Place attachment and perceived insecurity on the Mazatlán malecón

Place attachment: definition and dimensions

Place attachment is generally understood as the cognitive and affective bond that a person or a group establishes with a meaningful environment. Scannell and Gifford (2010) proposed organizing the diversity of definitions into a tripartite framework that distinguishes the person who bonds (individual or collective), the psychological process through which this occurs (affective, cognitive and behavioral) and the place to which the person bonds, with its physical and social components. This framework is a classic reference used here because it offers the conceptual vocabulary with which recent literature continues to define the construct. Lewicka (2011), in a review of four decades of research, documented the heterogeneity of definitions and instruments and the concentration of studies on the place of residence, and noted the need to address other scales and the processes through which the bond is formed.

The measurement of place attachment remains a developing field. Boley et al. (2021) warned that the variety of scales and items makes it difficult to compare results across communities and accumulate evidence, and presented a short six-item scale, three on place identity and three on place dependence, whose measurement invariance they supported in samples of residents and visitors from four countries. Erdoğdu (2025), when developing a scale for adolescents, obtained through exploratory factor analysis, supported by parallel analysis, a structure of three subscales—cognitive, emotional and behavioral—that corresponds to the process dimensions of the tripartite framework. In the Chinese-language literature, Xiang et al. (2022) studied the attachment of residents of a historic neighborhood in Yining, Xinjiang, and grouped its components into emotional, social and material bonds, which shows that the content of the dimensions varies according to the type of place and population.

For the present study, these contributions have two implications. The first is that the items of the Bond with the malecón dimension combine contents of personal importance (R1), pride (R2), integration into daily life (R3), anticipated loss (R4) and indifference (R5, reverse), that is, they mix affective, identity and behavioral components that have been separated in other scales. The second is that the analysis of the internal structure must be interpreted in light of the place studied: a publicly and touristically used seaside promenade is not equivalent to the residential neighborhood, which has been the predominant object of research. The review by Kamani Fard and Paydar (2024) on place attachment in urban environments points out, precisely, that the scale of the place and its function are related to the intensity of the bond, and that the social interaction that the place facilitates contributes to forming it.

Perceived insecurity and fear of crime in public space

Perceived insecurity and fear of crime have been conceptualized as experiences with emotional, cognitive, and behavioral components. In Mexico, Clément and Piaser (2021) constructed a fear-of-crime index with these three dimensions, which distinguish the feeling of insecurity, risk assessment, and avoidance or protective behaviors. This distinction is pertinent to the scale studied, whose Fear on the malecón dimension includes items on night-time insecurity (R6), avoidance of areas (R7), fear of a specific crime (R8), calmness in family company (R9, reverse), and vigilance over belongings (R10). The heterogeneity of these contents anticipates that the dimension might not behave as a homogeneous block.

Recent research has emphasized the situated nature of fear of crime. Song et al. (2023), in a survey of residents of communities in Guangzhou, found that fear of robbery varies significantly according to the type of place, with higher values at transport stations and wholesale markets and lower values in leisure spaces. In the same city, Gu and Chen (2024) observed that perceived social and physical disorder in the community is associated with greater fear of crime, whereas formal social control is associated with less fear. In Barcelona, Pérez-Tejera et al. (2022) found that perceived safety in public parks was related to sex and years of residence in the neighborhood, but not to recorded crimes, which confirms that perceived insecurity is not a direct reflection of crime incidence.

The design of space and its conditions of night-time use are also part of this experience. In an image-based randomized trial, Navarrete-Hernandez et al. (2021) showed that the removal of blind walls raised perceived safety, with larger differences among women, and in a neighborhood of Santiago de Chile they documented that perceptions of insecurity are concentrated in specific places and that simulated regeneration of public space was accompanied by greater perceived safety (Navarrete-Hernandez et al., 2023). On a riverfront of New Belgrade, Rakonjac et al. (2022) observed that lighting is related to the distribution of users, to their sense of safety, and to the duration and frequency of night-time use. In Mexico, Ortiz-Brunel et al. (2025) found in parks of Jalisco that men had a higher probability of feeling safe than women and that users of linear parks were more likely to feel safe than those of parks with controlled access.

A gender perspective runs through much of the Mexican evidence. Almanza Avendaño et al. (2022) described, in a city near the northern border, how street harassment is experienced as part of a continuum of threats that includes disappearance and femicide, and how this restricts the mobility of young women. Garfias Royo et al. (2023), in Corregidora, Querétaro, identified lighting, narrow access points, and the condition of sidewalks as infrastructure elements linked to perceived vulnerability, and proposed transferring to urban planning a burden of self-protection that currently falls on women themselves. These studies guide the interpretation of the avoidance and vigilance items, which may reflect coping strategies rather than fear in the strict sense.

Place attachment and perceived insecurity: a relationship under discussion

The relationship between place attachment and perceived insecurity is not unambiguous. On the one hand, perceived safety appears as a positive correlate of attachment: in Guangzhou, the sense of safety was positively associated with place identity, although social ties were the strongest predictor (Chang et al., 2023), and in a large sample of Japanese cities, perceived safety from crime was positively associated with attachment to the place of residence (Li et al., 2025). On the other hand, the experience of victimization in the neighborhood was associated with weaker attachment, although frequent interaction with neighbors attenuated that relationship (Zahnow and Tsai, 2021), which indicates that the bond can persist despite negative experiences when the place retains its social value.

For a space such as the malecón, this discussion suggests three possible scenarios: that attachment and fear are negatively associated, as two poles of the same valuation of the place; that they are independent, so that a person can feel a strong bond with the malecón and, at the same time, take precautions when walking along it; or that their relationship depends on conditions such as the time of day, the stretch walked, or gender. The design of the instrument studied, with two separate dimensions, implicitly assumes the second scenario, and examining its internal structure allows assessing whether the data are compatible with that separation. Given the cross-sectional nature of the study, any relationship observed will be interpreted as an association and not as a relationship of dependence between the variables.

The malecón of Mazatlán as coastal public space

The malecón of Mazatlán brings together the features that the literature associates with the complexity of coastal public space: it is linear and extensive, its character changes from one stretch to another, it concentrates residential, tourist, and sports uses, and its nighttime use is conditioned by lighting and by commercial activity. Qualitative studies conducted in the city show that waterfront renovations and the unequal distribution of security devices shape different experiences for residents and tourists, and for men and women (Cruz-Coria, 2024; Cruz Coria, 2024). Added to this is a state context in which the perception of urban insecurity is subject to periodic statistical monitoring (INEGI, 2026a, 2026b).

Within this framework, the habitability of the malecón cannot be reduced to its facilities. It also encompasses the degree to which residents recognize it as part of their lives and the extent to which its use is conditioned by fear. A brief instrument, with documented evidence of validity and internal consistency, would make it possible to monitor both dimensions and relate them to characteristics of the space, such as the section, the time of day, or the lighting. This study is the first step in that process: the empirical review of the instrument with data from the pilot survey, with the cautions imposed by its size.

Design

A quantitative, non-experimental, cross-sectional study with a descriptive and exploratory scope was conducted, aimed at gathering preliminary evidence on the internal structure and consistency of an instrument and on the association between its dimensions. The data correspond to survey LEV-0008, "Piloto malecón · Apego e inseguridad", of project FZ-HAB-0003 of the Factoriza UAS platform, and to the frozen dataset DST-0013, with a cutoff date of 9 October 2026. The survey is an independent sample and the only one in the project, so data from other surveys were not combined.

Population, sample and data cut

The survey population was defined as persons aged 18 years or older who reside in Mazatlán and visit the malecón. The territory corresponds to the malecón of Mazatlán, Sinaloa, in sections selected by the project lead. According to the Factoriza record, 18 persons opened the questionnaire and 14 completed it; no cases were excluded for quality control, so 14 cases were analyzed (Table 2).

The data cut analyzed is preliminary: it brings together 14 cases out of a target of 100, and the survey remained open, in field status, as of the cutoff date. This condition is central to interpreting all the results. The ratio of 1.4 cases per item is far below what a stable item factor analysis requires, and the adequacy of the data and of the sample is the first condition that the methodological literature calls for verifying (Ferrando et al., 2022; Goretzko et al., 2021), so the internal structure analyses are presented as a preliminary exploration intended to review the instrument, not as conclusive evidence about its dimensionality. Factoriza did not record sociodemographic variables in this dataset.

Instrument

The Scale of attachment to the malecón and perceived insecurity was applied, consisting of ten items with a five-point response format, from Strongly disagree (1) to Strongly agree (5). The items are organized into two dimensions of five items each: Bond with the malecón (R1 to R5) and Fear on the malecón (R6 to R10). Two items are worded in the reverse direction, R5 ("The malecón is a matter of indifference to me") and R9 ("I feel calm when visiting the malecón with my family"), and are recoded in the data matrix, so that in both dimensions a higher score indicates more bond or more fear, respectively. Table 1 presents the code, the dimension, the text, and the direction of each item.

Procedure

Responses were collected through the survey questionnaire managed in Factoriza UAS, which records express online informed consent before the first item and does not collect identifying data. Factoriza froze the DST-0013 dataset with a cryptographic fingerprint, generated the matrix with the reverse-scored items recoded, and produced the descriptive statistics and internal consistency analyses (runs RUN-00061 to RUN-00065). Using the data file and the item dictionary exported by the platform, the complementary analyses described in the following section were run.

Data analysis

The analyses were carried out in two stages. In the first, Factoriza UAS calculated, for each item, the mean, standard deviation, skewness, kurtosis and the percentages of responses at the extreme values, as well as the internal consistency of each dimension using Cronbach's alpha (Cronbach, 1951), with the corrected item-total correlation and alpha if the item is deleted. In the second, the complementary analyses were run with the program analisis_psicometrico.py, version 1.0, in Python 3.13.16 (Python Software Foundation, 2026), with the libraries NumPy 2.5.3 (Harris et al., 2020), SciPy 1.18.1 (Virtanen et al., 2020) and pandas 3.0.5. The program applies fixed, declared procedures, with no manual intervention in the calculations.

In this second stage, McDonald's omega total (McDonald, 1999/2013) was estimated for each dimension from a one-factor model, without reversing items, as a complement to alpha, which assumes equivalent loadings across items. Internal structure was explored through exploratory factor analysis on the Pearson correlation matrix of the 10 items and the 14 complete cases, after estimating the Kaiser-Meyer-Olkin measure of sampling adequacy (Kaiser, 1974) and Bartlett's test of sphericity (Bartlett, 1950); both are cited from their original source because they are the methodological references for these indices. The number of factors was determined through parallel analysis (Horn, 1965) with 100 simulations and seed 2026; extraction was by minimum residual and rotation was oblimin, which allows correlation between factors. The full pattern matrix loadings, communalities, item-level sampling adequacy and explained variance are reported. Relationships between dimensions were estimated with Spearman's coefficient between the mean scores of each dimension, with a 95% confidence interval based on Fisher's transformation and the standard error of Bonett and Wright (2000).

The reference values used to interpret the indices, such as the .30 threshold for considering a loading relevant, the .50 threshold for item sampling adequacy or the .60 threshold for the overall Kaiser-Meyer-Olkin index, are treated as orientative references from the literature and not as decision rules (Ferrando et al., 2022; Kaiser, 1974). The program does not automatically remove items: it flags those with insufficient loading, cross-loading, opposite sign or low sampling adequacy, and leaves the decision to the authors. No confirmatory factor analysis was run, because with a preliminary data cut of this size and with the same sample as the exploratory analysis it would not be defensible; it is proposed as future work with the final data cut or with an independent sample.

Ethical considerations

Participation was voluntary and anonymous. Factoriza UAS records each participant's express online informed consent and does not collect names, emails or other identifying data, and the analyses were performed on the frozen matrix with no possibility of linking responses to individuals.

Statement on the use of artificial intelligence

During the preparation of this manuscript, the following were used: the Factoriza UAS platform, for data management and descriptive statistics and internal consistency analyses; a Python analysis program with fixed and validated procedures, for omega, exploratory factor analysis and correlations between dimensions; and the Claude Opus language model (Anthropic), to run that program, assist in the literature search, draft a preliminary version and assemble the document from the results. The authors reviewed, corrected and approved the entire text, verified every reference and assume full responsibility for the content of the article.

Sample composition

Descriptive statistics

Table 3 and Figure 1 present the descriptive statistics for the items. In the Bond with the malecón dimension, the highest means were for R4, “I would miss the malecón if I left Mazatlán” (M = 3.93, SD = 1.00), and R1, “The malecón is an important place for me” (M = 3.86, SD = 0.95), while the lowest was for R2, concerning pride in the malecón (M = 3.14, SD = 0.95). In the Fear on the malecón dimension, the highest mean was for R10, which refers to vigilance over belongings (M = 3.43, SD = 0.76), and the lowest was for R6, feeling unsafe when walking at night (M = 2.57, SD = 1.09). R9, tranquility when visiting the malecón with the family in its recoded version, showed the greatest dispersion in the set (SD = 1.25).

In descriptive terms, the bond scores fell above the midpoint of the scale and the fear scores around it, with the exception of R10. Skewness ranged from -0.93 (R7) to 0.32 (R1). The kurtosis of R1 (-1.99) and of R2 (1.63) indicates distributions far from normal in both directions, although with 14 cases these indices are unstable. In R1, Factoriza recorded 50 % of responses at the floor and 36 % at the ceiling, which describes a polarized distribution. The table note indicates a pending verification of how Factoriza defines the floor and the ceiling.

Internal consistency

Table 4 brings together the item analysis that Factoriza calculated for each dimension, and Table 5 presents the internal consistency coefficients. The Cronbach's alpha calculated by the complementary program matched that of Factoriza in both dimensions: .848 for Bond with the malecón and .812 for Fear on the malecón. The omega total was .870 and .839, respectively. No dimension presented items with a negative correlation with respect to the rest of its dimension, so the omega is considered interpretable in both cases.

The item-by-item analysis qualifies these overall coefficients. In Bond with the malecón, R2, "I feel proud of the malecón in Mazatlán," showed the lowest corrected item-total correlation (.340), and its elimination would raise the dimension's alpha to .896, above the value for the full scale. The remaining items in the dimension presented corrected correlations between .573 (R4) and .837 (R5). In Fear on the malecón, R6, "I feel unsafe walking along the malecón at night," had the lowest corrected correlation (.412) and its elimination would raise alpha to .835, also above the value for the full dimension; R8, "I fear being mugged on the malecón," presented the highest (.870). These results identify R2 and R6 as the items least coherent with their dimension in this data cut.

Exploratory factor analysis

The overall Kaiser-Meyer-Olkin measure of sampling adequacy was .283, far below the guideline reference of .60 and below the value of .50, under which Kaiser (1974) rated adequacy as unacceptable, whereas Bartlett's test of sphericity rejected the hypothesis of an identity matrix, χ²(45) = 161.88, p < .001 (Table 7). The discrepancy between the two indicators is to be expected with such a small sample: correlations of appreciable magnitude exist, but the partial correlations are also appreciable, and the matrix lends itself poorly to a stable factor solution. At the item level, an index that is useful for identifying inadequate items before interpreting a factor solution (Lorenzo-Seva and Ferrando, 2021), only R1 reached a sampling adequacy above .50 (.843); the remaining nine items ranged between .123 (R2) and .401 (R6), as Table 6 shows.

The parallel analysis indicated the retention of two factors (Figure 2). The plot shows that the first eigenvalue of the data clearly exceeds the simulated one, whereas the second exceeds it by a narrow margin, which indicates that the decision on the second factor is fragile and could change with more cases. The two factors together explained 66.67 % of the variance: 45.22 % the first, with a sum of squared loadings of 4.52, and 21.45 % the second, with 2.15 (Table 7). The correlation between factors estimated by the oblimin rotation was -.243.

The pattern matrix (Table 6) did not reproduce the expected separation between the two dimensions. The first factor grouped the five bond items with high positive loadings, from .509 (R2) to .963 (R5), but also two fear items with negative loadings: R7, "I avoid certain areas of the malecón" (-.707), and R9, the recoded calmness with family (-.806). It is, therefore, a bipolar factor in which the bond with the malecón is opposed to avoidance and to the lack of calmness. The second factor was defined by R10, "I am alert to my belongings" (.977), R8, "I fear being mugged" (.714), and R6, "I feel unsafe walking at night" (.508), that is, by the contents of vigilance, fear of a specific crime and night-time insecurity.

The program flagged several items that require a decision by the authors. R2 (.509 and .409), R4 (.672 and .307), R8 (-.488 and .714) and R9 (-.806 and .361) showed cross-loadings; R7 and R9 loaded with the opposite sign to that of their dimension on their main factor, and all items except R1 showed low sampling adequacy. In addition, four items reached communalities above .90 (R5, R8, R9 and R10), values that with 14 cases should be read as a symptom of overfitting rather than as evidence of exceptionally representative items. No item was eliminated; the analysis is reported with the ten original items.

Relationship between the dimensions

The Spearman correlation between the mean scores of Bond with the malecón and Fear on the malecón was negative and of moderate magnitude, ρ = -.367, with a 95% confidence interval of [-.759, .222] and p = .197, with 14 cases (Table 8). The interval includes zero and ranges from a strong negative association to a weak positive one, so the data from this cut do not allow precise estimation of either the sign or the magnitude of the relationship. Moreover, since the factor analysis showed that two fear items group with the bond items, the mean score of the fear dimension does not represent a homogeneous construct, and the correlation between dimensions must be interpreted with that caveat.

Discussion of results

The purpose of this study was to examine, in the preliminary data cut of the pilot survey, the initial behavior of a brief scale of attachment to the malecón and perceived insecurity. The main finding is a divergence between the internal consistency indicators and the evidence of internal structure. Both dimensions reached high alpha and omega coefficients, but the exploratory factor analysis did not reproduce the two expected dimensions, and the sampling adequacy indices indicated that the correlation matrix is poorly suited to analysis. This divergence illustrates a well-established principle: internal consistency is not in itself evidence of unidimensionality or of structural validity. Simulation studies also show that alpha and omega tend to behave similarly under usual conditions (Malkewitz et al., 2023), and in this case both were estimated with a number of cases far below the smallest of the scenarios simulated in that literature, so they must be read as provisional values.

The configuration of the first factor admits a substantive reading that must be tested against further data. That avoidance of areas (R7) and the lack of peace of mind when visiting with family (R9) oppose the bond with the malecón is consistent with studies that associate perceived safety with stronger attachment (Chang et al., 2023; Li et al., 2025) and with evidence that negative experiences in the place are related to a weaker bond (Zahnow and Tsai, 2021). In particular, R9 alludes to an experience of shared enjoyment of the place, closer to the affective benefits of attachment than to fear in the strict sense, so its location in the bond factor may reflect its content and not a measurement error. This interpretation is tentative and poses a content decision for the authors: whether R9 and R7 should remain in the fear dimension, be worded differently, or be relocated.

The second factor, made up of vigilance over belongings, fear of robbery and night-time insecurity, approximates the cognitive and emotional components of fear of crime described by Clément and Piaser (2021), whereas avoidance of areas would correspond to its behavioral component. The separation between these contents suggests that the Fear on the malecón dimension, as worded, combines facets that might require different subscales or a more homogeneous selection of items. The literature on the situated character of fear of crime reinforces this reading: fear varies according to the type of place and its conditions (Song et al., 2023; Gu and Chen, 2024), and on the malecón the night-time experience, lighting and the stretch walked could differentiate responses that the scale groups into a single score (Cruz-Coria, 2024; Rakonjac et al., 2022).

The behavior of R2 deserves particular attention. Pride in the malecón showed the lowest item-total correlation in its dimension, the lowest sampling adequacy of the set and a cross-loading. One possible explanation is that civic pride in an emblematic space can be sustained independently of personal bond; scales that distinguish place identity and place dependence (Boley et al., 2021) or cognitive, emotional and behavioral components (Erdoğdu, 2025) offer frameworks for reviewing whether this item belongs to a different facet of the construct. Something similar occurs with R6 in the fear dimension, whose elimination would raise alpha, which indicates that night-time insecurity does not covary in the same way with the rest of the fear contents in this sample.

As for the relationship between dimensions, the observed correlation was negative, in the direction suggested by part of the literature, but its confidence interval includes zero and is too wide to support a conclusion. With the survey target of 100 cases, the estimate would gain precision, although the magnitude of the association will also depend on how the composition of the fear dimension is resolved. The cross-sectional design prevents, in any case, establishing the direction of a possible relationship: a stronger bond could be accompanied by less fear, but it is also plausible that those who perceive the malecón as safe visit it more and develop a more intense bond. Finally, the review of the data matrix must consider the quality of the responses, given that invariant response patterns can inflate or distort correlations in small samples (Ward and Meade, 2023).