Sunday, June 19, 2016

Visualizing Anti-Shia Violence in Pakistan Part II: Where do Shias experience the most violence?

This is part of a continuing series on visualizing anti-Shia violence in Pakistan. As I mentioned in the first post, all analyses in this series have been conducted using the Tashaddud sectarian violence data set. In this post, I will be using this data set to determine where anti-Shia violence is most prevalent in Pakistan.

I analyzed differences in the number of anti-Shia attacks, Shias killed and Shias killed per attack across all provinces and cities listed in the data set using generalized linear models (GLMs). GLMs with Poisson, quasi-Poisson or negative binomial distribution were chosen based on AIC values. For province models, we examined differences between all provinces for each year. For the city models, I only included data from four cities; Karachi, Quetta, Peshawar and Dera Ismail Khan where attacks occurred in four or more years between 2001 and 2013, examining the differences between all cities for each year. The structure of the final models was determined using log likelihood ratio tests. I also assessed the importance of Year and Province/City using conditional inference trees (party package version 1.0-25). Condition inference trees allowed me to determine differences in the number of anti-Shia attacks, Shias killed and Shias killed per attack across all provinces, cities and years. Splits for each conditional inference tree were determined using Monte Carlo multiplicity adjusted p values with 10,000 permutations. As in my earlier post, significant refers to statistical significance; p values less than 0.05. All statistical analyses were performed in R (version 3.0.3). All plots were also created using R.

Note: Anti Shia violence data from Tashaddud was only available for a nine year period from 2001 to 2013. I added data for the number of ppl killed and injured for 2014 from the United States Commission on International Religious Freedom 2014 report on Pakistan. Before plotting any data or performing any analysis, I removed any data for violent attacks by Shias on LeJ/ASWJ/Sunni Tehreek members (108 of 723 data points). Therefore I only analysed data specific to anti-Shia violence from this dataset.
(added June 21st 2016 to address comments by @AizazAli) 

Here are my results:

#1)  Sindh had a greater number of Anti-Shia attacks as compared to other provinces (p = 0.033 for Sindh). While Balochistan had the next highest number of Anti-Shia attacks, this was not significantly different from other provinces. Anti-Shia attacks increased by 359% across all provinces after 2011 (p = 0.02)



#2) Karachi had the greatest number of Anti-Shia attacks in Pakistan (p = 0.011), followed by Quetta (p = 0.07, marginally non-significant). There was no difference in the number of Anti-Shia attacks between the other cities. After 2011, there was a 560% increase in the number of Anti-Shia attacks across these four cities (conditional inference trees, p = 0.037).


#3) Overall, there was no difference in the number of Shias killed across each province/region, with the exception of Gilgit-Baltistan where significanlty fewer Shias killed (p = 0.01). However, after 2011, the number of Shias killed increased by 195% for all provinces (p = 0.006, conditional inference trees).



#4)There was no significant difference in the number of Shias killed across all cities with the exception of Quetta, which had the greatest number of Shias killed (marginally non-significant increase for Quetta; p = 0.06), followed by Karachi (statistically non significant, p = 0.23). Additionally, there was a 283% increase in the number of Shias killed after 2011 (p = 0.019, conditional inference trees) across these four cities.


#5) Shias in the FATA region experience the most deadliest attacks across all of Pakistan (conditional inference trees, p = 0.0003), with significantly more Shias killed per attack than any other province. Anti Shia attacks in FATA are 430% deadlier than anti-Shia attacks in any other province or region. This suggests that there are more acts of large scale anti-Shia violence (e.g. bomb blasts in public spaces) than any other region in the country. An average of 26 people die in an anti-Shia attack in FATA, while an average of 3 people die in anti-Shia attacks in Sindh. 


#6) There was no significant difference in the number of Shias killed per attack in Karachi, Quetta, Peshawar and Dera Ismail Khan suggesting that the types of attacks in each of these cities is similar.


My results seem to suggest that Karachi(Sindh) and Quetta(Balochistan) are the most dangerous places in Pakistan for Shia Muslims. This is not surprising given the near daily threats, attacks or assassinations on members of the Shia intelligentsia in Karachi or the plight of the Hazara Shia community that have been forced to physically segregate in Quetta due to constant threats of suicide attacks and assassinations.

The roots of anti-Shia militia (and organized anti-Shia violence) lie in dynamics of wealth, power and class of rural Southern Punjab, a region dominated by Shia feudal lords and Sunni peasants (1). However, anti-Shia violence has shifted from its class conflict beginnings to exploiting ethnic tensions in both cities. In Karachi, most Shias tend to be descendants of migrants from the North Indian city states, known as Muhajirs. Muhajirs (both Shia and Sunni) are in conflict with Pashtun immigrants (mostly Sunni) from Khyber-Pakhtunkhwa and Afghanistan for political dominance of the city. In Quetta, ethnic Hazaras are seen as agents of Iran and therefore implicated in oppression of Baluch nationalists/separatists by both Iran and Pakistan.

No city in Punjab has experienced violence comparable to Karachi/Quetta  despite producing the most violent and well established anti-Shia militias (LeJ/SSP/TTP). Possible explanations for this discrepancy may be the ease at which militarization is possible in both Karachi and Quetta as compared to rural Punjab, a greater opportunity for networking with other Sunni supremacist organizations in Karachi and multi-state extremist groups (e.g. Haqqani network & Taliban) in Quetta, the ease of visually identifying Hazara Shias in Quetta, a greater availability of prominent Shia targets in both cities and/or greater media/public exposure for anti-Shia violence in Karachi. Additionally, the importance of Karachi as Pakistan's trading hub, and the physical proximity of Quetta to both Iran/Afghanistan may allow anti-Shia militias to network with like-minded groups internationally, making Shias in these two cities especially vulnerable to violence.

 Footnotes

(1)Waseem, Mohammad in association with Kamran, Tahir; Ahmed Ali, Mukhtar; Riikonen, Katja ‘Dilemmas of Pride and Pain: Sectarian Conflict and Conflict Transformation in Pakistan’, Working Paper 48- 2010, Religions and Development Research Program

3 comments:

  1. Janab, good data and visualization! Would you be able to see how this data of the violence against the Shias correlates to the violence against everyone else? So as per #2, for example, Karachi had the greatest number of anti-Shia attacks. How does this relate to Karachi as the most violent city (for everyone) in Pakistan (if, indeed, it is? I am going off hearsay...)? Similarly, does such a relative analysis for all other cities yield a predictable pattern or no? Thanks!

    ReplyDelete
    Replies
    1. The Data can be downloaded from here
      http://tashaddud.org/data.html

      The stats include all the figures collected from newspaper irrespective of the sect.

      Delete
  2. @AizazAli If you read Part I of this series (see link: http://kissmyroti.blogspot.ca/2016/02/visualizing-anti-shia-violence-in.html), you will see that I excluded attacks on other sects before analyzing the data. Therefore my stats were only performed for Shias.

    ReplyDelete