Abstract
Although social support is typically associated with a number of health benefits, for some individuals support worsens outcomes, likely because receiving support can undermine feelings of competence. Some have argued that invisible support (i.e., support that recipients do not recognize as help) can reduce negative support-related health consequences (Bolger, Zuckerman, & Kessler, 2000); still, the physiological benefits of invisible support have yet to be established and likely differ as a function of self-efficacy. The purpose of this study was to investigate how visibility of support and self-efficacy interact to affect the cortisol reactivity of a support recipient. In a 2 (self-efficacy: high vs. low) × 2 (support visibility: visible vs. invisible) between-subjects, experimental design, 74 undergraduate students were primed for either high or low self-efficacy using false feedback. Participants then received either visible or invisible support from a confederate while preparing a speech as part of the Trier Social Stress Task (Kirschbaum, Pirke, & Hellhammer, 1993). A series of repeated measures analyses of variance revealed that individuals primed to have high self-efficacy experienced cortisol increases in response to the speech task when they received visible support but not invisible support. By contrast, individuals primed to have low self-efficacy experienced increased cortisol when they received invisible support but not visible support. This research suggests that although invisible support can effectively buffer stress, it is not always the best support strategy.
The provision of social support is a natural and important part of relationships with friends, family, and romantic partners (e.g., Marroquín, 2011; Rafaeli & Gleason, 2009). For many people, social support is associated with a number of benefits, including lower mortality rates as well as improved adjustment to cancer and cardiovascular disease (Heavey, Layne, & Christensen, 1993; House, Landis, & Umberson, 1988; Schwarzer & Leppin, 1989). However, depending on the needs and individual characteristics of the recipient, there are circumstances under which support can have unintended negative consequences (Grimes, Overall, & Simpson, 2013; Rafaeli & Gleason, 2009). As such, it is essential to better understand for whom and under what circumstances support benefits health.
It is now well established in the literature that social support, particularly support that the recipient is aware of receiving (i.e., visible support), can “backfire” leading to increased stress and negative psychological outcomes (Rafaeli & Gleason, 2009; Taylor et al., 2010). For example, unreciprocated support, or receiving support when you are not also providing support, results in increased negative affect (Gleason, Iida, Bolger, & Shrout, 2003). Further, receiving support that is belittling to the recipient’s abilities is more stressful than not receiving any support at all (Bolger & Amarel, 2007). Of particular interest to the current study, when the type of support provided does not match the type of support that the recipient desired (e.g., recipient desired emotional support but received instrumental support), support is not perceived as helpful (Cutrona, Shaffer, Western, & Gardner, 2007). Indeed, even when social network members are intending to be supportive, their support can be perceived negatively (Lehman, Ellard, & Wortman, 1986) and result in adverse psychological consequences (Bolger et al., 2000; Newsom & Schulz, 1998).
Several theories have been proposed to explain why support can hurt the recipient. Some have argued that support givers can be unskilled at providing support (Rafaeli & Gleason, 2009) and provide support in a way that does not match the recipient’s needs (Bolger et al., 2000; Taylor et al., 2007). Others have argued that support hurts individuals’ self-esteem by highlighting their inability to manage a stressor on their own (Bolger et al., 2000; Fisher, Nadler, & Whitcher-Alagna, 1982). Consistent with this argument, receiving support can threaten autonomy and create feelings of dependence on social networks that undermine individuals’ confidence in their ability to solve their own problems (Revenson, Wollman, & Felton, 1983).
Given the theorized links between received support and negative outcomes, some have argued that invisible support (i.e., support given outside of the recipients’ awareness) may be an especially effective support type (Bolger et al., 2000; Rafeali & Gleason, 2009; Shrout, Herman, & Bolger, 2006). Because invisible support goes unnoticed, it does not undermine self-confidence and can therefore induce the desired positive effects of support while avoiding negative consequences (Bolger & Amarel, 2007; Bolger et al., 2000; Shrout et al., 2006). In a particularly relevant study, Bolger, Zuckerman, and Kessler (2000) used daily diaries to examine a sample of bar examinees and their partners in the period leading up to the exam. On days when support recipients did not recognize the support of their partners (i.e., received invisible support), recipients reported lower depression and anxiety than on days when they did recognize their partners’ support (i.e., visible support). Later experimental research confirmed that invisible support reduced negative affect more than did visible support (Bolger & Amarel, 2007). As an additional benefit, invisible support can also increase self-efficacy (Howland & Simpson, 2010).
Despite evidence for the effectiveness of invisible support, the question remains whether or not invisible support is a ‘one-size fits all’ solution. Just as the benefits of support require that the type of support received match the type of support the recipient desired (Cutrona et al., 2007; Taylor, Welch, Kim, & Sherman, 2007), preferences for the visibility of support are likely need-specific. Consistent with this argument, the benefits associated with invisible support depend on responsiveness (Maisel & Gable, 2009). That is, invisible support only reduces negative affect on days when an individual feels understood, valued, and cared for by the support provider. Further, the effectiveness of invisible support is moderated by the recipient’s level of stress (Collins & Feeney, 2000; Grimes et al., 2013). When individuals experience more stress, they engage in direct support-seeking behavior (e.g., asking their partner for help; Collins & Feeney, 2000). It is perhaps unsurprising, then, that when individuals experience higher levels of stress, they perceive visible support as more effective and supportive than invisible support (Grimes et al., 2013). In other words, when experiencing stress, individuals desire and benefit from visible support.
Considering that preferences for support visibility can be contextually based, we argue that the effectiveness of invisible support is also moderated by self-efficacy, or an individual’s belief in their ability to accomplish a given task. Similar to individuals experiencing higher amounts of stress, individuals low in self-efficacy engage in direct support-seeking strategies (Feeney, 2004). Of particular interest to the current study, when individuals have high levels of self-efficacy, support interferes with autonomy and results in negative affect and poorer health behaviors (Warner et al., 2011). By contrast, when individuals have low self-efficacy, support can serve as a social resource that compensates for the individuals’ lack of personal resources (Warner et al., 2011). For such people, support can be used to maintain autonomy and increase feelings of competence (Reich & Zautra, 1995; Warner et al., 2011).
Taken together, it seems plausible that the reason invisible support has previously been associated with better psychological outcomes than visible support is because it does not challenge the self-concept of competence held by highly self-efficacious people (e.g., the senior law students in Bolger and colleagues’ study). By contrast, people who already doubt their ability to complete a task may not be further threatened by support. In fact, it may be that these people desire to receive support (and to know that they have received support) in order to increase their chances of being successful and to ease their minds about their chances of failure (Warner et al., 2011). To date, research has yet to experimentally test this theory. Further, although the psychological effects of invisible support are well established, past research has not tested the impact of invisible support on physiological well-being. As such, the current study investigated how support visibility and self-efficacy interact to affect a support recipient’s cortisol reactivity during a speech preparation task.
Cortisol is a stress hormone secreted by the hypothalamic–pituitary–adrenal (HPA) axis (Cohen, Kessler, & Gordon, 1995). Cortisol was chosen as an indicator of physiological health for several reasons. First, chronic stress is a significant predictor of immune functioning and cardiovascular health, with small- to medium-effect sizes (e.g., Lagraauw, Kuiper, & Bot, 2015; Segerstrom & Miller, 2004). Importantly, cortisol is one of the primary pathways through which prolonged stress negatively impacts health (Cohen et al., 1995; Dhabhar, 2009; Golbidi, Frisbee, & Laher, 2015). Furthermore, the HPA axis is particularly attuned to social stimuli, including social support (Kirschbaum, Klauer, Filipp, & Hellhammer, 1995). Finally, salivary cortisol can be assessed noninvasively, minimizing the burden of data collection for participants.
Given previous research suggesting that self-efficacy moderates the link between social support and autonomy (Warner et al., 2011), we hypothesized that there would be an interaction between support visibility and self-efficacy such that participants who were primed to have high self-efficacy would experience stress in response to the speech task when they received visible, but not invisible, support. By contrast, participants who were primed to have low self-efficacy would experience stress in response to the speech task when they received invisible, but not visible, support.
Method
Participants
The sample included 74 students (42 women and 32 men) from a university in the southwestern United States. The majority of participants were White/European American (78.1%; Hispanic American, 15.1%; multiracial, 2.7%; Asian American, 2.7%; African American, 1.4%). Participants’ ages ranged from 18 to 31 (M = 19.83). Participants were obtained by convenience sampling and were recruited from public areas around campus and through online notices (i.e., via Facebook, e-mail, and university campus notices). As compensation, participants chose either to be entered into a raffle to win one of two US$50 giftcards or to receive two extra credit points in an introductory psychology course. We excluded upper-level psychology majors from the study because these students were likely to be familiar with false feedback manipulations, which could make it difficult to manipulate their self-efficacy.
Design
The present study used a 2 (self-efficacy: low self-efficacy vs. high self-efficacy) × 2 (support visibility: visible support vs. invisible support) between-groups, experimental design, to investigate the impact of support visibility and support recipients’ self-efficacy on cortisol levels. To ensure a similar number of men and women were included in each condition, participants were randomly assigned within gender to conditions. After using a false feedback procedure to manipulate self-efficacy, a confederate provided either visible or invisible support to participants. Then, all participants took part in the Trier Social Stress Task (Kirschbaum et al., 1993), a public speaking task that reliably produces increases in cortisol levels (Dickerson & Kemeny, 2004). Cortisol was measured three times at approximately 22 min intervals; the first assessment was a baseline cortisol measure, the second assessment measured stress anticipating the speech task, and the third assessment measured stress during the speech.
Materials
Health assessments
Because many health variables affect cortisol reactivity, interested participants completed an online eligibility survey before coming into the lab. The screening survey asked participants to self-report health conditions that have implications for HPA axis functioning. Specifically, we asked respondents a series of yes/no questions including: “Are you currently diagnosed with depression,” “Are you currently diagnosed as having anxiety?,” “Are you pregnant or nursing at the current time?,” “Do you work between the hours of 11 p.m. and 6 a.m.?,” “Do you smoke?,” “Do you regularly use other tobacco products (i.e., dip, chew, patch, etc.)?,” “Do you have a medical condition that impacts your hormones?,” and “Are you currently taking any medications (including birth control)?,” If participants answered yes to having a hormone condition or taking medication, we asked participants to list their specific medical condition and/or medication, which allowed us to determine implications for HPA axis functioning. Participants were excluded from participation if they reported a current diagnosis of depression or anxiety, nursing or being currently pregnant, working night shifts, smoking or using other tobacco products, having a history of hormone problems that impact cortisol, or taking HPA axis relevant medications.
In the first questionnaire given during the lab session, we asked several questions about sleep schedule (e.g., “What time did you get into bed last night?”), alcohol use (e.g., “How many alcoholic drinks have you had in the last 24 hr?”), caffeine intake (e.g., “How many caffeinated beverages have you had in the last 24 hr?”), tobacco use (e.g., “How many cigarettes have you smoked in the past 24 hr?”), exercise (e.g., “At least once a week, do you engage in any regular activity like brisk walking, jogging, bicycling, yoga, Pilates, etc., long enough to build up a sweat?”), height, and weight so that we could account for these variables in analyses. For women, we also asked whether they were using hormonal birth control.
False speech aptitude test
As part of the self-efficacy manipulation, we created a 19-question, multiple-choice test and told participants this exam was a measure of their speech-giving aptitude. Questions were loosely based on exams from various public-speaking classes but were altered so that there was not an obvious right or wrong answer (e.g., “When giving a speech, it is important to come across to the audience as all of the following except: (a) respectable, (b) trustworthy, (c) moral, and (d) important”). Four items from the public-speaking self-efficacy inventory (Betz et al., 2003) were also included at the beginning of the exam. To increase the face validity of the exam, we labeled it the Miero-Shoyt Speech Aptitude Test, or M-SAT. Participants responded to questions on a scantron to increase believability in the immediate feedback we provided to participants.
Salivary cortisol
To assess circulating levels of free cortisol, participants provided a series of four saliva samples via the Salivette (Sarstedt, Germany). Saliva samples were taken at standard intervals (approximately every 22 min) because there is a 20-min time lag from when cortisol enters the bloodstream to when it is present in saliva (Dickerson & Kemeny, 2004). To reduce participant reactivity to the sampling procedure and to teach participants how to use a Salivette, we took the first saliva sample 5 min after participants entered the lab. Because this sample was assessing cortisol levels released approximately 15 min prior to entering the lab, this first sample was discarded; only the last three samples taken were used in analyses. We determined cortisol concentrations, reported in microgram per deciliter, using SalimetricsLLC expanded range high sensitivity salivary cortisol enzyme immunoassay kits. As stated in kit instructions, all samples were frozen at −20°C until assayed. Each participant’s samples were assayed in the same batch. To ensure reliability, high and low control samples provided by SalimetricsLLC were also included in each batch. Samples had an average intra-assay reliability of 7.16% and an inter-assay reliability of 9.42%. Cortisol concentrations in participant samples ranged from .01 μg/dL to .80 μg/dL at Time 1 (M =.14 μg/dL, SD =.13), from .02 μg/dL to .80 μg/dL at Time 2 (M =.14 μg/dL, SD =.14), and from .02 μg/dL to .73 μg/dL at Time 3 (M =.16 μg/dL, SD = .14). A log10 transformation was performed on obtained cortisol values prior to analyses to normalize their distribution. All samples more than 3 SDs from the mean (n = 1) were winzorized. 1
Other self-report measures
While in the lab, participants completed a number of other self-report measures that were not relevant to the current project. For a full list of measures we assessed, please see supplemental materials.
Procedure
Prescreening
We directed prospective participants to a secure website where they were able to read detailed information about the study, including the consent form. At this point, participants were told that the purpose of the study was to investigate how different types of speech preparation affected cortisol levels before, during, and after giving a speech. Interested participants then completed the online health assessment (previously described); eligible participants were scheduled to come into the study lab. Lab sessions lasted about 1 hr and 15 min, and only one participant was scheduled for each session time. To control for diurnal patterns in cortisol secretion, all lab sessions were scheduled between 2:00 p.m. and 8:30 p.m., Monday through Friday (Kirschbaum & Hellhammer, 1994). Following standard protocol for cortisol sampling, participants were instructed not to eat or drink anything (except for water), brush their teeth, or exercise for at least 1 hr before their scheduled lab session.
Laboratory session
For each session, a researcher met the participant and confederate in a student lounge outside of the lab. Researchers always wore white lab coats when meeting with participants. Unless otherwise noted, confederates were treated just as if they were participants.
Upon arrival to the lab, the researcher confirmed that the participant had not had anything to eat or drink except for water in the past hour. Participants who did not fulfill this requirement were rescheduled for a different lab session. The researcher then walked the participants to the lab and gave participants a copy of the study consent form to sign. Participants were instructed not to listen to music, sleep, read, or use their cell phones throughout the course of the study, as these activities would influence their cortisol reactivity; however, if participants finished a segment of the study early, they could look through a neutral picture book provided in the study rooms. To ensure compliance, participants were asked to leave all of their belongings (including cell phones) with the researcher. The researcher then sent the participant into one of two identical study rooms and the confederate into the other room, allegedly to protect the confidentiality of the participants while they completed the initial questionnaires.
The researcher followed the participant into their study room and administered the first saliva sample using a Salivette, which includes a sterile piece of dental cotton. Participants were told to place the cotton in their mouth and that the researcher would be back to collect the sample in 2 min. Although participants were told not to chew on the cotton, the researcher explained that they could make a chewing motion to increase saliva production. This first sample was used to familiarize participants with the sampling procedure. As such, it was discarded and was not used in any analyses. After returning to the room and collecting the first saliva sample, the researcher gave the participant the false speech aptitude test, a scantron, and the first questionnaire, which included a health assessment. Before leaving the room, the researcher instructed participants to complete the speech aptitude test first using the scantron and to proceed to the questionnaire if they finished early. Participants had 5 min to complete the speech aptitude test. After 5 min, the researcher returned to the room to collect the speech aptitude test and scantron and told the participant that they would score the test while the participant completed the questionnaire. The participants then had 12 min to complete the first questionnaire.
After 12 min (a total of 22 min since participants entered the lab), the researcher returned to the room to collect the first questionnaire and to give the participant a Salivette for the second saliva sample, which was used to measure baseline cortisol levels. When the researcher returned to the room after 2 min to collect the second saliva sample, he or she also brought a chart showing the participant’s false score on the speech aptitude test. If the participant was assigned to a high self-efficacy condition, they were told that they had scored in the top 25% of undergraduate students from other comparable universities who had taken the test and that the following speech task should be easy for them. If the participant was assigned to a low self-efficacy condition, they were told that they had scored in the bottom 25% of undergraduate students from other comparable universities who had taken the test and that they may need some more practice preparing and giving speeches. During these explanations, the researcher showed the participant a chart that corresponded to the false feedback they were receiving to increase credibility of the reported scores.
At this point, the confederate was brought into the room for the speech preparation task. The researcher explained the speech task, and participants were given 15 min to individually prepare a 3-min persuasive speech about the benefits of higher education. The researcher told participants that these speeches would be recorded and then sent to professors in the communications department for evaluation. Finally, adapted from the procedures established by Bolger and Amarel (2007), the researcher asked if either the participant or the confederate had any questions. If the participant had been assigned to an invisible support condition, the confederate responded to the researcher by asking if the researcher had any suggestions for creating a good speech. The researcher responded that they should include a catchy beginning to draw the audience in, support the points they want to make with evidence, and end with a summary of their thesis. If the participant had been assigned to the visible support condition, the confederate said that although he or she did not have any questions, they would like to give the other participant some advice about making a speech that they had learned in a public speaking course. The confederate then told the participant the same advice given by the researcher in the invisible support conditions.
After the 15-min prep time was finished, the researcher accompanied the confederate out of the room, allegedly to give their speech, while the participant took 3 min to fill out the second questionnaire. The researcher then returned to the room to collect the second questionnaire and to administer the third saliva sample (collected 22 min after the baseline cortisol sample). After 2 min, the researcher collected the saliva sample and took the participant to a room with a webcam that was used to record their 3-min speech. Participants were required to speak for the full 3 min and were asked to continue talking even if they ran out of material.
After giving their speech, the participant was taken back to their original room and given the third questionnaire. This questionnaire took 20 min to complete and included measures that are not relevant to the current study. After collecting the third questionnaire, participants provided a fourth saliva sample that measured cortisol immediately after the conclusion of the speech (23 min after the second sample). Finally, the researcher debriefed the participant and explicitly explained the use of a confederate and false feedback. In light of this information, the researcher asked participants to consent to the use of their data for a second time; all participants provided this second consent.
Results
Preliminary analysis
A 2 (support visibility: invisible vs. visible) × 2 (self-efficacy: high vs. low) analysis of variance (ANOVA) was used to ensure that there were no preexisting differences in participants’ baseline cortisol levels. There were no main effects for support visibility (p = .72) or self-efficacy (p = .79). There was also not a support visibility × self-efficacy interaction (p = .47).
Manipulation check
We conducted a t-test to determine whether our manipulation of self-efficacy was effective. After receiving false feedback about their performance on the speech aptitude test, individuals in the high self-efficacy condition reported being more self-confident in their ability to give a speech in front of a class (M = 3.79) than did individuals in the low self-efficacy condition (M = 3.36; t(72) = 1.92, p = .030). Given the directional prediction of these mean differences (i.e., that individuals in the low self-efficacy condition would have lower self-efficacy than individuals in the high self-efficacy condition), a one-tailed p-value is reported.
Control model
Before investigating our hypotheses, we ran a repeated measures ANOVA to determine whether any health-related variables covaried with cortisol level. Body mass index, caffeine intake, alcohol use, average number of hours spent exercising per week, hormonal birth control use, days since last period, and timing of cortisol assessment were all entered into the model as predictors of cortisol levels. Only the interaction between timing of cortisol assessment and average number of hours spent exercising was marginally associated with cortisol (p = .09); as a conservative test, hours spent exercising as well as the interaction of hours spent exercising and time were included in all subsequent models. Although alcohol use did not interact with timing of cortisol assessment to predict cortisol levels, two participants had consumed an exceptionally high number of alcoholic drinks in the past 24 hr (i.e., 9 and 15 drinks, respectively). Given the association between alcohol and cortisol demonstrated in previous literature (e.g., Thayer, Hall, Sollers, & Fischer, 2006), we also controlled for alcohol intake as well as the interaction of alcohol intake and time in subsequent analyses.
Primary analysis
To test our hypothesis, we ran a repeated measures ANOVA to determine the effect of self-efficacy (1 = high; 2 = low), support visibility (1 = invisible; 2 = visible), and cortisol assessment timing (1 = baseline; 2 = anticipation stress; 3 = speech stress) on participants’ cortisol levels. There was a marginal main effect of cortisol assessment timing on cortisol levels, F(2, 132) = 2.59, p = .079,
Of particular interest to study hypothesis, support visibility did not interact with timing of cortisol assessment to predict cortisol levels, F(2, 132) = 0.05, p = .952,
Simple main effects were used to determine which experimental conditions (i.e., invisible support with high self-efficacy, invisible support with low self-efficacy, visible support with high self-efficacy, and visible support with low self-efficacy) caused significant changes in cortisol. That is, we used repeated measure ANOVAs to determine the main effect of cortisol assessment timing in all four experimental conditions separately, but simultaneously. As hypothesized, for participants who were primed to have high self-efficacy, there was a main effect of cortisol assessment timing in the visible support condition, F(2, 30) = 3.77, p = .034,
Cortisol reactivity for individuals with high versus low self-efficacy who received visible versus invisible support.
Discussion
We experimentally manipulated visibility of support and self-efficacy to determine their effect on support recipients’ cortisol reactivity. Just as self-efficacy moderates the support visibility–autonomy link (Warner et al., 2011), we hypothesized that there would be an interaction between support visibility and self-efficacy, such that participants who had been primed to have high self-efficacy would show increased cortisol reactivity to a stressful task when they received visible support, but not invisible support. By contrast, participants who had been primed to have low self-efficacy would show increased cortisol reactivity to a stressful task when they received invisible support, but not visible support. Results were consistent with our hypothesis.
Invisible support most effectively buffered stress for individuals who were primed to have high self-efficacy. This finding is theoretically consistent with previous research documenting psychological benefits of invisible support (Bolger & Amarel, 2007; Bolger et al., 2000; Shrout et al., 2006). Indeed, when individuals had the skills needed to accomplish a task, being aware of support had physiological consequences, likely because it interfered with autonomy and self-confidence (e.g., Bolger et al., 2000; Warner et al., 2011). Importantly, our study is one of only two studies that assessed the impact of invisible support using experimental methodology (see also Bolger & Amarel, 2007). Further, beyond the previously established benefits invisible support has for mood and mental health (e.g., Bolger et al., 2000), our study is the first to demonstrate physiological benefits of invisible support. Given that cortisol reactivity has implications for immune system functioning (Lovallo, 2005), this study offers initial evidence linking invisible support to health. Consequently, our study adds to a growing body of literature demonstrating that social relationships can have a profound impact on health (for review see Holt-Lunstad, Smith, & Layton, 2010) and further implicates the HPA axis as one of the pathways through which relationships’ impact on health occurs (e.g., Slatcher, Seluk, & Ong, 2015).
Although invisible support buffered stress for individuals primed to have high self-efficacy, visible support most effectively buffered stress for individuals who were primed to have low self-efficacy. The fact that people with low self-efficacy showed increased cortisol reactivity when they received invisible support is consistent with past research demonstrating that invisible support is not always effective (Grimes et al., 2013; Maisel & Gable, 2009). Just as individuals under high stress benefit more from visible, emotional support (Grimes et al., 2013), our results suggest that individuals with low self-efficacy benefit more from visible, instrumental support. One possible reason for this benefit is that people who already doubt their abilities to perform a task do not experience further efficacy declines from receiving support. For such individuals, receiving information provides them with resources they need to be successful. Future research should explore this possibility.
Self-efficacy’s moderating role in the support visibility–health link has interesting implications for individuals in close relationships. It suggests that to effectively support a friend or partner, one must be aware of whether the support recipient has high as opposed to low self-efficacy in the domain for which they are receiving support. Just as support is most effective when it matches a recipient’s needs and is responsiveness to a recipient’s levels of stress (Cutrona et al., 2007; Neff & Karney, 2005), our study suggests that support visibility should vary as a function of a recipient’s domain-specific self-efficacy. This type of responsive support could lead to the support recipient feeling more satisfied in their relationship as well as experiencing more positive long-term health outcomes (Reis, Clark, & Holmes, 2004).
Although the implications of this study are noteworthy, the present findings must be considered in light of the study’s limitations. First, although we have measured both anticipatory stress and the stress response to the speech task itself, we do not know how long it took for participants’ stress levels to return to baseline. That is, we do not have a measure of cortisol that taps into recovery. Given that cortisol recovery has important implications for health (Seeman & McEwen, 1996), this is an important avenue for future research. Second, the sample used for the present study consisted entirely of undergraduate students, primarily between the ages of 18 and 22. During the years of emerging adulthood, many people are still somewhat dependent on parents or other adults, and it is culturally acceptable for young adults to not fully support themselves. By contrast, middle-aged and older adults are expected to be more independent. These social factors might make it easier for young adults to receive help when they don’t know how to perform a task than it is for middle-aged and older adults. Therefore, future studies should investigate possible age differences in people’s responses to receiving support. Finally, post hoc power analysis revealed that our study was underpowered; our power was in the .60 range. Thus, it is possible that the effect sizes in our study are slightly inflated (Button et al., 2013). Given that our effect sizes are comparable to others in the field, these concerns are minimized (e.g., Thorsteinsson & James, 1999). Still, future research should replicate this study with a larger sample.
Limitations notwithstanding, the present study presents interesting avenues for future research. First, future research should investigate the physiological effects of visible and invisible support when coping with different types of stressors. The present study used a mild, controllable stressor. That is, participants were free to choose whether they completed the speech and how they performed the task. Given that the effectiveness of different types of support (e.g., instrumental support vs. emotional support) varies as a function of the controllability of the stressor (e.g., Cutrona & Russell, 1990), it logically follows that the effects of support visibility would differ for a stressor not within the person’s control (e.g., after being laid off from work). Second, the benefits and costs of visible support vary when predicting immediate as opposed to long-term health outcomes. Indeed, although visible support was effective immediately after romantic partners were discussing goals for future behavior, it was not associated with long-term goal facilitation (Grimes et al., 2013). As such, it would be interesting to use diary methodologies to examine fluctuations in diurnal cortisol slopes as a potential way to illuminate long-term health consequences associated with receiving visible and invisible support. This future research would be particularly advantageous given that diurnal patterns of cortisol are a more robust predictor of health outcomes than acute cortisol changes (e.g., Adam & Kumari, 2009). Finally, to gain a deeper understanding of the impact support visibility and self-efficacy has on cortisol, it is important to consider the relationship context in which support is provided (e.g., Wills, 1991). In our study, support was provided by a stranger, who had no knowledge of whether the individual could accomplish the task on their own. In the context of a close relationship, however, the support provider is often intimately familiar with the recipient’s needs and abilities. In that context, support may communicate something about the recipient’s competency in a way that support provided by a stranger does not.
Conclusion
The present research cautions that in order to provide effective support, support visibility should vary as a function of individual’s self-efficacy in a given domain. When someone believes that they have the ability to manage a stressor on their own, invisible support buffers physiological stress in a way that visible support does not. By contrast, when someone does not believe they have the ability to manage a task, knowing they have received support has physiological benefits. In short, in order to successfully circumnavigate the costs of support, support providers must match support visibility to the individual characteristics of the recipient.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
Notes
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
