THE IMPACT OF EMPLOYER BRANDING ON EMPLOYEES: THE ROLE OF EMPLOYMENT OFFERING IN THE PREDICTION OF THEIR AFFECTIVE COMMITMENT

* Dorothée Hanin, Psychological Sciences Research Institute (IPSY), Université catholique de Louvain, Belgium; Florence Stinglhamber, Psychological Sciences Research Institute (IPSY), Université catholique de Louvain, Belgium; Nathalie Delobbe, Louvain School of Management Research Institute (ILSM), Université catholique de Louvain, Belgium. Correspondence concerning this article should be addressed to Dorothée Hanin, Psychological Sciences Research Institute (IPSY), Université catholique de Louvain, Place Cardinal Mercier, 10 bte L3.05.01, B-1348 Louvain-la-Neuve. E-mail: dorothee.hanin@uclouvain.be THE IMPACT OF EMPLOYER BRANDING ON EMPLOYEES: THE ROLE OF EMPLOYMENT OFFERING IN THE PREDICTION OF THEIR AFFECTIVE COMMITMENT


Introduction
Within an employment environment that is becoming increasingly competitive, many companies are struggling to attract, recruit, motivate and retain the best possible human talent. Accordingly, they use employer branding practices as a relatively new approach to differentiate their characteristics as employer from those of their competitors and to advertise the benefits they offer, including training, career opportunities, and challenging jobs (Backhaus & Tikoo, 2004). Demonstrating why the company can be considered as an employer of choice becomes a new role for the HR department, with implications for staff recruitment and retention (Edwards, 2005). Lloyd (2002) (cited in Berthon, Ewing, & Hah, 2005, p. 153) defined employer branding as "the sum of a company's efforts to communicate to existing and prospective staff that it is a desirable place to work". With employer branding, the branded product is therefore the unique and particular employment experience that the company claims to offer to current and potential employees (Edwards, 2010). The "employment offering" that the organisation refers to in its communication thus lies at the very core of employer branding. This employment offering (Edwards, 2010, p. 6) may describe the employment experience by referring to both the tangible and intangible features that the organisation offers to its personnel and the elements of the character of the organisation itself (e.g., the organisational values and guiding principles). Accordingly, Ambler and Barrow (1996, p. 187) suggested that the employment offering consists of the "package of functional, economic and psychological benefits provided by employment, and identified with the employing company".
Since the primary focus of employer branding was to develop a positive image of the organisation as an employer among applicants (Mosley, 2007), some studies have examined the influence of this global employment offering on applicants' attitudes and reactions (e.g., Lievens & Highhouse, 2003;Lievens, Van Hoye, & Schreurs, 2005). Over and beyond this external employer branding, many organisations have more recently begun to evolve towards an approach that also integrates the internal employees (Mosley, 2007). Promoting a favourable image of the organisation among employees has become part of these organisations' employer branding strategy (Backhaus & Tikoo, 2004;Lievens, 2007). This internal perspective aims at motivating and retaining employees who live the brand and act accordingly as ambassadors of their organisations with both clients and prospective employees (Mosley, 2007;Van Hoye, 2008).
Research that was conducted into this latter perspective has been dedicated so far to the influence of employees' lived experience of the benefits really provided by employment within the organisation (i.e. lived employment experience) on their attitudes and behaviours. This stream of research has thus left unexplored the impact of the employment offering as portrayed through organisational communications on employees' work attitudes or behaviours. Filling this gap, the objective of the present study was to examine the role played by employment offering in the determination of employees' affective commitment to the organisation. First, we suggested that the employment offering interacts with employees' lived employment experience in the prediction of employees' affective commitment. Second, we analysed the mechanisms underlying these relationships and argued that two variables related to the social exchange theory play a crucial role, i.e. perceived organisational support and psychological contract violation.

The double perception of current employees
Several authors (e.g., Lievens, Van Hoye, & Anseel, 2007) have examined the influence of the global employment offering of an organisation as perceived by applicants on their attitudes and intentions using the instrumentalsymbolic marketing framework (e.g., Keller, 1993;Lievens & Highhouse, 2003;Park, Jaworski, & MacInnis, 1986). Evidence has shown, for instance, that both instrumental (i.e. job and organisational characteristics such as task diversity and job security) and symbolic (i.e. the extent to which applicants ascribe to the organisation intangible attributes such as being honest and highly regarded) benefits had an influence on applicants' attraction to the organisation (Lievens & Highhouse, 2003;Lievens et al., 2007;Lievens et al., 2005).
Turning to studies on employer branding that used samples of current employees, they showed that the benefits provided by an organisation influence positively various employees' attitudes and intentions such as employees' organisational attractiveness (Lievens, 2007), their satisfaction with the job and the organisation (Davies, 2008;Schlager, Bodderas, Maas, & Cachelin, 2011), their affinity and loyalty with the organisation (Davies, 2008), their intentions to recommend it as an employer (Van Hoye, 2008) and their identification with the organisation . This research studied employer branding by examining employees' perceptions of the package of benefits really provided by employment within the company, and not the package of benefits that the organisation claims to offer through its communication campaigns. For instance, using the same unifying instrumental-symbolic framework as for samples of applicants, Lievens and his colleagues (2007) asked participants to rate the extent to which their organisation provides its personnel with instrumental benefits and the degree to which they perceive the organisation as characterised by symbolic characteristics. By doing so, they showed the influence of employees' lived employment experience on their identification, leaving unexplored how current employees react to the employment offering as portrayed by their organisation.
In agreement with Edwards (2010), we considered that employer branding practices could be perceived differently by prospective and current employees. While applicants are only confronted with the employment offering of an organisation through its communication campaigns, current employees are facing a double perception. On one hand, employees identify the employment offering (EO) communicated by their organisation and, on the other hand, they simultaneously develop a lived experience of the benefits really provided by employment within the organisation (i.e. lived employment experience or LEE). We argued that the literature on employer branding would gain in taking this double perception into consideration when the target of the research is current employees. In contrast with previous research which focused on LEE only (e.g., Lievens et al., 2007), we assumed that EO might also play a role in the determination of employees' attitudes and behaviours in the workplace. By integrating employees' perceptions of EO along with their perceptions of what they experience in-house (i.e. LEE), the present study extended previous research on the influence of employer branding practices on employees' attitudes. More precisely, we focused in this research on one specific attitude, i.e. employees' affective commitment toward the organisation.

Lived employment experience, employment offering, and employees' affective commitment
Commitment can broadly be defined as "a force that binds an individual to a course of action of relevance to one or more targets" (Meyer & Herscovitch, 2001, p. 301). In studies on organisational commitment, much emphasis has been placed on the distinction between affective, continuance, and normative commitment (see e.g., Allen & Meyer, 1990). Particularly, affective commitment (AC) can be defined as "an emotional attachment to, identification with, and involvement in the organisation" (Meyer, Stanley, Herscovitch, & Topolnytsky, 2002, p. 21). AC has been chosen as the outcome variable of this research because one of the purposes of employer branding is to motivate and retain employees by communicating to existing staff that it is a desirable place to work. Yet, affective commitment has been found to be among the best predictors of employee voluntary turnover and is also strongly related to increased in-role and extra-role performance (Meyer et al., 2002). With regard to its determinants, research has shown that AC is positively and significantly associated with favourable job conditions and work experiences (Meyer & Allen, 1997;Meyer et al., 2002). Based on the social exchange theory (Blau, 1964) and, more precisely, the norm of reciprocity (Gouldner, 1960), employees who encounter positive experiences at work or who receive desired outcomes from their organisation would be more likely to reciprocate this favourable treatment by getting committed to their organisation (Meyer & Allen, 1997).
In line with these findings, we assumed that more employees perceive their LEE as composed of beneficial job conditions, more they would be likely to get affectively committed to the employer providing this package of benefits and vice versa. Furthermore, this relationship should be even stronger when EO emphasises these specific favourable job conditions. Being exposed to communications from his/her company about the favourable job conditions that it offers to its personnel should make the employee more aware of how lucky he/she is to benefit from this package if he/she really does. In other words, EO should make salient these favourable job conditions so that employees should be even more willing and likely to reciprocate and develop an emotional attachment to this company. Salience refers to the degree to which a stimulus stands out from its immediate context (Fiske & Taylor, 1984). In contrast, when LEE is not composed of favourable job conditions though the EO claims that it is, we should observe an even lower affective commitment to the company, in comparison with a situation where these unfavourable characteristics would not have become salient through the EO. In sum, EO as perceived by the employee should strengthen the positive relationship between LEE and AC.
As stated above, one mechanism frequently suggested to explain the association between favourable job conditions and AC is related to the social exchange theory (Meyer & Allen, 1997). Furthermore, as claimed by Aselage and Eisenberger (2003, p. 491), "for many years, organisational theorists have alluded to employment as the exchange of employees' effort and loyalty for the organisation's provision of material and socioemotional benefits". Yet one of the purposes of employer branding precisely is to motivate and retain employees by communicating to existing staff that the organisation is a desirable place to work at, providing an interesting and distinctive package of benefits (Backhaus & Tikoo, 2004). Social exchange theory was thus used in the present study as the theoretical framework to further examine the relationships between LEE and EO on the one hand, and AC on the other hand.
Exchanges among employees and their employing organisation have among other things been investigated through organisational support (Eisenberger, Huntington, Hutchison, & Sowa, 1986) and psychological contract (Rousseau, 1989) theories. If perceived organisational support (POS) and psychological contract are conceptually similar in that they are based on the same underlying principles (Aselage & Eisenberger, 2003), they are distinct from each other in that POS is often considered to capture the quality of the employee-organisation social exchange relationship as a whole whereas psychological contract theory focuses on the kept and broken promises (Aselage & Eisenberger, 2003;Dulac, Coyle-Shapiro, Henderson, & Wayne, 2008). The relationships between these two constructs are still unclear in the literature. Whereas some authors have suggested or demonstrated that psychological contract is an antecedent of POS (Coyle-Shapiro & Conway, 2005;Coyle-Shapiro & Kessler, 2000), others have envisaged or established that POS is a determinant of psychological contract (Dulac et al., 2008;Tekleab, Takeuchi, & Taylor, 2005). These findings suggest that the direction of causality in the relationship between psychological contract and POS (e.g., their possible reciprocal link) needs further investigation using cross-lagged panel design. These two concepts are argued to play a mediating role in the relationships investigated in the present research.

The mediating role of perceived organisational support
Organisational support theory (Eisenberger et al., 1986;Eisenberger & Stinglhamber, 2011) states that employees develop a general perception concerning the extent to which the organisation values their contributions and cares about their well-being (i.e. perceived organisational support or POS). Based on the reciprocity norm (Gouldner, 1960) and the socio-emotional need fulfillment (Armeli, Eisenberger, Fasolo, & Lynch, 1998), high POS would strengthen AC (e.g., Eisenberger et al., 1986). In agreement with this view, numerous studies reported POS and AC to be strongly related (cf. Rhoades & Eisenberger, 2002). More precisely, Rhoades, Eisenberger and Armeli (2001, study 2) have demonstrated, using a panel design, that POS led to a temporal change in AC, and not the reverse, indicating that POS is an antecedent of AC.
According to Eisenberger et al. (1986), the development of POS is afforded by the natural tendency of employees to personify their organisation by ascribing humanlike characteristics to it. Employees would then view the treatment received from their organisation as an indication that it favours or disfavours them. Accordingly, a variety of favourable job conditions and work experiences such as opportunities for recognition, high pay, promotions, autonomy, and training have been found to be positively and significantly related with a high POS (cf. Rhoades & Eisenberger, 2002). Furthermore, several authors have found that, if favourable job conditions and work experiences enhance AC, that is because they elicit among employees a feeling of being cared about and supported by the organisation, so that POS mediates the relationship between job conditions and work experiences and AC (Rhoades et al., 2001, study 1;Stinglhamber & Vandenberghe, 2003).
In accordance with these findings, we posited in the present study that one of the mechanisms through which LEE and EO will exert an interactive influence on AC is POS. Favourable job conditions pertaining to the LEE would elicit feelings of being supported and cared by the organisation and an EO emphasising these specific favourable job conditions would reinforce this relationship by making even more salient the positive treatment. In contrast, a LEE composed of unfavourable job conditions would produce low feelings of POS and even lower POS if the organisation claims through its communication campaigns that it offers favourable job conditions. Finally, in line with numerous studies which showed a significant relationship between POS and AC, we suggested that this interactive effect on POS would extent to AC. Accordingly, we hypothesised the following: H1a: The relationship between lived employment experience and perceived organisational support is moderated by employment offering.
H1b: The interaction of lived employment experience by employment offering on affective commitment will be mediated by perceived organisational support.

The mediating role of psychological contract violation
Psychological contract, defined as "the terms of an exchange agreement between individuals and their organisations" (Rousseau, 1995, p. 9), has received a lot of consideration in the last two decades. A component of psychological contract theory is the concept of breach, defined as "the cognition that one's organisation has failed to meet one or more obligations within one's psychological contract in a manner commensurate with one's contributions" (Morrison & Robinson, 1997, p. 230). More relevant for the present study, psychological contract violation (PCV) is defined as an emotional distress and feelings of anger, bitterness, indignation and betrayal deriving from the perceived failure to receive something that is both desired and expected (Morrison & Robinson, 1997;Rousseau, 1989). Studies showed that feelings of PCV make the employees less affectively committed to the organisation (Dulac et al., 2008;Raja, Johns, & Ntalianis, 2004;Zhao, Wayne, Glibkowski, & Bravo, 2007). As a whole, these findings indicated that, based on the norm of reciprocity (Gouldner, 1960) and to maintain equity in-between contributions and rewards (Adams, 1965), employees decrease their contributions to the organisation in response to the violation with plausible consequences for AC (Robinson, Kraatz, & Rousseau, 1994).
With regard to its antecedents, Aselage and Eisenberger (2003) and Tekleab et al. (2005) pointed out that psychological contract theory did not consider the possible effects of favourable treatment that would be provided by the organisation in the absence of any promise or obligation. Nevertheless, several researchers have shown that fairness of treatment and high-quality relationships with and within the organisation contribute to reduce perceptions of PCV. For instance, Dulac et al.'s study (2008) showed that the quality of an employee's dyadic social exchange relationship with his/her leader (i.e. leader-member exchange or LMX; Graen & Uhlbien,1995) is significantly and negatively related to violation (see also Coyle-Shapiro & Conway, 2005;Tekleab et al., 2005). They concluded that it may be that the supportive relationships that comprise high-quality LMX relationships reduce negative feelings and emotions as incarnated in the PCV construct. Generalising this view, it is reasonable to assume that positive job conditions and work experiences generally exert a negative influence on PCV. In addition, several studies on psychological contract suggested that a core determinant of PCV is the discrepancy between what employees experience on the job and what they expected to encounter given organisational promises (Morrison & Robinson, 1997;Suazo, 2009).
Building on these findings, we suggested that PCV would be the second mediating mechanism in the relationship between the interactive term composed of LEE and EO on the one hand and AC on the other hand. When LEE is composed of favourable job conditions, it would decrease feelings of PCV and EO would strengthen this negative relationship by making even more salient the positive experienced conditions. On the contrary, when the job conditions composing the LEE are unfavourable, it would produce high PCV and even higher PCV if the organisation claims via its communication campaigns that it offers these favourable job conditions. Finally, in agreement with the meta-analytic finding showing a significant relationship between PCV and AC (Zhao et al., 2007), we proposed that this interactive effect on PCV will extend to AC. Accordingly, we hypothesised the following: H2a: The relationship between lived employment experience and psychological contract violation is moderated by employment offering. H2b: The interaction of lived employment experience by employment offering on affective commitment will be mediated by psychological contract violation.
For a better understanding of the hypotheses, Figure 1 provides with an overview of our conceptual model.

Figure 1
Theoretical model specifying the hypothesised structural linkages among lived employment experience, employment offering, perceived organisational support, psychological contract violation and affective commitment.

Sample and Procedure
We surveyed 897 department managers of a large multinational retailer of food and everyday products and services established in Belgium. This company has been approached by the research team because it was involved in employer branding practices. Over the last few months, this organisation had indeed extensively communicated about favourable job conditions it offered as an employer such as development opportunities in the job, challenging jobs, the opportunity to take initiatives and the opportunity to take responsibilities. This information was communicated through the organisation's intranet and website. In agreement with this context, we measured LEE and EO in the present study by selecting items that corresponded to the benefits claimed to be offered by the organisation. These items were finally validated by the organisation's HR department and by external consultants collaborating with the organisation on employer branding issues. Participants were sent a message through the company's intranet system explaining the purpose of the study, informing them that their answers will be anonymous, and providing them with a website address to take the questionnaire online. A total of 212 employees (response rate = 24%) completed the questionnaire. We excluded 26 respondents from the data analyses because they did not provide responses on all the variables of interest, leaving a sample of 186 employees. Of this final sample, 34.9% were females, 55.4% were males and 9.7% did not answer to the gender question. Average age was 39.6 years (SD = 8.4) and average organisational tenure was 14.5 years (SD = 10.5). Note that none of these demographic variables display a significant relationship with the dependent variables of interest in the present investigation. In order to reduce model complexity, we thus did not control for any demographic in the data analyses that we ran to test our hypotheses.

Measures
To assess the extent to which employee's LEE was composed of favourable job conditions, we relied on the 7-item scale developed by Stinglhamber and Vandenberghe (2004;study 2, table 4). Respondents were asked to rate the extent to which each of the seven favourable job conditions was typical of their job. This was assessed through a 5-point Likert-type scale ranging from not at all characteristic (1) to totally characteristic (5). All the items used in this study are displayed in Table 1. To assess employees' perceptions of the EO disseminated by the organisation through its communication campaigns, we used the same 7-item scale developed by Stinglhamber and Vandenberghe (2004;study 2, table 4). Precisely, respondents were asked to rate to what extent their organisation explicitly portrays in its communications that it offers these favourable job conditions to its personnel. This was assessed through a 5-point Likert-type scale ranging from not at all (1) to fully (5).
Given the unidimensionality and the high internal reliability of the survey generally used to assess POS (cf. Eisenberger et al., 1986), we selected 4 items from this scale. These items were selected (a) on the basis of the loadings resulting from the exploratory factor analysis conducted by Eisenberger et al. (1986), and (b) to assure non-redundancy and adequate representation of the construct's content (Rhoades & Eisenberger, 2002). For this and the remaining measures, respondents rated their agreement with each statement using a 5-point Likert-type scale (1 = strongly disagree; 5 = strongly agree). The 4-item scale developed by Robinson and Morrison (2000) was used to measure employees' perception of PCV. According to several authors (Dulac et al., 2008;Morrison & Robinson, 1997;Robinson & Morrison, 2000), PCV can be considered as a concept theoretically and operationally distinct from other concepts in the literature on psychological contract (e.g., psychological contract breach). Finally, to measure employees' AC to the organisation, we relied on a revised version of Meyer, Allen, and Smith's (1993) scale which has been previously validated (Stinglhamber, Bentein, & Vandenberghe, 2002;Vandenberghe, Stinglhamber, Bentein, & Delhaise, 2001). For use in this study, we retained the three items displaying the highest loadings on the intended factor, in the confirmatory factor analysis results reported by Stinglhamber et al. (2002).

Results
We followed the two-stage process recommended by Anderson and Gerbing (1988) to analyse our data using Lisrel 8.8 (Jöreskog & Sörbom, 1993).

Discriminant validity of the constructs
We conducted confirmatory factor analyses to assess the distinctiveness of the five constructs included in our study. Using chi-square difference tests (Bentler & Bonnett, 1980;James, Mulaik, & Brett, 1982), we compared the fit of nine measurement models, ranging from the hypothesised five-factor model to a single-factor model.  individual items loaded reliably on their hypothesised factors (cf. standardised loadings in Table 1). The results of these confirmatory factor analyses led us to treat the five constructs separately in the subsequent statistical analyses that were conducted to examine our hypotheses.

Relationships among variables
Means, standard deviations, internal reliabilities and correlations among variables are displayed in Table 3. We tested a structural equation model in which LEE, EO, and their interaction led to POS (Hypothesis 1a) and PCV (Hypothesis 2a) which, in turn, led to AC (Hypotheses 1b and 2b). Note that disturbance terms of POS and PCV latent variables were allowed to covary in order to take into consideration the potential link between these two constructs (see above). The approach of Marsh, Wen, and Hau (2004) was used to test the moderating influence of EO on the relationship between LEE and POS and on the relationship between LEE and PCV. Since we are dealing with latent constructs, indicators are needed for the interaction term (i.e. LEE × EO) as well as the main effects (i.e. LEE and EO). Following Marsh et al.'s method (2004) to lessen multicollinearity, we first centered the indicators of each latent variable included in the interaction (i.e. the indicators of LEE and EO). We then Note. N =186. The results are described in the text. LEE = lived employment experience; EO = employment offering; POS = perceived organisational support; PCV = psychological contract violation; AC = affective commitment; df = degrees of freedom; ∆χ² = difference in chi-square from the five-factor model; RMSEA = root-mean-square error of approximation; CFI = comparative fit index; NNFI = non-normed fit index. *** p < .001.
paired these indicators to form indicators of the latent interaction term. Because the same items were used to measure both LEE and EO, each indicator of LEE was multiplied by its content-equivalent indicator for EO. Then, we used Lisrel to estimate the effects of LEE, EO, and their latent product on POS and PCV, and their subsequent effect on AC. In this model, we allowed for autocorrelated error variances by freeing the error covariances of contentequivalent items used to measure LEE and EO. Table 4 presents fit indices for the hypothesised structural model (Model 1), along with those for three alternative models (Model 2 to Model 4). Model 1 (depicted in Figure 1) showed an acceptable fit to the data (χ² (444) = 953.65, p < .001; CFI = .96; NNFI = .96; RMSEA = .08). However, as indicated by the chi-square difference test, adding a direct path from LEE to AC (Model 2) resulted in a significant improvement in model fit (∆χ² [1] = 98.99, p < .001) with a RMSEA of .07, a CFI of .97, and a NNFI of .97. This suggests that LEE exerts a significant influence on AC, both indirectly (see indirect effect below) and directly.
To assess whether Model 2 was the best depiction of the data, we compared its fit to that of two alternative models (Model 3 and Model 4) containing additional paths that were theoretically plausible. As Model 2 was nested within each of these alternative models, it could be compared to them using the chi-square difference test procedure (Bentler & Bonnett, 1980;James et al., 1982). As can be seen from Table 4 (p. 70), none of these alternative models improved significantly over Model 2. We thus retained Model 2 as the best fitting model. Standardised parameter estimates for Model 2 are shown in Figure 2. For ease of presentation, we show the structural model rather than the full measurement model. As shown in this figure, LEE had a significant positive relationship with POS and AC (respectively, γ = .39, p < .01, and γ = .61, p < .001), and POS and PCV are significant predictors of AC (respectively, β = .27, p < .001, and β = -.11, p < .05). Note that LEE has no significant main effect on PCV, and EO has no significant main effect on both POS and PCV. As predicted (cf. Hypotheses 1a and 2a), the interactive influence of LEE and EO on both POS and PCV was also significant (respectively, γ = .33, p < .001, and γ = -.34, p < .001).
In order to further examine the interactive effect of LEE and EO on POS and PCV, lines representing the relationships between LEE and (1) POS and (2) PCV were plotted, at high and low levels of EO (plus and minus 1 SD) Note. N =186. LEE = lived employment experience; EO = employment offering; AC = affective commitment; df = degrees of freedom; ∆χ² = chi-square difference tests between the best fitting model (Model 2) and alternative models; RMSEA = root-mean-square error of approximation; CFI = comparative fit index; NNFI = non-normed fit index. ***p < .001.

Figure 2 Completely standardised path coefficients for Model 2 (final model).
For the sake of clarity, only structural relationships are shown.
( Figure 3). Results of simple slopes tests indicated that the relationships between LEE and (1) POS and (2) PCV were statistically significant when EO was high (t(182) = 4.09, p < .001, and t(182) = -2.03, p < .05, respectively) and not significant when EO was low (t(182) = 1.20, p >.05, and t(182) = 0.52, p >.05, respectively). The slopes were also significantly different from each other for POS (t(182) = 3.73, p < .001) and PCV (t(182) = -3.07, p < .01). As shown in Figure 3, the combination of a high LEE and a high EO induces the highest level of POS and the lowest level of PCV. In contrast with our predictions, however, the combination of a low LEE and a high EO does not produce deleterious effects, i.e. a lower level of POS and a higher level of PCV. As a whole, these findings partly support our Hypotheses 1a and 2a.
Note. High and low EO are, respectively, 1 SD above and 1 SD below the mean. LEE = lived employment experience; EO = employment offering; POS = perceived organisational support; PCV = psychological contract violation.

Figure 3
The relationship between lived employment experience and both perceived organisational support and psychological contract violation as a function of the employment offering.
To examine whether the interactive effect of LEE and EO on POS and PCV would extend to AC (cf. Hypotheses 1b and 2b), we relied on the results of Sobel tests. They indicate that the indirect effects of the interactive term LEE × EO on AC through (1) POS and (2) PCV were significant, respectively (Indirect effect = .09, z' = 2.58, p < .05; critical z-prime value for statistical significance = .97; see MacKinnon, Lockwood, Hoffman, West, & Sheets, 2002) and (Indirect effect = .04, z' = 1.64, p < .05; critical value = .97), thus supporting Hypotheses 1b and 2b. As a whole, these results provide evidence that the interactive influence of LEE and EO on AC is totally mediated by both POS and PCV. Finally, the indirect effect of LEE on AC through POS was also significant (Indirect effect = .12, z' = 2.24, p < .05; critical value = .97). These results suggest that the LEE-AC relationship is partially mediated by POS.

Discussion
By investigating the impact of employer branding on current employees instead of applicants, the present study is part of a limited (albeit growing) research since, to the best of our knowledge, few studies examined this issue (e.g., Lievens et al., 2007;Van Hoye, 2008). Building on this prior work, we proposed in this research that employees represent a specific target of employer branding practices. In agreement with Edwards (2010), we suggested that given their experience and knowledge of the organisation and its functioning, current employees (in contrast with applicants) are confronted with a double perception: the package of benefits that their organisation claims to offer through diverse communication campaigns (i.e. EO) and the package of benefits that their organisation is perceived to really provide inhouse (i.e. LEE).
To our knowledge, the present study is the first research on employer branding which examines the effects of LEE and EO as well as their interaction in order to predict employees' attitudes. First, our findings suggested that the positive relationship between a LEE which is composed of favourable job conditions and POS increases with EO, supporting Hypothesis 1a. However, the pattern of the interaction (cf. Figure 3) did not entirely fit with our expectations. As it can be seen in Figure 3, in accordance with our expectations, the relationship between a LEE composed of favourable job conditions and POS is reinforced when the organisation claims that it offers these favourable job conditions (i.e. a high EO). A LEE composed of unfavourable job conditions produce low feelings of POS but, in contrast with our predictions, they do not engender even lower POS if the organisation claims that it offers favourable job conditions (i.e. a high EO). Furthermore, although we would have expected that the relationship between LEE and POS is positive even though EO is low, results showed that the slope representing a low EO is not significant. Given our reduced sample size and therefore a potential lack of statistical power, these intriguing results certainly warrant replication. They suggest that it would be necessary that the organisation makes salient the positive job conditions that it offers to its personnel for eliciting a POS among its employees.
Findings on the moderating role of EO on the positive link between a LEE composed of favourable job conditions and POS might be interpreted in connection with a well-known mechanism in the development of POS, i.e. employee attribution of the discretionary nature of favourable treatment. A central tenet of organisation support theory (Eisenberger et al., 1986) is that favourable treatment contributes to POS to the extent that it is considered discretionary rather than being impelled by circumstances. Discretionary actions are a much stronger indication of the organisation's favourable or unfavourable orientation toward employees than actions that appear influenced by external constraints such as union contracts or government regulations. Accordingly, research has shown that favourable treatment that is considered highly discretionary on the part of the organisation contributed more to POS than treatment over which the organisation had little control (Eisenberger, Cummings, Armeli, & Lynch, 1997;Stinglhamber & Vandenberghe, 2004). In the present study, conducted in a strongly unionised industry [1] , it may be that a LEE composed of favourable job conditions are attributed per default to the action of unions and not to the goodwill of organisation. However, by communicating on the package of benefits that it pretends to offer, an organisation implicitly acknowledges that offering these favourable job conditions to its personnel depends on its willingness to do so or not. This attribution explanation, not directly tested in the present study, would be worth investigating in further research.
With regard to PCV, findings indicated that neither LEE nor EO have a significantly negative effect on PCV. In contrast with Dulac et al.'s suggestion (2008), our results did not show that positive job conditions and experiences at work reduce negative feelings and emotions comprised in the PCV construct. Only the interactive term between LEE and EO did negatively influence PCV. Even though this finding supported our Hypothesis 2a, we must acknowledge that, again, the pattern of this negative interaction (cf. Figure 3) did not entirely fit with our expectations. As it can be seen in Figure  3, results do not show a deleterious effect of a LEE composed of unfavourable 1. Trade union density rate was 51.9% in Belgium, when the data have been collected (source: OECD ;Visser, Martin, & Tergeist, 2011). And this rate was 76.13% in the industry in which the study was conducted, i.e. the retail of food and everyday products and services (source: Annual Report of Social Fund, 2008). job conditions which would be combined with a high EO. In contrast with our expectations, PCV is not higher when lived employment experience is composed of unfavourable job conditions and the organisation claims via its communication campaigns that it offers favourable job conditions, than when the organisation does not pretend that it offers favourable job conditions to its personnel. Again, this surprising effect certainly warrants replication and extension. Furthermore, the graph representing the significant interaction indicates that the relationship between LEE and PCV is not significant when EO is low (cf. Figure 3). This finding is in agreement with attributions as parts of the sense-making process leading to psychological contract violation. Following Morrison and Robinson (1997), assigning the responsibility of unfavourable or unexpected outcomes to the organisation or its representatives is indeed a prerequisite to the development of perceptions of psychological contract violation. This assignment of responsibility would be based on an examination of the following factors: causality, control, foreseeability, and intentionality (Morrison & Robinson, 1997). Again, this attribution explanation was not directly tested in the present research and further investigations are needed to test it. However, the strongly unionised context of the present study may imply that it is only by communicating on the package of benefits it offers that the company emphasises its control over the job conditions provided to its personnel. Finally, Figure 3 also indicates that, when EO is high, the more the LEE is perceived as composed of favourable job conditions, the less employees feel that their psychological contract has been violated. Or, in other words, the psychological contract is less likely to be perceived as violated only when favourable job conditions are confirmed through organisational communications (i.e. EO). Making salient the positive experienced conditions seems to be a boundary condition of the effect of LEE on PCV. These findings might be explained by the fact that EO facilitates the identification of the terms of the psychological contract.
Finally, the present study showed that the moderating influence of EO on the relationship between LEE and both POS and PCV extended to AC. As hypothesised, POS and PCV thus represent two relevant mechanisms to understand the impact of employer branding on affective commitment. These findings extend prior results revealing that (a) positive work experiences and favourable job characteristics enhance both AC (Mathieu & Zajac, 1990;Meyer & Allen, 1997;Meyer, Bobocel, & Allen, 1991) and POS among employees (Rhoades & Eisenberger, 2002), (b) POS is an important determinant of AC (Rhoades & Eisenberger, 2002;Rhoades et al., 2001;Stinglhamber & Vandenberghe, 2003), and (c) PCV is negatively related to employees' AC (Raja et al., 2004).

Limits of the Research
There are limits that must be acknowledged when interpreting the results of the present study. First, we used a cross-sectional design which prevents us from making strong conclusions about the direction of causality in our model. Even if prior research has established that POS is an antecedent of AC (Rhoades et al., 2001), longitudinal and panel research is needed to strongly infer the direction of causality among most of the variables included in this study. Second, this study might be criticised for relying on self-reported variables (i.e. problem of common method variance). However, we believe that self-reported measures are most appropriate for assessing the constructs included in this study as we were here more interested in perceptions than in reality. Additionally, we were able to partially address the concern over method bias by assuring participants of the anonymity of their responses and by performing analyses showing that a method factor accounted for only a small proportion of the variability in the data. Indeed, confirmatory factor analyses showed that a single-factor solution provided an extremely poor fit to the data (i.e. Harman's single-factor test; Podsakoff, MacKenzie, Lee, & Podsakoff, 2003). Furthermore, some of our predictions involved an interaction, which would not likely rise from common method variance. As a whole, we can state that the relationships found were robust to common method effects and, therefore, common method variance is not a pervasive problem in this study (Podsakoff et al., 2003;Spector, 2006).
Third, obtaining data from a single organisation may have raised more problems than it has solved. Surveying a single organisation offered the clear advantage of focusing on this specific organisation's EO. In the surveyed organisation, all employees were exposed to the same communication message. Items capturing this specific and unique EO were thus selected for the objectives of the present research. Despite the consistency of the exposure to the organisational communications, the EO was nevertheless perceived differently by employees based on their interest for, attention to and/or awareness of the organisational communications. Though existing, the variability in EO perceptions is however limited (SD = .75). Conducting the present study in a single company may certainly have lessened the variance of variables such as EO but also PCV and LEE. Furthermore, the characteristics of the sample itself may have caused restrictions of range on several variables. Only the department managers of the organisation were surveyed. This may have provided us with a limited picture of the plurality of views and perceptions that exists within the whole organisation. As a result, these restrictions of range may have decreased the likelihood to find significant relationships.
Thus, the present results certainly warrant replication using samples pertaining to a variety of occupations, industries, and organisations. In particular, the use of multi-organisation samples in future research would have the advantage of considering EO as an organisation-level and more objective construct, and not solely as an individual perception. Conceptualising EO as an organisational characteristic would allow examining a cross-level model where the interaction between EO at the organisational level and LEE at the individual level on the mediating and outcome variables is tested. In addition, comparing the effects of EO at both levels -i.e. individual and organisational -should provide very interesting and challenging findings.
Furthermore, this replication is all the more necessary since the reduced sample size might have cause a lack of statistical power. Also, given the limited response rate, it is possible that responses of people not choosing to participate could have differed from those who answered the questionnaire. However, as the measures used in the present study were part of a larger survey, the low response rate is most probably due to the length of the questionnaire rather than a systematic bias. The replication of our results in further studies is even more needed given the floor effect that may have also plague the data related to psychological contract violation (M = 1.5). The low mean score for PCV can be due to the high organisational tenure that employees composing our sample have on average (M = 14.5 years). As the presence of these effects can undermine the interpretation of statistical analyses by increasing the likelihood of Type II error, we may have wrongly concluded that LEE has no influence on PCV.
Finally, several limitations regarding the measures used to assess some of our constructs (e.g., EO and LEE) should also be noted. The introductory sentence used to measure the occurrence of seven job conditions in communication campaigns (i.e. EO measure) is undoubtedly hard to comprehend. Despite its good psychometric properties, it would be worthwhile to refine this measure in the future. Additionally, in spite of the results of the confirmatory factor analyses, we cannot ignore the strong correlation that exists between EO and LEE (r = .76, p < .001). Even though this correlation is probably artificially inflated due to the fact that these two measures are composed of the same job conditions, it cannot be entirely explained by the similarities between the two measures. Future research might examine how these variables are related to each other over time. We cannot rule out the possibility of a reciprocal relationship. Finally, the items used to measure EO and LEE in the present study mainly referred to intrinsic job conditions. This type of job conditions derives from the very nature of work experiences (e.g., job responsibilities, use of one's abilities) and does not include job conditions capturing external reinforcers such as pay or relationships with coworkers. As explained above, we focused on this specific kind of job conditions because they correspond to the benefits that the surveyed organisation communicated on.

Perspectives for Future Research
In addition to our former suggestions, we propose several perspectives for future research. First, as has just been pointed out, we operationalised LEE and EO by focusing on a specific kind of (un)favourable job conditions (which equate a specific kind of instrumental benefits) that fitted with the organisational context of our study. Yet, as explained above, the package of benefits that an organisation can offer to its personnel and communicate on in its campaigns may also include symbolic organisational attributes (e.g., being successful) or organisational values (e.g., interest for human relations) (Backhaus & Tikoo, 2004;Edwards, 2010;Lievens & Highhouse, 2003). Future research should examine whether the interactive effect of LEE and EO in the prediction of employees' attitudes and behaviours is also relevant when organisation's symbolic characteristics and values are captured through the LEE and EO measures.
Second, future studies on the topic should consider the influence of the importance that employees give to the job conditions included in the employer branding campaigns. In line with Conway and Briner's study (2002) which showed that the importance of a promise is an important predictor of emotional reactions following the non fulfillment of this promise, we think that the relationship between the interaction (i.e. LEE × EO) and POS or PCV might even be strengthened if employees value the job conditions that are emphasised in the employment offering. This suggests a three-way interaction. This suggestion is in agreement with that of Moroko and Uncles (2008) who argued that successful employer branding practices are "characterised as having a value proposition that is relevant to, and resonant with, their prospective and current employees" (p. 162). In a similar vein, Gouldner (1960) suggested that the strength of an obligation to repay depends on the value of the benefit received, so that highly-valued benefits generate stronger obligation to reciprocate (Coyle-Shapiro & Conway, 2004).
Third, future research should investigate other mechanisms that may play a crucial role in the relationships between employment offering and employees' attitudes and behaviours. In particular, as the goal of communication campaigns is to stress the positive and distinct aspects of a specific organisation as an employer, we may assume that they will influence employees' organisational identification (Edwards, 2010). Organisational identification is a perception of oneness with or belongingness to the organisation (Ashforth & Mael, 1989). It is determined by the distinctiveness of the group's values, the group's prestige, and the salience of the out-group (Ashforth & Mael, 1989), which can be emphasised in the communication campaigns of the organisation.

Practical Implications
As many companies use employer branding to create an image of the organisation as a good place to work, it is important for experts in human resources, marketing and communication to understand the influence of these practices on employees' attitudes and behaviours. First of all, this study demonstrates the interest for an organisation to promote itself as a good place to work. The results suggest that the EO communicated by the organisation represents a valuable opportunity to reinforce employee-employer relationships by increasing positive employees' attitudes and behaviours toward their organisation. More precisely, the results of this study encourage practitioners to keep in mind that alone communication campaigns do not have an influence, either positive or negative, on employees' attitudes. To be effective and to have a return on investment, the content of the communication campaigns should explicitly and accurately describe the employment experience provided to employees. In that case, employees perceive to be supported by their organisation and feel less that their organisation violates the psychological contract. As a consequence, their AC is strengthened. By enhancing this emotional link among their employees, organisations are more likely to reduce their employees' intentions to quit the organisation and, in fine, their turnover rate (Meyer et al., 2002). Bentein, Vandenberghe, Vandenberg and Stinglhamber (2005) concluded from their study that "sustained reductions in turnover must be accomplished through sustained, not only one-time, efforts to promote high levels of affective commitment (...) over time" (p. 479). This suggests that it would be worthwhile to develop sustained efforts via an employer branding strategy in the long term.
Above and beyond personnel retention, practitioners have to be aware that, by playing a role in the development of AC, EO indirectly contributes to other employees' favourable work behaviours. For instance, since AC is strongly related to increased in-role and extra-role performance (Meyer et al., 2002), we could assume from the present study that EO may also impact employees' performance and, more globally, organisational effectiveness. In the same vein, employees whose favourable attitudes toward the organisation are strengthened by the EO of their company might be more willing to act as ambassadors of their organisation with customers or prospective employees.
Another implication is that organisations involved in employer branding practices should probably evaluate them on a regular basis. Given that EO has a beneficial influence on employees' attitudes when it makes salient favourable job conditions that employees perceive to have, they should among other things assess their employees' perceptions. By doing so, organisations will be able to adapt the content of communication campaigns so that their EO fits with the day-to-day favourable job conditions as perceived by employees.
More generally and still related to the content of the EO, our results suggest that job conditions are relevant characteristics to communicate in employer branding campaigns in order to foster employees' favourable attitudes.
Last but not least, the findings of this research also suggest that employer branding practices dedicated to current employees should not emphasise job characteristics or benefits that employees perceive to not have. Even though our results surprisingly showed that such misleading communications would not have the expected deleterious effect on employees' attitudes, it is worthwhile to highlight that they do not have either beneficial effects on employees' attitudes.