A common method for classifying occupations in studies on parliamentary elites is the one used by the EurElite Project ( Best e Cotta, 2000b ).

Since the mid-1980s, the EurElite Project has gathered a network of researchers interested in investigating political recruitment patterns by examining the social background and career pattern of individuals elected to national parliaments in 11 European democracies ⁸ . EurElite ’s objective was the historical comparison of trends in the profile of the European parliamentary elite profile from 1848 to 1999. The main problematic was in how to gather information on so many years of parliamentary history from several countries, each with its unique trajectories. Besides the difficulty in collecting such information, there was also the need to operationalize it into a manageable database, that is, one able to express with some clarity the interdependence between changes within the social structure and transformations in parliamentary representation over the course of 150 years. This meant producing generalizations based on the effects that the extension of suffrage, the emergence of mass parties, the new means of communication, and the crisis regarding the Nation-State had on parliamentary representation on a supranational level.

The solution adopted was the DataCube . Its purpose was to homogenize in a database a set of identical indicators (or equivalents) that would encompass the analyzed countries. The DataCube is no more than a representation of the data matrix, capable of organizing the information produced by two dozen researchers in just three dimensions. The first dimension covers the set of countries or party families which parliamentarians belong to; the second dimension is time, while the third dimension is the set of variables used in the research. Thereby, the information from each measured variable could be crossed per country or party over time. The variables contain data on the parliamentarians’ social background, information concerning their education level, prior occupation, age when they came into office, but above all it collects data on their political career prior to the exercise of their mandate, including local jobs and positions, party leadership positions, rate of parliamentary renewal, etc. ⁹

With the DataCube , the EurElite Project was able to circumvent a major difficulty: data dispersion. It then became possible to refine the understanding of the interrelationship between social changes and changes in each country’s structure of political opportunities towards a greater democratization of opportunities. As expected, a major variance was verified in how the located indicators behaved over the course of 11 years.

Nevertheless, when one adopts a more general point of view, that is, by observing the results in longer temporal series, it becomes possible to compare the peculiarities of national cases with regular changes in the social configuration of the parliamentarian board during the course of four major periods. The first period runs from 1848 until the 1880s, when European parliamentarians displayed a social profile very close to the ancien régime aristocrats (extremely high educational level when compared to the rest of the population as well as being landowners). They are the “dignitaries”. The second period encompasses roughly four decades (1880 until 1920) and is a time of change within the political elite. Nobility bonds cease to be a prerequisite for arriving to parliament and the old rural aristocracy gradually leaves the scene with an increase in the group of “semi-professional” political representatives: lawyers. However, this is a very gradual process since the parliamentary renewal rate in most elections remains under 40% ( Cotta e Best, 2000 , p. 512). The third period goes from 1920 until the 1960s, the era of mass democracy. From the 1920s onwards the parliamentary elite renewal process becomes clearer. Recruitment becomes more representative of the social structure with the arrival of the working-class newcomers. The percentage of blue-collar-worker parliamentarians increases in all countries and it is in this period that activism and participation in major partisan and union organizations becomes a precondition for a successful parliamentary career. It is the time for “party employees”. Eventually, from the 1970s onwards the social configuration of the European elites converges towards a sort of “middle class”, one that possesses a medium/high education level, arrives into parliament circa aged 40, has significant political life experience, and usually makes a living from the political profession. They are the “professional politicians”.

The story of the transformation of the parliamentary representation profile may be told from four key indicators: i ) level and type of education (diploma); ii ) belongingness to the nobility; iii ) economic sector of origin (primary, secondary, etc.); and iv ) professional or social background. The last variable was deemed a social position indicator at the time of debut in a political career. According to the model suggested by the EurElite Project ( Best e Cotta, 2000b , pp. 25-26), the 15 occupations considered were as follow:

Social Background Indicators

We based ourselves on this classification of occupations and applied two tests to our database to investigate their analytical performance. We included an additional variable, “professional politician”, i.e., an individual who has never practiced any effective occupation outside of politics. This variable was absent from the original model, but we consider it vital due to the particularities of our population.

The most glaring obstacle here is data dispersion due to the high number of occupations. This, however, is the least of troubles. The variable “Professions other than the law”, a kind of residual variable, has very high values in our case. This is because it encompasses professions that have traditionally delivered many politicians in Brazil, such as medical professions (an average of 11.45% in the considered period). Moreover, the aggregation of a plethora of liberal professions under a single label hides, for example, “engineers” and “economists”, and is thus unable to document the entrance rate of more technical and less traditional crafts into the national political class. The recurrence of ‘Lawyers’ throughout the whole series (an average of 21.6%), the constant presence of “Journalists and other writers,” and the vigorous appearance of businessmen (urban and rural) in more recent legislatures could lead us to conclude, based on the conventional classification, that there is an excessive closure of the political market to new occupations and new qualifications. Therein lies the difficulty in this aggregation of occupations to grasp, for the Brazilian case, the confluence between modernization (of society), democratization (of the political market), and professionalization (of political agents). The most significant presence of professional politicians since the 1990s may only be documented due to its inclusion in the set of variables, as it was not envisaged in the original list. The value of 11% of politicians among Senators in the past two decades could even be underreported due to the difficulty of obtaining accurate information from sources. In any case, it is a number greater than, for example, “Lawyers”, which has been decreasing at each legislative term (9% in the 2000s decade).However, another set of problems exists, not regarding the empirical inadequacy of the variables to our universe, but rather methodological obstacles.

Despite the analytical advantages of descriptive categories for shorter time diagnostics, these conventional classifications present a special obstacle for diachronic analyses. This difficulty in grasping large temporal changes is due to, among other things, i) the omission from the descriptive classification of the political weight of each profession in each historical moment (certain professions in certain periods are true incubators for political vocations ( Offerlé 1999 , p.10), but not always), and ii) the relational prestige of each occupation towards another. This is because of the change in status of a certain occupation in a given social space. For example: from the descriptive classification, we could be led to believe that “Journalists and other writers” had the same social importance in the 1920s as in the 2010s. Being a journalist at the end of the Old Republic (an eventual writer, publicist, or even editor for a partisan paper) meant something completely different, and much more valued, than being a radio broadcaster or TV host in our current democracy. The same reasoning applies to pastors and priests, landowners, and so forth ¹⁰ . However, the descriptive categorization of the EurElite Project would lead us to conclude towards similarities rather than differences among professions of origin across different chronological periods. Therefore, this portrait may be a superficial smudge (from a sociological point of view) as to who these individuals are across different historical moments.

Even the “Lawyer” category, an omnipresent social type in Western parliaments and whose characterization could be peaceful at first, cannot be taken anti-historically. Dezalay and Garth underlined the deep transformations within this profession during the XX century. In Brazil and in other important Latin American countries, the gentleman politician of the law gave way to the technocrat-specialist-globalized-lawyer ( Dezalay e Garth 2002 , pp.18-21). Within the Brazilian parliament, the “lawyer” from the 1950s is not the same as the 2000s “lawyer”. To further complicate matters, this diachronic variation also requires that the aggregation of professional categories must be contextualized by considering parameters, which for most situations, are not identical across different countries or different regional configurations.

Let us see how EurElite Project ’s classification for professions behaves in the case of individuals elected to the Brazilian Federal Senate between 1918 and 2010 if we divide this interval into 4 different political periods (and thus introducing, if not a , at least a tangible temporal reference). This precaution is necessary since between 1890 and 2010 Brazil had three coups d’état (1930, 1945, 1964), six constitutions (1891, 1934, 1937, 1946, 1967, and 1988), two different party systems – multi-party system between 1889 and 1937 (with the parties being mainly regional) and 1945-1965 and 1980 until today (with national parties), and bipartisanship between 1965 to 1980 (during the military dictatorship) – and an interval with no political parties or elections (1937-1945). Institutional ruptures and constant changes in the rules defining the political game had effects on the process of selecting political leaders.

In our timeline, the first period goes from 1918 to 1937, that is, from the height of the oligarchic politics until the coup d’état, which abolished all political parties in 1937, closed down the legislative houses, and terminated parliamentary mandate terms. The second period, from 1945-1962, encompasses an important cycle in Brazilian politics: the “populist” democracy, which ends with the 1964 military coup. The third period ranges from the military dictatorship until the liberalization of authoritarian control over the political system (1964-1982). Lastly, the fourth period is from 1982 until 2010 and spans the final political transition period, the consolidation of liberal democracy in Brazil until the current regime. The contingency coefficient in Table 2 indicates the average association between EuroElite Project ’s classification categories and the periods established by ourselves. The standardized residues show the difference between the expected values and the ones found.

TABLE 2

	I		II		III		IV
	First Republic (1890/1937)		Populist Democracy (1945/1962)		Military Dictatorship (1966/1982)		New Democracy (1986/2010)
	%	sr	%	sr	%	sr	%	sr
1.Noblemen legislators	–	–	–	–	–	–	–	–
2.Teachers/Professors	2,6	-4,2	8,3	1,1	6,9	0,3	15	6
3. Journalists and other writers	5,6	-1,2	7,1	0,3	10	1,5	7,8	0,8
4. Political party/trade union employees	–	–	–	–	–	–	–	–
5. Civil servants	0,6	0,4	0,6	0,2	–	0,8	0,3	-0,4
6. Public sector employees	6,9	1	3	-1,6	3,1	-1,4	6,8	0,5
7. Military persons, all levels	15,9	3,2	13,1	0,4	8,5	-1,2	2	-4,9
8. Priests, all clergymen	0,8	-0,2	1,8	1,2	–	-1,1	1	0,2
9. Lawyers, practising	22	1,1	24,4	1,2	20,8	0,1	12,9	-2,8
10. Judges, Prosecutors	8,2	2,2	7,7	0,8	1,5	-2,2	2	-2,9
11. Primary sector, agriculture	2,7	-1,1	6,5	2,2	2,3	-0,7	4,1	0,6
12. Blue-collar workers	–	-2,2	–	-1	–	-0,9	2,7	5
13. Managers, “businessmen”	3,9	-3,8	7,7	0,1	12,3	2	15,3	4,9
14. Professions other than the law	20	-0,4	19	-0,4	27,7	1,8	20,1	-0,2
16. “Professional politician”	10,9	1,7	0,6	-3,7	6,9	-0,8	9,9	0,4

Contingency coefficient = 0,376 (sig = 0,000).

In order to analyze Table 2 we must compare the percentage values of the occupations in the four columns indicative for each period. For the majority of the professions one may notice a great difference between columns, and therefore a heterogeneity among the periods, which is confirmed by the positive and negative values of the standardized residuals. This holds particularly true for ‘Teachers/Professors’, ‘Lawyers’, ‘Military persons’, ‘Managers, businessmen’, and ‘Blue-collar workers’, in which there exists a high variation at each interval and notable opposition in the underlined standardized residual values.

Nevertheless, what does this entail, in terms of sociological relevance, beyond the openly stated? That is, that the total number of lawyers plummeted from 22% at the beginning of our series to 13% almost one hundred years later? Or that we rarely find ex-military members in the Brazilian Senate? Or that businessmen have significantly increased their presence within the political elite?

As we have previously suggested, descriptive professional categories fail to clearly recognize morphological changes in the political class over time. Even if the variables are precise (their meanings immediately recognizable), they are less revealing of long-term transformations within this population, since they describe activities whose names remained constant over time, but for which their social status and political weight have changed across the political space according to different historical periods. Therefore, it is not possible to transport these occupational labels from one historical period to another without considering the different political and extra-political resources mobilized by the holders of these professions. Likewise, these resources are neither identical or their influence the same throughout a significant interval over time.

We may find ‘Lawyers’ in all political periods (especially in the pre-1986 period). If we look exclusively at occupations, we could believe that parliamentarians elected in recent periods (where we have a larger number of ‘businessmen’) are increasingly recruited from the high society, in the “social elite”. In reality, the Old Republic ‘Lawyers’ may belong to the high society (and in fact they did), while businessmen in current legislatures may have a more wealthy origin, albeit alongside an extensive political career and political party activism. Therefore, the values in each professional category, while correct, may fault to recognize the measurements and decline verified in the contingent of lawyers. In turn, the increase of businessmen may hide, for example, the politic professionalization of those elected for the Brazilian Senate.

The above classification and the brief interpretation of the tests presented in Tables 1 and 2 allows us to discuss the difficulties of unreflectively importing and applying occupational categories in elite studies. A large share of the potential inconsistencies from such occupational categorizations happens because of its all-encompassing and blunt usage.

TABLE 1

	1890	1900	1910	1920	1930	1940	1950	1960	1970	1980	1990	2000
1. Noblemen legislators	–	–	–	–	–	–	–	–	–	–	–	–
2. Teachers/professors	4	3,1	2,3	–	1,5	5,7	11,9	5,3	7,9	9,7	13,6	17,8
3. Journalists and other writers	6,6	3,4	3,4	7,2	8,8	5,7	5,1	12,3	7,9	12,5	7,3	7,4
4. Political party/trade union employees	–	–	–	–	–	–	–	–	–	–	–	–
5. Civil servants	0,7	1,1	–	–	–	–	–	1,8	–	–	–	0,7
6. Public sector employees	10,2	8,4	2,3	1,3	5,9	4,3	1,7	3,5	2,2	6,9	5,5	7,4
7. Military persons, all levels	17,2	18,7	21,6	11,1	2,9	15,7	11,9	12,3	6,7	4,2	1,8	2,2
8. Priests, all clergymen	0,7	1,9	–	–	–	1,4	1,7	1,8	–	1,4	–	1,5
9. Lawyers, practicing	15,7	17,2	13,6	39,9	36,8	24,3	22	24,6	23,6	16,7	15,5	8,9
10. Judges, Prosecutors	4,4	10,7	15,9	7,8	4,4	12,9	6,8	–	2,2	2,8	0,9	2,2
11. Primary sector, agriculture	4,4	3,4	–	1,3	–	8,6	5,1	5,3	2,2	4,2	3,6	3,7
12. Blue-collar workers	–	–	–	–	–	–	–	–	–	1,4	2,7	3,0
13. Managers, “businessmen”	5,5	3,1	4,5	1,3	5,9	7,1	10,2	5,3	13,5	11,1	20	13,3
14. Professions other than the law	18,2	16	15,9	28,8	27,9	14,3	23,7	24,6	27	23,6	18,2	20,7
15. “Professional politician”	12,4	13	20,5	1,3	5,9	–	–	3,5	6,7	5,6	10,9	11,1
TOTAL	100	100	100	100	100	100	100	100	100	100	100	100

N mandates = 1,437 (+ 8 missing).

TABLE 3

			POLITICAL TYPES
			Type 1	Type 2	Type 3	Type 4	Type 5	Type 6	Type 7	Type 8	Total
POLITICAL PERIODS	1 st Republic (1890-1937)	n	62	79	148	272	42	54	95	99	851
		%	68,1%	63,2%	71,5%	70,5%	42,9%	37,8%	59,4%	42,1%	58,9%
		SR	1,1	,6	2,4	3,0	-2,1	-3,3	,1	-3,3
	4 th Republic (1946-1962)	n	13	18	24	36	16	16	14	31	168
		%	14,3%	14,4%	11,6%	9,3%	16,3%	11,2%	8,8%	13,2%	11,6%
		SR	,7	,9	,0	-1,3	1,4	-,2	-1,1	,7
	5 th Rep ublic (1966-1982)	n	8	14	14	24	8	10	22	30	130
		%	8,8%	11,2%	6,8%	6,2%	8,2%	7,0%	13,8%	12,8%	9,0%
		SR	,0	,8	-1,1	-1,8	-,3	-,8	2,0	1,9
	6 th Republic (1986-2010)	n	8	14	21	54	32	63	29	75	296
		%	8,8%	11,2%	10,1%	14,0%	32,7%	44,1%	18,1%	31,9%	20,5%
		SR	-2,5	-2,3	-3,3	-2,8	2,7	6,2	-,7	3,9
	Total	n	91	125	207	386	98	143	160	235	1445
		%	100,0%	100,0%	100,0%	100,0%	100,0%	100,0%	100,0%	100,0%	100,0%

N mandates = 1437 (+8 Missing).

The classification suggested below is an attempt to escape the traps of categories conventionally used in elite studies. This typology aims to contextualize the social distance between two variables identical in their terminology – ‘Teachers/Professors’, ‘Journalists and other writers’, Civil servants, etc. –, but apart in historical time. Therefore, the solution hereby outlined prioritizes a categorization that considers, at the same time, the weight of social and political status of political agents (in accordance with Matthews, 1961 ). This resource should allow for a contextual comparison of relative positions among occupations (or among their holders), that is, according to their time and place. These two dimensions (as well as educational status , which will be left out) are the dimensions mobilized by studies that rely on the debate on political professionalization for outlining changes in the parliamentary recruitment pattern. The chief inspiration for our model was Eliassen e Pedersen (1978) .

According the Weber’s classic formulation, the substitution of the notable for the professional politician is concomitant to the decrease in importance of the social status (prestige, possessions) and the increase of the individual’s political status (career, positions, specific vocation) as vectors for establishing an elite elected in a given society. We adopted both of these dimensions, adapting these two vectors to specific social types throughout Brazilian political history. The model indicators were selected inductively based on the peculiarities of federal senators. Since the senatorial political elite was, during the I Republic, highly connected to prestigious liberal professions and held an extensive political career, we noticed that simple occupational indicators would lead to the conclusion that in Brazil politicians “de-professionalized” themselves over time, since Parliament begins to admit not quite the typical figure of the professional politician (such as the one found in the EurElite Project ), but instead businessmen, public employees, teachers.

Our social dimension model seeks to capture the variations between senators from the indicator “occupation before entering a political career”, but with one difference in the descriptive typology of occupations. Occupations were divided into two groups in function of their social values in Brazil. Some professions are “more noble” than others. Thus, the assumption is that lawyers, businessmen, doctors, and diplomats (all of which “elite” professions in Brazil) arrive into Senate with a higher social status than those in less prestigious occupations (economists, engineers, and other non-traditional liberal professions).

In turn, the political dimension combines two indicators. The first refers to the profile or political career orientation, measured from the number of municipal, state, and national offices occupied by an individual before arriving into Senate. If the majority of an individual’s offices are at the national level ¹¹ , we infer that their ambition is focused on the national political game. The reverse situation aggregates individuals whose ambition is directed towards state/local politics.

The second political dimension indicator is political vocation. To assess the occurrence of this quality in the biography of Brazilian senators we measured their age at the time of their first political office and the period they devoted to political life before being elected to the Federal Senate. Thus, we hope to identify the vocation for politics in those who started younger and who have devoted longer time to this activity ¹² .

The proposed analytical model consists of three dummy variables:

The logical combinations from the six qualities lead us to eight “political types”:

To what extent are such logical types also historical types and capable of explaining the change in the profile of individuals who reached the Brazilian Senate since 1918? Are these types concentrated within a particular political period or randomly distributed throughout the XX century? Are these indicators – profession, career orientation, and political vocation – indeed suitable to capture the historical changes of the national political class?

In order to answer these questions, we conducted two tests, as described below.

First, we tested for any coincidence between these eight ideal types obtained by a logical combination of both dimensions (political and social) and the Brazilian senators’ biographical data within the studied period.

Table 3 distributes political types by the same referred periods as Table 2 . These periods encompass, roughly speaking, four political regimes: oligarchic regime (1st and 2nd Republics.) ¹³ , populist democracy (4th Republic.), Military dictatorship (5th Republic.), and new democracy (6th Republic.). The “Third Republic” does not appear on the list since it refers to the Estado Novo (1937-1945) when parliaments were closed, elections abolished and political careers interrupted.

The first thing we should stress is the concentration of more accentuated positive and negative standardized residues in the First and Sixth Republics. This trend towards a concentration of type-cases in regimes at the beginning or end of the studied period suggests that the proposed typology is sensible to modifications that affect the Brazilian senatorial elite over time, even if not wholly adequate.

We ran a correspondence analysis to verify the contiguity between logical types and historical regimes (Graf 1).

The test revealed a correspondence between points, marked in the graph in the colors white for logical types and black (filled) for political regimes. Type 1 (national career, political vocation, and elite profession) and type 3 (national career, elite profession, but no political vocation) and type 4 (individual with a national career, but with no vocation or elite profession) were closer to the 1st Republic. Type 5 (vocation and elite profession, but local career) and type 6 (political vocation, local career, but no elite profession) remained closer to the 6th Republic.

We then replicated the initial typology and aggregated types 1 and 3 in a new category named new type 1, and types 5 and 6 into a new category titled new type 2 (the remainder were also aggregated in new types 3 and 4, but they will be disregarded as they were not statistically significant). Thus, the final rankings arrived to the following new types:

The first new type fixes the values of the social dimension , measured by the occurrence of elite professions, and alternates the values of the political dimension . The second new type maintains fixed the vocation for politics and careers oriented to a state level, but alternates the values of the social dimension , combining into the same group senators who practiced and who did not practice elite professions in Brazil. New types 3 and 4 revealed themselves statistically irrelevant as well as from a sociological standpoint.

The following test presents the correlation results between new types and national political regimes (Graf 2).

The correspondence analysis of equidistant points between the two distributions reveals that new type 1 is closer to the 1st Republic and that new type 2 is closer to the 6th Republic. The opposite can be said for new types 3 and 4 : despite the proximity in the graph (which is only by chance) both show no statistic correspondence (presented in appendix 2). This means that the combination is not a mere logical deduction from the model, since it revealed itself able to follow the populace socio-political variations and its affinity with at least two regimes, the oligarchic and the democratic regimes.

Nevertheless, we may attempt a further sociological understanding of the possible connections between the political profile and political period.

Our model indicates that there are two characteristic types of professional politicians at the beginning and end of the analyzed period rather than a progression or replacement of types (the notable for the professional, the oligarch for the professional, etc.) as one could be led to believe solely with the variations of the amount of professions. Thus, there does not seem to be a less professional politician at the beginning of this series in contrast to a more professional one at the end. Instead, we find a change in the attributes of the professional politician in Brazil.

Four of our findings are worth underlining:

We recall here how the codification of the politicians’ prior professions may lead not only to indexations that distribute individuals per occupations and from thereon produce a socio-graphical mapping of the studied universe, but also how they orient their own explanations as to their recruitment processes and professionalization.

We present here a proposal of a new classification by confronting a logically deduced typology with the Brazilian contexts or political regimes of the XX century. It has revealed itself to be potentially capable of assimilating variations in the profiles of politicians’ careers as well as the social values ( status ) assumed by professional occupations throughout the national parliamentary history. Therefore, we understand that a classification of such individuals based on a double (political and social) allows for a better understanding of the historical recruitment dimension for the Brazilian case.

In more general terms, this exploratory work on income and the logical and methodological difficulties of using the variable ‘profession’ in studies of parliamentary elites allowed us to advance three conclusions.

First, the categorization of occupations prior to a political life seems to be necessarily ad hoc , i.e., its employment depends on the historical context.

The second conclusion asserts that purely descriptive classifications provide interesting resources for interpreting specific category variations in short periods of time. However, this same classification would produce, if it were used for the case of Brazilian senators, a mistaken inference on the evolution and changes in the politicians’ occupational origin types.

In this case, the third conclusion of this study points towards the potential return of a strategy that unites a long term theoretical lens with indicators more sensitive to the history of the analyzed cases, that is to say, contextual. In order to verify the strength of this proposal the same indicators suggested by our model should be tested in a wider range of cases.

GRAPH 1 Singular Value in dimension 1= 0,303 (sig 0,000). Singular Value in dimension 2 = 0,101 (sig 0,000). Source: Nusp/UFPR; The observatory of social and political elites of Brazil http://observatory-elites.org/

GRAPH 2 Singular Value in dimension 1= 0,243 (sig 0,000). Singular Value in dimension 2 = 0,064 (sig 0,000). Source: Nusp/UFPR; The observatory of social and political elites of Brazil http://observatory-elites.org/.