Islamist Militancy in Pakistan: A View from the Provinces
July 24, 2009
By C. Christine Fair
In Pakistan’s struggles with Islamist militant groups–foreign and homegrown–the country’s provinces have experienced levels of violence that differ widely by type, severity, and consequences. An analysis of a recent WorldPublicOpinion.org poll of Pakistan reveals that these different provinces vary significantly in their views of militant groups and the recent government’s actions to an extent that is relevant for policymakers. Of special interest, support for the Pakistani government’s efforts and concern about specific militant groups prove to be strong in some unexpected places.
Pakistan is home to numerous ethnic groups, who tend to be clustered in specific provinces, e.g.: Punjabis tends to be clustered in the Punjab, Pashtuns in the North West Frontier Province (NWFP), the Baluch in Baluchistan and Sindhis in Sindh. Of course, there has been tremendous migration within Pakistan. Karachi, in Sindh, is now to home to more Pashtuns than any other city in Pakistan and ethnic Punjabis are settled throughout Pakistan. Indeed, these internal demographic changes have often been a source of conflict among different ethnic groups.
While many U.S. analysts focus upon “Pakistan,” and “Pakistani opinions,” it is useful to consider sub-national analyses of data. For example, provinces differ by voting patterns, access to education and other public services, representation in government institutions (including the armed forces), and literacy among other factors. These enormous inter-provincial differences tend to be overlooked by policy makers.
This paper begins with the opinion survey’s timing, its methodology, and a few caveats on its provincial comparisons. It then offers a brief summary of provincial differences in attitudes. It then considers the provincial responses to specific questions on three topics: beliefs about Islamist militants’ activities and objectives; views of the Pakistani government and its response to the militants, especially in Swat; and attitudes toward foreign militant groups, the Afghan Taliban and al Qaeda.
The Timing of the Study
In March 2009, the efforts of the NWFP provincial government to strike a deal with the Pakistani Taliban–offering Shari’a courts in Swat in exchange for peace–was formally endorsed by Pakistan’s Parliament and President Zardari. In the last week of April the Pakistani Taliban extended its reach, moving into Buner, and the government and army reacted forcefully soon afterward with a major offensive. The survey was fielded in late May during a key phase of the army’s offensive against the Pakistani Taliban, coinciding with heavy street-to-street fighting in the Swat Valley’s main town, Mingora.
Methodology and a Caveat
The differences between provinces selected for discussion in this paper are very large ones. The smallest variations reported here are in the 20-point range; most are in the 30-50 point range. The variations in each question offered are at the p<.001 level of significance. The paper seeks to discuss and interpret only these very robust differences.
The survey of the national population of Pakistan was developed by WorldPublicOpinion.org, with the author contributing to the study’s conceptualization and the design of the survey instrument. It was carried out by SEDCO (Socio-Economic Development Consultants, Islamabad, Pakistan). All interviewing was conducted in Urdu. A total of 1,000 face-to-face interviews were conducted across 64 primary sampling units in rural areas and 36 in urban areas. In order to properly capture opinion in Baluchistan (a multi-ethnic, sparsely populated province), it was oversampled, using 15 primary sampling units; results were then weighted back to reflect true proportions among provinces. Interviews were conducted between May 17 and 28, 2009. Sampling error for a sample of this size is approximately +/-3.2 percentage points.
Even so, given the sample size, there are limits to how fine disaggregation can be, due to Pakistan’s population distribution. For example, Baluchistan is the largest province in geographical size, but it is home to only 5 percent of the country’s population, according the most recent census in 1998. Moreover, Baluchistan is ethnically diverse, including Baluch, Pashtuns and Punjabis with Baluch and Pashtuns concentrated in different parts of the province. While the survey over-sampled Baluchistan, it is still possible that the Baluchistan sample does not perfectly reflect ethnic distributions within the province. Thus tabulations for Baluchistan in particular must be viewed with these caveats in mind. (Similar concerns obtain with Sindh to a lesser degree.)2
That said, it is clear that the study’s questions about the Swat crisis, the government’s response, and various Islamist militant organizations elicited differences across provinces so large and robust that they merit consideration.
A Brief Summary of Provincial Differences
In general, the NWFP and Punjab have experienced most intensively the ravages of violence from the Pakistani Taliban and allied militant groups (including foreign militants) while also being closest to the state’s military and police efforts to counter these militant groups. In contrast, both Sindh and Baluchistan have experienced other kinds of violence in the past, but they have been relatively spared the predations of the Pakistani Taliban. Perhaps for these reasons, we see considerable differences across the provinces. This is true for Baluchistan even though the Afghan Taliban has long used Baluchistan territory as a sanctuary, without making Baluchistan itself a focus of operations.
1 C. Christine Fair is Senior Political Scientist at RAND Corporation and Assistant Professor, Georgetown University. Dr. Fair collaborated on the development of the survey on which this article is based.
2 While here we focus upon inter-provincial differences, we also looked at differences in sample means based upon whether the respondent lived in rural or urban areas. However, we found few notable differences. Thus we do not present those results here.