Political responsiveness is a cornerstone of representative democracy, yet access to representation is often unequal. Minority and disadvantaged groups frequently face inadequate legislative responsiveness (Grossman and Slough, Reference Grossman and Slough2022). Building on this concern, we investigate whether Mexican Federal legislators exhibit differential responsiveness to constituents based on racial identity, focusing on Indigenous, Mestiza, and European-White individuals. Specifically, we ask: Do legislators respond differently to Indigenous constituents compared to Mestiza and European-White ones? Are there differences in responsiveness across political coalitions? How does the proportion of the Indigenous population in a legislator’s state affect their responsiveness?
Previous research highlights unequal legislative responsiveness across racial and socioeconomic lines, often revealed through audit experiments (Butler and Broockman, Reference Butler and Broockman2011; Driscoll et al., Reference Driscoll, Cepaluni, Guimaraes and Spada2018; McClendon, Reference McClendon2016). While these studies underscore a pattern of racial and class-based bias in legislative systems (Costa, Reference Costa2017), the role of Indigeneity remains understudied, particularly in Mexico. The concept of mestizaje, which promotes a post-racial narrative of racial mixture, often obscures the systemic discrimination Indigenous populations face (Moreno Figueroa and Saldívar Tanaka, Reference Moreno Figueroa and Saldívar Tanaka2016). Our study addresses this gap by exploring legislative responsiveness to Indigenous constituents, who comprise 19.4% of Mexico’s population (Instituto Nacional de Estadísticay Geografía (INEGI), 2020).
We conducted an audit experiment with all 626 members of Mexico’s 65th Legislature (2021–2024), randomly assigning each to receive an email from a simulated Indigenous, Mestiza, or European-White constituent. Using block randomization by political coalition and Indigenous population percentage, we measured the likelihood of legislators responding to a financial-related inquiry, as well as the quality of those responses. Specifically, we examined whether legislators used the constituent’s name in the greeting, expressed warmth throughout the reply, and offered substantive help, three common indicators of relational responsiveness in constituent communications (Costa, Reference Costa2021).
Contrary to expectations of bias against Indigenous-named constituents, legislators were significantly more likely to respond to emails from Indigenous-coded senders, by an estimated 7.5 percentage points, compared to their European-White counterparts, with no significant difference between Mestiza and White constituents. This higher responsiveness toward Indigenous constituents extended across several dimensions of response quality.
These patterns weakened in states with larger Indigenous populations, where the use of named greetings and warm tone by legislators decreased significantly, suggesting that symbolic responsiveness was more likely in states with less indigenous constituents. In addition, we found no evidence that any of these responsiveness patterns varied by political affiliation. However, due to limited statistical power for interaction effects, we cannot rule out modest partisan differences.
These findings suggest that Mexican legislators may be engaging in selective responsiveness toward Indigenous constituents, potentially as a form of symbolic alignment with global anti-discrimination norms and in line with the political environment in Mexico. The following sections examine the racial politics of representation in Mexico, outline the experimental design, and present our empirical results.
Mestizaje, indigeneity, and policy responsiveness
Mexico’s racial dynamics are shaped by the ideology of mestizaje, originating in the Spanish Colonial Period and institutionalized after the Mexican Revolution of 1910. Mestizaje promotes the notion that all Mexicans are racially mixed, fostering a ’post-racial’ narrative that masks systemic inequalities experienced by indigenous populations and the privileges enjoyed by Mexicans of European descent (Lund, Reference Lund2012). Despite this equalizing narrative, racial hierarchies persist, with Indigenous individuals facing widespread discrimination (Sue, Reference Sue2013). Surveys by INEGI and CONAPRED report that 24% of respondents experience discrimination due to their skin color (Instituto Nacional de Estadísticay Geografía (INEGI), 2023). Economic and social advantages remain tied to whiteness, while Indigenous and darker-skinned individuals face disadvantages (Hernández et al., Reference Hernández, GuzmánAyala and Pérez-Salgado2018; Vargas Cervantes, Reference Vargas Cervantes2024).
In response to these deep-rooted disparities, contemporary political movements have sought to address the exclusion and marginalization of Mexico’s racial and ethnic minorities. One such movement is the rise of the MORENA (National Regeneration Movement) party, which has positioned itself as a champion of the marginalized, including Indigenous populations. MORENA, which translates to brown woman and is reference to the Virgin of Guadalupe, the patron saint of Mexico, emerged as a response to the failures of traditional parties to address Mexico’s deep-rooted frustration with economic inequality, crime, and corruption (Greene and Sánchez-Talanquer, Reference Greene, Sánchez-Talanquer and Mainwaring2018). Its name and platform carry racial and religious weight, aligning the movement with indigenous and darker-skinned populations that have been left behind by the economic and political elites of Mexico (Castro Cornejo, Reference Castro Cornejo2023).Footnote 1
Despite MORENA’s rise, the Indigenous population continues to be disadvantaged in comparison to their Mestiza and European-descended counterparts. Given these persistent racial and socioeconomic disparities, it stands to reason that inequalities may also be reflected in political interactions between legislators and constituents. The current political context, including the rise of movements like MORENA, forms part of a complex landscape where the national discourse of mestizaje and social justice rhetoric coexist, but may not fully alleviate the realities of Indigenous populations. This study seeks to explore whether these racial inequalities exist in legislative responsiveness, where racial biases could further marginalize Indigenous people in their interactions with public officials.
Building on this framework, we propose the following hypotheses concerning how legislators’ responsiveness varies by constituents’ racial identity, political coalition, and indigenous population proportion. We define responsiveness as encompassing both whether a legislator replies and the quality of that response, given that these elements signal how constituents are valued and legitimized by their representatives (Costa, Reference Costa2021): (H1) legislators will be less responsive to Indigenous constituents than to those perceived as Mestiza or European-White; (H2) there will be no difference in responsiveness between Mestiza and European-White constituents; (H3) legislators affiliated with the ruling coalition (MORENA and allies) will be more responsive to Indigenous constituents than opposition legislators, reflecting the party’s stated commitment to addressing social and racial inequalities; (H4) responsiveness to Indigenous constituents will increase with the proportion of the Indigenous population in a legislator’s state, due to electoral incentives and demographic relevance.
Experimental design
This study employs an email-based audit experiment to examine how racial identity influences both the likelihood and quality of responses from Mexican federal legislators. Legislators were randomly assigned to receive an email from a simulated constituent whose name was racially coded to signal Indigenous, Mestiza, or European-White identities. In addition to recording whether legislators responded, we analyzed three indicators of response quality: the use of a personalized greeting, response warmth, and the provision of assistance relevant to the request. The goal is to assess whether Indigenous constituents receive fewer and lower-quality responses than their Mestiza and European-White counterparts, with the latter group serving as the baseline throughout the analysis.
The sample includes all 626 members of Mexico’s 65th Legislature (499 deputies and 127 senators), providing full coverage of the federal legislative body. Emails were sent to legislators’ official inboxes, and substitutes were contacted when principals were on leave. Automated replies were excluded, and no distinction was made between replies from legislators or their staff, as legislative offices are treated as unified actors in this context (Butler and Broockman, Reference Butler and Broockman2011). Consultations with legislative staffers confirmed that this type of communication is common and that the content of the email passes as real constituent input. The treatment variable is the racial identity signaled by the sender’s name: Itzayana Pech (Indigenous), Sofía Hernández (Mestiza), and Julie Betancourt (European-White). All names are female, consistent with findings that female constituents are more likely to elicit responses from public officials in the context of Mexico (Magni and Ponce de Leon, Reference Magni and Ponce de Leon2020). Also consistent with Magni and Ponce’s analysis, we expect a low response rate. They had a total response rate of 5.7 for Mexico in their audit on gendered constituents (Magni and Ponce de Leon, Reference Magni and Ponce de Leon2020).
Names were selected based on national naming data and their recognizability as racial cues. Sofía Hernández was the most common name in Mexico between 2017 and 2020 (Instituto Nacional de Estadística y Geografía (INEGI), 2020). Itzayana Pech, of Mayan origin, was selected for its strong Indigenous association. Julie Betancourt signals a European identity, with the French surname and a first name spelled differently from the Spanish “Julia,” emphasizing perceived whiteness and foreignness.
All legislators received identical messages, asking for information about programs supporting the elderly—a topic selected for its political neutrality, low salience, and broad support across Mexican coalitions and society (Bárcena Juárez and Kerevel, Reference Bárcena Juárez and Kerevel2021; Béland and Medrano, Reference Béland and Medrano2025). Legislative staff confirmed the issue is common and non-sensitive in constituent communication.
The emphasis on selecting a low-salience topic was deliberately chosen to help minimize the risk that response behavior would be influenced by ideological, partisan, or strategic considerations, strengthening the study’s internal validity. The email, translated into English, reads as follows (see Appendix C for the original Spanish email):
Dear Deputy/Senator X,
I am a voter in the district/state for which you were elected. I am writing to ask about the funds available to support the elderly. I have a family member who desperately needs assistance, and we do not know whom to contact.
Respectfully,
Itzayana Pech/Sofía Hernández/Julie Betancourt.
To account for possible confounders, legislators were block-randomized by coalition affiliation (MORENA vs. opposition) and the proportion of Indigenous population in their state, categorized as Low, Medium, or High using INEGI data (see Appendix A). Block randomization ensured equal distribution of treatments across strata, and each legislator received only one email. Descriptive covariates are presented in Table 1, with conducted chi-squared tests showing no significant differences across treatment groups (p > 0.05) (see Appendix D).
Table 1. Descriptive statistics of experimental sample
We analyze four binary dependent variables for this analysis. The primary outcome is whether the legislator responded, while the remaining three capture response quality through the following: the use of the constituent’s name in the salutation, response warmth (e.g., gratitude, supportive language), and whether the inquiry was directly answered. These indicators reflect key dimensions of relational and substantive engagement in public service communication (Costa, Reference Costa2021).
Following recommendations from experimental literature on avoiding post-treatment bias, all non-responses and low-quality replies were coded as 0 on the quality measures (Coppock, Reference Coppock2019).
We estimate treatment effects using ordinary least squares (OLS) regressions with White heteroskedasticity-robust standard errors. Given random assignment, linear models are widely accepted for estimating average treatment effects in experimental research (Grose et al., Reference Grose, Lopez, Sadhwani and Yoshinaka2022). We estimate three models: our main treatment effects, interactions with coalition affiliation, and interactions with Indigenous population size. Each model is estimated on the full sample and includes all four outcomes (see Appendix E for full specifications).
Results
Table 2 presents descriptive results examining the response rates across different constituent identity groups, providing an initial view of how legislators respond to requests from constituents with names suggesting different racial backgrounds.Footnote 2 Contrary to our initial expectation, Mexican legislators were more responsive to Indigenous-named constituents, with a response rate of 9%, compared to 3% for Mestiza-named and only 1% for European-White.
Table 2. Response rates across treatment groups
Given the overall response rate of just 4%, it is important to note that such rates are not uncommon in the Mexican context. For example, an audit experiment conducted by Animal Político in 2011 targeting 500 deputies via physical letters also yielded a 4% response rate (Bárcena Juárez and Kerevel, Reference Bárcena Juárez and Kerevel2021). Similarly, Magni and Ponce de León (Reference Magni and Ponce de Leon2020), in a large-scale cross-national audit experiment on gender bias, found a response rate of only 5.7% from Mexican legislators. These findings suggest that limited engagement with constituent communications – particularly via email – is typical in Mexico.
To assess whether our study was capable of detecting meaningful differences in responsiveness across identity groups, we conducted a post hoc power analysis. The results (Appendix F) indicate that our study sample has sufficient power to detect moderate-to-large effects, but is underpowered to reliably detect small effects. We therefore interpret all null results in the analyses that follow with caution, especially in cases where effect sizes may fall below that threshold.
Main treatment effects
Building on these descriptive findings, we report the main results of this experiment in Table 3, where the dependent variable is coded as 1 if the legislator exhibited a specific form of responsiveness—replying to the email inquiry, using a name greeting, expressing warmth, or resolving the inquiry—and 0 otherwise. The table displays the estimated treatment effects for Indigenous- and Mestiza-named constituents, with the European-White-named group serving as the reference category.
Table 3. Main analysis results for all dependent variables
Note: Robust standard errors in parentheses. *p < 0.1, **p < 0.05, ***p < 0.01
The results revealed a consistent and statistically significant positive effect for the Indigenous treatment across nearly all measures of responsiveness. Legislators were 7.5 percentage points more likely to respond to an Indigenous-named constituent compared to the baseline (p < 0.01). These responses were also more likely to include a personalized greeting (p < 0.05), a warm tone (p < 0.01), and a resolution to the constituent’s request (p < 0.05). These patterns suggest that the observed bias in favor of Indigenous constituents extends not only to reply rates but also to the tone and substance of communication.
In contrast, the Mestiza treatment produced no statistically significant differences in response rate or quality measures. The only exception is a marginal increase in name personalization (p < 0.1), although this effect is modest. Moreover, a direct comparison between the Indigenous and Mestiza treatments reveals no statistically significant difference in the likelihood of receiving a name greeting, suggesting the two treatment effects are statistically indistinguishable on this particular dimension. Given the design constraints discussed earlier, we interpret these null results with caution, as they may reflect limitations in statistical power rather than definitive evidence of equivalence.
Together, these results contrast with our initial expectation that legislators would exhibit bias against Indigenous constituents (H1). Instead, we find consistent evidence of positive responsiveness, both in reply likelihood and quality. At the same time, the absence of significant differences in the Mestiza condition remains inconclusive due to lower power. These findings suggest that Indigenous-coded constituents were more likely to receive engaged and helpful replies, while Mestiza-coded constituents were treated similarly to their European-White counterparts (H2), but with greater uncertainty due to the aforementioned limitations.
Effects by coalition affiliation
In Table 4, we present the results from coalition-conditional models, testing whether the effects of constituent race vary by a legislator’s affiliation with the MORENA coalition versus the opposition (H3). Consistent with earlier models, the Indigenous treatment exhibits a positive and statistically significant effect on overall responsiveness (p < 0.05), as well as on warm tone (p < 0.05). These findings suggest that Indigenous-named constituents receive more engaged replies, regardless of partisan alignment. The effects on name personalization and inquiry resolution are positive but not statistically significant. Meanwhile, the Mestiza treatment continues to yield no consistent or significant effects across most responsiveness dimensions, with the exception of a modest increase in name greeting usage (p < 0.1).
Table 4. Coalition-conditional results for all dependent variables
Note: Robust standard errors in parentheses. *p < 0.1, **p < 0.05, ***p < 0.01
Crucially, none of the interaction terms between treatment condition and coalition affiliation reach statistical significance across the four outcome measures. That is, MORENA legislators were neither more nor less likely than opposition legislators to respond—or to respond warmly, personally, or substantively—to either Indigenous or Mestiza-named constituents. These results indicate no evidence of partisan divergence in responsiveness patterns, thus offering no support for H3. Given the study’s limited statistical power to detect small interaction effects (see Appendix F), the absence of statistically significant interactions does not constitute evidence of no effect.
Effects by indigenous population proportion
Table 5 presents the results of models examining whether the effects of constituent racial identity vary depending on the proportion of Indigenous population in a legislator’s state (H4). The analysis distinguishes between low-, medium-, and high-Indigenous population contexts, allowing us to assess whether legislators respond differently to Indigenous- or Mestiza-named constituents depending on the demographic makeup of their constituency.
Table 5. Indigenous proportion-conditional results for all dependent variables
Note: Robust standard errors in parentheses. *p < 0.1, **p < 0.05, ***p < 0.01
The results provide evidence of a meaningful interaction between Indigenous treatment and population context, particularly for overall responsiveness and response tone. In districts with a low Indigenous population, Indigenous-named constituents were 18.2 percentage points more likely to receive a reply than European-White-named constituents (p < 0.01), along with significantly higher rates of warm tone (p < 0.05) and use of a personalized greeting (p < 0.05). However, these effects consistently diminish as the proportion of Indigenous residents increases. For each increase in Indigenous population category (from low to medium to high, as described above and presented in Appendix A), the effect of the Indigenous treatment decreases by 5.6 percentage points in reply likelihood (p < 0.05), 3.0 points in name personalization (p < 0.05), and 3.4 points in warm tone (p < 0.1). While the pattern for providing a substantive resolution follows the same negative trend, it does not reach statistical significance.
These results suggest a potential reactive effect of Indigenous population size. Legislators may be more likely to respond to Indigenous constituents in districts where they represent a smaller portion of the electorate, but that heightened responsiveness may taper off in areas with larger Indigenous populations.
Meanwhile, the Mestiza treatment yields no significant effects across any of the four outcomes, nor do the interaction terms between Mestiza treatment and Indigenous population. Given the study’s overall power limitations, especially in detecting subtle interaction effects, it remains possible that modest variation in response to Mestiza constituents across population contexts exists but was not detectable within the constraints of our experiment.
To assess robustness, we conducted a leave-one-out (LOO) sensitivity analysis across all models and dependent variables, confirming that our Indigenous treatment effects were stable (See Appendix G).
Discussion
The results of this audit experiment offer a counterintuitive view of legislative responsiveness in Mexico. Contrary to our initial hypothesis (H1), legislators were significantly more likely to respond to constituents with Indigenous-sounding names than to those coded as Mestiza or European-White, who showed no significant differences in reply likelihood (H2). Moreover, this positive bias toward Indigenous constituents extended beyond reply rates to include warmer tone, greater use of personalized greetings, and more frequent substantive assistance.
Importantly, we were unable to determine whether these patterns were consistent across political coalitions. Across all outcome measures—response likelihood, tone, greeting, and resolution—we found no evidence that legislators affiliated with the MORENA coalition were more responsive to Indigenous constituents than their opposition counterparts. These null interaction effects indicate no partisan divergence in behavior, at least within the scope of this experiment on replies to inquiries about social programs for the elderly. However, leave-one-out sensitivity checks revealed that these estimates were relatively unstable, with several significant changes across iterations. Combined with the study’s low statistical power to detect interaction effects, this instability suggests that even if modest coalition-based differences exist, our design was not adequately powered to detect them. As such, the results provide no empirical support for (H3), though we cannot entirely rule out small partisan effects, particularly positive ones in the MORENA-Indigenous constituent interaction.
The interaction between responsiveness and Indigenous population size, on the other hand, yielded a robust and theoretically informative pattern. While legislators in districts with low Indigenous populations were more likely to respond warmly and substantively to Indigenous constituents, this effect declined significantly as the proportion of Indigenous residents increased, contradicting our expectation in (H4). This negative conditional effect was consistent across multiple dimensions of response quality and remained robust under the leave-one-out sensitivity analysis.
Taken together, these findings suggest that Mexican legislators may be selectively signaling responsiveness to Indigenous constituents, particularly when the political costs of doing so are low. This aligns with arguments from the World Polity framework, which posits that global norms, such as those codified in the United Nations Declaration on the Rights of Indigenous Peoples (UNDRIP), can influence domestic institutions seeking legitimacy through symbolic compliance (Meyer et al., Reference Meyer, Boli, Thomas and Ramirez1997). Legislators may respond more favorably to Indigenous constituents not only to signal alignment with international norms but also to capitalize on reputational benefits in contexts where Indigenous populations are electorally marginal. Another possible interpretation is that in areas with few Indigenous constituents, legislators respond more simply because such emails are rare.
The Mexican political context likely reinforces this dynamic. MORENA’s rise as a populist force with rhetorical ties to marginalized groups, including Indigenous populations, may have elevated the symbolic value of responding inclusively. However, the lack of differential treatment across coalitions and the decline in responsiveness in high-Indigenous districts suggest that electoral incentives and local power dynamics can override party rhetoric or global norms (Hafner-Burton and Ron, Reference Hafner-Burton and Ron2013).
Future research should move in two directions: first, expanding beyond welfare inquiries to include topics like policy or constitutional reforms; second, exploring how global pressures interact with local structures, especially at the state level and with names tied to specific Indigenous groups. Doing so could illuminate how legislators weigh symbolic responsiveness against strategic considerations and help clarify the mechanisms driving both inclusion and exclusion in democratic representation. In addition, qualitative interviews with legislators or targeted surveys across congressional districts could further enrich this line of inquiry by uncovering how representatives perceive constituent communications and whether they recognize or respond differently to specific cues given local dynamics. These approaches would complement the findings of this experiment by offering deeper insight into the motivations and perceptions shaping observed behaviors.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/XPS.2025.10008
Data availability
The data, code, and any additional materials required to replicate all analyses in this article are available at the Journal of Experimental Political Science Dataverse within the Harvard Dataverse Network, at: https://doi.org/10.7910/DVN/VS3GL8
Acknowledgements
The authors thank the legislative staff from Baja California’s municipal circuit for feedback on our email inquiry draft. For their helpful comments and suggestions related to the discussion section, the authors thank their departmental colleague, Therese Franklin. A previous version of this paper has been presented at the University of Southern California’s Center for International Studies Workshop. The authors thank the organizers and participants of this workshop. The authors are especially grateful to the anonymous reviewers at the journal, whose thoughtful critiques and suggestions during the revise-and-resubmit process substantially improved the quality and clarity of the paper.
Competing interests
None.
Ethics statement
This study was approved by the authors’ university Institutional Review Board (#UP-23-00147). The requirement for informed consent was waived due to the audit design and the exemption status of public officials at the time of the study. This research adheres to APSA’s Principles and Guidance for Human Subjects Research. Section H in the supplemental appendix expands on the items above.
Open access
