Scholars and theoreticians have long argued about the compatibility of Islam and democracy. Bush administration support for export of democracy to the Middle East has brought the debate to the forefront of policy circles. As the discussion continues, statistical models can be useful to interpret the historical record. While recent studies have reached contradictory conclusions, as more data is considered, a nuanced relationship between Islam and democracy emerges: In all but the poorest countries, Islam is associated with fewer political rights.


Scholarly studies disagree about whether Muslim countries as a whole are less democratic than non-Muslim countries. For political scientists and statisticians, the compatibility between Islam and democracy remains controversial, and studies have reached opposite conclusions. Part of this is due to limited sample size. Research has focused either on particular countries or subcategories such as Arab countries which are not representative of majority Muslim states. Indeed, none of the four most populous Muslim countries—Indonesia, Pakistan, Bangladesh, and Turkey—is Arab.

As the World Bank and other international organizations have produced more standardized data on each country's economy, researchers have been able to perform broader statistical analyses. Freedom House's quantification of political liberalism has further expanded the breadth of comparison and correlation. Scholars can now use statistical techniques to examine patterns across a greater number of Islamic countries. Since 2000, there have been four sets of conflicting empirical analyses.

All of the studies considered below are by political scientists, and most employ regression analysis. This is a common statistical technique that predicts the value of a variable based on the measured values of other variables. In practice, this means finding the best fitting curve to a series of points so that one can see the causal impact of each variable on the variable to be explained. Both Michael L. Ross, University of California-Los Angeles, and M. Steven Fish, University of California-Berkeley, use a regression technique to disentangle the impact of a number of causal variables on a single outcome and separately show that Muslim countries score lower on indices of democracy than non-Muslim countries.[1] Yale University political scientists Daniela Donno and Bruce Russett present two sets of regressions. One yielded a similar conclusion, but the other which examined cross-sections of all major Muslim countries at different points in time showed mixed results.[2] In contrast, Columbia University professor Alfred Stepan and his University of North Carolina colleague Graeme B. Robertson use a different statistical technique to argue that differences in the level of democracy between Muslim and non-Muslim countries in the developing world is not significantly different.[3] The discrepancies among studies suggest that statistical techniques, countries considered, and other variables included in the analysis shape the results. So what conclusions does the collective evidence support?

Constructing a Model

The four studies use the "political rights" variable of Freedom House[4] and the "democracy" variable of Colorado State University professors Monty G. Marshall and Keith Jaggers in their Polity IV data set.[5] Both are highly correlated definitions of political democracy. Because both variables yield the same general results, use of the Freedom House data set is preferable since it includes more countries. However, while Freedom House uses a descending scale from seven (fewest political rights) to one (highest), the scale here is reversed for clarity. A new, more complete model can also replicate many of the variables and data sources used in previous studies.[6]

Other data included are per capita gross domestic product (GDP) with data supplied by the World Bank.[7] While some previous studies considered per capita GDP in a linear fashion, this may miss certain relationships. Those countries in Europe, North America, as well as Japan, South Korea, and Australia, which belong to the Organization for Economic Cooperation and Development, are all industrialized. They vary in per capita GDP although they maintain similar political rights. To minimize this difficulty, Donno, Russett, and Ross all used the logarithm of per capita GDP.

This study explores differences in democracy among "very poor," "poor," and "medium/high income" nations. These categories are based on their 2000 per capita GDP compared to the same year's U.S. per capita GDP. "Very poor" nations had a per capita GDP of less than 15 percent of the U.S. figure while "poor" nations had a per capita GDP of less than 25 percent, and the medium/high income nations had a per capita GDP greater than 25 percent of that of the United States.

In his essay, Ross explores the characteristics of countries in which raw materials such as oil are an important source of state income. Regardless of their dominant religion, these states often have lower indices of political rights. Accordingly, the analysis below also includes World Bank statistics measuring the proportion of income derived from types of export averaged over more than a decade.

Donno and Russett observe that war erodes political rights, and this hypothesis is easily tested.[8] They also suggest that present and former communist countries have a freedom deficit as their citizenry's political rights often have not advanced. This too was factored into the analysis using dummy variables (either 0 or 1).

Finally, since large countries are more difficult to govern than small countries, especially in poor countries, another explanatory variable is the logarithm of the population. As expected, this variable turns out to be statistically significant only in the sample of developing economies.

The study sample includes 162 countries of which 39 have a Muslim majority.[9] The "Muslim variable" in this study is the percentage of Muslims in the population rather than an artificial variable of either zero or one, which statisticians use to indicate whether any country's population reaches an arbitrarily assigned proportion of Muslims. The latter approach, used by all of the previously cited studies with the exception of Ross, loses useful information.

While conclusions drawn by earlier investigators about the relationship between Islam and democracy conflict, such disagreements result primarily from the sample of countries chosen and secondarily from how the equations and variables are specified. A broader set of regressions with different samples of nations clarifies this discrepancy.

The statistical results in Table 1 show that in all but very poor countries, Islam has a significantly negative impact on political rights.

Table 1: The Political Impact of Islam in 2000

Variable to be explained: political rights

1 2 3 4 5 6 7
All All richer All poor All very poor All non-Arab Richer non-Arab Poor non-Arab
Explanatory variables:
% of Muslims -1.241*
Arab culture
(0-1 variable)
--- --- ---
Former communist
(0 or 1)
Now communist
(0 or 1)
--- -3.249*
--- -3.252*
Log per capita GDP 0.828*
Log population -0.045
Raw material export as % of total exports -1.633*
Years at war, 1990-2002 -0.095
Constant 7.830*
# observations
Adj. R square

Note: T-tests of statistical significance of the calculated coefficients are given in parentheses and statistical significance at the 0.05 level is marked with an asterisk.

This conclusion is reflected in the fact that, with the exception of the regression in column four, the greater the percentage of Muslims in the population, the lower are the political rights in the country. This Muslim variable is not, however, statistically significant when the very poor nations in the sample are examined separately. Such a result makes sense. Citizens of very poor nations often enjoy few rights to begin with, and so Islam does not have much room to create a downward impact. For the entire sample, the countries with an Arab culture also have lower political rights than other Muslim nations although in the other results with different sub-samples, there is not an indication of such a relationship. The correlations also support previous work that shows a positive correlation between political rights and per capita income but a negative relationship both between the relative importance of raw material exports and whether the nation currently has a communist government. For example, most countries which derive the bulk of their income from oil exports suffer in political freedoms. Communist countries such as China, Cuba, and North Korea also do not compare well in political freedoms with many other states.

Other concrete examples illustrate the statistical results. Among pairs of oil exporting states with roughly the same per capita GDP, one predominantly Muslim, the other not, the Muslim country compares unfavorably with the non-Muslim state. For instance, Algeria is less democratic than Gabon; Syria has fewer political freedoms than Sri Lanka, and even Iran does not compare well to Venezuela. Of the 58 nations classified as having medium or high per capita GDP, only the 6 Gulf Cooperation Council states—Bahrain, Kuwait, Oman, Qatar, Saudi Arabia, and United Arab Emirates—are Muslim. These 6 countries have an average political freedom score roughly one third that of the remaining 52 nations in this group. When their oil exports are taken into account, they score roughly half that of the other nations.

There are some exceptions to the general conclusion that Islam has a negative impact on political rights. In the sample of 39 predominantly Muslim countries, Bangladesh, Mali, Senegal, and Turkey have political freedom scores in the middle range. Even more perplexing, the first three are found among the "very poor nations" where low political rights are the norm. Whether these poor Muslim countries can maintain their democratic edge as they develop economically is unclear since this edge depends on the intricate interactions between Islamist and democratic forces during modernization. The constitutional crisis surrounding the aborted May 2007 presidential elections in Turkey and speculation about military interference provides an example of the uncertainties of democracy in a developing Muslim nation. The parliamentary elections in July 2007, however, proceeded smoothly, and relative calm prevailed in the first few weeks afterwards.

Some critics might object to the general conclusion about the deficit in political rights in Muslim countries because exact results differ somewhat across models, and moreover, are dependent upon the quality of data used in the calculations. Nevertheless, in statistical experiments not reported here that use smaller samples of nations having more reliable data, the major conclusion about the negative relationship between Islam and political rights does not change. Similarly, experiments using other explanatory variables such as population heterogeneity did not change the general conclusions. Interestingly, although Islam has a considerable influence on the polity, it has had relatively little impact on the economy.[10]

The empirical results do not answer the question of why Muslim states usually have fewer political rights than non-Muslim countries. In part, historical factors are responsible. Islam found root primarily in countries with low per capita GDPs and a deficit of political rights. But Islamic doctrine also places little emphasis on individual rights, especially by those Muslim sects such as the Shi‘a whose leaders have stressed a theocratic approach toward government. Although the leaders of some Muslim nations have severed the bonds between mosque and state, many of these same rulers still maintained the politically repressive traditions of their countries to enhance their own power.

Three additional conclusions appear clear. The causes underlying the deficits in political freedom in most Islamic nations can only be disentangled by a detailed analysis of the special features of each nation. Moreover, this democracy deficit is likely to remain for many years in the future. Finally, the fact that some predominantly Muslim nations have relatively respectable degrees of political freedom suggests that this deficit may not be a permanent condition. These conclusions imply, in turn, that exporting democracy to the Middle East or other Muslim countries may not be impossible but will certainly have a very distant horizon.

Frederic L. Pryor is an economist and a senior research scholar at Swarthmore College. He thanks Daniela Donno, Phillip Jefferson, Zora Pryor, Michael L. Ross, and Bruce Russett for their helpful comments on an earlier draft. A more technical version of this essay can be found at

[1] Michael L. Ross, "Does Oil Hinder Democracy?" World Politics, Apr. 2001, pp. 325-61; M. Steven Fish, "Islam and Authoritarianism," World Politics, Jan. 2002, pp. 4-37.
[2] Daniela Donno and Bruce Russett, "Islam, Authoritarianism, and Female Empowerment: What Are the Linkages?" World Politics, July 2004, pp. 582-607.
[3] Alfred Stepan and Graeme B. Robertson, "An ‘Arab' More than a ‘Muslim' Electoral Gap," Journal of Democracy, Oct. 2003, pp. 30-44.
[4] "Methodology: General Characteristics of Each Political Rights and Civil Liberties Rating," Freedom in the World 2004 (New York: Freedom House, 2004); "Freedom in the World: Country Ratings," Freedom House,, accessed June 10, 2006.
[5] "Polity IV Project Variables: Indicators of Democracy and Autocracy (Composite Indicators)," Monty G. Marshall and Keith Jaggers, Polity IV Project, Center for International Development and Conflict Management, University of Maryland, College Park, Md., accessed June 15, 2006.
[6] For political rights and civil liberties, see Freedom in the World, 2004, pp. 714-8.
[7] Per capita gross domestic product (GDP) in U.S. dollars from the year 2000; population data, and raw material export figures come from the World Development Indicators, 2000 (New York: Oxford University Press, 2000).
[8] For data for the period a country was at war between 1990 and 2002, see Faten Ghosn, Glenn Palmer, and Stuart Bremer, "The MID3Data Set, 1993-2001," Conflict Management and Peace Sciences, 21(2004): 133-54.
[9] David B. Barrett, George T. Kurian, and Todd M. Johnson, World Christian Encyclopedia: A Comparative Survey of Churches in the Modern World (New York: Oxford University Press, 2001) report 41 nations with a Muslim majority, but the sample in this study does not include the Palestinian Authority or the Cocos Islands.
[10] Frederic L. Pryor, "The Economic Impact of Islam on Developing Nations," World Development, forthcoming 2007.