An opinion widely expressed in the literature on political communication is that the media play a fundamental role in shaping the political agenda ( JONES and BAUMGARTNER, 2005 ; WALGRAVE and AELST, 2006 ). Case studies ( SEVENANS et al., 2015 ; SOROKA, 2002 ), comparative studies ( MAURER, 2011 ; SEVENANS, 2018 ) and specific topic studies ( LIU et al., 2008 ) have demonstrated that politicians respond and are sensitive to media priorities. Legislators operate with a significant degree of voter preference uncertainty ( BUTLER and NICKERSON, 2011 ). The media play a paramount role as a source of information for mapping public opinion, thereby connecting voters and elected representatives ( BAUM and POTTER, 2008 ).

The issues that receive the most media attention are likely to have a higher status on the political agenda ( SEVENANS, 2018 ). To use a quantitative parameter, a meta-analysis of 90 empirical studies found an average correlation of r = .53 between the media agenda and the public agenda ( WANTA and GHANEM, 2007 ). Media coverage influences which issues the public perceives to be relevant and shapes the collective view of how issues should be resolved ( IYENGAR and KINDER, 1987 ). Based on a case study that considered members of the Israeli parliament, Cohen et al. (2008) demonstrated the ‘influence of the presumed influence of the media’, where legislators’ perceptions of the power of the media increase their efforts to appear in media coverage, thus reinforcing the media’s prominence in the political aspirations of legislators. In view of the realization of the importance of the media in the public exposure of legislators, ministerial portfolios that get more coverage in the country’s mainstream media vehicles is of greater interest than others to a party in any horse-trading with the president.

From an empirical point of view, media coverage of a ministry is often used to measure the prestige and reputation of State bureaucracies ( CARPENTER, 2001 ; GILAD et al., 2013 ; MAOR and SULITZEANU-KENAN, 2015 ). The prestige of a bureaucracy can be defined as a set of beliefs held by the external public about the exclusive capabilities, roles and obligations of a given ministry ( CARPENTER, 2010 ). In the case analyzed here, the external public is the media, due to their relevance to the political parties and legislators mentioned above. To assess the prestige component of a ministry in isolation, it would be necessary to use sentiment analysis of media content by calculating the frequency of positive and negative words associated with the names of bureaucracies ( CARPENTER and KRAUSE, 2012 ). As we describe below, we analyze media exposure and add the frequencies of mentions of ministries, be they positive, negative or neutral. In this way, the quantification proposed here adds both the prestige of a ministry and the degree to which it receives public attention, thus capturing a construct exogenous to the state institutional structure. We call this new variable ‘media’. In the following section we present the description of this variable as well as the results of its insertion into the ministerial importance index comprising the variables ‘positions’, ‘budget’ and ‘legislative initiatives’.

In order to determine the degree of media exposure of ministries, we chose to use the front pages of the three Brazilian print newspapers with the largest readerships, the Folha de São Paulo (hereinafter ‘FSP’), O Estado de São Paulo (hereinafter ‘Estadão’) and O Globo. The front page of a print newspaper highlights the primary content of each daily edition by offering greater public exposure to that content ( MEDEIROS et al., 2010 ). The front page has the advantage of being available online to readers who do not subscribe to the newspaper, in addition to its physical exposure to non-purchasers at points of sale. The front pages of print newspapers also have an important influence on the formation of the most-commented topics on social media ( OFFENHUBER and SCHECHTNER, 2013 ), which increase their dissemination further still. These elements confirm the capacity of front pages to reach a significant number of people, for which reason they are frequently used as markers in media coverage studies ( DI TELLA and FRANCESCHELLI, 2011 ). Sampei and Aoyagi-Usui (2009) , in an analysis of the exposure of the topic of global warming in the three main newspapers with Japan-wide circulation, demonstrate that the number of articles on their front pages had a significant impact on the degree to which the Japanese public was concerned about global warming. Thus, the frequency with which ministries appear on the front pages of major newspapers offers an excellent source of data to measure the degree of public exposure of a given portfolio, as it is a variable that can be counted daily in an ongoing, consistent and easily accessible manner. As mentioned earlier, our data cover three presidential terms between 2003 and 2014 ³ and essentially consist of accounting for the absolute frequency with which government ministries appear on the front pages of newspapers. In Graph 01 below, we represent the standardized values of the averages of the three newspapers (solid points) and their respective standard deviation (vertical bars) by ministry. In Annex , Table 04 (supplementary document), we present descriptive media exposure statistics disaggregated by newspaper and ministry.

Graph 01 Source: Elaborated by the authors, based on data from FSP, Estadão and O Globo.

Graph 01 shows that the ministry with the highest degree of exposure in the media is Finance, followed by Education, Justice, Health and Foreign Affairs. As expected, given the centrality of these ministries in the creation of policies and goods of high sensitivity to civil society, their exposure in the media is higher than others, indicating greater relevance. For parties and politicians seeking greater visibility on the national scene, these are the ministries that dominate the public agenda. The ministries that appear least in national media are Sports, Tourism, Social Development, Agrarian Development, Fisheries and Science and Technology, so they attract less interest to politicians interested in the visibility of their portfolios.

As our primary purpose was to add the variable ‘media’ into an index composed of the three original variables, ‘positions’, ‘legislative initiative’ and ‘budget’ ( BATISTA, 2017 ), we proceeded with factor analysis as a method for composing a single index. Thus, with all standardized variables, we estimated a new factor analysis that included the four variables. The estimated factor analysis showed the utility of the statistical model, showing slightly better values than Batista’s original model (2017). The KMO test (Kaiser-Meyer-Oklin) in the original model scored 0.513 and 0.690 in the new index with the variable ‘media’. The Cronbach’s Alpha coefficient also showed an improvement, scoring 0.688 and 0.742 respectively. The Barllets test for sphericity also demonstrates adequate significance in the estimation of factor analysis with media. The test values are available in Annex , Table 05 (supplementary document).

The presentation of the results of the factor analysis follows the same structure and steps performed by Batista (2017) in order to ensure the best comparability of the original results with those including the media exposure of the ministries. Table 01 below shows the ‘communalities’ of the estimation. This indicator returns the proportion of variances associated with the factors estimated in the factor analysis by variable. The closer to ‘one’, the better represented the variables are in the common factorial space. Variables with an extraction indicator greater than 0.5 must be included in the final index.

Table 01

Variable	Original		With Media
	Initial	Extraction	Initial	Extraction
Positions	1.0	0.829	1.0	0.786
Legislative Initiative	1.0	0.649	1.0	0.653
Media	-	-	1.0	0.612
Budget	1.0	0.394	1.0	0.258

Source: Batista (2017) and FSP, Estadão and O Globo.

As we can see, the media variable has significant relevance in the dimension estimated by the factor analysis, presenting a value of 0.612, slightly below legislative initiative capacity and the availability of commissioned positions in the ministries. As in the original estimate, ‘positions’ and ‘legislative initiative’ are relevant variables in the composition of the index, whereas ‘budget’ is not. In fact, with ‘media’ included in the index, the importance of ‘budget’ decreased further still. Nonetheless, we kept it in the aggregate index in view of Batista’s argument (2017, p. 22) that the importance of ‘budget’ lies in its direct connection to the power to spend money in ways that can maximize electoral return, which makes it central in theoretical terms. Table 02 depicts the variances of the four components.

Table 02

Component	Initial eigenvalues				Extraction sums of squared loadings
	Total	% Variance	Accumulated %	Total		% Variance	Accumulated %
1	2,309	57,713	57,713	2,309		57,713	57,713
2	.885	22,033	79,74
3	.515	12,478	92,225
4	.312	7,775	100,000

Source: Batista (2017) and FSP, Estadão and O Globo.

The values of the four components estimated in the factor analysis are indicated in the component column. The total column returns the eigenvalues. The higher they are, the greater the variance in the estimation. The next columns return the absolute and accumulated percentage of each component in the total variance. Following the parameters established by Batista (2017) for extracting the components in the construction of the index, the eigenvalue must be greater than ‘one’ and the percentage of variance greater than 60. As in the original estimation, which did not include ‘media’, only the first component survived the criterion and made it into the composition of the final index. Unlike our other estimates, the original model reveals a slight improvement in utility with respect to the percentage of variation of the first factor with 62.3% against 57.7% of the new index with ‘media’. Even so, as the first component has a high value, 2.309, and 57.7% is very close to 60%, it is reasonable to consider the utility of the new estimated index.

Table 03 below shows the results in the component matrix. The results are the correlations between the factor and each variable. Again using the same criterion adopted by Batista (2017) , results greater than 0.30 are considered the minimum requirement for the variable to be inserted into the final index. All four variables exceed this minimum requirement, showing high correlations with the factor, indicating that the extracted factor represents the original variables well. In addition, all four variables show positive correlations with the factor by varying in the same direction.

Table 03

Variables	Component
	Original	With Media
Posts	0.91	0.88
Legislative Initiatives	0.80	0.80
Media	-	0.78
Budget	0.62	0.51

Source: Batista (2017) and Folha de São Paulo, Estadão and O Globo.

The results of the factor analysis with ‘media’ included are satisfactory in the composition of a single index. Graph 02 below shows the final arrangement of Federal ministries according to their relevance in the four dimensions combined.

Graph 02 Source: Batista (2017) and FSP, Estadão and O Globo.

The results show the strong relevance of the Ministry of Finance in the composition of the final index and its prominence in relation to the other ministries. The Ministry of Finance handles such matters of great importance to civil society and government as macroeconomic and microeconomic policy, tax and revenue collection, the international representation of the country in financial affairs and specific policies for sectors such as agriculture and energy. Thus, the Ministry of Finance has a transversal role vis-à-vis other ministries and is the most important ministry in respect of national policy. Besides Finance, such ministries as Planning, Health, Justice and Education are also important. The index shows that the ministries with the least relevance and importance in terms of political negotiations are Fisheries, Sports, Tourism and Communications. In fact, the Fisheries, Sports and Tourism Ministries scored close to zero in the importance index. To compare the new index including ‘media’ and the original Batista index (2017) we used Graph 03 , shown below, with a comparison of the averages by ministry in both indexes estimated via factor analysis.

Graph 03 Source: Batista (2017) and FSP, Estadão and O Globo.

The first aspect that stands out in Graph 03 is the lower score that the index with ‘media’ gives almost all ministries than in the index without ‘media’. In the ministries of Planning and Welfare, for example, scores with ‘media’ are much lower than scores without ‘media’, which increases the distance of these ministries from the Ministry of Finance in the index with ‘media’. Low scores in the estimate with ‘media’ show that the media exposure of these ministries is low, in contrast with potentially high scores for ‘positions’ and ‘budget’. The index with ‘media’ makes a kind of adjustment. Some ministries in the middle range of the graph swapped positions when ‘media’ was inserted, as in the case of Education and Social Security. However, the importance ranking for most ministries changed little from one index to the other. This indicates the high correlation between the media and the other variables directly related to the government. Ministries with larger budgets, positions and legislative activity are also those that appear more in the national media, and that makes it possible to aggregate these variables in a single dimension.

The empirical utility of inserting media exposure in the same dimension composed of variables of different natures allows for theoretical gain in measurement of the importance of ministries. As expected, there are no profound differences between the indexes with and without ‘media’ in Graph 03 . The main contribution of the incorporation of the media into the final index of ministerial importance is its conceptual expansion, which adds a theoretically subjective element to the understanding of political behavior. The media exposure of a ministry expresses a component of ministerial importance related to the perception among the political elite of the role of the media in influencing public opinion. The index with ‘media’, in this sense, extrapolates the variables objectively connected to government structure and its relationship with the legislature. The theoretical support for the influence of the media on the political elite, as well as its empirical correlation with the other variables, gives the index with ‘media’ a greater conceptual breadth of ministerial relevance.

The Ministry of Foreign Affairs (‘MRE’), highlighted in red in the graph, is one case where the correlation between variables endogenous to the government and the media is lower. As shown in Graph 03 , the value in the index with ‘media’ exceeds the value of the index without it. The MRE has a limited budget, low investment capacity and few staff positions, which are usually held by the State (i.e. civil-service-examination-selected) bureaucracy and is therefore not very attractive for political parties. However, the MRE is usually described as very prestigious ( CHEIBUB, 1989 ; BURGES and BASTOS, 2017 ). Its high media exposure (the fifth largest) is an objective indicator of this prestige. As Batista (2017) says: “[...] the Ministry of Foreign Affairs is among the most important ministries, and in our ranking, this ministry is ranked among the least important ones. Is this a major flaw in the measure? It is possible” ( BATISTA, 2017 , pp. 18-19). Using the framework of Batista’s index (2017) while adding the media exposure of the ministries fills this conceptual gap and proves its empirical utility here, in a single dimension.

Finally, in keeping with Batista’s analysis (2017), we include Graph 04 below with the variability of the index with ‘media’ in ministerial portfolios by presidential term in order to assess the stability of the index over time. Very large variations in the distribution of ‘positions’, ‘budget’, ‘media exposure’ and ‘legislative initiative’ could hinder analyses that use ministerial importance in the study of political bargaining between the executive and the legislative branches. In Graph 04 below, the vertical axis represents the values of the importance index estimated by the factor analysis with the presence of ‘media’. The presidential terms, corresponding to the ministries’ averages, are represented by black circles (Dilma I) and gray circles (Lula I and Lula II).

Graph 04 Source: Batista (2017) and FSP, Estadão and O Globo.

The most obvious and useful aspect to bear in mind when analyzing of the utility of the aggregate index is that it presents no significant variability between ministries by presidential term. In other words, the measure ‘importance of ministries’ is satisfactorily stable over the years. The Ministries of Finance and Social Security showed the greatest variability, with higher scores in Lula’s first term. The stability of the index over the three presidential terms reveals low sensitivity to external shocks, such as economic crises and corruption scandals. However, in the specific case of the variable ‘media’ it would be extremely relevant to consider the effect of external shocks in the quantification of media exposure of the ministries in future research. Corruption cases, for example, can increase media exposure, affecting the performance of the variable media. In general, the results described throughout this research note demonstrate the empirical utility and theoretical value of inserting a value corresponding to ‘media exposure of ministry’ in an aggregate index of their relevance. The insertion of ‘media’ in the composition of the final index incorporates a dimension exogenous to the institutional apparatus of the Brazilian State, and this increases its conceptual scope.

Measuring the relevance of ministries to political parties is an empirical element of great value for studies on Brazilian multiparty presidentialism, especially vis-à-vis negotiations between the Presidency and Congress in the search for a legislative majority. The inclusion of media exposure in the composition of the ‘relevance index’ of ministries ( BATISTA, 2017 ) offers a contribution with high theoretical relevance, by incorporating subjective elements such as the prestige and public projection of a given ministry. The media play a central role in determining the political agenda and public perceptions and are important actors in determining the politicians’ prospects of electoral success. Greater media exposure of a ministry, and of the person who heads that ministry, means greater public projection and potentially enhanced electability. This research note has described and tested a method for creating the variable ‘media exposure’ and added it to Batista’s ministerial relevance index (2017), previously published in this magazine. Our results demonstrate the utility of this variable, thereby contributing to the theoretical improvement of an indicator with high potential for empirical and conceptual contributions to studies on the relations between the executive and legislative branches of Brazil’s government.