Attitudes of Police Recruits Towards Offenders: the Impact of the Police Training on Attitudinal Change

The traditional prevailing perspectives on attitudes assume that attitudes, once formed, are relatively stable over time. However, research has revealed that attitudes may be influenced by different factors that may change them. The present pre-post-test study aims to understand if the police recruits’ attitudes towards offenders change after police training and analyse the impact of sociodemographic and personality variables on recruits’ attitudes towards offenders. The Attitude Scale towards Offenders (ATO) and the NEO-FFI Personality Inventory were used in a sample of 74 recruits of a Portuguese Police Force at the beginning and the end of the police training program. Results revealed that, at the end of the training, recruits hold more negative attitudes towards offenders. Those who had previous professional experience in the army hold more positive attitudes, but only at the beginning of the training. Only attitudes towards offenders at the beginning of the police training program predict recruits’ attitudes towards offenders at the end of the training. These results emphasize the importance of initial and continuous training of recruits and police officers.


Introduction
Attitudes are evaluations of an object, concept or behaviour, based on personal beliefs with an appreciative meaning (e.g. good-bad; pleasant-unpleasant, harmful-harmless; Ajzen and Fishbein 2000). They can be formed through personal experience, information received from various means or through others' beliefs or statements (Garner 2005). As Hogg and Vaughan (2005) referred, an attitude is "a relatively enduring organization of beliefs, feelings, and behavioural tendencies towards socially significant objects, groups, events or symbols" (p. 150).
The prevailing view has been that there is a strong correspondence between a person's attitude and behaviour (e.g. Garner 2005;Myers 2013;Robinson and Chandek 2000) and that an individual's behaviour is consistent with their attitudes. However, research has found that discrepancies between attitudes and behaviours may occur (e.g. Myers 2013) and that expressed attitudes may change over time (e.g. Garner 2005). Many factors could influence the variation (e.g. Myers 2013), such as context, the subject's information about the attitude object and the salience of one's thoughts, feelings or memories at the time of their past actions towards the attitude object (e.g. Garner 2005).
Nonetheless, the study of attitudes remains of particular importance, especially among professionals who deal with specific groups of individuals daily, which is the case of police officers. Their attitudes are critical since prejudice and stigma might lead to negative attitudes toward offenders (e.g. Dovidio et al. 2005) and, consequently, influence how they respond to the incidents (e.g. Logan et al. 2006). However, police officers' attitudes toward offenders and law enforcement may be subject to significant changes throughout their careers (e.g. Garner 2005). The present pre-posttest study examines attitudinal changes related to offenders among police recruits of a Portuguese Police Force after * Olga Cunha olga.cunha@ulp.pt police training. We also aim to understand which sociodemographic and personality variables impact recruits' attitudes towards offenders.

Police Officers' Attitudes Towards Offenders
Literature has revealed that police officers' attitudes toward offenders and crime might influence how they respond to criminal events (e.g. Gaines and Kappeler 2005). Muir (1977), one of the first authors theorizing about how attitudes predict police officers' behaviour, proposed two attitudinal dimensions: perspective (tragic or cynical) and passion (integrated or conflicted). Police officers with a tragic perspective consider all the sides and recognize the circumstances and motives that lead offenders to act. Cynical officers do not understand offenders' circumstances and are not empathetic. These two perspectives were then analysed regarding officers' perspectives on the use of coercive tactics. Officers who agree with coercive tactics have an integrated morality and those who do not have a conflicted morality (Muir 1977). Subsequent research revealed that the attitudes towards offenders might impact police officers' responses to crime and offenders. For example, Parker et al. (2004) found that police officers with a more empathetic and tolerant attitude towards young offenders were more likely to divert them. Similarly, Schaible et al. (2020), using data collected during a Law Enforcement Assisted Diversion training session for officers, found that officers who hold an optimistic view towards offenders' rehabilitation were more likely to divert offenders. Research has been consensual in stating that police officers are more likely to hold negative attitudes regarding offenders than other groups, such as community members, students, criminal justice professionals and correctional officers (e.g. Fielding and Fielding 1991;Hogue 1993;Melvin et al. 1985;Terpstra and Kort 2016). Melvin et al. (1985) found that police officers hold a significantly more negative view towards prisoners than university students, community members, correctional officers, prisoners and reform/rehabilitation workers. Likewise, Cunha and Gonçalves (2017) found that Portuguese police officers had less favourable views towards offenders than graduate students and correctional officers.
Nonetheless, police officers' attitudes towards offenders seem to vary according to the type of crime. Hogue (1993) found that police officers presented more negative perceptions about sex offenders than about other groups of offenders. Further, Hogue and Peebles (1997) found that British police officers had significantly less favourable attitudes towards sex offenders than other British professionals, including mental health workers, social workers, probation, or parole officers. The high visibility and censorship of some crimes and the punitive criminal policies towards some crimes might lead to the formation of negative perceptions, beliefs and attitudes towards specific offenders, such as sex offenders (e.g. Church et al. 2011).
Different studies also have concluded that compared to other groups, police officers hold less favourable attitudes towards offenders' rehabilitation (e.g. Chen and Einat 2015;Terpstra and Kort 2016). Fielding and Fielding (1991) found that a large portion of police recruits did not feel that criminals needed treatment and felt that prison inmates are a threat to the community if released on probation or parole. Similarly, Furnham and Alison (1994) observed that police officers advocate harsher sentences for criminal acts and are more prone to view crime as a deviation. Terpstra and Kort (2016) also found that only a small percentage of police officers believe that rehabilitation is an essential function of criminal justice. They also found that officers felt that pretrial detention should be used more frequently, viewing it as an essential tool for preserving order in neighbourhoods and as a first and provisional sentence for an assumed offense.
The negative attitudes shown by police officers may impact offenders' rehabilitation and their consequent reintegration into society. When a professional promotes the involvement of the offender in rehabilitation and adopts a more tolerant and understanding approach (e.g. Gakhal and Brown 2011), it might influence offenders' search and integration in rehabilitation programs and, therefore, change offenders' behaviour (e.g. Beech and Fordham 1997). Contrarily, professionals who show a more positive attitude towards offenders revealed a greater involvement, satisfaction and commitment to work, translating into higher effectiveness of the rehabilitating strategies (e.g. Craig 2005), when compared to professionals who show negative attitudes (e.g. Misis et al. 2013). Thus, although police officers do not have the role of rehabilitating the offender, the more positive the offenders' contact with the judicial system, including police officers, the better the prognosis for success in future rehabilitation.

Correlates of Attitudes Towards Offenders
Identifying variables associated with police officers' attitudes towards offenders may improve these professionals' knowledge, reduce stigma and facilitate offenders' rehabilitation (e.g. Rade et al. 2016). Several factors have been related to police officers' attitudes. Prior research has found that police training program impacts police officers' attitudes exposing them to both the theoretical constructs and realities of policing (e.g. Andersen 2006). Besides, during the training program, recruits develop strong peer bonds and a sense of group identity (e.g. Obst and Davey 2003) and learn to model the behaviours and attitudes of their mentors (e.g. Haas et al. 2015;Miles-Johnson 2019), which may also significantly impact their attitudes towards offenders. Research also revealed that individuals' direct experience with the object of attitude might influence attitudes towards that object (e.g. Cunha and Gonçalves 2017). Similarly, interpersonal contact theory (e.g. Pettigrew and Tropp 2006) advocates that increased contact with a group member is associated with decreasing prejudice and/or stigma towards that group. Consistent with this theory, individuals who report more contact with offenders tend to hold more favourable attitudes towards them (e.g. Hirschfield and Piquero 2010;Rade et al. 2016;Viki et al. 2012). Besides, police officers who were more experienced and had more years of service tended to adopt more positive attitudes towards offenders than less experienced police officers (e.g. Cunha and Gonçalves 2017;Gakhal and Brown 2011). Literature also found that police officers who received training about offenders and who had a greater knowledge of the law demonstrated a significant improvement in their attitudes towards offenders and crime (e.g. Craig 2005;Hogue 1993;Kleban and Jeglic 2012;Lin et al. 2020).
Several studies have analysed the relationship between gender and attitudes, concluding that although both genders demonstrate punitive attitudes, male officers tend to score higher than their female counterparts (e.g. Chen and Einat 2015;Unnever and Cullen 2007). Besides, female officers tend to have a more optimistic view towards offenders' rehabilitation (e.g. Schaible et al. 2020). Nonetheless, other studies found that male and female individuals hold similar attitudes towards offenders (e.g. Hirschfield and Piquero 2010;Rouhani et al. 2019), while others found that women tend to be more punitive than men (e.g. Mancini et al. 2010;Willis et al. 2013).
Other demographic variables, such as age and education, revealed mixed findings. Age has been an important factor that influences police officers' attitudes towards offenders, with older professionals showing positive attitudes more frequently (e.g. Cunha and Gonçalves 2017;Hirschfield and Piquero 2010) and younger ones demonstrating a higher tendency to opt for detention than rehabilitation (e.g. Misis et al. 2013). However, other authors found that older individuals are more punitive than younger ones (e.g. Hough and Moxon 1985; Klein and Cooper 2019), while others found no association between age and punitive attitudes (e.g. Chen and Einat 2015;Mancini et al. 2010;Shafiq et al. 2016). Concerning education, there is no consistency in literature, with some studies demonstrating a correlation between higher education and a lower level of punitive attitudes (e.g. Chen and Einat 2015;Willis et al. 2013), while others pointing to the opposite direction (e.g. Cunha and Gonçalves 2017), and others suggesting no effect of education level on attitudes (e.g. Hirschfield and Piquero 2010;Mancini et al. 2010).
Personality variables also seem to be related to attitudes towards offenders. For instance, Olver and Barlow (2010) concluded that Agreeableness and Openness to Experience positively correlate to attitudes towards sex offenders and sex offenders' rehabilitation. Extraversion was also correlated to more negative attitudes towards sex offenders and their rehabilitation. This result might be explained by the fact that these individuals value the will and wish for the offenders to rehabilitate, believe in the effectiveness of treatments and the offenders' right to change and reintegrate society (Olver and Barlow 2010). Studies on personality variables have also found a positive association between conservative attitudes and punitiveness (e.g. Huang et al. 1996) and between conservative attitudes and negative attitudes towards offenders and offenders' rehabilitation (e.g. Rosselli and Jeglic 2017).

Current Study
Examining police officers' attitudes is crucial since attitudes might influence how police officers interact with offenders and respond to criminal incidents (e.g. Melvin et al. 1985;Robinson and Chandek 2000). The present prepost-test study aims to assess recruits' attitudinal change during their training, testing if the police training influences recruits' attitudes towards offenders. Participants, police recruits from a Portuguese Police Force -Guarda Nacional Republicana (GNR) -were assessed at the beginning and the end of their training program. We also aim to understand which sociodemographic and personality variables influence police recruits' attitudes towards offenders and which variables predict recruits' attitudes.
GNR is a Portuguese military police force constituted by militaries organized in a special body of troops. Its functions are to keep peace and public order, ensure obedience to the law, prevent criminality, develop criminal investigations and supervise and resolve traffic problems, among others (Cunha and Gonçalves 2017). The GNR training program is a residential program (i.e. recruits live at the training centre) consisting of three phases: (i) general military training (150 h) that includes subjects related to the military component (e.g. decision-making process; techniques, tactics, and operations; justice and discipline); (ii) educational training (800 h) which is composed of themes that are associated with the different areas of performance (e.g. police legislation; law; human, material and financial resources; communications, public relations, and protocol; sociology, psychosociology) and (iii) in-service training (480 h) that involves practical training in the territorial units. The total length of the GNR training program is 1430 h, distributed by 42 weeks. The training programs take place once a year at one of the two GNR training centres.

Participants
The participants of this study were selected through a convenience non-probabilistic sampling process. Data were collected in two different moments: at the beginning and the end of the training program. After pairing the collected data between the first and the second moments, the final sample was composed of 74 police recruits from GNR.

Instruments
A Sociodemographic Questionnaire was developed for the present study to collect information about gender, age, educational qualifications, marital status and previous professional experience.
An adaptation of the "Attitudes toward Prisoners" scale (ATP; Melvin et al. 1985) was used, the "Attitudes toward Offenders" (ATO) scale (Cunha and Gonçalves 2017), where the word "prisoner" was replaced by "offender" in all items. The ATP is a self-reported measure consisting of 36 items assessing attitudes towards prisoners. The items are rated on a five-point scale, ranging from 0 (strongly disagree) to 4 (strongly agree). The higher the total score on the scale, the more positive attitudes towards inmates. The instrument presents good psychometric properties. In the original study, the ATP had split-half reliability ranging from 0.84 to 0.92 in five different samples (i.e. students, correctional officers, police officers, prisoners and personal involved in prisoner rehabilitation), and the test-retest reliability was 0.82 (Melvin et al. 1985). The Portuguese ATP version has 34 items since two items were negatively correlated with the total score and were removed (Gonçalves and Vieira 2007). Cronbach's alpha was 0.91 for the Portuguese version (Gonçalves and Vieira 2007). For the present sample, the internal consistency of the total scale was 0.92 for the first moment and 0.86 for the second moment.
The NEO Five-Factor Inventory (NEO-FFI; Costa and McCrae 1992;Portuguese version Magalhães et al. 2014) is a 60-item self-report instrument to assess personality according to the Five-Factor Model: Neuroticism, Extraversion, Openness to Experience, Agreeableness, and Conscientiousness. The NEO-FFI comprises self-descriptive statements responded in a five-point Likert scale from 0 (strongly disagree) to 4 (strongly agree). Both the original and the Portuguese version reveal adequate internal consistency values, varying between 0.68 and 0.86 for the original version (Costa and McCrae 1992)

Procedures
Data was collected through an online survey. Initially, the research team contacted the Police Direction to request authorization for the police recruits to participate in the study. After obtaining permission, the Police School was contacted through the institutional supervisor to clarify the objectives and define the data collection procedures. A link to the protocol was then sent to the institutional supervisor, who shared it with the police recruits through the institutional email, inviting them to participate in a study about their attitudes towards offenders. The participants could withdraw from the study at any time, under ethical board regulations. Data were collected in two different moments: at the beginning and the end of the police training program. An identification code was requested at the beginning of the protocol. This code did not allow the participants' identification as its purpose was only to match the participants' identity in the two moments. Thus, the study's participation was anonymous and voluntary, and no financial assistance, compensation or incentives were provided, as well as no harm could result from refusing to participate.
This study was conducted in 2018. All ethical procedures established by the University of Minho Ethics Commission and Portuguese legislation were followed.

Statistical Analysis
All the analyses were conducted using the Statistical Package for the Social Sciences (SPSS -version 27.0). Descriptive statistics were performed to describe participants' sociodemographic characteristics. Wilcoxon tests were used to analyse the differences in attitudes between moment 1 and moment 2, and Mann-Whitney tests were performed to examine differences in attitudes according to sociodemographic and personality variables. Pearson's correlation tests were also conducted. A linear regression analysis was performed to determine the predictors of recruits' attitudes.

Attitudes Towards Offenders
In ATO, police recruits obtained an average score of 70.24 (SD = 15.57) at the beginning of the police training program and 68.12 (SD = 11.09) at the end. Although there was slight decrease in the average score (i.e. 3%), the Wilcoxon test revealed statistically significant differences in attitudes towards offenders between moment 1 and moment 2, Z = − 2.402, p = 0.016. Thus, police recruits held more negative attitudes towards offenders at the end of the training program than at the beginning (see Table 2). Results also revealed a significant decrease of almost 29% in the standard deviation from moment 1 to moment 2, which indicates that during the training period, the recruits' attitudes towards offenders became more consensual. An item analysis of the ATO revealed a significant difference at the beginning of the police training and after the training on 11 items: "offenders never change", "offenders have feelings like the rest of us", "it is not wise to trust an offender too far", "I think I would like a lot of offenders", "trying to rehabilitate offenders is a waste of time and money", "there are some offenders I would trust with my life", "offenders are just plain immoral", "most offenders can be rehabilitated", "I would like associating with some offenders", "offenders respect only brute force" and "if a person does well in prison, he should be let out on parole". Results are presented in Table 2. These differences represent a significant worsening in attitudes towards offenders post police training.

Sociodemographic Variables and Attitudes Towards Offenders
The results concerning sociodemographic variables are presented in Table 3. Results revealed statistically significant differences in attitudes towards offenders between recruits with and without previous professional experience in the army but only in moment 1, U = 228.500, p = 0.012. Recruits with experience in the army hold the most positive attitudes towards offenders. There were no statistically significant differences in recruits' attitudes towards offenders in the other variables (i.e. age, gender, and educational level), neither in moment 1 nor moment 2.

Dimensions of Personality and Attitudes Towards Offenders
The correlations between the recruits' attitudes toward offenders, both in moment 1 and moment 2, and dimensions of personality are presented in Table 4. Results showed no statistically significant correlations between the variables, neither in moment 1 nor in moment 2.

Predicting Attitudes Towards Offenders
Linear regression was conducted to examine if the attitudes towards offenders in moment 1 (the only statistically significant variable) predict recruits' attitudes toward offenders in moment 2. Table 5 presents the linear regression model.

Discussion
This study aimed to increase our understanding of the attitudes of Portuguese police recruits towards offenders. We intended to identify potential changes in attitudes of police recruits towards offenders during their initial police training program. We also examined which sociodemographic variables and personality dimensions impact recruits' attitudes towards offenders and which variables predict recruits' attitudes. The current study extended previous research on police officers' attitudes towards offenders using a pre-posttest design and a sample of police recruits.
Results revealed that, although slightly, police recruits changed their attitudes towards offenders during their training, holding more negative attitudes at the end of the police training program. These results seem to contradict previous research that point that greater knowledge and contact with offenders are related to more positive attitudes towards offenders (e.g. Craig 2005; Hirschfield and Piquero 2010; Kleban and Jeglic 2012;Rade et al. 2016;Viki et al. 2012). However, since the in-service training is the only moment during the training program that recruits contact with offenders and only consists of 480 h, it might not be enough time to form a clear, informed and consistent opinion about offenders. The analysis of the ATO items revealed that this negative change was more evident in items related to offenders' rehabilitation and a view of the offenders as deviant individuals. Although the attitudinal change was in the opposite way that we expected, these findings are in line with previous studies that revealed that police officers tend to hold more negative attitudes towards offenders' rehabilitation, advocate harsher sentences for criminal acts and are more prone to view crime as a deviation (e.g. Fielding and Fielding 1991;Furnham and Alison 1994). This attitudinal change in a more negative way after police training may be related to the nature of police training and job requirements. The aim and content of the police training were mainly focussed on the acquisition of military skills (e.g. general staff doctrine, decision-making process; techniques, tactics and operations) and the theoretical and practical knowledge on the topics directly related to the job (e.g. police legislation; law; communications, public relations and protocol) and less on the acquisition of knowledge about crime's causes and offenders' characteristics and motivations. As the police officers' main functions are controlling the offenders' behaviour and, when necessary, arresting them, they must focus on the negative characteristics of the offenders  (Kjelsberg et al. 2007). Thus, it is possible that during the training, negative implicit messages about offenders might be delivered and worsen the attitudes of recruits towards offenders (e.g. Craig 2005). Another possible explanation may be related to the fact that for most recruits, this was their first contact with the law, the institution, the offenders (during the in-service training) and with their duties as law enforcement agents. Therefore, this "confrontation with reality" and the confrontation with their professional demands and their job risks might impact their personal and professional lives (Ribeiro and Surrador 2005). These difficulties might be perceived in the first contact with offenders, leading to disappointment and frustration and, consequently, triggering negative attitudes towards offenders since they can be perceived as the cause of their difficulties. Besides, previous research indicates that police officers may use a defence mechanism against stress in the first years of their career, the so-called "John Wayne Syndrome" (Chandler and Jones 1979). This mechanism described by emotional withdrawal and coldness, authoritarian attitudes and cynicism reflects the adaptation problems caused by the difficulties in carrying out their tasks or interpersonal difficulties when they face their professional demands (Chandler and Jones 1979). Literature suggests that the salience given to the crime and the offenders is a strong predictor of more punitive attitudes (e.g. Costelloe et al. 2009). Since police recruits are at the beginning of their careers, they might experience more stress, leading to a more negative view of offenders. However, to test this hypothesis, it would be necessary to conduct in-depth interviews with police recruits to explore their perceptions about offenders and their professional roles.
The attitudinal change might also be related to a social desirability effect and a desire for society's admiration and/ or supervisors' recognition. Although we tried to control this variable through anonymous participation, recruits may have responded negatively to show a more authoritarian and cynic attitude towards offenders, believing that this view makes them more valued and respected by the community and/or to please their supervisors. Research demonstrates that the supervisors are mediators between the communication of expectations to the recruits and implementing these expectations when dealing with crime; thus, the recruits may replace their priorities and expectations with the supervisors' priorities and expectations (Engel and Worden 2003). The leadership theory (Schein 2017) may also explain police recruits' attitudinal change. Attitudes can be formed through personal experience and influenced by others' beliefs and statements (e.g. Garner 2005). Those perceived as leaders within the organization may particularly influence recruits' attitudes and perceptions (Schein 2017). Thus, the increased negative perceptions/attitudes towards offenders may result from senior officers and instructors sharing experiences that influence recruits' perceptions (e.g. Haas et al. 2015;Miles-Johnson 2019).
In addition, an interesting finding relates to the significant reduction in the ATO standard deviation from moment 1 to moment 2, which suggests that, during the training period, the recruits' attitudes towards offenders increased in similarity. That is, recruits seem to develop a consensus in their view on offenders. This result is consistent with prior literature suggesting that police officers' attitudes tend to transform and become more homogeneous over time (e.g. Andersen 2006). Different factors may explain this result. First, as mentioned above, the formal police training and the modelling of attitudes of senior police officers may contribute to align recruits' attitudes since they are exposed to both. Second, as police training programs are physically and mentally challenging, recruits tend to develop strong peer bonds and a sense of group identity (e.g. Obst and Davey 2003). Third, the residential nature of the GNR training program can have a significant impact in transforming attitudes among recruits (e.g. Cooper-Thomas and Anderson 2002) due to the time they spend together. Thus, attitudinal changes seem to be explained by formal and informal socialization processes, and individual experiences occurred during police training program (Andersen 2006). Regarding the sociodemographic variables, recruits who served in the army held more positive attitudes towards offenders than those who did not, although this difference was only observed at the beginning of the police training program. However, it is essential to state that the only similarity between army and GNR is the military structure, which does not necessarily mean direct contact with offenders. In fact, in most cases, recruits who serve in the army do not contact with offenders and/or crime. In the other sociodemographic variables, no significant results were found. However, these findings were not unexpected since previous literature revealed mixed findings concerning demographic variables (e.g. Chen and Einat 2015;Cunha and Gonçalves 2017;Hirschfield and Piquero 2010;Klein and Cooper 2019;Shafiq et al. 2016). Nonetheless, one possible explanation for the absence of significant results is the small sample size and its sociodemographic characteristics (e.g. age, educational level), making it challenging to perform comparisons between groups.
Also, no significant statistical correlations were found between the recruits' attitudes towards offenders and their personality dimensions. The absence of results might suggest that perhaps police recruits share a set of common and homogeneous personality characteristics (e.g. Garbarino et al. 2012;Twersky-Glasner 2005). Those features could explain their predisposition to choose this specific job.
Alternatively, the selection process for police recruits, which in the GNR case includes a psychological interview, might promote a specific personality type. Therefore, there was a predominant set of personality features across police personnel (Twersky-Glasner 2005). Nonetheless, this assumption needs further examination. Another possible explanation for our findings is the type of questionnaire used to assess personality dimensions, i.e. a self-report instrument and, therefore, more vulnerable to social desirability. Besides, reliability values on the dimension of Openness to Experience values were unacceptable, revealing problems with internal consistency, which undermines our confidence in the results obtained in this domain. This is particularly problematic since literature identifies Openness to Experience as a dimension positively correlated to attitudes towards offenders (Olver and Barlow 2010).
A regression was performed to examine which variables predict the recruits' attitudes towards offenders. The regression confirmed that attitudes at the beginning of the police training program were positively correlated with attitudes at the end of the police training program. Thus, more negative attitudes in the first assessment predict more negative attitudes in the second assessment. These findings seem to suggest that police training -involving new experiences, skills and potentially new group influences -has a critical impact on recruits' attitudes (Garner 2005). Besides, recruits' contact with offenders, the information received from various channels and/or the contact with supervisors' beliefs and statements (Garner 2005), and the recruits' previous beliefs might influence recruits' attitudes towards offenders.

Practical Implications
Studies about police officers' attitudes towards offenders are essential, as they allow the development of appropriate interventions to improve attitudes toward offenders and, consequently, impact their decisions and how they respond to criminal events (e.g. Gaines and Kappeler 2005). Improving knowledge about recruits' attitudes can inform future policies and practices and training programs and improve overall crime responses. Besides, improving attitudes and reducing stigmatization might promote offenders' active search for help (Imhoff 2015). Positive attitudes towards the population with whom police officers deal daily are also associated with higher involvement and satisfaction with work, decreasing stress levels (Misis et al. 2013). Thus, the study of attitudes should be taught in the initial police training and continuous training, allowing reflection upon attitudes and expectations to change incorrect beliefs and prejudice that can negatively influence the job. Monitoring and self-awareness of their attitudes and beliefs have been shown to decrease the susceptibility to errant attitudinal and increases behavioural change (Garner 2005). Thus, police training programs could focus on modelling more positive attitudes and behaviours towards offenders. Despite the tactical and technical component of police training being vital for citizens' safety, the current training is essentially focused on law application, which is insufficient (Blumberg et al. 2019). Although the police training program already includes a psychosociology module, it focuses mainly on stress management, conflict resolution, crisis intervention and victim assistance. Thus, it is crucial to include a sociological and psychological module that accounts for an adequate understanding of crime and its causes, offenders, their surrounding circumstances, reasons to commit crimes, offenders' rehabilitation and the purposes and effectiveness of the distinct penal sanctions -namely to deconstruct unfounded punitive attitudes. Obtaining factual and reliable information about offenders and offenders' rehabilitation can change attitudes (Singer and Cooper 2009). Psychoeducational intervention dispelling myths about offenders could positively impact individuals' attitudes towards offenders (Kleban and Jeglic 2012) and towards their rehabilitation. Training should also encourage the more experienced police officers and supervisors to talk to their younger colleagues about these matters and promote an improvement of attitudes towards offenders (Harper 2012). Therefore, there are three critical components to include in police officers' training: the acquisition of skills that promote good performance in the police functions, the acquisition of knowledge and attitudes (mainly recognising their attitudes and their impact on the way they respond to the events) and the acquisition of skills to identify, analyse and solve problems. As Blumberg et al. (2019) discussed, it is critical that the police training promotes the psychological skills (i.e. cognitive, emotional, social, and moral) "essential for officers to meet the contemporary challenges of police work" (p. 1). It is also decisive to improve their initial training regarding in-service training, to allow a higher familiarization with their job and improve attitudes towards offenders.

Limitations
Despite the findings of the present study, some limitations should be mentioned. First, our sample is relatively small (and therefore the statistical power of our results) and homogeneous in sociodemographic and personality variables. The absence of significant results may be an artefact of the lack of statistical power and variability regarding sociodemographic variables, compromising more reliable and consistent conclusions. However, it is essential to note that this lack of variability might be explained by the fact that some requirements are needed to join GNR (e.g. age between 18 and 27 years old; 12th grade), and all the candidates are subject to a psychological interview in order to verify if they accomplish all the criteria required to perform the job. Second, the instrument used to assess the attitudes towards offenders was developed for the US population. Although the instrument was validated in Portugal (with good psychometric properties), differences in culture and crime policies between the USA and Portugal might make the scale less reliable to measure what it originally intended. Moreover, the ATO original version includes 36 items, while the Portuguese version only includes 34 items. The version used in this study was modified and adapted to police officers' samples; hence, some discrepancies from previous studies may be due to measurement variability. Third, in this study, only self-report measures were used. A more reliable and comprehensive evaluation of the professionals' attitudes towards offenders should adopt a broader and more diversified approach (e.g. using in-depth interviews; exploring other variables such as political ideologies, motivations for choosing a police officer career, etc.). Fourth, participants only were assessed at the beginning and the end of the training program, lacking a follow-up assessment. As recruits' attitudes should be assessed not only at the end of their training program but also throughout the professional career, a longitudinal approach to the study of attitudes would be valuable to examine if they maintain their negative views of offenders or adopt a more tolerant view as their experience and contact with offenders increase (e.g. Cunha and Gonçalves 2017;Hirschfield and Piquero 2010;Rade et al. 2016;Viki et al. 2012). Finally, other factors that potentially mediate or moderate police officers' attitudes could be considered, such as the offenders' characteristics or supervisors' issues.

Conclusion
This study aimed to improve the knowledge about attitudes of police recruits towards offenders, specifically, the impact of the police training program, which includes military, educational and in-service training, on such attitudes and the influence of the police officers' features (sociodemographics and personality) over these attitudes. The results revealed that police recruits changed their attitudes towards offenders during their police training program in a negative direction, holding slightly more negative attitudes at the end of the training. Sociodemographic and personality features were not related to the attitudes towards offenders at the end of the training. Given the impact of attitudes on police officers' responses to situations and offenders, and the role that police officers have on offenders' future rehabilitation, the police officers' initial and continuous training should include the necessary psychological and relational dimensions, with a focus on grounding police officers' attitudes on evidence and knowledge, instead of previous common sense backgrounds. Thus, a focus on attitudes should be taught in the initial police training and be part of continuous training, allowing an in-depth reflection on attitudes and, therefore, a change of incorrect beliefs that can negatively influence their job.

Declarations
Ethics Approval The present research was approved by the University of Minho Ethics Commission. All procedures were in accordance with the ethical standards of the institutional ethics commission and with the 1964 Helsinki declarations and its later amendments.

Consent to Participate
All participants were informed about the content and aims of the study and were informed about their anonymity and voluntary participation. All participants signed an online informed consent.

Competing Interests
The authors declare no competing interests.