Signaling theory (Spence, 1973; Connelly et al., 2011; Bangerter et al., 2012) is widely used to explain how applicants' attraction to an organization can be influenced by information or signals during the recruitment and selection process. This theory suggests that the signals to which applicants are exposed have the potential to influence how applicants feel about themselves and the organization (Celani and Singh, 2011). When applicants receive information about the job and the work environment, it helps them decide whether or not they are excited about the job and whether they want to work for the organization (Farago et al., 2013). An important source from which applicants receive signals are employment interviews, which are characterized by a dynamic exchange between applicants and interviewers (Celani and Singh, 2011; Levashina et al., 2014).

Many theoretical frameworks suggest that the signals interviewers send to applicants influence applicants' attitudes, intentions, emotions, and behavioral reactions, but relatively little theoretical work has focused on applicants' perceptions of these signals. For example, as Celani and Singh (2011) point out, “an improved understanding of how applicants are influenced by recruitment signals will help to address outstanding research needs” (p. 231). Particularly, an issue that remains unresolved regarding signaling theory is how the signals that interviewers use to induce favorable reactions in applicants ultimately lead to these reactions.

The signaling timeline model by Connelly et al. (2011) specifies the process of how signals ultimately influence individual reactions, but has not been tested sufficiently in the field. According to the model, the most basic form of a signaling system consists of a sender (e.g., the interviewer), a signal (e.g., IM behavior), and a receiver (e.g., the applicant). In addition, the model suggests that there is a chronologic process through which signals have an effect: Signals are sent to the receiver, the receiver observes the signals, and eventually, the receiver reacts to the signals, for example, with an attitude, a behavioral intention, or an emotion. In the context of the interview, this means that the signals sent by interviewers are an effective means for recruitment when, first, interviewers send signals to achieve their intended outcomes, second, applicants receive these signals, and third, applicants react in the desired way. Based on the signaling timeline model by Connelly et al. (2011), we suggest that a better understanding of the effects of interviewer IM can be achieved by differentiating the signals sent (interviewer IM), the signals received (perceived interviewer IM), and the reaction to the signals (recruiting outcomes).

While earlier recruitment research suggested that interviewers play only a minor role in influencing recruiting outcomes (e.g., Powell, 1984; Taylor and Bergmann, 1987; Rynes and Barber, 1990), recent work shows that interviewers actually play an important role in attracting and retaining applicants (e.g., Chapman et al., 2005; Carless and Imber, 2007; Wilhelmy et al., 2016). In addition, interviewers are well-aware of their role as representatives of the organization and of the emotional reactions they can elicit from applicants. In a qualitative study, for example, Wilhelmy et al. (2016) found that interviewers can have a multitude of goals, but that interviewers' primary concern was about representing the organization. Most interviewers reported that they tried to signal an attractive image of the organization to applicants to influence applicants' acceptance intention and organizational reputation, for example. Furthermore, in that study, interviewers' secondary concern centered on their personal interaction with applicants. Most interviewers reported that they tried to signal closeness to applicants, for example, to make applicants feel comfortable and elicit positive emotions in applicants.

To achieve their goals, interviewers can send signals to applicants in the form of verbal IM, that is, interviewers use the content of what they are saying to influence applicants' impressions. Furthermore, the IM literature makes a fundamental distinction between self-focused IM and other-focused IM (Tsai and Huang, 2014). From the perspective of the interviewer, a central self-focused IM behavior is organization-enhancement, sometimes used interchangeably with the term enhancement of the organization or self-enhancement. Organization-enhancement refers to stressing the positive qualities that one's organization possesses (e.g., pointing out strengths of the organization, Stevens et al., 1990; Wilhelmy et al., 2016). A central other-focused IM behavior of interviewers is applicant-enhancement, sometimes used interchangeably with the term ingratiation or other-enhancement, which refers to flattering the applicant (e.g., acknowledging past accomplishments of the applicant, Stevens et al., 1990; Wilhelmy et al., 2016).

In the present study, we chose to focus on these two interviewer IM behaviors because research indicates that they are relevant for applicants' positive attitudes, intentions, and emotions. Specifically, we assume that organization-enhancement is related to the goal of representing the organization such as creating an attractive image, whereas applicant-enhancement is related to the goal of putting applicants at ease such as creating a close relation image. For example, qualitative interviews with applicants revealed that applicants were impressed by interviewers who knew how to pitch the job and the organization and felt enthusiastic when interviewers encouraged them during the interview (Rynes et al., 1991). Similarly, Wilhelmy et al. (2016) found that interviewers from the health services field such as hospitals put emphasis on selling their organization by pointing out the advantages of the job and their respective hospital. In addition, in that study, it was found that interviewers reported trying to ensure that applicants would leave the interview room feeling good about themselves by paying attention to what applicants were saying and encouraging them during the interview. Furthermore, a laboratory experiment conducted by Stevens et al. (1990) found initial evidence that both organization-enhancement and applicant-enhancement elicited overall positive reactions in applicants.

Although research has shown that interviewers use IM to influence applicant impressions, an important question that remains is whether applicants discern interviewers' IM. On the one hand, the selection interview is a highly stressful event that can trigger strong emotions such as anxiety (McCarthy and Goffin, 2004). In contrast to interviewers, who are known to be responsive to applicants' IM, applicants are often nervous and anxious, which might not only lower their performance but also their attention span and the cognitive capacity to receive all the signals sent by interviewers (Tsai and Huang, 2014). On the other hand, signaling theory (Spence, 1973; Connelly et al., 2011; Bangerter et al., 2012) proposes that interviewers are an important source of information for applicants because applicants lack information about the job and the organization. As a consequence, applicants should try to pay attention to interviewers' behavior and to every piece of information that interviewers provide.

In line with the latter suggestion, previous studies indicated that applicants often perceive interviewer IM behavior. Wilhelmy et al. (2016) found that applicants were able to report examples of verbal interviewer IM behaviors that they had perceived in prior selection interviews. Thus, applicants seem to notice these behaviors. In addition, in Stevens et al.'s (1990) laboratory study, recruiters who praised the applicant were perceived as putting more persuasive effort into the interview compared to recruiters who used other IM behaviors. Furthermore, in the same study, participants reported a high degree of flattery when they were exposed to a recruiter who used applicant-enhancement. Therefore, given applicants' ability to report and comment on interviewer IM behavior (Stevens et al., 1990; Wilhelmy et al., 2016), and rooted in Connelly et al.'s (2011) signaling timeline model, we expected that organization-enhancement and applicant-enhancement used by interviewers during the interview would be perceived by applicants (see Figure 1).

The signaling timeline model by Connelly et al. (2011) suggests that when a signal is received, the receiver interprets the signal and reacts to it. In the context of recruitment, these reactions come in the form of recruiting outcomes, that is, applicants construe interviewer behavior as signals of how interesting a job and an organization is for them and also emotionally react to these signals (Rynes et al., 1991). Prior research indicates that important recruiting outcomes encompass applicant attitudes such as organizational prestige, which refers to the degree to which an organization is perceived as being well-regarded and reputable (Highhouse et al., 2003), applicant intentions such as offer acceptance intentions, which captures applicants' willingness to accept an offer for a job or a place at a university (Chapman et al., 2005), and applicant emotional reactions such as positive affect, which reflects applicants' positive emotional state (Watson et al., 1988), and interview self-efficacy, which captures the extent of their belief in their ability to succeed in an interview (Bauer et al., 1998).

In addition, previous work indicates that organization-enhancement plays a central role for applicants' attitudes and intentions, whereas applicant-enhancement is more important for applicants' emotions. Qualitative findings suggest that organization-enhancement is primarily used with the intention to create an attractive image to “sell” the organization to the applicant. By engaging in organization-enhancement, the interviewer attempts to quickly recruit applicants, enhance the organizations' reputation, and increase applicants' intention to accept an offer (Wilhelmy et al., 2016). Furthermore, in Stevens et al.'s (1990) laboratory study, participants watched three videos of interviewers, each of whom applied a different IM behavior to describe a hypothetical study program: self-promotion, which is in line with the conceptualization of organization-enhancement in the present study, other-enhancement, which is in line with the conceptualization of applicant-enhancement in the present study, and opinion-conformity, which combined aspects of organization-enhancement such as advantages of the study program and aspects of applicant-enhancement such as emphasizing the applicants' qualifications. The authors found that organization-enhancement was perceived as persuasive by some participants. However, the authors did not find full support for their assumption that compared to other IM behaviors, organization-enhancement would be most effective in making participants choose the hypothetical study program, particularly when organization-enhancement was used early in the interview. The authors interpreted this finding by suggesting that applicants may not always take organization-enhancement seriously or may not always pay attention to this kind of information. However, an important limitation of Stevens et al.'s (1990) study is that it did not consider whether applicants actually perceived the IM behavior that was presented in the videos. Therefore, given that organization-enhancement has been linked to interviewers' intention to create an attractive image (Stevens et al., 1990; Wilhelmy et al., 2016), we expect that perceived organization-enhancement would increase organizational prestige and applicants' intention to accept a potential offer from the organization.

Given the positive association that can be expected between the degree of IM sent by interviewers and the degree to which applicants perceive interviewer IM, and based on Connelly et al.'s (2011) model, we also aimed to test the whole signaling timeline mechanism for organization-enhancement. Extending the arguments above, we assume that interviewers' organization-enhancement during the interview has an indirect effect on organizational prestige and applicants' intention to accept a potential offer through perceived organization-enhancement (see Figure 1).

Although applicants' attitudes and intentions are central to an organizations' success in recruiting new employees, researchers have recognized that applicants become emotionally invested. Thus, applicant emotions play a central role in the recruitment process (Truxillo et al., in press). This is particularly true for job interviews, which, unlike selection tests, involve a social exchange and can therefore trigger emotional reactions (McCarthy and Goffin, 2004).

So far, applicant reactions research has mainly focused on negative affective reactions such as interview anxiety (e.g., McCarthy and Goffin, 2004) or perceived strain (e.g., Merkulova et al., 2014). Several researchers, however, have called for research that considers applicants' affective reactions from another point of view, that of positive emotional reactions such as positive affect and self-efficacy (McCarthy et al., 2017; Truxillo et al., in press). In a recent framework on applicant perspectives in selection, McCarthy et al. (2017) emphasize the central role of applicant affect. They propose that treating applicants favorably during selection procedures enhances applicants' cognitive processing and triggers positive emotional reactions like positive affect. Specifically, applicants may feel more comfortable and self-confident when recruiters and interviewers show interest in their past achievements (Truxillo and Bauer, 2011). In this sense, perceived applicant-enhancement, which includes showing interest in applicants' past achievements, flatters, and reassures applicants, which makes them feel inspired and excited, and strengthens their beliefs in their own interview ability.

Despite the theoretical indications for a positive relationship between interviewers' applicant-enhancement and applicants' interview self-efficacy and positive affect, empirical evidence remains scarce. Regarding general self-efficacy, meta-analytic findings show that recruiters' attentiveness and friendly interpersonal treatment is related to applicants' self-perceptions such as self-efficacy after the selection process (Hausknecht et al., 2004). In addition, Stevens et al.'s (1990) study provided evidence that applicant-enhancement can be perceived as “buttering the ego” (p. 1087) by some applicants, but can also make applicants feel desired and reassured.

Regarding positive affect, Wilhelmy et al. (2016) found indications suggesting that applicant-focused IM is used with the aim of creating a close relation image and making applicants feel good in terms of creating a positive affective state. For example, interviewers compliment applicants with the intention to cheer them up. In addition, Stevens et al. (1990) proposed that in contrast to organization-enhancement, applicant-enhancement should provide weak information about prestige and suitability of the organization, but should provide a stronger basis for positive emotional reactions. In line with this assumption, they found that compared to other IM behaviors, applicant-enhancement had a positive influence on applicants' ratings of recruiter likeability, but no other affective reactions were assessed in the study. In addition, the study was limited to comparative effects of IM behaviors, that is, there was no variation in the degree of applicant-enhancement so that no main effect of applicant-enhancement on recruiting outcomes was assessed.

Given preliminary evidence regarding the positive link between perceived applicant-enhancement and feelings of reassurance and self-efficacy (Stevens et al., 1990; Hausknecht et al., 2004), along with interviewers' aim of creating a close relation image (Wilhelmy et al., 2016), and applicants' favorable affective reactions to applicant-enhancement (Stevens et al., 1990), we expected that perceived applicant-enhancement would increase applicants' positive affect and interview self-efficacy:

Again, given the positive association that can be expected between the degree of IM sent by interviewers and the degree perceived by applicants based on Connelly et al.'s (2011) model, we also wanted to test the whole signaling timeline mechanism for applicant-enhancement. Extending the arguments above, we assume that interviewers' applicant-enhancement during the interview has an indirect effect on applicants' positive affect and interview self-efficacy through perceived applicant-enhancement (see Figure 1).

Unlike prior studies on interviewer characteristics and behaviors, we wanted to directly observe interviewer IM behaviors in a real selection setting by videotaping interviews and behaviorally coding these videotapes. Our approach to code IM behaviors was in line with previous studies that focused on applicants' IM (Stevens and Kristof, 1995; McFarland et al., 2005; Peeters and Lievens, 2006). Specifically, six I/O psychology graduate students served as coders (two male and four females; mean age = 24.7 years, SD = 1.5 years). They had gone through 5-h of frame-of-reference (FOR) training (Bernardin and Buckley, 1981) to recognize and record the frequency of interviewer IM behaviors. To record these behaviors, they used the INTERACT video coding software, which is a software to assess the frequency of behaviors by marking video sections with pre-assigned codes (Version 9, Mangold, 2010). Using this software, the amount of interviewer IM behaviors per interview was assessed based on the number of specific keystrokes: Coders watched the video recording of an interview and any time they identified an IM behavior, they pressed a key that was programmed to represent the specific interviewer IM behavior, respectively.

During the FOR training, coders were provided with definitions and examples of IM behaviors. Interviewers' organization-enhancement included statements promoting the university and the study program (e.g., “We have partner universities all over the world, you know”). Interviewers' applicant-enhancement included statements flattering and encouraging applicants (e.g., “that's impressive,” “very interesting,” “nice”). In addition, we also coded for defensive IM behaviors, which included interviewers' justifications and apologies (e.g., “sorry for making you wait”). However, these codes were neither included in the conceptual model nor in the data analyses¹. As part of the FOR training, several video sections (overall 25 min) that each focused on different IM behaviors were coded independently by each coder to practice how to recognize and record each IM behavior. Based on these frequency-codings, coders received feedback. Any coding discrepancies were discussed to enhance coders' understanding of the IM behavior categories.

After the FOR training, ten 10-min video sections representing different sections of the interview (e.g., opening sentences, interview questions, closing sentences) were frequency-coded independently by each coder to evaluate the reliability of the codings. The overall percentage of agreement was found to be good (0.81 for organization-enhancement, 0.72 for applicant-enhancement, and 0.71 for defensive IM). The median of the interrater agreement of different pairs of coders was reasonable (κ = 0.66) considering that organization-enhancement and defensive IM occurred rather seldom (see Landis and Koch, 1977). Furthermore, the level of interrater agreement found in the present research is comparable to previous studies on interviewees' IM (e.g., Stevens and Kristof, 1995; McFarland et al., 2005).

Afterwards, the actual coding took place. As agreement between the coders was shown to be good, each video of the 153 selection interviews was coded by one coder. The frequency of organization-enhancement, applicant-enhancement, and defensive IM per video were assessed as the sum of the IM behaviors of the two interviewers because overall, that was what applicants were exposed to during the interview. We then calculated the relative frequencies of IM behaviors (i.e., IM behavior use divided by interview duration in minutes) to control for interview duration.

Applicants were asked to complete two surveys at two different points during the interview process. The first survey (Time 1) was mailed to applicants 1–2 weeks prior to the interview along with an informed consent form and a cover letter. This survey also assessed demographic information such as age, gender, and years of prior work experience. The interviews were video recorded and directly after the interview, a second survey was handed to the applicants (Time 2).

Unless noted otherwise, five-point rating scales ranging from 1 = strongly disagree to 5 = strongly agree were used in this study. The items from the different scales used in the final model of this study are listed in Appendix B in Supplementary Material. Table 1 presents internal consistency coefficients for the measures for which they are applicable. All scales had adequate reliabilities.

Table 1 presents descriptive statistics of the variables used in this study. To test our hypotheses in a methodologically rigorous manner and in one analytical model, we applied structural equation modeling (SEM). We analyzed the data using the statistical environment R (Version 3.3.2, R Development Core Team, 2016) and Mplus (Version 7.4, Muthén and Muthén, 2015). Given that applicants were nested in interviewer teams, we first explored the degree of potential non-independence of the observations. We examined the variances in IM behavior within and between interviewer teams by inspecting the intraclass correlations coefficients (ICCs). These indicated substantive dependency of interviewers' organization-enhancement on the interviewer teams, ICC(1) = 0.28, ICC(2) = 0.71, as well as a comparable dependency of interviewers' applicant-enhancement on the interviewer teams, ICC(1) = 0.26, ICC(2) = 0.69. In other words, ~30% of the variance in the IM behaviors that interviewers used was dependent on the interviewer team. Because this dependency is not taken into account by classical SEM (e.g., Heck and Thomas, 2015; Kline, 2016), we used the COMPLEX procedure and the MLR estimator in Mplus (Muthén and Satorra, 1995; Muthén and Muthén, 2015). This procedure corrects the χ²-test of model fit, the resulting fit indices, and the standard errors for non-independence of observations to ultimately deliver unbiased parameter estimates.

In our study, some variables could only be measured with single indicators. Specifically, interviewers' organization-enhancement and applicant-enhancement were each measured as cumulative frequencies (i.e., number of IM behavior coded per interview), interview experience was measured as the number of prior interviews that applicants have had, and acceptance intention was measured as applicants' self-reported likelihood to accept an offer from the university. Given that the residuals of single indicators cannot be estimated in SEM and thus potential measurement error of these variables cannot be taken into account, we followed recommendations by Hayduk and Littvay (2012) and estimated the residual variance for the single indicators (see also Little, 2013; Kline, 2016). By doing so, we estimated that 34% of the total variance of the interviewer IM variables was residual variance and 25% of the total variance of the single-item-indicators of acceptance intention and interview experience was residual variance. In addition, in light of the rather complex analytical procedure, we tried to reduce model complexity whenever possible (see Landis et al., 2000). We therefore followed the construct-to-item-balance approach (Little et al., 2002; Williams and O'Boyle, 2008) and created two parcels for the organizational prestige as well as for the positive affect measures.

To establish our postulated measurement model, we conducted three confirmatory factor analyses (CFAs; see Table 2). Residuals were not allowed to covary because there was no theoretical reason to assume they would (Cortina et al., 2016). Following recommendations by Little (2013) regarding acceptable model fit, we evaluated our models in light of five fit statistics: (1) absolute test of fit, χ², (2) CFI ≥ 0.90, (3) TLI ≥ 0.90, (4) RMSEA ≤ 0.05, (5) SRMR ≤ 0.05. First, we estimated a one factor CFA in which all variables of interest loaded onto a single factor. This CFA did not converge and was therefore considered to be a misfit to the data. Next, we estimated a four-factor model where the items of each category of variables (predictors, mediators, outcomes, and controls) loaded onto a corresponding category variable. This four-factor model converged but displayed inacceptable fit to the data. Ultimately, we estimated the postulated measurement model with 13 factors, that is, we specified all theoretically postulated variables as separate entities. This model displayed acceptable fit, χ(138)2 = 191.93, p < 0.01, CFI = 0.96, TLI = 0.93, RMSEA = 0.05 [90% CI: 0.03 −0.07, p = 0.44], SRMR = 0.04. Therefore, this measurement model was well-suited to test the SEM in the next step.

To test our hypotheses in a rigorous manner, we estimated the theoretically postulated structural model, which follows the one depicted in Figure 1, but also tested all potential direct as well as indirect effects. In addition, control variables³ were taken into account by including paths from applicants' prior interview experience to the mediators and the outcome variables, and by also including paths from the respective baseline value (Time 1) of an outcome variable to its respective outcome at Time 2. This structural model displayed acceptable fit and was thus considered to be a valid representation of the data, χ(117)2 = 221.30, p < 0.01, CFI = 0.95, TLI = 0.93, RMSEA = 0.05 [90% CI: 0.03 −0.07, p = 0.44], SRMR = 0.06.

Table 3 and Figure 2 provide unstandardized as well as standardized path coefficients of the structural equation model. Our first two hypotheses referred to the question of whether applicants would notice the IM behaviors that interviewers used during the interview. Consistent with Hypothesis 1, interviewers' organization-enhancement in the interview was positively related to applicants' subsequent perceptions of organization-enhancement, b = 1.06, SE = 0.51, p < 0.05. Furthermore, and in line with Hypothesis 2, interviewers' applicant-enhancement in the interview was positively related to applicants' subsequent perceptions of applicant-enhancement, b = 0.70, SE = 0.27, p < 0.05.

In support of Hypotheses 3a, applicants' perceptions of organization-enhancement were positively related to organizational prestige, b = 0.14, SE = 0.06, p < 0.05. However, with regard to Hypothesis 3b, the association between applicants' perceptions of organization-enhancement and applicants' acceptance intention after the interview failed to reach significance at the conventional level (p = 0.06).

Hypothesis 4 specified a positive indirect effect of interviewers' organization-enhancement on (a) organizational prestige and (b) acceptance intentions through applicants' perceptions of organization-enhancement. Although the COMPLEX procedure of Mplus corrects the standard errors and consequently provides unbiased parameter estimates, it does not provide confidence intervals (Muthén and Muthén, 2015). Therefore, we applied the distribution-of-product method for building 95% confidence intervals for the indirect effects (MacKinnon et al., 2002) based on the corrected parameters using the R package RMediation (Version 1.1.4, Tofighi and MacKinnon, 2011). Indirect effects and confidence intervals are shown in Table 4. Consistent with Hypothesis 4a, the 95% confidence interval excluded zero and therefore indicated a significant positive indirect effect of interviewers' organization-enhancement via applicants' perceptions of organization-enhancement onto organizational prestige after the interview, b = 0.15, SE = 0.10, p < 0.05. With regard to Hypothesis 4b, the 95% confidence interval included zero, which means that the indirect effect of interviewers' organization-enhancement via applicants' perceptions of organization-enhancement onto acceptance intention after the interview was not significant.

Consistent with Hypotheses 5a, applicants' perceptions of applicant-enhancement were positively related to their positive affect, b = 0.10, SE = 0.05, p < 0.05. Furthermore, in support of Hypothesis 5b, applicants' perceptions of applicant-enhancement were positively related to their interview self-efficacy after the interview, b = 0.17, SE = 0.07, p < 0.05.

Finally, Hypothesis 6 specified a positive indirect effect of interviewers' applicant-enhancement on applicants' (a) positive affect and (b) their interview self-efficacy after the interview through their perceptions of applicant-enhancement. In line with Hypothesis 6a, the 95% confidence interval excluded zero and therefore indicated that the indirect effect of interviewers' applicant-enhancement through applicants' perceptions of applicant-enhancement was positively related to applicants' positive affect after the interview, b = 0.07, SE = 0.04, p < 0.05. Similarly, with regard to H6b, there was a positive indirect effect of interviewers' applicant-enhancement onto applicants' interview self-efficacy via applicants' perceptions of applicant-enhancement because the 95% confidence interval excluded zero, b = 0.12, SE = 0.07, p < 0.05.

This study has several theoretical and practical implications. Our study relates to signaling theory (Spence, 1973; Connelly et al., 2011; Bangerter et al., 2012) by proposing and testing a signaling timeline model to explain the effects of interviewer IM. It sheds light on the effects of interviewer IM by considering both the interviewer's perspective in terms of sending signals and the applicant's perspective in terms of receiving and responding to signals. Specifically, we found that for interviewer IM (i.e., signals sent) to have an effect on recruiting outcomes, applicants need to perceive interviewers' IM behavior (i.e., signals received), and respond to it (i.e., reactions to signals). This finding emphasizes that in the context of the interview, it is important to conceptually and empirically distinguish the signals sent from the signals received because to have an effect, signals that are sent by interviewers need to be received by applicants.

Furthermore, we expand previous theoretical frameworks on IM (e.g., Gilmore et al., 1999; Tsai and Huang, 2014) by providing evidence that different kinds of interviewer IM (organization-enhancement and applicant-enhancement) are conceptually related to different goals of interviewers (representing the organization and making applicants feel good): Organization-enhancement seems to be particularly effective at influencing applicants' positive attitudes toward the organization, whereas applicant-enhancement seems to be particularly effective at influencing applicants' positive emotions. This is in line with and further corroborates prior propositions that one kind of IM behavior may not be relevant for every purpose (Tsai and Huang, 2014; Wilhelmy et al., 2016). In addition, our findings stress that it is crucial to consider various recruiting outcomes, including those that reflect organizations' perspective such as organizational reputation and prestige, but also criteria that reflect applicants' perspective such as their positive affect and belief in the ability to succeed with an interview.

The findings of the present study do not only benefit the research community but also have implications for practitioners. Our finding that interviewers' organization-enhancement and applicant-enhancement can influence recruiting outcomes provides indications on the opportunity of increasing organizations' competitive advantage through signals that are sent to applicants in the course of the interview process. For instance, organizations could conduct training sessions to enhance interviewers' IM skills. Specifically, organization-enhancement could help interviewers highlight strengths of the organization and attract the best applicants. Applicant-enhancement, in contrast, could help interviewers to make applicants feel wanted and thus foster positive emotions and feelings of self-confidence, which is, for example, important for word-of-mouth recommendations. Nonetheless, the effect sizes in the present study were relatively small, which potentially limits the practical relevance of the findings. The small effect sizes, however, could also be due to the large number of factors that influence recruiting outcomes. In addition, a phenomenon can be important despite small effect sizes (Cortina and Landis, 2009). We therefore believe that interviewers' IM has the potential to be effectively used by organizations, but the findings presented in this paper should be bolstered by insightful future research before more specific practical recommendations can be made.

Despite these potential benefits, interviewer IM may lead to conflicts between the selection and recruitment needs of employers. Interviewer IM may enhance the recruitment function of the interview but may adversely affect the psychometric qualities such as inter-rater reliability (because of deviations from standardization) and criterion-related validity (because of the introduction of systematic error) and thus impede the selection function. For example, Marr and Cable (2014) found that interviewers' selling-orientation reduced their judgment accuracy and the interview's predictive validity. However, as Tsai and Huang (2014) pointed out, this does not mean that employers have to sacrifice their recruitment needs for their selection needs. For example, in order to achieve a balance between the recruitment and selection functions, interviewers may consider conducting two separate interviews, “one designed strictly for prediction and the other designed to allow an informal question-and-answer session to meet the needs of applicants” (Kohn and Dipboye, 1998, p. 839). However, as separate interviews involve additional costs, another solution might be to conduct highly structured interviews but incorporate a more personal interview stage, for example at the beginning and the end of the interview (Schuler and Funke, 1989; Chapman and Rowe, 2001). In addition, some researchers and practitioners may argue against the use of interviewer IM because interviewers might exaggerate information and try to deceive applicants when applying IM behaviors. Unrealistic expectations can ultimately result in negative affective reactions on the part of employees (Wanous et al., 1992). Thus, interviewers should avoid pursuing the wrong applicants on the wrong terms. A practical solution may be to combine IM with realistic job previews (RJPs, Wanous, 1976) in terms of presenting positive attributes while also informing applicants about possible downsides of the job and the organization (see Wilhelmy et al., 2016).

Although results of the present study provide valuable insights into the effects of interviewer IM, the study is not without limitations. For example, our data was based on real selection interviews that were conducted as part of the selection process for one specific academic organization. This may call the generalizability of our study into question, but it does not diminish the relevance of this sample because the challenges that universities face regarding their recruitment efforts are similar to those in private enterprises in terms of competition with other universities and the need to balance selection and recruitment aims (e.g., Colarelli et al., 2012). Furthermore, it is important to note that even though we examined applicants who applied for a Bachelor's study program, our sample does not represent a student sample. In fact, in our study, applicants had work experience and were markedly older than high school graduates with an average age of 25 years. The reason for this is that the study program is specifically designed for individuals with prior work experience who want to complement their training and education with an additional Bachelor's degree.

Moreover, the present study was conducted in the field using pre-interview and post-interview applicant survey data and video coding of interviewers' IM behavior, but mediators were measured at the same time as the outcomes. This raises common method bias concerns. However, these concerns are alleviated to some extent by using two measurement points (before and after the interview) and two data sources (video recordings and applicant surveys) to separate the data collection on predictors and outcomes (following recommendations by Podsakoff et al., 2003) and to incorporate baselines of the outcome variables (following recommendations by MacKinnon et al., 2012). Nonetheless, future research should use a more rigorous research design with multiple measurement points to obtain predictors, mediators, and outcomes at different points in time to strengthen causal inferences (see Ployhart and MacKenzie, 2015).

A strength of the current study, which is also a potential limitation, is the fact that we focused on interviewer IM behavior that was actually applied during interviews and used an extensive behavioral video coding procedure to gain valuable insights into the frequency of these behaviors. Even though this methodology is in line with previous IM research (e.g., Stevens and Kristof, 1995; Ellis et al., 2002; McFarland et al., 2005; Peeters and Lievens, 2006), it remains unclear what intentions interviewers had when they used these IM behaviors. Future research should put more emphasis on interviewer IM intentions, for example, by combining video ratings as used in the present study with self-ratings of IM behavior (Levashina and Campion, 2007).

In the present study, interview structure, interview questions, and the relation of selection and recruitment goals were held constant across interviews. However, interview structure and the ratio of recruitment vs. selection goals may be important boundary conditions of interviewer IM and should therefore be studied. Specifically, researchers may focus on the possibility that there might be a minimum level of freedom regarding the interview content and a minimum level of recruitment objectives needed for interviewer IM to occur and to be effectively applied. Relatedly, the present study focused on the signals that were sent by a team of two interviewers who alternated in asking the interview questions. To understand how interviewer IM evolves in the first place, it is important to also consider the individual interviewer as the unit of analysis. It could be that some interviewers use the same kind and degree of IM in every interview and that others adapt their IM behavior to every interview situation and every applicant. Thus, we encourage future research to study larger samples of interviewers and examine the degree to which an interviewer's IM behavior remains stable across interviews (e.g., depending on interview guidelines of the organization and the interviewer's personality), and to what degree interviewers adapt their IM from interview to interview (e.g., depending on how much they want to win an applicant over).

The current study provides indications that interviewers' IM is indeed perceived by applicants, and that it can influence applicants' attitudes, intentions, and emotions. However, as a starting point for this line of research, we only focused on two central IM behaviors, namely organization-enhancement and applicant-enhancement. Qualitative findings revealed that interviewers use a broad range of different verbal IM behaviors that go beyond organization-enhancement and applicant-enhancement such as demonstrating humor and depreciating applicants (Wilhelmy et al., 2016). In addition, interviewers not only use verbal IM but also artifactual IM (i.e., using appearance and visual information to influence applicant impressions such as displaying application documents) and administrative IM (i.e., using timing of communication and providing services to influence applicants' impressions such as incorporating future colleagues). Future research should consider this broad range of interviewer IM behavior to provide more comprehensive insights into its effects.

In addition, future research should consider interactive effects and different combinations of interviewer IM. The current study examined the effects of organization-enhancement and applicant-enhancement on recruiting outcomes, but there could be different configurations of interviewer IM used by an interviewer team (e.g., mainly organization-enhancement vs. mainly applicant-enhancement vs. both organization-enhancement and applicant-enhancement vs. neither organization-enhancement nor applicant-enhancement), and these different configurations could have different effects on recruiting outcomes. Therefore, to examine interactive effects with increased statistical power to detect those effects, we encourage future research to use experimental study designs to specifically vary and combine interviewers' IM behaviors. In addition, in real selection settings with large samples of interviewers, the prevalence and effects of different configurations of interviewer IM could be examined through latent profile analyses. Furthermore, Stevens et al. (1990) found that organization- and applicant-enhancement can have a positive influence on recruitment by increasing applicants' perceptions of interviewer competence, but can also backfire by appearing presumptuous to applicants. This raises the question of what kind of IM may tip the scale between perceiving organization-enhancement as valuable information and as a signal of interviewer competence vs. arrogance. Future research exploring the interactive effects of different kinds of interviewer IM behavior and the role of perceived interviewer competence vs. arrogance is warranted.

Another potential limitation is that the current study examined the effects of interviewer IM on applicants' attitudes, intentions, and emotions, but not on behavioral outcomes. For example, we do not know the extent to which applicants' job choice decisions would be influenced by interviewer IM. Recent findings from a military context indicate that perceived recruiter characteristics such as personableness can influence applicants' acceptance decisions (Harold et al., 2016). However, as Tsai and Huang (2014) point out, it is possible that applicants' attitudes and intentions after the interview are influenced by interviewer IM, but that their actual job choice behavior will be primarily based on information provided by other sources than the interviewer such as friends or insiders that work at the organization. Future research should therefore employ longitudinal designs in selection settings where there is enough variance in applicants' job choice decisions to answer this important question.

Finally, several unexpected findings warrant attention. In the current study, applicants' perceptions of organization-enhancement were not significantly related to their intention to accept a potential offer. Stevens et al. (1990) found that only a few of the study participants opted for the university that was presented by a recruiter who used organization-enhancement, and that this was because the arguments presented were not seen as convincing. Thus, future research should consider how applicants interpret organization-enhancement, and how much weight this information carries for different applicants. Another interesting finding is that applicants' perceptions of organization-enhancement were related to applicants' positive affect after the interview. This may be because strengths and advantages of the study program and the organization that were pointed out by interviewers may increase applicants' excitement and enthusiasm for the interview and thus elicit positive affective reactions in applicants. Lastly, the mean level of the recruiting outcomes decreased after the interview. This is in line with prior studies: When comparing means before and after the interview, applicants' acceptance intention and ratings of organizational attractiveness were often found to decrease (Powell, 1991; Carless, 2005; Carless and Imber, 2007). In another study, applicants' test taking self-efficacy was found to decrease after the selection process (Bauer et al., 1998). A possible explanation for this decrease in recruiting outcomes is that after the interview, applicants are often overwhelmed and buried in information that they need to process. In addition, it could be that the interview helped applicants decide against the organization and select themselves out of the process, which might have decreased the mean level of recruiting outcomes. Furthermore, we do not know how much the mean level of the recruiting outcomes would have decreased without any interviewer IM during the interviews. Future research could therefore use longitudinal, quasi-experimental designs with several groups that differ in the amount of interviewer IM to examine main effects of interviewer IM on recruitment outcomes.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.