Print this page

by Phil Kelly

Background to the Fitzgibbon Democracy Survey

The origins of the democracy survey date back to 1945 when Professor Russell H. Fitzgibbon, a UCLA political scientist, asked a panel of ten distinguished U.S. latinamericanist scholars to rank the twenty Latin American republics according to a set of criteria that he felt would measure the extent of democracy in each of the countries. His criteria for assessing the strength of democracy, fifteen in all, encompassed the following describers (complete definitions for each of the criteria were furnished panelists in addition to the survey instrument):

On a five-point evaluation, panelists were to rate the republics separately according to each of the criteria, and the poll results were tallied later.
Fitzgibbon replicated his canvass at regular five-year intervals through 1970, adding more panelists than his original ten but maintaining the original fifteen criteria. Kenneth Johnson became associated with the project in 1960 and he assumed sole authorship of the 1975 and 1980 polls after Fitzgibbon's retirement. As the present director of the democracy project, Phil Kelly assisted Johnson in 1985 and administered the instrument alone for the three most recent evaluations, 1991, 1995, and 2000. In total, twelve democracy surveys, taken every five years and all adhering to Fitzgibbon's original format, have been conducted since 1945. Over one-hundred panelists contributed to the 2000 survey. ¹
All three project directors, Fitzgibbon, Johnson, and Kelly, experimented with the poll; most changes were tried only once and not continued. For example, Fitzgibbon gave certain criteria more weight than other criteria, and he also attempted a "self-assessment as to the respondent's familiarity with both [Latin American] states and [the fifteen] criteria" (Fitzgibbon 1967, 155). Both attempts were inconclusive and dropped. Likewise, Fitzgibbon tested for statistical associations between the democracy scales and an assortment of national attributes, but he found none. Johnson composed a separate political scale drawn from five "select criteria" among the total fifteen (Fitzgibbon 1976, 131-132), and he and Miles Williams created a "Power Index" that sought to measure various groups' impact on politics (Johnson and Williams 1978, 37-47). But again, neither innovation was kept. Nor did a later "Attitudinal Profile" of panel respondents' backgrounds by Johnson and Kelly enjoy long life. In sum, Fitzgibbon's original 1945 survey has continued for the past fifty-five years without significant adjustment.
The most notable legacy of the Fitzgibbon democracy survey is its long life, fifty-five years and twelve different polls since 1945. No other surveys can boast of such longevity and repetition over a time span that has seen so many changes and perhaps improvements in democracy and government in Latin America. Also, this survey is the only panel-of-experts technique for gauging the extent of democracy, as other assessments of democracy, described below, rely on census and other secondary statistical data or a variety of subjective measures. As stated by Fitzgibbon, "[Panel] Specialists are likely to introduce desirable nuances and balances which are impossible in the use of cold statistical information, even of the most accurate sort" (Fitzgibbon 1967, 135). In addition, the canvass possesses both conceptual and operational definitions of democracy, the former rendered in the fifteen criteria and the latter in the survey method itself, such that ordinal and interval data measurements become available and hence statistical analysis can be performed between the democracy ranking scales and an assortment of independent variables. More on the results of the author's statistical analysis below. Finally, despite the absence of major overhauling of the project's approaches since 1945, the panel procedure remains open to adjustment and to replication by others (Kelly 1998, 3-11).
Of course, problems are inherent to the Fitzgibbon approach as well. Are panelists indeed sufficiently familiar with enough elements of political life in Latin America to gauge accurately the depth of democracy in each one of the twenty republics? Most likely not. Accordingly, are panelists'"images" and perceptions of the Latin communities strong enough indicators of constitutionalism? Despite yearly comparisons made among countries over the years, we do not know in general if democracy itself is more grounded in the Southern Hemisphere's political culture. Likewise, are certain reformist and/or radical states, such as Cuba and Nicaragua, given higher scale rankings because the majority of survey participants reflect a "liberal" bias as was seen in the1985 Johnson-Kelly Attitudinal Profile? Should the poll be made annually or at two-year intervals instead of each five years, and might certain of the democracy criteria be eliminated and others given more weight?

Analysis of the Survey

Table One shows the democracy standings for each of the twelve Fitzgibbon canvasses (1945-2000) plus the cumulative ratings when the ordinal positions of each country are summed and the totals ranked (extreme right-hand column).

TABLE ONE

FITZGIBBON-JOHNSON INDEX: SPECIALISTS' VIEW OF

DEMOCRACY IN LATIN AMERICA, 1945-2000

Table Two separately presents the cumulative democracy rankings of the twenty republics. Costa Rica places first in the total figures, having scored first since the 1965 poll. Uruguay stands some way below in second place, and a clustering arises among the next five republics, only ten points separating Chile and Mexico from Colombia. Cuba and Ecuador match each other's position in the scale's middle, and a bunching again appears among Nicaragua, the Dominican Republic, Guatemala, and Honduras. Paraguay and Haiti occupy the lowest positions on the democracy survey.

Table Two - 2000 Survey Rankings of the Original Twenty Latin American Republics

* Total ranking responses of the one-hundred and three survey participants

In the 2000 poll Kelly added thirteen newly-independent Caribbean countries, as revealed in Table Three, enlisting the original Fitzgibbon survey instrument. But, just thirty-nine of the 103 panel respondents felt sufficiently knowledgeable to evaluate these states. He then integrated four of the Caribbean countries (those with the greatest panelists' responses) within the 2000 ratings of the twenty original Latin American republics (Table Four).

Table Three - 2000 Survey Rankings of the Newly-Independent Caribbean Countries

* Average of ranking responses of thirty-nine survey participants

Table Four - 2000 Survey Rankings of Twenty Original Republics plus Four Caribbean States

*Average of ranking responses

Kelly assembled an array of fifty independent variables (taken from Kurian 1979) to test for possible statistical linkages between the 1945-2000 cumulative democracy rankings (as his dependent variable) and certain national traits of the twenty Latin American states. Enlisting first a Spearman rho coefficient as a preliminary barometer of associational strength, he found the following ten independent variables exhibiting very high bivariate associations with the democracy rankings: telephones per capita, urbanization, electrical energy consumption per capita, general energy consumption per capita, physical quality-of-life index (life expectancy, infant mortality, and literacy), daily newspaper circulation per capita, tractors per hectare, public education expenditures per capita, gross national product per capita, and steel consumption per capita.
To carry these comparisons further, Kelly next grouped all eleven variables (one dependent and ten independent) together within a stepwise regression procedure in order to assay a more concise prediction of democracy in Latin America. This technique possesses the advantages of controlling for spuriousness and of reducing the number of variables into a model for simpler utilization. From this endeavor, four of the independent variables proved to be particularly strong in associating with the democracy rankings, in order, newspaper circulation per capita, tractors per hectare, energy consumption per capita, and public education expenditures per capita. The best statistical fit among these four, that being the strongest predictor model, were the first two variables, newspaper circulation per capita and tractors per hectare, together producing an R² score of .87. Consequently, these two variables accounted for 87 percent of the variance among the ten traits in forecasting Latin American democracy. Interestingly, a similar regression was performed between these variables and the 1945-1995 democracy rankings, with a .84 rho score for the newspapers and tractors variables (Kelly 1998, 10). In sum, democracy seems to be present wherever substantial per capita newspaper circulation and widely mechanized or tractor-oriented agriculture exist, according to the Fitzgibbon democracy surveys.

Comparisons of Other Surveys of Democracy

No other of the extant democracy surveys focus exclusively on Latin America. However, this author is aware of four other polls that use different sorts of data and criteria to assess democracy, and rating tabulations of the Latin America states can be gleaned from their total listings. Accordingly, in this section of the article the author will briefly describe these four democracy rankings and make comparisons between them and the Fitzgibbon democracy scales.
For example, Kenneth Bollen (in Inkeles 1991, 3-20, especially 10, 16-19) combined six "subjective" indicators for democracy into single indexes for the years 1960 and 1965: "press freedom, the freedom that political parties have to organize and oppose the government, and the extent of government sanctions imposed on individuals and groups [in addition to] fairness of elections, whether the chief executive came to office via an election, and the effectiveness and elective/nonelective nature of the national legislative body," taking this data from several secondary sources. His tabulations for Latin America when culled from his universal data set quite closely corresponded to those years of the Fitzgibbon surveys, a spearman rho association of .85 in 1960 and of .82 in 1965 (Table Five).

Table Five - Kenneth A. Bollen

(Rank orders, lower scores = most democratic)

Rho Correlation Score = .85 Rho Correlation Score = .82

Compiled from: Bollen, Kenneth 1991. Political democracy: conceptual and measurement traps. In On Measuring Democracy: Its Consequences and Concomitants, ed.Alex Inkeles, 16-19). New Brunswick and London: Transaction Publishers.

The "Freedom Ratings" of Raymond Duncan Gastil (in Inkeles 1991, 21-46, especially 23, 38-42 ) offer another comparison to Fitzgibbon. For the years 1978 and 1988, he evaluated each country of the world "against a reference book description. . . of following news about a country in a variety of sources, and occasionally changing ratings when the news did not fit the established rating level. In effect, the author developed rough models in his mind as to what to expect of a country at each rating level, reexamining his ratings only when current information no longer supported this model." The Gastil system for 1988 closely paralleled Fitzgibbon's scale with a rho figure of .94, although this was not the case for 1978 with a much lower comparable figure of .45 (Table Six).

Table Six - Raymond Duncan Gastil

(Rank orders, lower scores = most democratic)

Rho Correlation Score = .45 Rho Correlation Score = .94

Source: Raymond Gastil. 1991.The Comparative Survey of Freedom: Experiences and Suggestions. In On Measuring Democracy: Its Consequences and Concomitant, ed. (Alex Inkeles, 38-42). New Brunswick and London: Transaction Publishers.

A third democracy index, a "Polyarchy Scale" of 1984 again of all nations by Michael Coppedge and Wolfgang Reinicke (in Inkeles 1991, 47-68, especially 49, 59-62), assembled data from a variety of "collection efforts" by the Department of State, Freedom House, and elsewhere into five rating variables: extent of suffrage, freedom of expression, freedom of organization, existence of alternative sources of information, and free and fair elections. Their scaling results exhibited fairly strong rho correlations (.68) when compared to Fitzgibbon's ratings as shown in Table Seven.

Table Seven - Michael Coppedge and Wolfgang H. Reinicke

Rho Correlations = .68 Rank Orders (lowest = most democratic)

Compiled from: Coppedge, Michael and Wolfgang Reinicke. 1991. Measuring Polyarchy. IN On Measuring Democracy: Its Consequences and Concomitants, ed. Alex Inkeles, 59-62. New Brunswick and London: Transaction Publishers.

Finally, Ted Robert Gurr, Keith Jaggers, and Will Moore (in Inkeles 1991, 69-104, and especially 72-73; 82-83) expanded and updated a POLITY I data set of others into their own POLITY II and compared democracy and autocracy tendencies for 1978 of states from two world regions, one being Latin America. Their rankings differed significantly from Fitzgibbon's, revealing a low rho score of .34 (Table Eight).

Table Eight - Gurr, Jaggers, and Moore

Rho Correlations = .34 Rank Orders (lowest = most democratic)

Compiled from: Gurr, Ted Robert, Keith Jaggers, and Will Moore. 1991. The transformation of the western state: the growth of democracy, autocracy, and state power since 1800. IN On Measuring Democracy: Its Consequences and Concomitants, ed. Alex Inkeles, 82-83. New Brunswick and London: Transaction Publishers.

Accordingly, some tabulations of the other four approaches of measuring democracy closely approximated the Fitzgibbon survey and others correspondingly did not. But, again the strength of Fitzgibbon in contrast comes from its expansive life span of fifty-five years and the repetition of an every-five-year-cycle of surveys, the participation of a significant number of expert panelists, the wider variety of its democracy criteria, the flexibility of its definition and operationalization, and the ease of enlisting statistical measures for testing likely factors associated with democracy.

Conclusions

Much more could be said about democracy in Latin America and in general. Indeed, is democracy the most efficient form of government, the best protector of citizens' rights and of the natural environment, or the most peaceful (the "democratic peace" thesis)? Can the concept and practice of democracy be accurately defined and compared? Which type of environment is most fertile for the rise and maintenance of constitutionalism? What role has the United States played in the institutionalization of democracy in Latin America? How can any country or regional association like the Organization of American States promote political stability and democracy elsewhere? Such queries understandably are very difficult to answer, and obviously are well beyond the scope of this article.
Yet, it is asserted that the Fitzgibbon democracy survey project has made a positive contribution to the study of Latin American government and politics. We know, roughly at least according to the panelists' images, the most and the least democratic states, how these rankings have or have not changed over the past fifty-five years, and certain environmental attributes (per capita newspaper circulation and tractors per hectores, for instance) that could be statistically associated with constitutionalism.

¹ 2000 Fitzgibbon Democracy Survey Panel Participants: Juan del Águila, Emory University; Marvin Alisky, Arizona State; José Álvarez, University of Georgia; Christopher Anderson, University of Kansas; Craig Auchter, Butler University; John Bailey, Georgetown University; Steven Barracca, University of Texas - El Paso; Lorraine Bayard de Volo, University of Kansas; Marc Becker, Truman State University; Robert Biles, Sam Houston State University; Jan Knippers Black, Monterey Institute of International Studies; Alvaro Félix Bolaños, University of Florida; Dallas Browne, Southern Illinois University - Edwardsville; Winfield Burggraaff, University of Missouri; David Bushnell, University of Florida; Damarys Canache, Florida State University; Henry Carey, Georgia State University; John Carey, Washington University; Jack Child, American University; Richard Clinton, Oregon State University; Michael Coppedge, University of Notre Dame; Irasema Coronado, University of Texas - El Paso; Brian Crisp, University of Arizona; Alfred Cuzán, University of West Florida; Lee Daniel, Texas Christian University; David Dent, Towson University; Henry Dietz, University of Texas; Gary Elbow, Texas Tech University; Julio Fernández, State University of New York at Cortland; Cornelia Butler Flora, Iowa State University; David Foster, Arizona State University; Bill Furlong, Utah State University; Connie García-Blanchard, Fort Lewis College; John Garganígo, Washington University; Michael Gold-Biss, Saint Cloud State University; Louis Goodman, American University; Yvon Grenier, St. Francis Xavier University; Claudio Grossman, American University; John Hart, University of Houston; Richard Hillman, St. John Fisher College; Kathryn Hochstetler, Colorado State University; Jamie Elizabeth Jacobs, West Virginia University; Mark Jones, Michigan State University; Phil Kelly, Emporia State University; Harvey Kline, University of Alabama; Michael Kryzanek, Bridgewater State College; William LeoGrande, American University; Tom Leonard, University of North Florida; Todd Lutes, University of Arizona South; Don Mabry, Mississippi State University; Scott Mainwaring, University of Notre Dame; Christian Maisch, American University; Gabriel Marcella, U.S. Army War College; Jennifer McCoy, Carter Center; Terry McCoy, University of Florida; Ron McDonald, Syracuse University; J. Michael McGuire, University of the Incarnate Word; Frank O. Mora, Rhodes College; Stephen Mumme, Colorado State University; David Myers, Pennsylvania State University; Fred Nunn, Portland State University; Harley Oberhelman, Texas Tech University; Guillermo O'Donnell, University of Notre Dame; Salvador Oropesa, Kansas State University; David Scott Palmer, Boston University; John Passé-Smith, University of Central Arkansas; Neale Pearson, Texas Tech University; John Peeler, Bucknell University; Orlando Pérez, Central Michigan University; Anibal Pérez-Liñan, University of Notre Dame; Robert Peterson, University of Texas - El Paso; David Pion-Berlin, University of California - Riverside; Guy Poitras, Trinity University; Nancy Powers, Florida State University; Gary Reich, University of Kansas; Steve Ropp, University of Wyoming; Mark Ruhl, Dickinson College; Henry Schmidt, Texas A&M University; Cathy Schneider, American University; Friedrich Schuler, Portland State University; Mitchell Seligson, University of Pittsburgh; Eduardo Silva, University of Missouri- St. Louis; Shawn Smallman, Portland State University; Paul Sondrol, University of Colorado - Colorado Springs; Charles Stansifer, University of Kansas; Dale Story, University of Texas - Arlington; Dean Talbott, University of Northern Iowa; Robert Tomasek, University of Kansas; John Tuman, Texas Tech University; Roberto Villarreal, University of Texas - El Paso; Richard Walter, Washington University; Christopher Welna, University of Notre Dame; Joseph Werna, Southeast Missouri State University; Howard Wiarda, University of Massachusetts; Marvin Will, University of Tulsa; Edward Williams, University of Arizona; Miles Williams, Central Missouri State University; Philip Williams, University of Florida; Larman Wilson, American University; Ralph Lee Woodward, Texas Christian University; Eduardo Zayas-Bazán, Middle Tennessee State University; Daniel Zirker, Montana State University; Clarence Zuvekas, Annandale, Virginia.

Bibliography

Bollen, Kenneth. 1991. Political democracy: conceptual and measurement traps. In On Measuring Democracy: Its consequences and concomitants, ed. Alex Inkeles, 3-20. New Brunswick: Transaction Publishers.
Coppedge, Michael and Wolfgang Reinicke. 1991. Measuring polyarchy. In On Measuring Democracy: Its consequences and concomitants, ed. Alex Inkeles, 47-68. New Brunswick: Transaction Publishers.
Fitzgibbon, Russell. 1967. Measuring political change in Latin America . Journal of Politics 19: 129-166.
Gastil, Raymond. 1991. The comparative survey of freedom: Experiences and suggestions. In On Measuring Democracy: Its consequences and concomitants, ed. Alex Inkeles, 21-46. New Brunswick: Transaction Publishers.
Gurr, Ted, Keith Jaggers, and Will Moore. 1991. The transformation of the western state: The growth of democracy, autocracy, and state power since 1800. In On Measuring Democracy: Its consequences and concomitants, ed. Alex Inkeles, 69-104. New Brunswick: Transaction Publishers.
Johnson, Kenneth. 1976. Scholarly images of Latin American political democracy in 1975. Latin American Research Review 11: 129-140.
Johnson, Kenneth and Miles. Williams. 1978. Democracy, Power, and Intervention in Latin American Political Life: A study of scholarly images. Tempe: Center for Latin American Studies, Arizona State University, Special Studies no. 17.
Kelly, Philip. 1998. Measuring democracy in Latin America. In Assessing Democracy in Latin America, ed. Philip. Kelly, 3-11. Boulder: Westview Press.
Kurian, George. 1979. The Book of World Rankings. New York: Facts on File.