The Communication of Fear: Factors of Crime News Impacting Engagement on Social Networks

Abstract

This research analyzes the impact of crime news on users’ digital engagement on social networks. Specifically, this study reviews the influence of presentation format, crime details, and news or discursive values on people’s interaction with media content. To achieve the study’s objective, 1000 posts from the social network Instagram about crimes in the main media outlets of Colombia, Mexico, Paraguay, and Ecuador were reviewed. Content analysis was employed to code the variables, while negative binomial regression models were used to assess their impact on engagement, measured through likes and comments received on the posts. The findings show that shorter videos and image collections generate more engagement than other formats, while the type of crime did not show significant differences in interaction, suggesting a possible normalization of violence among the analyzed viewers. Among the news values, inseparability had a positive effect on engagement, while consonance and references to elite figures demonstrated negative effects. The original value of this research lies in empirically verifying how the characteristics of crime news influence engagement, providing relevant information for understanding the interaction between electronic media and the perception of criminality among Latin American viewers.

1. Introduction

News involving crimes and offenses drastically influences viewers’ perception of security, generating a wide range of emotions and reactions (Carrasco Paillamilla & Disi Pavlic, 2023). Specifically, events involving violence often provoke feelings such as fear, anger, and sadness in people, impacting their emotional well-being and contributing to a generalized sense of insecurity (Silva & Guedes, 2022).

On the other hand, the way these events are reported by the media plays an important role in how viewers interpret and respond to them. The media not only reports the facts but also, through their presentation and narrative, can enhance or minimize the emotional and social impact of the news, especially in those linked to crimes (Intravia et al., 2017; Bai, 2023). In this sense, the textual characteristics of the publications, the emphasis on certain details of the crime, the news presentation formats, and the discursive news values (DNV) become significant elements that shape the viewers’ perception and can impact their attitudes and behaviors towards the information (Lischka, 2021).

Within digital ecosystems and particularly on social media platforms, the influence that news about crimes and offenses can have on the perception of security is a central research topic. Viewers not only passively consume this type of information but can also actively participate in the dissemination of news through digital engagement indicators, such as commenting or reacting to a post (Osorio-Andrade et al., 2022). These audience behaviors can not only drastically increase the potential reach of the news but also amplify the impact of the information on the viewers’ emotions, for example, contributing to a greater sense of collective insecurity or increasing levels of social stress (Silva & Guedes, 2022).

Considering the above, understanding the influence exerted by the characteristics of the news—such as its presentation format, the textual content of the information, and the incorporation of discursive news values (DNV)—on the behavior and attitudes of viewers becomes an essential gap in contemporary media research. This gap is particularly crucial in the Latin American context, where high crime rates and the penetration of social media configure a volatile media landscape capable of rapidly shaping public opinion and security policies (Afrouzi, 2024). Investigating these factors is pertinent not only for their academic relevance but also for their sociopolitical implications in often fragile democracies (Cheung, 2021). While consumer engagement expressed through comments, likes, reactions, and other types of statements is an indirect indicator, its study is fundamental as it can have considerable influence on various audience beliefs, such as the collective perception of security and trust in law enforcement. Clarifying the factors that drive this interaction is the first step in understanding its subsequent effects.

To address this gap in academic research, this study aims to analyze the influence of different independent variables related to news about insecurity disseminated on social media on the digital engagement of viewers. Clarifying the scope of this study, our focus is on the factors that drive engagement, not on the direct measurement of their effects on audience perception, although the implications of these findings will be discussed. Specifically, we analyze the presentation format of the news, for example, whether it has been disseminated in image or video format. The details of the crime highlighted by the reviewed digital media, such as the type of crime (theft, homicide, kidnapping, etc.) and the personal characteristics of the perpetrator, such as gender or nationality, are also examined. The analysis of DNVs, which are the discursive resources used by the media, such as consonance, proximity, personalization, the use of superlative resources, and references to authority figures or elites within the news, is also included. On the other hand, the audience’s digital engagement was measured by the volume of likes and comments each of the retrieved news articles received.

To guide this research, three explicit research questions are posed. First, we seek to investigate: How does the format of the news, whether static image, image collection, short video, or long video, influence the level of user engagement, measured separately in terms of “likes” and comments on publications related to crimes? Second, we examine: How do specific details of the crime, such as the type of crime (e.g., homicide or kidnapping) and the gender of the aggressor, affect the volume of reactions these crime news receive? Finally, we analyze: Which discursive news values (DNVs), such as unexpectedness, the presence of elite figures, or thematic consonance, are associated with an increase or decrease in audience interaction?

Thus, this manuscript is developed in six sections, beginning with this introduction. The second section presents the theoretical framework that supports the formulated hypotheses. The third section outlines the methodological process applied to achieve the study objectives. The fourth section presents the descriptive and empirical results through the formulated models, while the fifth section discusses the main findings obtained. Finally, the sixth section of the article addresses the study’s conclusions and discusses the main limitations and research opportunities.

2. Theoretical Framework

2.1. User Digital Engagement in Crime News

How the media shares criminal news and how users interact with this content has undergone a significant transformation with the emergence of social media (Ruffio & Hubé, 2022). Unlike traditional media, where communication was mostly unidirectional, on social media, users have the ability to actively participate in discussions, engaging not only in the interpretation of information but also in its dissemination (Mayr & Statham, 2021). In this way, a cycle of emotional amplification is produced, where the comments and reactions that publications receive can intensify the psychological and social impact of the information, particularly that related to criminal or delinquent news (Silva & Guedes, 2022).

Moreover, the way criminal news is presented in digital ecosystems can directly influence the level of engagement they receive from the audience. The visual, textual, and narrative elements used in the construction and dissemination of information can generate different levels of emotional response and audience participation, that is, different levels of engagement (Lischka, 2021). Other aspects to consider are the speed at which news spreads and the constant updating of content by digital media. These are distinctive characteristics of social media and can exert an influence on sustained attention and greater interaction from the viewer in the discussion of specific criminal news and topics.

An important point to highlight is that digital engagement in the context of criminal news is also impacted by other psychosocial factors. For example, the perception of proximity or geographical closeness to the reported events, the cultural resonance of the events, and personal identification with the victims or the circumstances of the crime can trigger a greater impact on the viewer, which in turn could lead to greater interaction with news content (Makki, 2019). Similarly, the viral nature of social media can focus the viewer’s attention on certain aspects of the reported criminal situations, maximizing their reach and contributing to the construction of collective narratives and imaginaries about crime and security in certain regions and contexts.

In general terms, understanding the relationship between criminal news and digital engagement metrics becomes a novel and significant field of study. This is so that media and communication professionals can develop more effective yet ethical content dissemination strategies. Understanding this interaction, in addition to optimizing the informative impact, provides a broader framework to anticipate and manage the emotional and social repercussions that criminal news can have on the audience.

In the following sections, the specific characteristics that the media imprints on criminal news are analyzed, and the hypotheses that guide the research are formulated.

2.2. News Format and Its Effect on Digital Engagement

The format refers to the structural and presentation elements used by the media to disseminate news content. These formats can include various characteristics, such as static or fixed images, videos, textual content, and all their possible combinations. Additionally, each of these formats possesses different levels of representational richness that can impact the consumer in various ways, including their digital behavior (Gkikas et al., 2022).

The influence of the representational richness of formats on the consumer’s digital engagement can be explained through complex multilevel cognitive processes. For example, media and formats that present greater representational richness, understood as a greater number of sensory signals, can simultaneously activate various brain regions associated with information processing. While a static image exclusively activates the visual areas of the brain, a video can stimulate both sight and hearing, enriching the viewer’s experience by incorporating movement, sound, and, in some cases, even complementary textual elements. This combination of sensory stimuli not only makes the message more attractive but also facilitates the understanding and retention of information (Sreejesh et al., 2020; Leung et al., 2017).

This multimodal activation triggers several cascading effects. First, the need to process different types of information simultaneously increases the demand for cognitive resources, naturally elevating the user’s levels of attention and concentration towards the content (Leung et al., 2017). Second, the parallel processing of different types of sensory information favors the creation of deeper connections in memory, improving both the encoding and subsequent retrieval of information (Ji et al., 2019). Third, the more immersive experience generated by multisensory stimulation is more likely to trigger emotional responses, as it more faithfully simulates the experience of a real interaction (Molina et al., 2020). This chain of cognitive and emotional processes culminates in a greater motivation to interact with the content. When the user processes information more deeply, establishes richer connections in their memory, and experiences a more intense emotional response, their propensity to engage with the content increases, manifesting through concrete actions such as liking or commenting on a post (Eslami et al., 2022). Based on the above, the following research hypothesis is proposed:

H1. 

Message formats with greater representational richness, such as videos, foster the consumer’s digital engagement.

2.3. Crime Details and Their Impact on Digital Engagement

The coverage that the media gives to crime profoundly influences how society perceives and understands violence. According to Sosa et al. (2019), the selection of specific details presented in crime news disseminated on social media, such as the type of crime and the traits of the perpetrator, can determine both the public’s emotional response and their level of interaction with digital posts. For example, scientific research has identified that not all crimes capture the viewer’s attention equally. According to Kesić et al. (2020), extraordinary or significantly violent events arouse much greater interest in the audience. Crimes such as homicides, kidnappings, and terrorism are events that provoke more intense emotional responses. In other words, this type of criminal activity not only dominates the headlines but also triggers interactions on social media precisely because of its emotional impact (Chermak, 1998). This phenomenon is partly explained because we perceive violent crimes as rare and surprising events, which increases their news value in digital media (Rios & Ferguson, 2020).

On the other hand, economic crimes or crimes against property, although more frequent in our society, generate less curiosity and emotion. Some authors, such as Kort-Butler and Habecker (2018), confirm this trend by pointing out that financial crimes such as fraud or property theft, despite their high frequency, rarely generate the same level of media coverage or digital interaction as violent acts.

The contrast between these types of crimes highlights that digital engagement does not depend exclusively on the actual prevalence of the crime but on the crime’s ability to capture public attention through its extraordinary or impactful nature. This trend can also be understood through the “cultivation” theory (Gerbner, 1976), which suggests that the media tends to present violent crimes with a much higher frequency than other crimes, creating a distortion in the public’s perception of the prevalence of violence in society. In this way, the media plays an active role in constructing a “media reality” in which violent crimes appear to be more common than they actually are, reinforcing the focus on these crimes and generating greater digital engagement (Callanan & Rosenberger, 2015). However, it is important to note that this premise could present a theoretical tension when applied to contexts of endemic violence, such as those in several Latin American countries. This is where the concept of the “normalization of violence” arises, a form of desensitization or social habituation that occurs after constant and prolonged exposure to news about crimes (Silva & Guedes, 2022). From this perspective, the high frequency of violent crimes could paradoxically reduce the differential impact of each individual event, leading the audience not to react more intensely to a homicide compared to a robbery, as both are part of the everyday landscape of insecurity. Despite this tension, the predominant literature maintains that the severity of the crime remains a key factor. Based on these arguments, the following hypothesis is proposed:

H2A. 

The type of crime presented in the news influences the digital engagement of the audience.

On the other hand, the characteristics of the perpetrator, such as gender and nationality, can also influence digital engagement. The gender of the perpetrator has been a recurring topic in research on the effects of crime news. Studies show that crimes committed by men usually receive more prominent coverage than those committed by women, and public attitudes can vary depending on whether the perpetrator is male or female (Wozniak & McCloskey, 2010). In this sense, the media often tends to highlight crimes committed by men, which can result in greater media interest and digital interaction (Martins, 2021). Additionally, the gender of the perpetrator influences public responses due to existing gender stereotypes, which affect how the crime and victimization are interpreted. In particular, crimes committed by men tend to generate more intense responses due to biases that associate men with violence or aggression (Savage et al., 2022).

Regarding the nationality of the perpetrator, its relevance has gained attention in recent migratory contexts, particularly in Latin America, where Venezuelan migration has altered social and political dynamics in several countries (Afrouzi, 2024). The presence of immigrants as perpetrators of crimes can generate intense emotional reactions in the public, especially when the crime is presented in a context that reinforces negative stereotypes about migrants (Menjívar, 2016). In this sense, the media tends to associate immigrants with criminal activities, which increases polarization and fear towards these communities (Pan et al., 2019). This dynamic can affect digital engagement, as news involving immigrants as perpetrators can generate greater interaction due to the emotional and social load they carry (Seate & Mastro, 2017). Furthermore, the nationality of the perpetrator, when associated with migrant groups in high-visibility situations, can contribute to feeding distrust and xenophobia in the audience, which translates into a higher level of discussion and sharing of news on social media (Asakitikpi & Gadzikwa, 2020).

In these cases, the nationality of the perpetrator can increase polarization and feed fear or distrust towards immigrants, affecting digital engagement with the news, especially when crimes are presented in a context that amplifies tensions about migration. This highlights how national stereotypes can play an important role in how news is perceived and the level of interaction it generates on digital platforms (Kunst, 2021).

Based on these considerations, the following research hypotheses are proposed:

H2B. 

The gender of the perpetrator presented in the news influences the digital engagement of the viewer.

H2C. 

The nationality of the perpetrator presented in the news influences the digital engagement of the viewer.

2.4. Discursive News Values and Their Impact on Digital Engagement

Discursive news values are the characteristics that determine the newsworthiness of an event and are constructed through specific linguistic resources in journalistic text (Caple & Bednarek, 2016). These values act as criteria that influence how news is perceived and processed by the audience (Makki, 2019).

Among the main discursive news values are several aspects that influence how news is constructed and presented. Consonance, for example, is discursively constructed through the use of words and expressions that evoke stereotypes, such as those suggesting similarities with the past, such as “once again,” “just like before,” or “previously” (Khan & Anjum, 2022). These references help connect the event with the public’s common knowledge, establishing a link with a shared understanding and reinforcing the collective perception of certain groups or situations. Elitism, on the other hand, is manifested through references to authorities, institutions, or high-status individuals, which gives the information a weight of credibility or relevance.

Personalization, for its part, is constructed by giving a human face to the news, highlighting ordinary people to make it more relatable and understandable for the audience (Bednarek, 2016). Proximity, another important value, refers to the geographical or cultural closeness of the event to the audience, which increases the interest and relevance of the news (Wu & Pan, 2021). Superlativism is constructed through the use of quantifiers and intensifiers that highlight the magnitude of the event, generating a sense of awe or importance (Bednarek, 2016). Unusualness, or unexpectedness, is related to unusual or surprising events that capture immediate attention (Baissa et al., 2024). Finally, temporality is constructed through temporal references that emphasize the currentness of the event, ensuring that the news is seen as relevant at the present moment (Li et al., 2019).

Discursive news values are closely related to digital engagement through various psychological and social mechanisms. Firstly, these values function as “cognitive triggers” that capture the reader’s attention and facilitate the processing of information (Bednarek, 2016). For example, when crime news incorporates elements of personalization, such as details about the life of the victim or the perpetrator, mechanisms of empathy and emotional connection are activated, which can motivate greater interaction with the content (Makki, 2019).

The process of influence of news values on digital engagement develops at different levels. At the cognitive level, values such as unexpectedness and impact can increase cognitive activation, which favors a deeper processing of information and, therefore, a greater interest in the content (Poutvaara & Ropponen, 2018). On the emotional plane, values such as proximity and personalization can intensify emotional responses, a key factor for engagement, as the closeness of the event or the humanization of those involved generates stronger responses (Anastasio & Costa, 2004). At the social level, the presence of values such as elitism or consonance fosters discussion and debate on digital platforms, especially when they align with stereotypes or social expectations (Makki, 2019). Finally, at the contextual level, temporality and proximity increase the perceived relevance of the news, which can result in greater digital engagement due to the perception of immediate relevance to the audience (Kwon et al., 2017).

Based on these considerations, the following research hypothesis is proposed: H3. The presence of discursive news values in crime news positively influences the digital engagement of the audience.

Below, Figure 1 presents the formulated research model. To create an illustration for the research model you described, I will generate an image that represents the key components and relationships discussed. This will include elements such as consonance, elitism, personalization, proximity, superlativism, unexpectedness, and temporality, and how they relate to digital engagement. Let’s proceed with generating this image.

3. Materials and Methods

3.1. Sample Description

The sample for this study consisted of Instagram posts shared by traditional media outlets with a national presence and the highest number of followers on this social network in the four most dangerous countries in Hispanic America, according to the ranking by World Population Review (2024). These countries were Colombia, Mexico, Paraguay, and Ecuador. For each country, the five media outlets with the highest number of Instagram followers were selected (see

Table 1. Selected media outlets by country.

Country	Media	Followers on IG
Colombia	Noticias RCN	1.7 M
	Noticias Caracol	4.3 M
	El Tiempo	2.9 M
	Revista Semana	2.3 M
	El Espectador	2.1 M
Mexico	Fuerza Informativa Azteca	1.1 M
	El Universal	887 K
	Milenio	798 K
	Revista Proceso	576 K
	Reforma	435 K
Paraguay	Telefuturo	1.3 M
	ABC Digital	1.1 M
	Ultima Hora PY	855 K
	La Nación	299 K
	Onda Crónica	208 K
Ecuador	TC Televisión	1.9 M
	El Universo	1.2 M
	El Comercio de Ecuador	1.3 M
	RTS Ecuador	684 K
	Ecuavisa Noticias	599 K

), and the last 50 posts related to crimes shared by these outlets were analyzed. The collection was conducted retrospectively until 50 posts were reached for each outlet, regardless of their age. The data collection period spanned from May to September 2024, while the compilation and processing of the information were carried out between October 2024 and January 2025.

3.2. Data Collection

The data collection to analyze the 1000 gathered posts was conducted through content analysis. This technique involves identifying, organizing, and quantifying patterns present in textual or visual data and has proven highly useful for analyzing both digital and traditional media (Krippendorff, 2018). To ensure the rigor and replicability of the analysis, two coders previously trained in the theoretical and operational definitions of each variable included in the study were employed.

Regarding coding, the format of the news was operationalized into exclusive categories, distinguishing between static image, collection of images, short-duration video (less than 30 s), and long-duration video (more than 30 s). The details of the crime were coded using a non-exclusive categorical variable with six possible values: assault, homicide, robbery, sexual crimes, kidnapping, and others. This last category included other types of crimes such as torture, terrorist attacks, threats, and extortion. The gender of the perpetrator was established as a dichotomous variable (1 for male and 0 for female), while nationality was coded as 0 for national offenders and 1 for immigrant offenders.

Discursive news values were operationalized following the recommendations of Makki (2019), using a non-exclusive categorical variable that included the following values: consonance, elite, personalization, proximity, superlativeness, unexpectedness, and temporality. This characteristic allowed a news item to present a combination of several values, such as a post that simultaneously highlighted the proximity and unexpectedness of the events. Finally, the viewer’s digital engagement was operationalized as a quantitative variable that included the number of comments each post received as well as the volume of likes.

The coders independently analyzed 20% of the posts to assess the reliability of the operationalization using Krippendorff’s alpha statistic. This statistic measures the degree of agreement between coders on ordinal categorical variables and is particularly useful for evaluating consistency in content analysis (Krippendorff, 2018). All variables achieved values above the recommended threshold (greater than 0.66), ensuring the reliability of the coding. In cases where discrepancies arose, the coders and researchers discussed until reaching a consensus, ensuring consistency in the coded data.

Table 2. Coding manual.

Variable	Operationalization	Krippendorff’s Alpha
News Format	Exclusive Category: Static image, image collection, short video (<30 s), long video (>30 s).	0.98
Type of Crime	Non-exclusive Category: Assault, homicide, robbery, sexual crimes, kidnapping, others.	0.87
Gender of the Perpetrator	Dichotomous: 1 for male, 0 for female.	0.97
Nationality of the Offender	Dichotomous: 0 for national, 1 for immigrant.	0.95
Discursive News Values	Non-exclusive Category: Consonance, elite, personalization, proximity, superlativeness, unexpectedness, temporality (a news item can present various categories).	0.79
Consumer Digital Engagement	Quantitative: Number of comments and number of likes.	0.97

presents a summary of the coding manual and the Krippendorff’s alpha values achieved for each variable.

3.3. Data Analysis

After collecting the information through content analysis, an empirical model was used to evaluate the relationship between the characteristics of crime news and the digital engagement of consumers, measured by the number of comments and “likes” each post received. The first step was to ensure that the independent variables (news format, crime details, and discursive news values) met the assumption of no multicollinearity, meaning they did not exhibit excessive correlation with each other. For this, an Ordinary Least Squares (OLS) regression model was used, and the Variance Inflation Factor (VIF) was calculated. In this analysis, all variables showed acceptable values according to the criteria of Hair et al. (2010), i.e., they ranged between 0.02 and 4.

In the second stage, the appropriate statistical model was selected to analyze the data. Since the dependent variables corresponded to counts (number of comments and “likes”) and did not meet the normality assumption, Poisson models and Negative Binomial Regression (NBR) models were considered as alternatives. The NBR model, however, is usually preferred over the Poisson model due to its ability to handle overdispersion in the data. Considering this characteristic, both models were estimated and compared using the Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC). Both statistics supported the superiority of the NBR model over the Poisson model. Additionally, a robust NBR model was applied to control for the heteroscedasticity present in the observations. Consequently, the NBR model was adopted, with its specific equation being:

$$\begin{gathered} \mathit{L}\mathit{o}\mathit{g}\left({\mathit{Y}}_{\mathit{j}}\right)=\mathit{\alpha }+{\mathit{\beta }}_1\mathit{i}\mathit{m}\mathit{g}+{\mathit{\beta }}_2\mathit{c}.\mathit{i}\mathit{m}\mathit{g}+{\mathit{\beta }}_3\mathit{v}\mathit{i}\mathit{d}\mathit{e}\mathit{o}\mathit{c}+{\mathit{\beta }}_4\mathit{v}\mathit{i}\mathit{d}\mathit{e}\mathit{o}\mathit{l}+{\mathit{\beta }}_5\mathit{g}\mathit{e}\mathit{n}+ \\ {\mathit{\beta }}_6\mathit{a}\mathit{s}\mathit{s}\mathit{a}\mathit{u}\mathit{l}\mathit{t}+{\mathit{\beta }}_7\mathit{h}\mathit{o}\mathit{m}+{\mathit{\beta }}_8\mathit{t}\mathit{h}\mathit{e}\mathit{f}\mathit{t}+{\mathit{\beta }}_9\mathit{s}\mathit{e}\mathit{x}+{\mathit{\beta }}_{10}\mathit{s}\mathit{e}\mathit{c}+{\mathit{\beta }}_{11}\mathit{o}\mathit{t}\mathit{h}\mathit{e}\mathit{r}\mathit{s}+ \\ {\mathit{\beta }}_{12}\mathit{c}\mathit{o}\mathit{n}\mathit{s}+{\mathit{\beta }}_{13}\mathit{e}\mathit{l}\mathit{i}+{\mathit{\beta }}_{14}\mathit{p}\mathit{e}\mathit{r}+{\mathit{\beta }}_{15}\mathit{s}\mathit{u}\mathit{p}+{\mathit{\beta }}_{16}\mathit{s}\mathit{o}\mathit{r}+{\mathit{\beta }}_{17}\mathit{t}\mathit{e}\mathit{m}+\mathit{L}\mathit{o}\mathit{g}\left(\mathit{s}\mathit{e}\mathit{g}\right) \end{gathered}$$

where $\mathit{L}\mathit{o}\mathit{g}\left({\mathit{Y}}_{\mathit{j}}\right)$ represents the natural logarithm of the expected number of comments and “likes” on post j; α is the intercept term; ${\mathit{\beta }}_{\mathbf{1}}$ to ${\mathit{\beta }}_{\mathbf{17}}$ are the coefficients associated with the independent variables; “img” denotes a static image, “c.img” a collection of images, “videoc” a short-duration video, and “videol” a long-duration video. The variables “gen”, “assault”, “hom”, “theft”, “sex”, “sec”, and “others” represent the gender of the perpetrator and the different types of encoded crimes, respectively. On the other hand, “cons”, “eli”, “per”, “sup”, “sor”, and “tem” correspond to the considered discursive news values, such as consonance, elite, personalization, superlativeness, surprise, and temporality. It is worth noting that the nationality of the perpetrator and the temporality value were excluded from the equation; as shown later in the descriptive statistics, both variables presented a very low frequency of appearance, less than 5%. Additionally, Log(seg) was included, which represents the exposure effect of the number of followers of each media outlet on Instagram. This term captures the impact that the audience base of each media outlet can have on digital interaction, recognizing that media with a higher number of followers could generate higher levels of engagement, regardless of the characteristics of the posts. This inclusion ensures that the model controls for inherent differences in reach and exposure among the analyzed media, providing more precise and comparable results.

4. Results

To begin, the descriptive results of the quantitative variables are presented in

Table 3. Descriptive statistics.

Var	M	SD	Min	Max
IG Followers	1,327,050	968,825.6	208,000	4,300,000
Likes	3784.2	4824.87	0	40,773
Comments	223.7	412.9	0	4243

, and the qualitative variables in

Table 4. Frequency distribution of qualitative variables.

Var		Freq	%
News Format	Image	522	52.2%
	Image Collection	147	14.7%
	Short Video	136	13.6%
	Long Video	195	19.5%
Crime Details	Male Perpetrator	757	75.7%
	Female Perpetrator	243	24.3%
	Foreign Perpetrator	17	1.7%
Type of Crime	Assault	141	14.1%
	Homicide	357	35.7%
	Robbery	157	15.7%
	Sexual Crimes	149	14.9%
	Kidnapping	139	13.9%
	Others	329	32.9%
News Values	Consonance	226	22.6%
	Elite	555	55.5%
	Personalization	301	30.1%
	Proximity	26	2.6%
	Superlativeness	169	16.9%
	Unexpectedness	262	26.2%
	Temporality	73	7.3%

. The data shows that the analyzed accounts have an average of 1,327,050 followers on Instagram (SD = 968,825.6), with a range from 208,000 to 4,300,000 followers. Regarding interaction, the posts receive an average of 3784.2 likes (SD = 4824.87) and 223.7 comments (SD = 412.9), although there is significant variability in these metrics, as demonstrated by their maximum values of 40,773 likes and 4243 comments, respectively, with both metrics starting from a minimum of 0 interactions.

Regarding the qualitative variables, it is observed that the most common format of the news is the individual image (52.2%), followed by long video (19.5%), image collection (14.7%), and short video (13.6%). Regarding the details of the crime, the majority of reported cases involve male perpetrators (75.7%) compared to females (24.3%), with a low presence of foreign perpetrators (1.7%). In the distribution by type of crime, homicide represents the most frequent category (35.7%), followed by robbery (15.7%), sexual crimes (14.9%), assault (14.1%), kidnapping (13.9%), and other crimes (32.9%). Regarding the news values present in the posts, the presence of elite (55.5%) stands out as the most common value, followed by personalization (30.1%), unexpectedness (26.2%), consonance (22.6%), superlativeness (16.9%), temporality (7.3%), and proximity (2.6%).

Subsequently,

Table 5. Results model.

Var		Likes	Comments
Format	Image	0.392 (***)	0.141
	Image Collection	0.575 (***)	0.825 (***)
	Short Video	0.723 (***)	0.618 (***)
	Long Video	0.026	−0.103
Crime Details	Male Perpetrator	0.104	0.668
	Assault	−0.046	0.033
	Homicide	−0.090	−0.055
	Robbery	−0.037	−0.111
	Sexual Crimes	0.875	−0.213
	Kidnapping	−0.049	−0.287 (*)
	Others	−0.195	−0.060
News Values	Consonance	−0.162 (*)	−0.215 (**)
	Elite	−0.162 (**)	−0.113
	Personalization	0.014	0.043
	Superlativeness	−0.077	−0.098
	Surprise	0.246 (***)	0.430 (***)
	Temporality	−0.053	0.162
Followers		Exp	Exp
Constante		−6.110 (***)	−8.864 (***)
Wald Chi2 (17)		132.89 (***)	115.24 (***)
Pseudo R2		9%	10%
LR Test Alfa		2.9 × 106 (***)	2.9 × 106 (***)

p < 0.10 (*), p < 0.05 (**), p < 0.01 (***). Asterisks indicate the statistical significance of each coefficient.

presents the results of applying the Negative Binomial Regression (NBR) models. As can be seen, the models are jointly significant since the Wald Chi2 statistic is significant at 99% confidence for both likes (132.89, p < 0.01) and comments (115.24, p < 0.01). Likewise, the LR test of the alpha parameter confirms the suitability of using an NBR model instead of Poisson. Finally, the pseudo R2 values indicate that the models explain approximately 9% and 10% of the variance in likes and comments, respectively.

Regarding specific effects, the format of the posts emerges as a determining factor in digital engagement. Posts using individual images (β = 0.392, p < 0.01), image collections (β = 0.575, p < 0.01), and short videos (β = 0.723, p < 0.01) generate significantly more likes than long videos. However, this pattern varies when analyzing comments, where only image collections (β = 0.825, p < 0.01) and short videos (β = 0.618, p < 0.01) show significant positive effects.

Regarding the characteristics of the crime, the results reveal an interesting pattern of relative independence between the type of crime and digital engagement. Most variables related to crime details, including the gender of the perpetrator and different types of crimes (assault, homicide, robbery, sexual crimes, and others), do not show statistically significant effects. The only exception is kidnapping, which shows a significant negative effect on comments (β = −0.287, p < 0.1), suggesting that audiences tend to comment less on posts related to this type of crime.

Concerning news values, significant and contrasting effects are observed. Consonance shows a negative effect on both likes (β = −0.162, p < 0.1) and comments (β = −0.215, p < 0.05), while the elite news value presents a significant negative effect only on likes (β = −0.162, p < 0.05). In contrast, surprise emerges as a significant positive factor for both likes (β = 0.246, p < 0.01) and comments (β = 0.430, p < 0.01), suggesting that audiences tend to interact more with content that presents unexpected or surprising elements. The values of personalization, superlativeness, and temporality do not show significant effects in any of the engagement metrics.

5. Discussion

Based on the findings obtained from the applied models, significant elements emerge that clarify the influence of the characteristics of crime news on the digital engagement of the audience. These results are interesting because, while they confirm some of the research hypotheses, they also challenge other initially formulated propositions. Below, these findings and their practical and theoretical implications are presented in detail.

Regarding the first independent variable analyzed, namely the format of news presentation (H1), the findings partially support the hypothesis regarding the positive influence that representational richness can have on digital engagement. However, contrary to what was proposed in previous studies, long-duration videos do not seem to be the main format contributing to engagement actions, such as commenting or liking an informative post (Gkikas et al., 2022).

On the contrary, both collections of images and short-duration videos demonstrated a more positive and significant impact on the analyzed engagement metrics. This result may suggest that representational richness does not have a linear effect within the context of crime news, but rather that this effect could be mediated or moderated by other relevant factors not considered in the present study, such as the audience’s attentional capacity and the immediacy characteristic of social media consumption. In other words, it could be thought that on these digital platforms, users prefer content that represents a lesser cognitive challenge and, consequently, allows for quicker comprehension and adoption.

Regarding the types of crimes specified in the news, which corresponded to the second research hypothesis (H2A), the findings reveal a result that contradicts the consulted literature. Unlike what was proposed by authors such as Chermak (1998) or Rios and Ferguson (2020), no statistically significant differences were identified in the engagement metrics derived from the type of crime emphasized in the news; the only exception was the crime of kidnapping, which had a marginal and negative effect specifically on the number of comments received. This finding is interesting because it challenges the pre-existing assumption that certain crimes, particularly those of a violent nature, could trigger greater digital interaction from the audience. A plausible explanation for this result, which aligns with the previously mentioned theoretical tension, could be found in a phenomenon of normalization and adaptation to crime by audiences in the countries under study. In other words, constant exposure to various types of crimes may have generated a kind of fatigue or habituation in the viewer towards these events (desensitization), which consequently reduces the differential influence that different types of crimes have on their digital behavior.

Regarding the gender of the perpetrator (H2B), the results allow us to reject the hypothesis that gender influences the digital engagement of the audience. This contradicts the findings of previous studies, such as those by Wozniak and McCloskey (2010) or Martins (2021), who propose greater media attention towards crimes perpetrated by males. The absence of a significant effect could have several contextual explanations. On one hand, on a highly visual platform like Instagram, it is possible that the rawness of the image or the impact of the headline prevails over demographic attributes of the perpetrator, such as gender. On the other hand, this result could reflect an evolution in social norms where the criminal act itself is the focus of the public’s reaction, beyond the gender stereotypes traditionally associated with the perpetrator.

Regarding the nationality of the perpetrator (H2C), the hypothesis could not be tested, as the low prevalence of foreign criminals in the reviewed news limited the capacity for analysis. In fact, only 1.7% of the perpetrators were presented in the news as foreigners, which is striking considering the current migratory situation in Latin America and the findings of previous studies where there is a strong tendency for the media to highlight crimes committed by immigrants (Afrouzi, 2024; Pan et al., 2019). This result could indicate that the prevalence of crimes by foreigners is more a matter of public perception, or it could also suggest a transformation in the practices of the media towards less biased coverage.

Regarding discursive news values (H3), the results show a complex picture that partially supports the hypothesis. The most notable finding is the positive and significant effect of unexpectedness on both likes and comments, which confirms Bednarek’s (2016) theories about the importance of the unusual in capturing attention. However, and in contrast to what was expected according to Makki (2019), consonance showed a negative effect on both engagement metrics. This finding suggests that, in the context of crime news, the digital audience may show greater interest in events that diverge from established patterns rather than those that confirm their pre-existing expectations.

It is pertinent to note the lack of statistically significant effects regarding values such as personalization and superlativeness. This observation is notable, as it contradicts the postulates of the preceding literature that emphasized the importance of these values in promoting digital engagement. In light of this research, the findings could be interpreted as an indication that the operational dynamics of these news values differ within the specific contextual framework of crime news disseminated through social media. In this scenario, it becomes plausible that other factors may acquire greater preponderance in determining the interaction patterns exhibited by the audience.

An aspect of particular interest within the results concerns the negative effect of the elite news value on likes. This finding contrasts with the traditional theory that suggests that references to elite figures increase interest and interaction. A possible explanation for this could lie in a certain skepticism or fatigue on the part of the digital audience towards official sources or authority figures in the context of crime news, perhaps leaning towards narratives that are more direct or community-centered.

6. Conclusions

Based on the applied research, relevant conclusions can be drawn regarding the relationship between the presentation of crime news on social media and digital engagement, measured through digital participation metrics. Although this study focuses on the factors driving interaction and does not directly measure their effects, the conclusions offer important implications for both the media and the audience, as well as for official entities.

From the perspective of the media and the effectiveness of journalistic strategies, the results of this study show that focusing on shorter, more easily consumable, and dynamic formats, such as short videos or image collections, can have a positive effect on capturing and retaining audience attention. However, it is essential that the presentation format of crime news maintains a balance with informative richness so that objectivity and professionalism are not sacrificed. On the other hand, the inclusion of elements that allude to surprise or unexpected events seems to be a good strategy to foster user interaction, rather than other news resources such as references to authority figures or excessive personalization of the actors involved in the news.

Regarding the social component, the apparent normalization of various types of crimes, evidenced by the lack of significant differences in engagement based on the type of crime, suggests a concerning phenomenon of social desensitization to violence in the studied countries. This “normalization” could be contributing to the perpetuation of cycles of violence and the erosion of the social fabric, as the lack of differentiated reaction to different types of crimes may indicate a decrease in society’s ability to recognize and adequately respond to different levels of severity in criminal acts.

Furthermore, the low presence of news about foreign perpetrators, contrary to the common perception of criminality and migration, raises important questions about the social construction of the “other” and the prejudices associated with migration. This finding could serve as a basis for questioning and rethinking xenophobic narratives that often associate criminality with the migrant population, thereby contributing to generating a more informed debate around migration policies and security in the current context.

Regarding future research avenues, several paths appear particularly interesting. Firstly, it would be beneficial to extend the analysis to other social networks with diverse interaction dynamics, such as TikTok, where the predominantly audiovisual format and extreme conciseness may be generating novel ways of communicating and processing information about criminality. The viral nature and recommendation algorithm of TikTok could be shaping informative bubbles with particular narratives about criminality that are well worth studying in detail.

Similarly, future research should consider including alternative and community media, including citizen collectives and grassroots organizations that use social networks to document and denounce criminality from non-institutional perspectives. These actors might be developing narratives and engagement modalities significantly different from those of traditional media, with the potential to be more effective in citizen mobilization and the construction of community responses to insecurity.

Another promising line of research would be the analysis of the content and tone of comments on crime-related posts, not just their quantity. This qualitative analysis could reveal patterns in public discourse about criminality and help understand how different characteristics of the news influence not only the quantity but also the quality of public debate on security. Furthermore, it would be valuable to explore how different strategies for presenting crime news affect not only digital engagement but also the offline attitudes and behaviors of the audience. This could include studies on how the consumption of crime news on social media is configured as a factor influencing participation in community security initiatives and support for different public security policies.

Finally, it is crucial to recognize some limitations of the study. The sampling methodology, by collecting the “last 50 posts” without a strict time frame, could weaken the analysis of news values such as “temporality” and mix current events with older ones. Additionally, as has been clarified, interaction is an indirect indicator of public perception; future studies should employ mixed methods to directly measure how engagement influences the offline attitudes and behaviors of the audience, thereby closing the gap between digital reaction and its tangible social impact.