Advisees' Expectations for Support as Moderator Between Advisor Behavior and Advisee Perceptions of Advisor Behavior

In the context of interpersonal relationships, an individual's expectations heading into the relationship can affect her or his later judgments about the value and experience of having been in the relationship (Propp & Rhodes, 2006). This can happen in at least three ways. First, according to the confirmation bias, initial expectations of how a target person will behave causes the observing individual to pay selective attention to the incidences in which the target behaves consistently with expectations while disregarding the cases in which the target disconfirms those expectations (Nickerson, 1998). Therefore, an individual may not give the same weight to behaviors that run contrary to expectations than to those consistent with expectations. In this case, as a result, the expectations do not affect the target's behavior, only the way the observer perceives it. In the context of our study, a confirmation bias would lead to a positive correlation between advisees' expectations and perceptions of specific types of support received, but we would find no correlation between expectations and observable behaviors consistent with such support. In a previous study, Young and Perrewé (2004) found that protégés' initial expectations for career and social support provided by their mentors were positively related to their later perceptions of these benefits. Thus, we formulated the following hypothesis:

In the prior study, Young and Perrewé (2004) did not examine relations between expectations and objectively coded mentor behavior. Thus, their findings may reflect either a confirmation bias or a change in behavior.

A self-fulfilling prophecy (Snyder, 1992) describes a situation in which a perceiver's pre-existing expectations of a target individual cause the observer to behave in ways toward the target that impact the target's behavior. Thus, expectations result in changes in the target's behavior. In a classic study, Rosenthal and Jacobson (1968) led teachers at the beginning of the school year to believe that some students were more cognitively capable than others. However, in reality, no differences in scholastic capability were evident among the students. At the conclusion of the school year, an analysis of the achievement test scores revealed that the students whom the teachers believed to be especially capable outperformed the other students, despite the fact that the children had not differed on these measures at the beginning of the school year. Presumably, teachers behaved differently toward those they expected to be more capable resulting in (as opposed to merely teacher perceived) effects on student performance.

Similarly, in a lab study, Snyder and Swann (1978) led perceivers to believe that the target whom they would be interviewing was either an extravert or introvert. The researchers then provided the perceivers with a list of potential questions to ask the target. The perceivers who believed the target to be extraverted were more likely to ask questions such as “In which situations are you the most talkative?” whereas perceivers who were told their target was introverted were more likely to ask questions such as, “What factors make it hard for you to open up to people?” Trained, third-party judges who were blind to the conditions listened to the conversations and rated the personality of the target. Not surprisingly, the raters assessed targets in the extravert condition as more extraverted than targets in the introvert condition.

Self-fulfilling prophecy may affect relationships either apart from or in tandem with confirmation bias. In our study, self-fulfilling prophecy would manifest itself as, over the course of interacting with their advisors, advisee behavior bringing about advisor behaviors consistent the advisees' preconceived notions. To test for this effect, coded ratings of the target's behavior (i.e., the advisor) are necessary because advisee ratings would prove insufficient if confirmation bias also characterizes the situation. That is, if an advisee simply attends to advisor's behaviors congruent with his or her own expectations and conveniently fails to notice incongruent behaviors, then the extent to which the advisee's perceptions correspond to reality remains unclear. However, if an advisee's initial expectations cause changes in advisor behavior, we would expect to see a positive correlation between an advisee's pre-program expectations and the coded ratings of advisor behavior. Considering these conditions, we developed a second hypothesis:

A third way to determine whether advisees' expectations affect perceptions of the support they receive involves differentially weighting observed behaviors consistent with that support. If one's high expectations of a particular type of support are fulfilled, the person likely reacts positively, and this sense of satisfaction could affect the individual's reports of the support received. Conversely, if an expected event or situation fails to materialize, a person may react negatively, which will manifest in the reported perceptions of his or her experience.

Heilman and Chen (2005) devised an experiment to understand how expectations associated with gender roles can impact evaluations of identical behaviors. They found that women who failed to perform an altruistic behavior (i.e., staying after hours to assist a coworker in meeting an important deadline) received lower performance ratings than did men who failed to perform the identical act. We suggest that the expectation that women would behave more altruistically then men strengthened the relationship between their behavior and evaluations of that behavior. In the same way, advisees with high expectations for psychosocial and career support from their peer advisors may reward the provision of such support with high ratings and punish the absence of such support to a greater degree than advisees with lower expectations for psychosocial and career support from advisors.

In their qualitative analysis, Eby and Lockwood (2005) noted that mentor failure to meet mentee expectations creates the most commonly reported problem for protégés in formal mentoring programs. Young and Perrewé (2000) found that the fulfillment of expectations mediated the relationship between mentor support and protégé perceptions of relationship effectiveness and trust. Based on these studies, we developed a third hypothesis:

In the current investigation, we employed a structured peer-advising program as a context for testing our three hypotheses. In the peer-advising program, incoming students paired with advanced senior students who aided them in their academic and personal adjustment. Students acclimating to the new college environment face both academic and social challenges in their first year (Tinto, 1993). Rosenthal and Shinebarger (2010) discussed “the gap between student needs and the type of mentoring they receive” (p. 24) and argued that peer advisors can bridge this gap. Peer advisors often have more flexibility than faculty advisors. Specifically, students may feel more comfortable talking with a peer about certain topics, such as roommate disputes or homesickness, than with a faculty advisor. The lack of hierarchy between students and peer advisors can increase interpersonal comfort and a sense of similarity when peer advisors discuss their own experiences and how they dealt with them. As Smith and Allen (2006) emphasized, researchers need to examine student characteristics that predict their preferences for different types of academic advising and their perceptions of advising effectiveness. With our study, we hope to shed light on the way advisee expectations about the quality of advising can influence their subsequent evaluations of advising relationship quality.

Initially 213 undergraduates receiving advising in a formal peer-advising program, established for the socialization of incoming students, participated in this study. However, we restricted the analysis (using listwise deletion) to the 179 participants who completed data on all of the variables used in this report. Peer advisor (32 females, 23 males) volunteers were either juniors or seniors with a minimum grade-point average (GPA) of 3.0, and advisees (117 females, 62 males) were freshmen or transfer students in their first semester at the university. The advisee pool consisted of 117 (65%) Caucasian, 24 (13%) African American, 24 (13%) Hispanic, and 14 (8%) Asian American students. The participating advisors were comprised of 45 (82%) Caucasian, 2 (4%) African American, 3 (6%) Hispanic, and 4 (7%) Asian American advisors; 1 (2%) was classified as other.

The core functions of quality mentoring typically discussed in the literature are psychosocial and career support. Our measures included advisee expectations, coded behavior, and advisee perceptions of the two types of support that they have received.

Before the first advising session, using a modified version of the Allen, McManus, and Russell (1999) 24-item measure of psychosocial and career support, we measured the degree to which advisees expected to receive career and psychosocial support through their advising relationships. The instrument was developed for measuring the quality of workplace mentoring; see Noe (1988) for the entire original scale. We modified these items in two ways. First, we reworded the items to reflect an expectation for the future. For instance, a pre-advising item stated, “I expect my advisor to share the history of his/her academic career with me.” Second, we recast the items to reflect an academic, rather than workplace, context. Respondents used a 6-point Likert scale (1 = strongly disagree to 6 = strongly agree) with coefficient α values for psychosocial support of .83 and career support of .83. Finally, we summed the items on all scales in the study and then calculated the mean score to use in our analyses.

We measured advisee-perceived receipt of support after the advising sessions ended using the Allen et al. (1999) 24-item measure (modified from Noe, 1988) that we adapted for use with academic populations. Advisees rated the degree to which their advisor provided psychosocial and career support to them during their advising sessions. A sample item stated, “My advisor encouraged me to prepare for academic advancement.” Respondents utilized a 6-point Likert scale (1 = strongly disagree to 6 = strongly agree). Advisee-reported psychosocial support and academic development yielded coefficient α values of .89 and .90, respectively.

We also examined career support and psychosocial support based on a third-party coded procedure. Specifically, we scrubbed advising-session transcripts of any identifying information in relation to gender and ethnicity to avoid any undue bias. Next, two trained raters, unaware of the study's hypotheses, coded the frequency of statements consistent with psychosocial support (e.g., “I think you will do great; just believe in yourself”) and career support (e.g., “I would talk to current nursing students to see what you can do to make yourself more competitive”). For example, if a peer advisor said, ‘‘I recommend that you set up an appointment with your professor to discuss becoming a research assistant,” this was coded as one statement of career support. Coders summed all statements of psychosocial and career support separately for each session transcript, and then we calculated the averages of all three advising sessions to arrive at overall scores for the two types of advising support. To learn the essence of the constructs of psychosocial and career support and gain familiarity with the content of the two categories (per Cascio & Aguinis, 2005), coders received frame-of-reference training with operational definitions consistent with the self-report measure collected. Interrater reliability was .77 for coded psychosocial support and .94 for career support.

Advisor and advisee gender were used as covariates in all analyses (1 for males and 2 for females). Furthermore, some participants communicated through electronic chat rather than verbally. Thus, we also controlled for communication mode in all analyses (coded as 0 for verbal and 1 for nonverbal).

Before participating in the formal peer-advising program, advisees indicated the extent they expected to receive psychosocial and career support from their peer advisors, and all participants completed demographic questions. All participants completed 30-minute face-to-face orientation sessions with program administrators, where they received suggestions for discussion topic areas (e.g., selecting a major, study tips, class scheduling and planning, roommate issues, getting involved on campus) and rules of conduct (e.g., no discussing illegal activity, no offensive comments). At this time, all participants learned that three advising sessions would take place at an experimental laboratory on campus and would be recorded for research purposes. In these sessions, participants could discuss any topic in line with the rules of conduct. Immediately following their third session, advisees rated the degree to which they received psychosocial and career support. Participants were then debriefed and given the opportunity to exchange information with their peer advisor, so they could continue to meet as they wished; 52% of the advisees continued to meet with their peer advisors after the research ended.

All session transcripts were coded by raters trained to identify statements of career support and psychosocial support. For example, a rater coded an advisor recommendation of “I took calculus and recommend that you take a couple other math classes first” as a statement of career support. Coders summed all statements of psychosocial support and career support separately for each session transcript. At the conclusion of the last session, advisees self- reported the degree to which they received psychosocial and career support from their advisors.

Hypothesized relationships were examined using Pearson product-moment correlation and multiple regression in IBM SPSS Statistics 19. An α level of .05 was used for all analyses. Pearson product-moment correlation results and descriptive statistics are reported in Table 1.

Advisee expectations of receiving psychosocial support were positively related to advisees' perceptions of having received psychosocial support (r = .37, p < .001). However advisee expectations of receiving career support were not significantly related to advisees' perceptions of career support received (r = .09, p = .24). In support of H1a, advisees' expectations for receiving psychosocial support was significantly related to their perceptions of psychosocial support received (β = .41, p < .001). However, in relation to H1b, advisees' expectations of receiving career support was not significantly related to their perceptions of career support received (β = .09, p = .19). Advisee expectations were not significantly related to coded ratings of psychosocial support (r = −.12, p = .11) or to coded ratings of career support (r = .03, p = .72). Thus, no support was found for H2a or H2b.

According to Baron and Kenny (1986), “a moderator is a qualitative (e.g., sex, race, class) or quantitative (e.g., level of reward) variable that affects the direction and/or strength of the relation between an independent or predictor variable and a dependent or criterion variable” (p. 1174). Thus, interaction terms were created in the present study to determine whether the relationship between the independent variable (coded advisor behavior) and dependent variable (perceived advisor behavior) changed at various levels of the moderator (expectations). A significant interaction term means that coded advisor behavior differentially predicted perceived advisor behavior based on advisee's expectations of receiving psychosocial and career support. Interaction terms were computed by multiplying expectations of a particular type of support by coded instances of behavior consistent with such support.

To test H3a, advisee perceptions of the psychosocial support they had received was regressed simultaneously onto advisee expectations of receiving psychosocial support, coded psychosocial support, advisor gender, advisee gender, communication mode, and the interaction of advisee expectations and coded psychosocial support. This equation was significant: R² = .21, F(6, 172) = 7.70, p < .001 (see Table 2). Results indicated that advisors who exhibited a greater number of behaviors consistent with psychosocial support were perceived by their advisees to have provided greater psychosocial support: β = .18, t(178) = 2.49, p = .01. In addition, advisees' expectations of receiving psychosocial support accounted for unique variance in postprogram perceptions of psychosocial support received: β = .41, t(178) = 5.84, p < .001. However, advisee expectations of psychosocial support did not significantly moderate the relationship between the frequency of behaviors coded as being consistent with psychosocial support and advisee perceptions of having received psychosocial support: β = −0.10, t(178) = −1.36, p = .17. Thus, H3a was not supported.

Next, advisee perceptions of having received career support from their advisor was regressed simultaneously onto advisee expectations of receiving career support, coded career support, advisor gender, advisee gender, communication mode, and the interaction between advisee expectations of receiving career support and coded career support. This equation was significant: R² = .15, F(6, 172) = 5.06, p < .001 (see Table 3). Results indicated that advisors who exhibited a greater number of behaviors consistent with career support were perceived by their advisees to have provided greater career support: β = .29, t(178) = 3.65, p < .001. However, as hypothesized, the relationship was stronger when advisees held higher expectations of receiving career support at the start of the program: β = .15, t(178) = 2.01, p = .05. As illustrated in Figure 1, the slope of the relationship between coded instances of career support and perceptions of career support received was more steeply positive for protégés with higher expectations of receiving career support. This pattern of results provides support for H3b.

Results demonstrated that advisees' subjective reports of the psychosocial and career support they had received from their peer advisors had been influenced by both the behavior of advisors and the advisees' preexisting expectations of support. Those who expected more psychosocial support prior to participating in a peer advising program indicated that they received more psychosocial support at the conclusion of the program, even after the frequency of advisor-demonstrated psychosocial support behaviors was controlled. Advisee expectations were not correlated with advisor behavior for either type of support. However, expectations of receiving career support moderated the relationship between coded and advisee perceived career support from their advisors. As predicted, coded behavior consistent with career support was a better predictor of advisee perceptions of career support for advisees who entered the program with higher expectations for career support (Figure 1).

We offer three propositions about the ways advisee expectations might affect the support they report having received. Prior expectations may emerge as confirmation biases with regard to perceptions of support received, the expressions of which, in this case, manifest as significant correlations between expectations and postprogram perceptions with no similar correlation between advisee expectations and advisor behaviors. Our findings provide evidence that bias affected the data of psychosocial support but not career support. Consistent with the findings of Young and Perrewé (2004), advisee expectations and perceptions regarding psychosocial support were positively related in our study. However, in an extension of prior research, we demonstrated that this bias did not affect the relationship between expectations and objectively coded advisor behavior.

In fact, expectations were not significantly related to either type of coded advisor support. These findings suggest that advisees' expectations did not lead to self-fulfilling prophecies as hypothesized. However, we found partial support for the proposition that expectations cause individuals to differentially weigh identical behaviors when assessing perceptions of their advising relationships. Specifically, the slope of the relationship between coded advisor behaviors is consistent with that for received career support. In addition, the slope for advisee perceptions of career support is steeper for advisees with initially high expectations for receiving such support. This pattern of results shows that fulfilled expectations led to accentuated positive ratings, in this case, of support advisees received from peer advisors; in the converse situation, dissatisfaction resulting from unmet expectations led to accentuated negative ratings.

Advising and mentoring programs are typically evaluated using subjective reports from participants. Results from our study demonstrate that such data can be colored by participants' initial expectations regarding the support they expect to receive. As a consequence, we recommend that program administrators recognize that advisee reports regarding the quality of their advising relationships may reflect their own expectations and not necessarily the behavior of their advisors. Assessment of both expectations and behaviors from the dyad provide a clearer picture of the situation. For example, according to our study, an advisee with low expectations for psychosocial support was less likely to perceive receiving such support regardless of his or her advisor's behavior. Thus, training advisors to provide psychosocial support, while affecting their practice, may not yield the desired positive effect on subjective reports from advisees with low expectations for such support.

Of course, raising advisees' expectations without training advisors to behave accordingly can backfire, as indicated by our results regarding career support. Specifically, advisees who expected to receive career support reacted more positively to support behaviors than did those with low expectations for it. However, they also reacted more negatively to advisors who demonstrated few career support behaviors. In sum, participants' expectations should be aligned with respect to the types and level of support provided through advisory relationships, and advisors should be trained to behave in ways that are consistent with these expectations.

Conducting the current study longitudinally in the laboratory allowed us to examine the natural progression of advisory relationships using a controlled environment (i.e., holding constant the frequency and length of dyad interactions) and third party coders. The use of time lapse and trained coders, not often utilized in advising or mentoring research, make this study a unique contribution to the existing literature. However, like all other studies, limitations remain and must be noted before generalizing the results.

First, to balance our need for experimental control with the rights of the participants to develop productive and satisfying relationships, our investigation lasted only 3 weeks. At the study's conclusion, participants could choose to continue their advising relationship informally. Thus, through the duration of the study each of the advisory relationships was likely in the first, initiation mentoring phase discussed by Kram (1983) and therefore more positively valenced, much like the honeymoon phase of a relationship. Kram surmised that these initial meetings tend to create and support positive expectations. However, early perceptions shape advisory relationships and provide the framework for their transition to a new phase (Kram, 1983). Therefore, the relationship between expectations and the amount of advising received may change as a relationship develops over time. For example, as the dyad members become more familiar with one another, initial expectations may become less predictive of advisee perceptions of the quality of advising.

Second, we did not examine the role advisor expectations might have played in the relationship. Previous studies on mentoring suggest that mentors and protégés hold unique and often different, reactions, reports, and memories of the same relationship (Eby, Durley, Evans, & Ragins, 2008; Wanberg, Kammeyer-Mueller, & Marchese, 2006). Accordingly, future research may delve into the influence of advisor expectations on their own and advisee perceptions of the degree of advising received and their link to external outcomes (e.g., academic success, personal growth, learning).

Finally, we acknowledge the quantitative nature of the coded measure of advisor behavior, so we cannot determine the degree to which the advisees emphasized quantity or quality when responding to items eliciting their perceptions of the support received (e.g., “My peer advisor provided advice”) because the scale anchors were strongly agree to strongly disagree. However, the data allowed us to understand the relative subjective worth that advisees placed on each additional advisor behavior, as recorded by the coder. The scale we used to measure advisee perceptions was used in prior research and is worded and scaled similarly to most of those used in research on advising and mentoring. However, whether (or how) the pattern of results would differ with identically operationalized constructs remains unclear; that is, we do not know whether advisee- and coder-counted frequencies of advisor behaviors would result in differing views of advisor support.

In summary, we demonstrated that advisee expectations of the support from their peer advisors can affect the manner in which they react to their advisors' behavior. The results yield important understanding in interpreting and appropriately responding to perceptual data of the advisory relationship. Because of the extensive use of advising and mentoring in academic and applied settings, we hope those using theoretical models and applied interventions in this area will heed our recommendations to include preprogram expectations and coded data to obtain a more representative picture of advising processes and outcomes. Every effort should be made to ensure that an advisee avoids assuming that the advisors automatically understand her or his needs.