejudiced” or “neutral” or “square” or “disinterested” or “uninvolved” or “uncommitted” or “detached” or “dispassionate” or “objective” or “objectively” or “objectivity” or “objectiveness” or “nondiscriminatory” or “indisputable” or “undeniable” or “demonstrable” or “provable” or “verifiable” or “neutralism” or “neutrality”))	And no (“impartial” or “equal” or “equitable” or “evenhanded” or “fair” “unbiased” or “unprejudiced” or “neutral” or “square” or “disinterested” or “uninvolved” or “uncommitted” or “detached” or “dispassionate” or “objective” or “objectively” or “objectivity” or “objectiveness” or “nondiscriminatory” or “indisputable” or “undeniable” or “demonstrable” or “provable” or “verifiable” or “neutralism” or “neutrality”)	((“not” or “no” or “little” or “minimal” or “few” or “ineffective” or “useless” or “nominal” or “negligible” or “unsuccessful” or “fruitless” or “unproductive” or “futile” or “purposeless” or “inefficient” or “inefficacious” or “inadequate” or “feeble” or “inept” or “impotent” or “weak” or “poor” or “scant” or “meager” or “thin” or “small” or “negative” or “oppose” or “dissent” or “harmful” or “damaging” or “unfavorable” or “disadvantageous” or “bad” or “adverse” or “detriment” or “detrimental” or “dismissive”) w/3 (“impartial” or “equal” or “equitable” or “evenhanded” or “fair” “unbiased” or “unprejudiced” or “neutral” or “square” or “disinterested” or “uninvolved” or “uncommitted” or “detached” or “dispassionate” or “objective” or “objectively” or “objectivity” or “objectiveness” or “nondiscriminatory” or “indisputable” or “undeniable” or “demonstrable” or “provable” or “verifiable” or “neutralism” or “neutrality”))
Justice	OR (“justice” or “just” “equitability” or “equitableness” or “evenhandedness” or “fair-mindedness” or “fairness” or “impartiality” or “goodness” or “righteousness” or “uprightness” or “accountable” or “accountability”) AND NO ((“not” or “no” or “little” or “minimal” or “few” or “ineffective” or “useless” or “nominal” or “negligible” or “unsuccessful” or “fruitless” or “unproductive” or “futile” or “purposeless” or “inefficient” or “inefficacious” or “inadequate” or “feeble” or “inept” or “impotent” or “weak” or “poor” or “scant” or “meager” or “thin” or “small” or “negative” or “oppose” or “dissent” or “harmful” or “damaging” or “unfavorable” or “disadvantageous” or “bad” or “adverse” or “detriment” or “detrimental” or “dismissive”) w/3 (“justice” or “just” “equitability” or “equitableness” or “evenhandedness” or “fair-mindedness” or “fairness” or “impartiality” or “goodness” or “righteousness” or “uprightness”))	And no (“justice” or “just” “equitability” or “equitableness” or “evenhandedness” or “fair-mindedness” or “fairness” or “impartiality” or “goodness” or “righteousness” or “uprightness”)	((“not” or “no” or “little” or “minimal” or “few” or “ineffective” or “useless” or “nominal” or “negligible” or “unsuccessful” or “fruitless” or “unproductive” or “futile” or “purposeless” or “inefficient” or “inefficacious” or “inadequate” or “feeble” or “inept” or “impotent” or “weak” or “poor” or “scant” or “meager” or “thin” or “small” or “negative” or “oppose” or “dissent” or “harmful” or “damaging” or “unfavorable” or “disadvantageous” or “bad” or “adverse” or “detriment” or “detrimental” or “dismissive”) w/3 (“justice” or “just” “equitability” or “equitableness” or “evenhandedness” or “fair-mindedness” or “fairness” or “impartiality” or “goodness” or “righteousness” or “uprightness”))

Endnotes — (click the footnote reference number, or ↩ symbol, to return to location in text).

1. 1.

   International Criminal Court, Report of the Court on the Development of Performance Indicators for the International Criminal Court (Nov. 12, 2015), available online, archived. ↩

2. 2.

   Id. at 1–6. ↩

3. 3.

   Id. at 7. ↩

4. 4.

   Office of the Prosecutor, International Criminal Court, Office of the Prosecutor: Strategic Plan 2016–2018, ICC-ASP/14/22 (Aug. 21, 2015), available online ↩

5. 5.

   Id. at 23. ↩

6. 6.

   Id. at 24. ↩

7. 7.

   Id. at 23. ↩

8. 8.

   Id. ↩

9. 9.

   Id. ↩

10. 10.

    Id. ↩

11. 11.

    Id. ↩

12. 12.

    See Catherine Happer & Greg Philo, The Role of the Media in the Construction of Public Belief and Social Change, 1 J. Soc. Pol. Psych. 321 (Dec. 16, 2013), available online. ↩

13. 13.

    See Connie L. McNeely, Perceptions of the Criminal Justice System: Television Imagery and Public Knowledge in the United States, 3 JCJPC 1 (1995), available online. ↩

14. 14.

    See Kenneth Dowler, Media Consumption and Public Attitudes Toward Crime And Justice: The Relationship Between Fear of Crime, Punitive Attitudes, and Perceived Police Effectiveness, 10 JCJPC 109 (2003), available online. ↩

15. 15.

    See Robert J. Gebotys, Julian V. Roberts & Bikram DasGupta, News Media Use and Public Perceptions of Crime Seriousness, 30 Canadian J. Criminology 3, 16 (Jan. 1988), NCJRS paywall HeinOnline paywall. ↩

16. 16.

    Mali: Situation in the Republic of Mali, ICC-01/12, ICC, available online (last visited Jul. 22, 2017). ↩

17. 17.

    Emanuele del Rosso, Why is the Mali Situation at the ICC?, Just. Hub (Dec. 2, 2015), available online. ↩

18. 18.

    Press Release, ICC, ICC Trial Chamber VIII Declares Mr Al Mahdi Guilty of the War Crime of Attacking Historic and Religious Buildings in Timbuktu and Sentences Him to Nine Years’ Imprisonment, PR1242 (Sep. 27, 2016), [hereinafter Al Mahdi Guilty], available online. ↩

19. 19.

    Id. ↩

20. 20.

    International Criminal Court, Case Information Sheet: The Prosecutor v. Ahmad Al Faqi Al Mahdi, Case No. ICC-01/12-01/15 (Oct. 7, 2016), available online (last visited Jul. 22, 2017). ↩

21. 21.

    Al Mahdi Guilty, supra note 18. ↩

22. 22.

    Id. ↩

23. 23.

    Id. ↩

24. 24.

    Id. ↩

25. 25.

    Id. ↩

26. 26.

    Id. ↩

27. 27.

    Id. ↩

28. 28.

    Id. ↩

29. 29.

    Id. ↩

30. 30.

    Id. ↩

31. 31.

    Id. ↩

32. 32.

    United Nations Security Council, Security Council Authorizes Deployment of African-Led International Support Mission in Mali for Initial Year-Long Period, SC/10870 (Dec. 20, 2012), available online. ↩

33. 33.

    Hugh Schofield, Mali and France ‘push back Islamists’, BBC News, Jan. 12, 2013, available online. ↩

34. 34.

    Press Release, ICC OTP, ICC Prosecutor Opens Investigation into War Crimes in Mali: “The legal requirements have been met. We will investigate”, PR869 (Jan. 16, 2013), available online. ↩

35. 35.

    Marlise Simons, Extremist Pleads Guilty in Hague Court to Destroying Cultural Sites in Timbuktu, N.Y. Times, Aug. 22, 2016, available online. ↩

36. 36.

    Id. ↩

37. 37.

    Marlise Simons, Prison Sentence Over Smashing of Shrines in Timbuktu: 9 Years, N.Y. Times, Sep. 27, 2016, available online. ↩

38. 38.

    The Prosecutor v. Ahmad Al Faqi Al Mahdi, Case No. ICC-01/12-01/15, Warrant of Arrest (Sep. 18, 2015) (fr.), available online. ↩

39. 39.

    The arrest warrant was first made available to the public on September 28, 2015. ↩

40. 40.

    Press Release, ICC, Situation in Mali: Ahmad Al Faqi Al Mahdi Surrendered to the ICC on Charges of War Crimes Regarding the Destruction of Historical and Religious Monuments in Timbuktu, PR1154 (Sep. 26, 2015), available online. ↩

41. 41.

    Press Release, ICC, Initial Appearance of Ahmad Al Faqi Al Mahdi Scheduled for 30 September 2015, MA189 (Sep. 28, 2015), available online. ↩

42. 42.

    International Criminal Court, Ahmad Al Faqi Al Mahdi Case: Initial Appearance, 30 September 2015, YouTube, at 6:07 (Sep. 30, 2015), available online. ↩

43. 43.

    Case Information Sheet, supra note 20. ↩

44. 44.

    Id. ↩

45. 45.

    International Justice Resource Center, ICC Convicts Al-Mahdi Of War Crime For Destroying Cultural Sites (Oct. 5, 2016), available online. ↩

46. 46.

    Al Mahdi Guilty, supra note 18. ↩

47. 47.

    Janet Anderson, Mali Jihadist First at ICC, Just. Hub, Aug. 22, 2016, available online. ↩

48. 48.

    Case Information Sheet, supra note 20. ↩

49. 49.

    Id. ↩

50. 50.

    Anderson, supra note 47 ↩

51. 51.

    Id. ↩

52. 52.

    Id. ↩

53. 53.

    The Prosecutor v. Ahmad Al Faqi Al Mahdim, Case No. ICC-01/12-01/15, Judgment and Sentence (Sep. 27, 2016), available online. ↩

54. 54.

    Case Information Sheet, supra note 20. ↩

55. 55.

    Id. ↩

56. 56.

    Judgment and Sentence, supra note 53 ↩

57. 57.

    Id. ↩

58. 58.

    Al Mahdi Guilty, supra note 18. ↩

59. 59.

    Id. ↩

60. 60.

    Id. ↩

61. 61.

    Happer & Philo, supra note 12; McNeely, supra note 13; Dowler, supra note 14; Gebotys et al., supra note 15. ↩

62. 62.

    Although the repetition of positive or negative mentions could help analyze the strength of a country’s media bias, it also had the potential of severely skewing the results. Many countries published articles that essentially re-stated the official opinions of important political figures or NGOs. These statements usually repeated the same phrases multiple times and could have had the effect of overstating the positive nature of news publications. ↩

63. 63.

    Matthew A. Baum & Yuri Zhukov, Filtering Revolution: Reporting Bias in International Newspaper Coverage of the Libyan Civil War (Harv. Kennedy Sch. 2014), available online. ↩

64. 64.

    John Wihbey, Bias in Reporting of International Conflict and War: Research on the Libyan Civil War, Journalist’s Resources (Mar. 3, 2015), available online (last visited Jul. 22, 2017). ↩

65. 65.

    The Economist, Democracy Index 2015: Democracy in an Age of Anxiety (2016), available online. ↩

66. 66.

    Id. at 4. ↩

67. 67.

    The World Factbook, CIA (2016), available online (last visited Jul. 22, 2017). ↩

68. 68.

    The States Parties to the Rome Statute, ICC, available online (last visited Jul. 22, 2017). ↩

69. 69.

    The Africa Question, ICC Forum (Mar. 17, 2013), available online. ↩

70. 70.

    The Withdrawal Question, ICC Forum (Nov. 15, 2016), available online. ↩

71. 71.

    Lexis News & Business Database, Lexis (2016), Lexis/Nexis paywall. ↩

72. 72.

    Richard Alan Nelson, Tracking Propaganda to the Source: Tools for Analyzing Media Bias, Global Media J. (Dec. 17, 2013), available online. ↩

73. 73.

    The broad, initial terms-and-connectors script was: (“Al Mahdi” or “Al-Mahdi” or “Ahmad Al Faqi Al Mahdi” or “Abu Turab” or “ICC-01/12-01/15”) and (“International Criminal Court” or “ICC” or “Cour Penale Internationale”) and not (“Sadiq” or “Sudiq” or “Sadeq” or “Sudeq” or “Stockly”). ↩

74. 74.

    Baum & Zhukov, supra note 63. ↩

75. 75.

    Some more information regarding the Logit Model in Table 4:
    Number of observations = 961;
    LR χ²(2) = 56.69;
    Prob > χ² = 0.0000;
    Log likelihood = −637.75568;
    Pseudo R² = 0.0426. ↩

76. 76.

    Some more information regarding the Logit Model in Table 6:
    Number of observations = 961;
    LR χ²(3) = 21.32;
    Prob > χ² = 0.0001;
    Log likelihood = −655.44231;
    Pseudo R² = 0.0160. ↩

77. 77.

    Some more information regarding the Logit Model in Table 7:
    Number of observations = 961;
    LR χ²(5) = 70.98;
    Prob > χ² = 0.0000;
    Log likelihood = −630.61315;
    Pseudo R² = 0.0533. ↩

78. 78.

    Some more information regarding the Logit Model in Table 9:
    Number of observations = 961;
    LR χ²(1) = 7.79;
    Prob > χ² = 0.0053;
    Log likelihood = −932.63974;
    Pseudo R² = 0.0042. ↩

79. 79.

    Some more information regarding the Logit Model in Table 11:
    Number of observations = 961;
    LR χ²(1) = 16.05;
    Prob > χ² = 0.0001;
    Log likelihood = −928.50814;
    Pseudo R² = 0.0086. ↩

80. 80.

    Some more information regarding the Logit Model in Table 13:
    Number of observations = 961;
    LR χ²(1) = 0.69;
    Prob > χ² = 0.4069;
    Log likelihood = −936.18904;
    Pseudo R² = 0.0004. ↩

81. 81.

    Some more information regarding the Logit Model in Table 14:
    Number of observations = 961;
    LR χ²(1) = 18.40;
    Prob > χ² = 0.0004;
    Log likelihood = −927.33327;
    Pseudo R² = 0.0098. ↩