Abstract

Purpose. The paper investigates how the presence of a female police officer relates to the exposure to suspect resistance in a LATAM country. Design/methodology/approach. We adopted a case-control design, using a random sample of 1,716 official reports of a Brazilian southern state’s police department. Reports were analyzed regarding the type of resistance, presence of female officers on the scene, municipality size, and hour of the day. Purpose. The paper investigates how the presence of a female police officer relates to the exposure to suspect resistance in a LATAM country. Design/methodology/approach. We adopted a case-control design, using a random sample of 1,716 official reports of a Brazilian southern state’s police department. Reports were analyzed regarding the type of resistance, presence of female officers on the scene, municipality size, and hour of the day. Findings. Our results showed that the presence of a policewoman doesn’t seem to be associated with an increase in the ratio of suspect resistance during encounters. Suspects seem to resist more in small towns regardless of female presence. Practical implications. Assembling teams with female officers shouldn’t be a safety concern for police leaders. Originality/value. Past studies indicate that local-level factors may affect suspect resistance. This study was the first investigation into the exposure of policewomen to resistance in a LATAM nation.

Introduction

Police and scholars worldwide are discussing the need to increase women's representation in police. In Brazil, women's recruitment by police institutions has grown over the last decades (Fórum Brasileiro de Segurança Pública, 2023; Musumeci & Soares, 2012; Secretaria Nacional de Segurança Pública, 2013). On the other hand, women still represent, on average, only 12% of police officers, with a significant variation among different states (Fórum Brasileiro de Segurança Pública, 2023). Even so, the most recent evidence suggests that, by 2013, little more than one-third of the female officers worked on operational (patrol) duties – and there is little motive to believe that this proportion has changed since then.

Many male and female officers believe the performance of policewomen on operational duties is worse than their male counterparts. Therefore, officers of both genders prefer to work with male partners, mostly due to safety concerns (Lopes et al., 2021; Secretaria Nacional de Segurança Pública, 2013). In that sense, is’s a commonplace among Brazilian police that policewomen are usually unable to elicit a deterrence effect on a suspect’s will to resist. That is to say that suspects are more prone to resist police actions when a female officer is present, thus increasing the risk of injuries.

There’s a vast literature regarding women and suspect resistance. Most of it, however, aimed to investigate the relation between officer gender and use of force (UoF), either number of incidents or level of force deployed. Some investigations suggest that male officers exhibit a higher propensity for employing force overall (Ba et al., 2021; Bolger, 2015; Morabito & Doerner, 1997) and are more inclined to employ more severe forms of force compared to their female counterparts (Garner et al., 2002). In the same direction, some research suggests that departments with more female officers tend to present fewer police shooting deaths (Carmichael & Kent, 2015). Nevertheless, contrasting findings indicate non-significant findings regarding the impact of an officer's gender (Cojean et al., 2023; Hine et al., 2018b; Hoffman & Hickey, 2005; Klahm & Tillyer, 2010; Lawton, 2007; Noppe, 2018; Paoline et al., 2021; Paoline III & Terrill, 2007; Rydberg & Terrill, 2010; Schell-Busey et al., 2023; Schuck & Rabe-Hemp, 2005; Stroshine & Brandl, 2020; Terrill et al., 2008).

A smaller fraction of the literature has investigated the relationship between an officer’s gender and the exposure to suspect resistance, i.e., the likelihood of a suspect presenting any level of resistance towards the officer or the team of officers. Notably, little research has investigated the relationship between the officer’s gender and the exposure to assaults by suspects. Most of the research indicates that an officer’s gender isn’t significantly related to the suspect’s resistant behavior (Santos et al., 2009), the resistance level (Bazley et al., 2007; Boivin, 2017; Rabe-Hemp & Schuck, 2007), or officers’ injuries (Hine et al., 2018a; Paddock et al., 2019; Rabe-Hemp & Schuck, 2007). On the other hand, some findings point in the other direction, suggesting that either female officers are more likely to face resistance from suspects (Crawford & Burns, 2002) or policemen are more exposed to arresting violent persons (Brown & Fielding, 1993). Additionally, Rabe-Hemp & Schuck (2007) suggested policewomen are more likely to experience resistance in domestic encounters. Those mixed results indicate that other variables such as local idiosyncrasies may influence the effect of officers’ gender on the exposure to resistance.

So far, research on that subject has been limited to the United States, Canada, the UK, and Australia. We have found no research regarding Latin American (LATAM) police. Due to social, economic, and cultural factors, results from LATAM countries may diverge from those regarding other countries.

That said, the current research aims to add to this discussion by investigating how the presence of a female police officer relates to the exposure to suspect resistance in a large southern Brazilian state police department. In addition, we used the collected data to explore the relationships between resistance and municipality size and hour of the day.

Method and procedures

We adopted a case-control study design, using the official reports database from the Secretary of Public Safety of the State of Paraná, a southern Brazilian state. We considered only encounters with the state’s uniformed police, the Military Police of the State of Paraná (PMPR – Polícia Militar do Paraná)1. PMPR is a large state department with almost 20,000 officers. We chose reports that included a criminal charge of resistance (resistência)2 as ‘cases.’ Reports that did not include those charges were selected as ‘controls.’ We considered ‘exposure’ the presence of at least one female police officer on the scene. It should be noted that in PMPR, almost all police officers work in double teams; it’s very unusual to have officers working solo.

First, we selected all reports (boletins de ocorrência unificados) of interventions by regular policing teams (radiopatrulha), during 2020, 2021, and 2022. All interventions summed 1,813,325 records (2020: 690,841; 2021: 598,071; 2022: 524,413). Subsequently, we filtered the interventions where at least one person was registered as a suspect or the author of a crime or misdemeanor3. This stage was needed to exclude from the report population all interventions in which there was no encounter with people who could resist police action (e.g., false calls or calls in which the police talked only to victims or witnesses). After filtering, 633,341 registers remained (2020: 227,366; 2021: 218,807; 2022: 187,168). From those, 8,550 (1.35%) included a resistance charge (2020: 3,310; 2021: 2,935; 2022: 2,305). We draw a random sample, stratified by year. We sampled 1723 reports (2020: 641; 2021: 594; 2022: 488), with approximately the same number of cases and controls. The number of sampled cases was limited to avoid violating the 10% rule. We restricted the number of controls to the number of cases to make data extraction feasible. From a spreadsheet containing the population of reports, we used a Google Sheets random number generator function to draw a systematic sample.

The following variables regarding the sampled reports were drawn from the reports database: type of resistance (outcome), number of female officers on the scene, municipality size, and hour of the day. Municipalities were classified into four categories, according to the 2022 Census4: (A) population > 1,000,000, (B) 300,000 < population ≤ 1,000,000, (C) 100,000 < population ≤ 300,000, and (D) population < 100,000. The resistance reports were classified into four categories (types), according to the officers’ reports: (1) vehicle pursuit, (2) potentially lethal aggression, no matter the results, (3) non-lethal assault (UoF initiated by the suspect or third-parties), (4) non-lethal resistance (UoF initiated by the officers). A team of 30 police officers, previously trained and supervised by one of the authors, were distributed into teams of two and coded all data. The researcher would decide the tie if the two-officer team eventually disagreed on the coding. The extracted and coded data were compiled into an electronic spreadsheet. The original data, extracted from the reports database, was discarded to ensure personal data confidentiality. After coding, the sample contained 1,716 reports (Table 1). Of those, 43 (2,5%) incidents were handled by a single officer, 1,332 (77,8%), by teams of two officers, 135 (7,9%) by teams of three officers, 123 (7,2%) by teams of four officers, and 83 (4,8%) by teams of five or more officers. Only four records were handled by all female teams.

The data was analyzed using R Studio and Jamovi softwares. We chose not to consider cases involving type 1 resistance for the analysis. During a vehicle pursuit, the suspect is unlikely to distinguish the officers’ gender before fleeing or resisting – since the team is inside the patrol car. Therefore, including type 1 resistance would generate noise in the analysis. After excluding type 1 cases, 1,646 reports remained.

Engaging in resistance results in distinct consequences for the individual involved, depending on the level of resistance displayed (non-lethal or lethal. In a behavioral interpretation, behavior encompasses not just the physical actions but also the alterations it brings about in the surrounding environment, i.e., its consequences (Skinner, 2013). From that perspective, the acts of resisting the police through non-lethal and lethal aggression are essentially disparate behaviors, each influenced by distinct variables. Therefore, it’s reasonable to separate lethal and non-lethal cases of resistance for a deeper analysis. That said, we conducted a three-phase analysis of the data. Initially, we examined data encompassing all levels of resistance. Subsequently, we narrowed the analysis down to non-lethal levels of resistance, followed by resistance on lethal level.

Table 1 – Resistance type and municipality size

Results

Frequency of resistance

Analyzing all records from 2020 to 2022, we found that 1.35% of encounters involved some form of suspect resistance (95%CI: 0.0132-0.0138). Of these cases of resistance, suspects resisted non-lethally (UoF initiated by the officers) in 48.14% (95%CI: 0.4475-0.5154), assaulted officers non-lethally (UoF initiated by the suspect or third-parties) in 39.42% (95% CI: 0.3614-0.4277), and exhibited potentially lethal aggression in 4.30% (95%CI: 0.0305-0.0588).

Number of officers on scene

One or two officers were on scene in 1,378 (65.45%) of the analyzed encounters. In 258 (21.64%), there were three or four officers, and in 83 (12,91%), there were five or more officers on scene. Considering only encounters where female officers were present, those number are, respectively, 213 (42.40%), 75 (26.10%), and 48 (31.50%).

Encounters per hour of the day

The distribution shown in Figure 1 suggests that most encounters happened during the night shift. There is a growing trend after 2 PM, reaching peak values between 8 PM and 10:59 PM. On the other hand, encounters seem to be less likely to happen between 5 AM and 6:59 AM.

Figure 1 – Encounters per hour of the day.

Encounters per municipality size

Figure 2 illustrates the percentage of incidents involving resistance based on the size of municipalities. The findings reveal that Class A municipalities (with a population exceeding 1,000,000) exhibited the lowest proportion of resistance cases at 36.6%. In contrast, Class D municipalities (with a population below 100,000) presented the highest proportion – 52.1%. A chi-square test indicated a significant association between city size and the incidence of resistance [χ²(3, N = 1646) = 21.5, p < .001].

Figure 2 – Proportion of encounters with resistance per municipality size

Female officers by the time of the encounter

Figure 3 presents the proportion of female officers on encounters per hour of the day. No more than 30% of the encounters happened with female officers. There was some variation, with greater participation of policewomen during late night and dawn. But, as a whole, data suggests that the presence of female officers at encounters is well distributed throughout the day.

Figure 3 – Female officers on encounters per hour of the day

Figure 4 shows the proportion of encounters evolving female officers per municipality size. There seems to be a similar proportion between encounters within municipalities with different sizes [χ²(3, N = 1646) = 2.47, p = .481].

Figure 4 – Proportion of encounters evolving female officers per municipality size

Female officers and suspect resistance

Figure 5 displays the ratio of incidents involving resistance compared to those without resistance, depending on the presence or absence of female officers at the scene. In instances where no female officers were present, 622 (47.2%) of the observed incidents encountered resistance. Conversely, 168 (51.4%) of the sampled incidents involving female officers also resulted in resistance. Despite the slight discrepancy, a chi-square test indicated that the presence of a female officer and the probability of suspect resistance are unrelated [χ²(1, N = 1646) = 1.87, p = .172].

Figure 5 – Encounters with resistance vs. presence of female officers

Non-Lethal Resistance

Encounters per hour of the day

Figure 6 displays the ratio of types 3 and 4 resistance incidents, depending on the hour of the day. From 4 AM to 6:59 AM, resistance seems to be slightly more likely. There are other peak values by 8AM, 11AM, and 8PM (20h on a 24-hour clock). Despite those variations, the ratio of non-lethal levels of resistance and the hour of the day weren’t significantly related [χ²(46, N = 1609) = 54.7, p = .177].

Figure 6 – Proportion of encounters with non-lethal resistance per hour of the day

Encounters per municipality size

Figure 7 shows the proportion of encounters with resistance per municipality size. Results show that Class A municipalities (population greater than 1,000,000) had the least proportion of resistance cases (35.2%), while Class D municipalities (population minor than 100,000) had the greatest proportion (50.9%). A chi-square test suggested that city size and the occurrence of resistance are likely related [χ²(3, N = 1609) = 21.6, p < .001].

Figure 7 – Proportion of encounters with non-lethal resistance per municipality size

Female officers and suspect resistance

Figure 8 shows the proportion of encounters with and without resistance when there are or aren’t female officers on the scene. When no female officers were on the scene, 593 (46.0%) of the sampled encounters resulted in resistance. Conversely, 160 (50.2%) encounters with female officers resulted in resistance. While that difference may suggest a marginal influence, a chi-square test failed to reject the null hypothesis that the variables are independent [χ²(1, N = 1609) = 1.80, p = .180, w = 0.0335]. The result suggests that there is likely no association between the presence of a female officer and the likelihood of suspect resistance.

Figure 8 – Encounters with resistance vs presence of female officers

Following, we individually analyzed both categories of non-lethal resistance: (3) non-lethal assault (UoF initiated by the suspect or third parties), (4) non-lethal resistance (UoF initiated by the officers). Contingencies are shown in Figure 9. A chi-square test suggested that the presence of a female officer isn’t related to exposure to assault events [χ²(1, N = 1609) = 1.93, p = .381].

Figure 9 – Type of non-lethal resistance vs presence of female officers

Logistic regression model

Using logistic regression methods, we adjusted a statistical model with the occurrence of resistance as a dichotomous response variable (1 - yes or 0 - no). During this phase, we considered only non-lethal forms of resistance, i.e. outcomes coded as 0, 3, or 4. The imbalance ratio for that sample was 1.085, indicating it isn’t significantly imbalanced according to the standards set by the literature (Zhu et al., 2020). For the first adjustment, we chose as explanatory variables: female presence (yes/no), hour of the day, and city size (category). The model suggests a statistically non-significant contribution of the presence of a female officer and the hour of the encounter to the occurrence of resistance. VIF indicates no substantial correlation between the independent variables.

Subsequently, we adjusted another model, using the backward elimination technique. The resulting model excluded the presence of a female officer and the hour of the encounter as explanatory variables. In the next step, we adjusted a third model, using the forward selection technique. Once again, the resulting model rejected the presence of a female officer and the hour of the encounter as explanatory variables, resulting in the same Model 2. The models are presented in Table 2.

Table 2 – Logistic regression models for non-lethal suspect resistance

The AIC value for the null model was 2227.21. Thus, the AIC values suggest that model 2 is the most suitable. Since the model excluded the presence of a female officer as an explanatory variable, we can conclude that the variable doesn’t yield a significant contribution. The residual plots, the Normal Q-Q Plot, and the HNP with simulation envelopes indicate that Model 2 is adequately fit and reliable.

Lethal Resistance

Encounters per hour of the day

Figure 10 presents the proportion of incidents involving lethal resistance in relation to the time of day. The occurrences appear to be spread across the entire day. Even though there are peaks observed at 5:00 - 5:59 AM, 7 - 7:59 AM, and 12 - 12:59 PM, the connection between the ratio of lethal resistance incidents and the specific hour of the day did not reach statistical significance [χ²(23, N = 1646) = 25.5, p = .325].

Figure 10 – Proportion of encounters with lethal resistance per hour of the day

Encounters per municipality size

Few of the sampled reports described lethal-level assaults – 2.3% (size A municipalities), 1.1% (size B), 2.5% (size C), and 2.5% (size D). Despite those differences, city size and the occurrence of deadly resistance don’t seem to be related [χ²(3, N = 1646) = 0.987, p = .804].

Female officers and suspect resistance

In instances where female officers were absent from the scene, lethal-level resistance occurred in 29 cases, constituting 2.1% of the sampled encounters. Conversely, in encounters involving female officers, 8 instances (2.38%) resulted in lethal resistance. A chi-square test indicated that the presence of a female officer is unlikely to be associated with the probability of suspect resistance at a lethal level [χ²(1, N = 1646) = 0.073, p = .787, w = 0.0067].

Logistic regression model

We adjusted a statistical model using the occurrence of resistance at the lethal level as the response variable (yes/no) and the same explanatory variables we tested before. The imbalance ratio for that sample was 1.0848, suggesting no significant imbalance of the data (Zhu et al., 2020). Once more, we built a first model using all selected explanatory variables: female presence (yes/no), hour of the day, and city size (category). The model suggests a statistically significant contribution of city size. VIF indicates no substantial correlation between the independent variables.

Following that, we fine-tuned an additional model utilizing the backward elimination technique. Notably, the resultant model omitted a female officer's presence and the encounter hour as explanatory variables. Although the model kept city size as an explanatory variable, its effect wasn’t statistically significant. Subsequently, we refined a third model by applying the forward selection technique. In this iteration, the resulting model continued to exclude the inclusion of a female officer's presence and the encounter hour as explanatory variables, essentially reproducing the outcomes of Model 2. The models are presented in Table 3.

Table 3 – Logistic regression models for lethal suspect resistance

The AIC value for the null model was 2282.5. Thus, the AIC values suggest that model 2 is the most suitable. Since the model excluded the presence of a female officer as an explanatory variable, we can conclude that the variable doesn’t yield a significant contribution. The residual plots, the Normal Q-Q Plot, and the HNP with simulation envelopes indicate that Model 2 is fit.

Discussion

This study primarily aimed to investigate how the presence of a female police officer relates to the exposure to suspect resistance in a Latin American country. We examined calls answered by a large southern Brazilian state police department. Our results showed that the presence of a policewoman doesn’t seem to be associated with an increase in the ratio of suspect resistance during encounters – at any level. This inference is supported both by hypothesis testing (chi-square tests) and regression models yielded by three different techniques. Those results are in phase with most of the research on the subject (Bazley et al., 2007; Boivin, 2017; Paddock et al., 2019; Rabe-Hemp & Schuck, 2007; Santos et al., 2009), suggesting that the behavior of Latin American suspects regarding female police officers is similar to other countries’.

According to our results, the common idea that police teams with female officers are more likely to face suspect resistance can be deemed a myth with a high level of confidence. Therefore, assembling teams with policewomen shouldn’t be a safety concern for police leaders. Teams composed only of female officers shouldn’t worry leaders either, although further research should address the effects of deploying policewomen-only teams – our sample comprised only four calls handled by policewomen-only teams. What should actually be a safety concern for leaders is proper UoF training encompassing all levels of force, equipment, weapons, and tactics available.

Our data showed a relatively small number of encounters involving female officers – 334 out of 1,716 reports, i.e., less than 20%. We should underscore that we took our sample from encounters registered by officers assigned to regular patrolling teams. Considering that our results suggested that female officers are no more likely to face suspect resistance than their male counterparts, the small representation of policewomen in encounters is probably a fruit of the underrepresentation of women in regular patrolling assignments, as noted by the Brazilian literature (Secretaria Nacional de Segurança Pública, 2013).

The municipality size statistically correlated with the likelihood of suspect resistance. In smaller municipalities (population < 100,000), the suspects are more prone to resist than in big cities (population > 1,000,000), whether there is or there isn’t a female officer on the scene. These results bear two implications. First, they suggest peculiar dynamics in smaller cities and towns with fewer officers and where backup may be tens of miles away. Also, social and cultural variables may influence the suspects’ willingness to resist or assault the officers. Second, the difference we’ve found between municipalities of different sizes supports the idea that the inconsistency in the literature might be related to local-level variables.

According to the local police culture, suspects would be more likely to display resistance during late night hours, especially due to alcohol consumption. Our results don’t support this claim since the association between the ratio of suspect resistance and the hour of the day wasn’t statistically significant.

By using an observational design this study couldn’t control for confounding factors. Future studies should adopt experimental or quasi-experimental designs for thorough assessment. The reduced number of cases involving female officers may limit the reliability of the results. A scarcity of data makes the identification of patterns and statistical relationships difficult. The results regarding suspect resistance at lethal levels should also be taken carefully. Since this study wasn’t designed to analyze this specific level of resistance or UoF, only a small number (37) of reported encounters were included. Therefore, the non-significant results may have been caused by an insufficient subsample size.

This study was the first to address the issue of the exposure of female police officers to suspect resistance in a Latin American country. It’s worth mentioning that Brazil represents a unique case in LATAM. Brazil has a continental size (comparable with the US), one of the world’s largest economies, and an extremely racially mixed population, with considerable cultural differences between regions due to massive immigration. Also, differing from the rest of the LATAM, Brazil was colonized by the Portuguese and was founded as an empire, keeping ties with its former metropole after its independence. Since local-level variables seem to influence suspect resistance, future research should study the exposure in other Latin American countries and other Brazilian regions.

Conclusion

The principal focus of this investigation was to examine the association between the presence of a female police officer and the likelihood of encountering suspect resistance in a Latin American country. Results from a large law enforcement agency in the southern region of Brazil suggest such a relationship between the variables doesn’t exist. As a side result, this study found an association between city size and suspect resistance.

Declaration Of Interest Statement

The authors report there are no competing interests to declare.

Acknowledgments

The authors thank the police officers of the 2022/2023 Class of the 20th Battalion Training Center for their contribution to the data extraction.

References

Ba, B. A., Knox, D., Mummolo, J., & Rivera, R. (2021). The role of officer race and gender in police-civilian interactions in Chicago. Science, 371(6530), 696–702. https://doi.org/10.1126/science.abd8694

Bazley, T. D., Lersch, K. M., & Mieczkowski, T. (2007). Officer force versus suspect resistance: A gendered analysis of patrol officers in an urban police department. Journal of Criminal Justice, 35(2), 183–192. https://doi.org/10.1016/j.jcrimjus.2007.01.005

Boivin, R. (2017). Correlates of subject(ive) resistance in police use-of-force situations. Policing, 40(4), 719–732. https://doi.org/10.1108/PIJPSM-06-2016-0081

Bolger, P. C. (2015). Just Following Orders: A Meta-Analysis of the Correlates of American Police Officer Use of Force Decisions. American Journal of Criminal Justice, 40(3), 466–492. https://doi.org/10.1007/s12103-014-9278-y

Brown, J., & Fielding, J. (1993). Qualitative differences in men and women police officers’ experience of occupational stress. Work & Stress, 7(4), 327–340. https://doi.org/10.1080/02678379308257072

Carmichael, J. T., & Kent, S. L. (2015). The Use of Lethal Force by Canadian Police Officers: Assessing the Influence of Female Police Officers and Minority Threat Explanations on Police Shootings Across Large Cities. American Journal of Criminal Justice, 40(4), 703–721. https://doi.org/10.1007/s12103-014-9283-1

Cojean, S., Deperrois, R., Taillandier-Schmitt, A., & Combalbert, N. (2023). Socio-demographic Variables and Impulsivity Can Affect the Decision to Use Legitimate Violence Amongst French Gendarmes. Em Journal of Police and Criminal Psychology. https://doi.org/10.1007/s11896-023-09605-2

Crawford, C., & Burns, R. (2002). Resisting Arrest: Predictors of Suspect Non-Compliance and Use of Force Against Police Officers. Police Practice and Research, 3(2), 105–117. https://doi.org/10.1080/15614260290033611

Fórum Brasileiro de Segurança Pública. (2023). 17^o Anuário Brasileiro de Segurança Pública. Fórum Brasileiro de Segurança Pública. https://forumseguranca.org.br/wp-content/uploads/2023/07/anuario-2023.pdf

Garner, J. H., Maxwell, C. D., & Heraux, C. G. (2002). Characteristics associated with the prevalence and severity of force used by the police. Justice Quarterly, 19(4), 705–746. https://doi.org/10.1080/07418820200095401

Hine, K. A., Porter, L. E., Westera, N. J., & Alpert, G. P. (2018a). The understated ugly side of police–citizen encounters: Situation, suspect, officer, decision-making, and force predictors of officer injuries. Em Policing and Society (Vol. 28, Número 6, p. 665–683). https://doi.org/10.1080/10439463.2016.1251430

Hine, K. A., Porter, L. E., Westera, N. J., & Alpert, G. P. (2018b). Too much or too little? Individual and situational predictors of police force relative to suspect resistance. Policing and Society, 28(5), 587–604. https://doi.org/10.1080/10439463.2016.1232257

Hoffman, P. B., & Hickey, E. R. (2005). Use of force by female police officers. Journal of Criminal Justice, 33(2), 145–151. https://doi.org/10.1016/j.jcrimjus.2004.12.006

Klahm, C. F., & Tillyer, R. (2010). Understanding police use of force: A review of the evidence. Southwest Journal of Criminal Justice, 7(2), 214–239.

Lawton, B. A. (2007). Levels of nonlethal force: An examination of individual, situational, and contextual factors. Journal of Research in Crime and Delinquency, 44(2), 163–184. https://doi.org/10.1177/0022427806297738

Lopes, C. D. S., Ribeiro, E. A., & Souza, M. A. D. (2021). Policiamento e gênero: Percepções entre policiais militares paranaenses. Opinião Pública, 27(1), 298–322. https://doi.org/10.1590/1807-01912021271298

Morabito, E. V., & Doerner, W. G. (1997). Police use of less-than-lethal force: Oleoresin Capsicum (OC) spray. Policing, 20(4), 680–697. https://doi.org/10.1108/13639519710368099

Musumeci, L., & Soares, B. M. (2012). Polícia e gênero: Participação e perfil das policiais femininas nas PMS brasileiras. Revista Gênero, 5(1). https://doi.org/10.22409/rg.v5i1.230

Noppe, J. (2018). Are all police officers equally triggered? A test of the interaction between moral support for the use of force and exposure to provocation. Policing and Society, 28(5), 605–618. https://doi.org/10.1080/10439463.2016.1199024

Paddock, E., Jetelina, K. K., Bishopp, S. A., Gabriel, K. P., & Reingle Gonzalez, J. M. (2019). Factors associated with civilian and police officer injury during 10 years of officer-involved shooting incidents. Em Injury Prevention. https://doi.org/10.1136/injuryprev-2019-043467

Paoline, E. A., Terrill, W., & Somers, L. J. (2021). Police Officer Use of Force Mindset and Street-Level Behavior. Em Police Quarterly (Vol. 24, Número 4, p. 547–577). https://doi.org/10.1177/10986111211025523

Paoline III, E. A., & Terrill, W. (2007). Police education, experience, and the use of force. Criminal Justice and Behavior, 34(2), 179–196. https://doi.org/10.1177/0093854806290239

Rabe-Hemp, C. E., & Schuck, A. M. (2007). Violence Against Police Officers: Are Female Officers at Greater Risk? Police Quarterly, 10(4), 411–428. https://doi.org/10.1177/1098611107304326

Rydberg, J., & Terrill, W. (2010). The effect of higher education on police behavior. Police Quarterly, 13(1), 92–120. https://doi.org/10.1177/1098611109357325

Santos, A., Leather, P., Dunn, J., & Zarola, A. (2009). Gender differences in exposure to co-worker and public-initiated violence: Assessing the impact of work-related violence and aggression in police work. Work & Stress, 23(2), 137–154. https://doi.org/10.1080/02678370903087934

Schell-Busey, N., Lewandowski, C., Houser, K., & Sorg, E. (2023). An Exploration of Officer Gender and Use of Force Incidents in a Transit Police Department. Em Women and Criminal Justice. https://doi.org/10.1080/08974454.2023.2271464

Schuck, A. M., & Rabe-Hemp, C. (2005). Women police: The use of force by and against female officers. Women and Criminal Justice, 16(4), 91–117. https://doi.org/10.1300/J012v16n04_05

Secretaria Nacional de Segurança Pública. (2013). Mulheres nas instituições de segurança pública: Estudo técnico nacional. Ministério da Justiça. https://dspace.mj.gov.br/bitstream/1/2308/1/4mulheres-na-seguranca-publica.pdf

Skinner, B. F. (2013). Contingencies of Reinforcement: A Theoretical Analysis. B. F. Skinner Foundation.

Stroshine, M. S., & Brandl, S. G. (2020). The use, effectiveness, and hazards associated with police use of force: The unique case of weaponless physical force. Police Practice and Research, 21(6), 591–608. https://doi.org/10.1080/15614263.2019.1582342

Terrill, W., Leinfelt, F., & Kwak, D.-H. (2008). Examining police use of force: A smaller agency perspective. Policing: An International Journal of Police Strategies & Management, 29(3), 541–563.

Zhu, R., Guo, Y., & Xue, J.-H. (2020). Adjusting the imbalance ratio by the dimensionality of imbalanced data. Pattern Recognition Letters, 133, 217–223. https://doi.org/10.1016/j.patrec.2020.03.004