Pedro Graterol
ENESPT

Pedro Graterol · Latinobarómetro · 1995–2024

For Researchers

This publication draws on a frozen research foundation. The figures below retain its definitions and restrictions.

Research materials and downloads ↓

Research design and analytical units

Pedro Graterol’s frozen working draft, A Political Trust Project: Institutions and Government in Latin America, 1995–2024 (November 2026), is the publication’s research foundation. The archive comprises 457,499 interviews, 24 observed waves, 413 country-wave samples and 18 countries. Spain is excluded. Counts describe archive coverage, not the sample for every analysis.

Latinobarómetro provides repeated cross-sections. National samples, confidence levels, paired institutional changes, individual-level associations and country-time relationships are different analytical units. The design does not follow respondents or identify within-person change. Countries are the primary comparative units; no regional population-weighted estimate is implied.

Harmonization, coding and eligibility

The frozen v0.3 dataset links source items through wording, categories, coding and institutional referents. A shared variable name is insufficient evidence of comparability. Incompatible formats remain in separate measurement segments or are excluded. Original provenance is retained; unavailable items and missing years receive no substantive value.

Positive institutional confidence combines categories 1 and 2; substantive categories 3 and 4 form the negative group. Primary percentages use substantive respondents. Nonresponse, refusal, not applicable and unavailable items are distinguished where source information permits. Government approval concerns the government headed by the nationally named president, not confidence in the presidency.

The trajectory inventory contains 2,028 eligible institution-country-wave estimates. Modern armed-forces comparisons exclude Costa Rica and Panama. Nicaragua’s record ends in 2020. Approval begins in 2002 with country-specific gaps: 319 eligible country-waves in the item inventory become 317 common-sample model country-waves and 1,549 coefficients after additional eligibility restrictions.

The archive begins in 1995, but a comparison needs more than an early observation. It needs the relevant institutions and countries observed at both chosen endpoints, with their measurement restrictions retained. The documented 2010–2024 comparison pairs 17 countries for Congress, parties, judiciary and police, and 15 for armed forces. It is one analytical window, alongside earlier histories and other endpoint comparisons; 2010 is not a universal threshold of survey quality. Brazil illustrates the distinction: its 1995 Congress observation is available, but the early sample had partial urban coverage. Comparing it with 2024 therefore describes the covered samples, not an unchanged national population. Earlier observations have not been discarded. They remain in the historical explorer where defensible, with missing waves and design boundaries preserved. A common comparison window makes the country set explicit; the full record shows what that window leaves outside.

Harmonization makes definitions explicit; it does not make three decades of surveys identical. These are different respondents at each wave. Sampling frames, territorial coverage and unavailable survey-design details still limit inference. The findings describe confidence and associations, not individual change or causal effects.

Weights, denominators and national comparisons

National estimates use original source weights within country-wave samples. Descriptive cells require at least 100 substantive responses and Kish effective N of 50. Kish N diagnoses weight dispersion; it does not reconstruct the complete survey design. Regional summaries, where used, give equal weight to eligible national estimates.

Available-country, balanced-country and matched-endpoint summaries answer different questions. The 2010–2024 comparison has 17 countries for Congress, parties, judiciary and police, and 15 for armed forces. Confidence changes are percentage-point differences. Pairwise gap change equals the change in institution A minus the change in institution B; its interpretation requires both components.

Observed points are not interpolated into annual data. Country trajectories retain their defensible historical coverage. Comparing alternative endpoints reveals historical dependence rather than assigning countries permanent types. Institution-specific confidence percentages may use different substantive denominators; pairwise comparisons use joint-response samples where specified.

Government-approval models

Source-weighted linear probability models estimate positive confidence separately by country, wave and institution. Government approval is the focal binary predictor, with age-group and recorded-sex indicators. The coefficient multiplied by 100 is the adjusted percentage-point confidence difference between approvers and disapprovers. It is an individual-level cross-sectional association, not a correlation of national averages.

Within country-wave, institutional comparisons use a common eligible respondent sample across five institutions, or four in Costa Rica and Panama. Estimates are not independent across institutions. No linear temporal trend is imposed. A more positive coefficient need not be a larger absolute association if it moves from negative toward zero.

Extended blocks add employment, interviewer socioeconomic assessment, ideology, political interest and interpersonal trust; B3 includes one evaluative predictor and B4 all four. Interviewer assessment is not income. B3 approval covers 96 cells; B4 only 53 across seven waves. Same-sample comparisons separate adjustment from sample-composition changes. Attenuation is not mediation.

Robustness and uncertainty

Sensitivity definitions include unweighted substantive shares, all-interview positive shares, adult-only estimates, fixed age/sex standardization, and weighted and unweighted joint-response contrasts. An all-interview positive share is not a recoding of nonresponse as negative opinion. Unsupported standardized cells remain unavailable.

The extension’s paired covariance and uncertainty calculations remain incompletely verified. Direction is not statistical significance. Historical Benjamini–Hochberg adjustments used 435 institution tests per sensitivity, pooling evaluative predictors; the 88 approval series were a subset. The paired family of 850 comparisons is separate.

Political Trust: What the Research Tells Us

Political trust is often discussed as though citizens give their political system a single verdict. Research instead distinguishes the objects and judgments involved. Easton’s (1965, 1975) distinction between specific and diffuse support remains a useful starting point: evaluations of political authorities and their performance need not move together with more enduring attachments to the political order. This framework does not make a survey question about confidence a pure measure of diffuse support. It asks researchers to identify what respondents are evaluating before deciding what an answer means.

Norris (1999, 2011) develops this distinction through the idea of critical citizens: attachment to democratic principles can coexist with dissatisfaction with the institutions through which democracy operates. Dalton (2004) likewise examines declining political support in advanced industrial democracies without equating every institutional disappointment with rejection of democratic government. These arguments complicate a simple crisis narrative, but they do not establish that distrust is inherently democratic. Critical evaluation, disaffection and rejection are different possibilities. Distinguishing them requires measures that address each attitude rather than treating institutional confidence as a substitute.

Government approval is another distinct evaluation. It concerns the incumbent administration, whereas confidence in Congress, courts or police concerns institutions whose existence extends beyond that administration. Their relationship is an empirical question. A positive association may be consistent with several processes, including shared evaluations of performance or political preferences; it cannot by itself establish which process occurred. Catterberg and Moreno (2006) place political trust within a wider set of individual and political evaluations across established and newer democracies. Their analysis cautions against explaining changes through a single account of increasingly demanding citizens.

Institutional confidence is also differentiated internally. Shared evaluations are possible: Mishler and Rose (1997) examine commonality in assessments of institutions in post-communist societies. Yet commonality does not make each institution interchangeable. Algan’s (2018) review discusses political and law-and-order dimensions of institutional trust, drawing on OECD measurement work. It does not establish that every composite index is invalid. More directly, Valgarðsson and colleagues (2025) identify different long-run global trajectories for representative and implementing institutions. Their findings make the choice of institutional object central to interpreting claims about declining trust, while leaving room for variation within those broad groupings.

Latin American research gives particular reasons to preserve this differentiation. Solar (2022) finds that the correlates of military confidence vary across Argentina, Brazil and Chile. Sung, Capellan and Barthuly (2022) examine police and military confidence separately in relation to insecurity and militarization. These studies do not justify treating security institutions as a single public-opinion object. Nor do they establish a common cause for the historical changes presented here. They show why institutional roles and national contexts belong in the research question, rather than being treated as incidental differences around a regional average.

This project contributes a comparative historical account of those distinctions using harmonized Latinobarómetro surveys. It follows five institutional confidence measures separately and examines adjusted approval–confidence associations within observed country-waves. Regional comparisons are decomposed into national trajectories, and alternative windows show how similar endpoints can conceal different histories. The contribution is descriptive and associational: it documents which institutions moved, in which countries and periods, and how approval-related differences varied. It does not identify individual attitude change or establish a causal theory of legitimacy. Satisfaction with democracy, support for democracy, institutional confidence and government approval therefore remain separate concepts throughout. The audit also separates the reproduced numerical core from additional analytical summaries whose calculation records remain incomplete.

Bibliography ↓

What the evidence can and cannot establish

Survey design. These are repeated cross-sections: different respondents are interviewed in successive waves. National changes do not reveal how particular people changed their minds. Historical territorial coverage, sampling frames and survey modes are not perfectly invariant.

Measurement. Harmonization assesses wording, categories, institutional referents and comparability; it cannot erase all differences. Unavailable questions and missing waves remain missing, never zero. Lines between observed waves are visual guides, not annual estimates. Institutional eligibility differs, especially for armed forces in Costa Rica and Panama.

Statistical inference. National source weights are used within country-wave samples. Kish effective N diagnoses weight dispersion, not the full survey design. The approval models use inherited respondent-level HC1 standard errors with weights treated as fixed; unavailable historical sampling units, strata and other selection uncertainty are not fully represented. Pointwise intervals are not simultaneous guarantees across institutions.

Causal interpretation. Approval-related differences compare groups of respondents after adjustment for age and recorded sex. They do not identify the effect of approving a government, an election, a crisis or institutional control. A larger positive difference does not establish higher average confidence, greater legitimacy or polarization.

Historical coverage. The archive spans 1995–2024, but countries and institutions enter at different times. Nicaragua ends in 2020. Early partial or urban frames limit long comparisons; Brazil’s 1995 observation is not evidence of an unchanged national population. The 2010–2024 window uses matched endpoints and is not a universal quality threshold. Nominal survey years can differ from fieldwork dates.

National implementation. Of 27 focal or adjacent national presidential-name insertions audited, two were independently verified: Brazil 2016, Michel Temer, and Brazil 2020, Jair Bolsonaro. The other 25 remain NOT VERIFIED because the national instruments were not independently recovered, not because they were proven incorrect. Venezuelan Congress wording remains incompletely verified. Costa Rica 2002 cannot support attribution to a single president.

Incomplete extensions. The full original calculation records for additional individual-level correlations and PCA, national approval summaries and paired-change uncertainty were not recovered. Their unsupported numerical summaries have been removed from the current explanatory account. Reproducing underlying estimates does not independently verify later aggregations or uncertainty calculations.

Audit provenance. Completed checks reproduced the core estimates, errors and sample sizes. Some temporary execution files were lost after environment restoration; completed checks are retained from recorded execution results, not claimed as rerun during publication preparation. Interrupted checks are not counted as completed. Numerical agreement does not certify measurement validity or every interpretation.

Data redistribution. Original and harmonized respondent-level files are not included in public downloads. Source-data access remains subject to Latinobarómetro’s terms. Publication of aggregate findings is not treated as unrestricted permission to redistribute a derived database; materials with unresolved redistribution status are withheld pending clarification.

Research Audit

The audit asked whether the published findings could be traced to the research record and whether the numbers matched the available analytical data. Completed independent checks reproduced 2,028 institutional-confidence estimates and 1,549 government-approval coefficients, with matching associated standard errors and analytical sample sizes. No material numerical discrepancies were found in those checks. The Argentine 2024 illustration is part of this reproduced core.

The verification is narrower than a certification of the whole project. The summaries 0.58 and 0.12 were reproduced as medians of archived correlation values from an earlier analysis; their full construction from respondent-level data was not repeated, and their exact equivalence to the final extension’s coverage was not established. The third reported median, 0.39, lacked a recoverable original calculation record. The three-number comparison has therefore been removed from the main story.

Precise national approval summaries, additional correlation/PCA summaries and extension uncertainty results remain incompletely verified. They are not presented as established publication findings. Bibliographic errors and overstated verification labels have been corrected. The dated correction history distinguishes removed unsupported claims from numerical corrections: the reproduced confidence trajectories and approval coefficients have not been changed.

The frozen draft, audit and corrected publication serve different purposes. The draft preserves the historical research record and may contain superseded claims. The audit records its evidence and exceptions. The current methodology, story and correction history state what this publication retains. Some completed audit results survive through their recorded execution outputs rather than all original temporary files; they were not all rerun for this release.

Data use and attribution

Original surveys: Latinobarómetro. Harmonization, analysis, interpretation, writing and visualization: Pedro Graterol. This is an independent research publication; it does not imply endorsement or verification by Latinobarómetro.

The official download and time-series pages describe noncommercial research, teaching and publication uses while restricting redistribution of data. Public availability is not an unrestricted license. Original microdata, harmonized respondent-level records and third-party questionnaires are not repackaged here. Researchers should obtain source files directly from Latinobarómetro under its applicable terms.

Publishing aggregate statistical findings and distributing a reusable database are different uses. Where permission to redistribute a derived dataset remains ambiguous, the download package withholds that dataset and explains the dependency. Code and documentation do not grant new rights over source data. Commercial reuse or broader redistribution requires separate clarification. This is a preliminary source-based assessment, not a definitive legal opinion.

Inspect the research

The download hub separates the current methodology, audit record, original analytical code and source-data access instructions. Each item states what it can reproduce and which inputs are required. A public script is not described as fully reproducible if it depends on unavailable or separately acquired data.

Fully reproducible from public materials means all required permitted inputs and procedures are supplied. Reproducible after obtaining source data means the code and instructions are available but source files must be acquired separately. Partially reproducible means a procedure or intermediate input is missing. Documented but not independently reproduced means provenance exists without an independent repeat. Not verified means available evidence is insufficient.

The audit matrix describes verification of research outputs; the download inventory describes what a reader can run now. These are related but different classifications. Consult the package README, input requirements and checksums before running scripts. No missing analytical recipe has been replaced with a guessed procedure.

Support for democracy

A person can disapprove of the government, distrust Congress and still prefer democracy to any alternative. These judgments may overlap, but they ask different questions. Government approval evaluates the administration in office. Institutional confidence concerns a particular institution. Satisfaction with democracy evaluates how democracy works; support for democracy concerns its desirability as a political system.

Research materials and downloads

SHA-256

Latinobarómetro ↗

Bibliography

  • Aydın Çakır, A., & Şekercioğlu, E. (2016). Public confidence in the judiciary: The interaction between political awareness and level of democracy. Democratization, 23(4), 634–656. DOI / URL ↗
  • Bourret Soto, M. (2026). “This is our legacy”: Youth exposure to repression and trust in institutions. Journal of Institutional Economics, 22, e47. DOI / URL ↗
  • Catterberg, G., & Moreno, A. (2006). The individual bases of political trust: Trends in new and established democracies. International Journal of Public Opinion Research, 18(1), 31–48. DOI / URL ↗
  • Dalle Nogare, C., Ferrante, C., Pontarollo, N., & Segovia, J. (2026). What explains political trust in Latin America? Understanding the role of the urban-rural divide. Global Challenges & Regional Science, 5, 100026. DOI / URL ↗
  • Dalton, R. J. (2004). Democratic Challenges, Democratic Choices: The Erosion of Political Support in Advanced Industrial Democracies. Oxford University Press. https://academic.oup.com/book/36196 DOI / URL ↗
  • Devine, D. (2024). Does political trust matter? A meta-analysis on the consequences of trust. Political Behavior, 46(4), 2241–2262. DOI / URL ↗
  • Diamint, R. (2015). A new militarism in Latin America. Journal of Democracy, 26(4), 155–168. DOI / URL ↗
  • Easton, D. (1965). A Systems Analysis of Political Life. Wiley. https://archive.org/details/systemsanalysiso00east/page/n5/mode/2up DOI / URL ↗
  • Easton, D. (1975). A re-assessment of the concept of political support. British Journal of Political Science, 5(4), 435-457. DOI / URL ↗
  • Estadieu, L., Klauer, K. C., Voßkuhle, A., & Thomaschke, R. (2025). Institutional trust in crisis? Conceptual and methodological challenges in measuring institutional trust. Sociology Compass, 19(6), e70073. DOI / URL ↗
  • Fuchs, D., & Klingemann, H. D. (1995). Citizens and the state: A changing relationship? In H. D. Klingemann & D. Fuchs (Eds.), Citizens and the State (pp. 1–24). Oxford University Press. DOI / URL ↗
  • Gueyap Kounga, B. R. (2026). How institutional trust changes during armed conflict: Evidence from Cameroon’s “anglophone crisis”. Afrobarometer Working Paper No. 233. DOI / URL ↗
  • Hetherington, M. J., & Husser, J. A. (2012). How trust matters: The changing political relevance of political trust. American Journal of Political Science, 56(2), 312–325. DOI / URL ↗
  • Huang, K.-p., Lee, F.-y., & Lin, T.-m. (2013). Partisanship and institutional trust: A comparative analysis of emerging democracies in East Asia. Taiwan Journal of Democracy, 9(1), 47–71. DOI / URL ↗
  • Kuhn, F., & Torgler, B. (2026). Empowering the principal: How direct democracy shapes trust in institutions. Public Choice, 208(3–4), 721– 751. DOI / URL ↗
  • Levi, M., & Sacks, A. (2009). Legitimating beliefs: Sources and indicators. Regulation & Governance, 3(4), 311–333. DOI / URL ↗
  • Martinangeli, A. F. M., Povitkina, M., Jagers, S., & Rothstein, B. (2024). Institutional quality causes generalized trust: Experimental evidence on trusting under the shadow of doubt. American Journal of Political Science, 68(3), 972–987. DOI / URL ↗
  • Mishler, W., & Rose, R. (1997). Trust, distrust, and skepticism: Popular evaluations of civil and political institutions in post-communist societies. The Journal of Politics, 59(2), 418–451. DOI / URL ↗
  • Murtin, F., Pipke, D., Grimalda, G., et al. (2026). What explains trust in institutions? Multi-disciplinary evidence from an international survey. European Journal of Political Research, 1–25. https://doi.org/10.1017/S1475676526100875 DOI / URL ↗
  • Newton, K., & Zmerli, S. (2011). Three forms of trust and their association. European Political Science Review, 3(2), 169–200. DOI / URL ↗
  • Norris, P. (Ed.). (1999). Critical Citizens: Global Support for Democratic Government. Oxford University Press. https://doi.org/10.1093/0198295685.001.0001 DOI / URL ↗
  • Norris, P. (2011). Democratic Deficit: Critical Citizens Revisited. Cambridge University Press. DOI / URL ↗
  • Algan, Y. (2018). Trust and social capital. In J. E. Stiglitz, J.-P. Fitoussi, & M. Durand (Eds.), For Good Measure: Advancing Research on Well-being Metrics Beyond GDP (Chapter 10). OECD Publishing. DOI / URL ↗
  • Bolaños Perez, N. P., & Saucedo-Acosta, E. J. (2022). Regional analysis of the Pacific Alliance countries on the relationship between income inequality and social capital. DIEM, 7(1), 1–9. DOI / URL ↗
  • Roniger, L. (1987). Coronelismo, caciquismo, and oyabun-kobun bonds: Divergent implications of hierarchical trust in Brazil, Mexico and Japan. British Journal of Sociology, 38(3), 310–330. DOI / URL ↗
  • Rothstein, B., & Stolle, D. (2008). The state and social capital: An institutional theory of generalized trust. Comparative Politics, 40(4), 441–459. DOI / URL ↗
  • Siegrist, M. (2021). Trust and risk perception: A critical review of the literature. Risk Analysis, 41(3), 480–490. DOI / URL ↗
  • Solar, C. (2022). Trust in the military in post-authoritarian societies. Current Sociology, 70(3), 317–337. DOI / URL ↗
  • Sung, H. E., Capellan, J., & Barthuly, B. (2022). Trust in the police and the militarisation of law enforcement in Latin America. Policing and Society, 32(3), 311–340. https://doi.org/10.1080/10439463.2022.2037551 DOI / URL ↗
  • Taylor, L. (2004). Client-ship and citizenship in Latin America. Bulletin of Latin American Research, 23(2), 213–227. DOI / URL ↗
  • Valgarðsson, V., Jennings, W., Stoker, G., Bunting, H., Devine, D., McKay, L., & Klassen, A. (2025). A crisis of political trust? Global trends in institutional trust from 1958 to 2019. British Journal of Political Science, 55, e15. DOI / URL ↗
  • Wiesehomeier, N., & Ruth-Lovell, S. P. (2024). Trust the people? Populism, trust, and support for direct democracy. Politics and Governance, 12, Article 8648. DOI / URL ↗
  • Inter-Parliamentary Union. (2015). Venezuela (Bolivarian Republic of): National Assembly, December 2015 election. Parline. https://data.ipu.org/parliament/VE/VE-LC01/election/VE-LC01-E20151206/ DOI / URL ↗
  • Biblioteca del Congreso Nacional de Chile. (2023). Ley 21.560. LeyChile. https://www.bcn.cl/leychile/navegar?idNorma=1191005&idParte=10421799&idVersion=2023-04-10 DOI / URL ↗

Archived draft: superseded claims. Read the correction notice.

Archive correction notice ↗
PDF

Archived draft: superseded claims. Read the correction notice.

PDF ↗