Relationship Between Gender,Stress,and Quality of Social Support

Abstract

Background

Research suggests that gender differences exist in both stress and how social support is utilized and that the relationship between stress and social support may not be linear.

Methods

An internet survey of n = 1080 participants was conducted evaluating quality and quantity of social support, gender, age, and perceived stress and coping.

Results

Reported quality of social support, gender, and age significantly predicted perceived stress and that there was a curvilinear interaction between the quality of social support and gender which significantly predicted perceived stress.

Conclusion

The current findings supported Taylor’s Tend and Befriend theory that females have higher reported stress levels, a larger support network, and report more quality in their social support.

Practitioner points: • Males and females may manifest stress differently in their relationships. • When working with males and females in practice it may be important to understand the depth and breadth of their social networks and how they utilize those networks. • Females indicate higher levels of stress and greater social support quality. • It is important to understand that one’s social network can be an important source of support (a coping mechanism) but that it can also serve as a stressor in some cases.

Keywords

Gender stress social support curvilinear

Introduction

Social support

Taylor et al. (2002) has proposed that social support is especially important for females who respond to stress with a ‘tend and befriend’ approach in which friendship and social closeness are key components of stress management for females. Social support describes different aspects of an individual’s social world (House et al., 1988) and includes the total number of individuals in a person’s support system and the structure of that network (House, 1987). One of the mechanisms through which social support may ameliorate stress is the quality of the social relationships or the degree to which an individual feels that there is adequate emotional and instrumental support (Eurelings-Bontekoe et al., 1995; House, 1987).

Research suggests that there are gender differences in social support. These differences are found in both the quality and quantity of social support (Cohen & McKay, 1984; Foorman & Lloyd, 1986; Misra & McKean, 2000; Roxburgh, 1996; Thoits, 1984; Wallston et al., 1983). Males and females also differ in how social support is utilized (Taylor et al., 2000; Thoits, 1995). Zhou et al. (2017) analyzed friendship dyads and found that male participants were less likely to provide or seek social support and when they did engage in social support it was less likely to involve emotion focused support. Kahn and Antonucci (1980) reported that women tended to have larger and more supportive networks with a greater number of close partners compared to males. Acitelli and Anontucci (1994) further found that women were more likely to have frequent contact with those in their support network compared to males. These results have been replicated in diverse samples from college students (Eshbaugh, 2008) to a multinational sample which confirmed that women reported a larger number of confidants as well as a greater sense of concern from others compared to men (Dalgard et al., 2006).

Gender, age, and stress

In addition to differences in social support utilization and size, gender differences appear in reported stress levels as well. Research suggests gender differences in perceptions of stress utilization of resources to moderate effects of stressors (McDonough & Walters, 2001; Misra & McKean, 2000; Roxburgh, 1996). Across diverse samples including patients with psychiatric diagnoses, multiple sclerosis, and adolescents, females have reported higher stress and distress levels compared to males (e.g., American Psychological Association, 2012; Cohen & Williamson, 1988; Gitchel et al., 2011; Hewitt et al., 1992; Martin et al., 1995). Biological evidence also supports differences between males and females in both reactions of the hippocampal morphology (McEwen, 2004; Sapolsky, 2002) and related behavior (Bowman et al., 2003). Gender differences span the type of stressors experienced and the perception of them. In a large study that evaluated gender differences in stress, women reported more chronic and minor daily stress than men and perceived events in their lives as more negative and less controllable (Matud, 2004). Furthermore, females reported most distressing things to include family, mental health, and health-related stressors (Matud, 2004; Zuckerman, 1989), academic stressors (Misra & McKean, 2000), and interpersonal issues (Narayanan et al., 1999) whereas men were more likely to report relationship, finance and work-related events (Matud, 2004; McDonough & Walters, 2001) as their greatest stressors.

Gender, social support, and stress

Research suggests that gender and age differences exist in reported stress as well as in the type and utility of social support; however, it is not clear exactly how these variables together predict stress (for review see, Graham et al., 2006). Nor is it obvious how gender might interact with these variables to produce a differential effect on perceived stress levels because of a lack of direct comparisons in the same models. There are a number of studies that have looked at the relationship between gender and social support in predicting variables related to stress such as health behaviors (Martin et al., 2013), adjustment (Srivastava & Barmola, 2012), physical health (Shumaker & Hill, 1991) or job stress (Bellman et al., 2003; Spielberger & Reheiser, 1994) or they report these two things separately (Day & Livingstone, 2003). In a recent study, emerging adult females benefited from social support when asked about physical health outcomes while emerging adult males benefitted from social support when asked about self-reported depressive symptoms (Lee & Dik, 2017). Another study evaluated gender differences in the role of social support using structural equation modeling to compare the use of social support and coping in males and females and found different patterns (Asberg et al., 2008). One of their conclusions was that social support was a key moderator of adjustment in females but not in males. They also found that males were more likely to seek out social support as a way of coping with stress whereas females were more likely to use emotion-focused coping or distancing themselves (Asberg et al., 2008). However, this study investigated young adults and did not directly evaluate males and females in the same model. While they did find different patterns in males and females, the lack of ability to use the same model limited the strength of their conclusions that social support is a key moderator of adjustment in females but not in males. Additionally, this study evaluated adjustment rather than stress specifically. A third study evaluated occupational stress and found that social support mediated different occupational stressors for females than for males and that for males social support actually increased stress for some variables (Bellman et al., 2003). This study also used separate models to evaluate males and females and assumed a linear relationship between social support and stress. A fourth study, junior high students were found to have a gender difference in the moderating effect of social support between perceived stress and depressive symptoms with girls reporting improvements in depression with positive social support and boys reporting no differences regardless of changes in social support (Zhang et al., 2015). In a fifth study, qualitative analysis of strategies utilized by male and female emerging adults found that males are more likely to engage in strategies to help achieve self-control while females are more likely to prioritize social support that will help them achieve problem awareness (Martínez-Hernáez et al., 2016). With only four quantitative studies available, only three of which directly compared males and females in a single model, and in very specific populations (young adults or children), a single analysis is needed to understand whether gender interacts with social support to differentially predict perceived stress levels and utilizes a model which is flexible enough to account for the type of interaction that is occurring.

Curvilinear contribution of stress

Mallinckrodt et al. (2012) hypothesized that the relationship between stress and social support was curvilinear with particular emphasis on the idea that social support is not always positive and that at a certain point it may become detrimental. This was confirmed in their study of breast cancer patients for whom those in the most distressed group benefitted most from social support but those with less distress did not have much additional benefit from social support (Mallinckrodt et al., 2012). The argument has been made (see Mallinckrodt et al., 2012; Varvel et al., 2007) that the limitation of examining social support as a linear variable in bivariate correlations and linear regression analysis limits full understanding of how this variable works in mediating the stress response. A second study (Varvel et al., 2007) found a curvilinear relationship between social support from peers and perceived stress among a small sample of firefighters. Their results suggest that there may be a threshold for social support for which at a certain point, additional support does not continue to reduce stress levels. A separate but related study also found that social support demonstrates a curvilinear relationship with self-reported health (Borg et al., 2000).Taken together, it is possible that the scant literature on the influence of social support in male and female stress mediation may be enhanced by an investigation of these variables together in one analysis as well as evaluation of the possibility of a curvilinear relationship between social support and gender. The research finding both non-linear relationships between social support and adjustment or stress as well as the evidence that males and females differ in both social support and reported stress suggests that there may be a non-linear interaction between gender and social support as it relates to perceived stress.

The research primarily hypothesized that social support would interact in a non-linear way with gender to predict self-reported stress levels. To test this hypothesis, a stepwise multiple regression analysis was conducted which included a curvilinear interaction between social support and gender as a predictor along with linear terms. Since a simple multiple linear regression analysis with gender and social support as predictors would not be able to capture the essential component of the hypothesis that gender interacts with social support in predicting stress, a curvilinear component was needed. In addition, the curvilinear social support variable helps determine whether an increase in social support is beneficial to a point and then detrimental to perceived stress. The stepwise method allowed for the precise identification of which of the possible independent variables were adding to the predictability of perceived stress. A secondary hypothesis was that age would contribute to perceived stress which was included in the model as this is an important variable in understanding the relationship between stress and social support in males and females. Lastly, it was hypothesized that females would report more perceived stress, larger support networks and more social support quality. Because most of the previously mentioned studies were conducted in the United States, this study also will be conducted with participants from the United States.

Materials and method

This research was approved by the Institutional Review Board.

Participants

The survey was constructed in survey monkey (surveymonkey.com) and posted on several social psychology research sites (including www.socialpsychology.org, http://beta.in-mind.org/online-research, and http://psych.hanover.edu/research/exponnet.html). Generally, individuals accessing these websites are able to scroll through multiple different types of research to participate in and choose the survey(s) they want to complete based on the title and a short description of the research. Sites used for this research include the investigator’s name and affiliation. Some sites categorize the survey by type (i.e. gender issues). This survey garnered n = 2099 clicks through to the web survey but only n = 1203 subjects’ data were useable for reasons described below. First the responses were screened for missing data. Non-responders were defined as those responding to less than 50% of the survey questions. This identified n = 857 surveys that had the majority of questions not answered. Because internet research risks multiple attempts by single users, two methods were used to prevent possible duplication of individuals filling out the surveys. The first was a question at the beginning of the survey that asked subjects to indicate if they had taken the survey previously which no subjects reported. The second step involved analyzing repeat IP addresses based on the recommendation of Reips (2002). Though n = 128 identical IP addresses were identified, careful screening of the responses led to none of these being deleted. Because the survey contained some open-ended questions where subjects filled in the name and relation of support people, data from the same IP address could be visually screened for repeat answers. Careful inspection of repeated IP addresses revealed that none of the repeat IP addresses that were retained had similar answers on the open-ended or likert scale responses. Research has found that it is uncommon for an individual to take an online survey multiple times (Konstan et al., 2005). Several studies have compared in laboratory results with internet research results and found that results were the same but with larger sample sizes and power available in the internet studies (i.e. see Klauer et al., 2000). Finally, because multiple regression analysis was used, casewise deletion of missing data was most prudent. A total of n = 39 subjects were missing either gender, race, or age data and were subsequently deleted from the regression analysis. As only n = 135 participants were from non-US based countries and the assessments had not been standardized to other languages or cultures, only participants who indicated their country of residence as the United States were included. Most of the 135 countries had 1 to 5 participants and included countries such as Signapore (9), Pakistan (2), United Arab Emirates (2), Romania (1), Turkey (4), Spain (1), India (7), Bangladesh (5), Greece (2), etc. Because of the low counts in each country and lack of control as well as uncertainty about language or cultural differences, those responses were not included in the final sample. After screening, the final sample size then was n = 1080.

Participants mostly self-identified as European, Caucasian, or White-American (63.3%) with a representation of African-American or African decent (9.6%) and Hispanic or Latino (10.6%) and Asian or Asian American (8.5%) subjects as well. A very small portion identified as Native American (0.6%) while 6.4% identified as either “other” or as belonging to more than one category. While fewer males (n = 244) participated than females (n = 817), no significant difference in the distribution of males and females across ethnic categories existed, χ²(5, n = 1052) = 5.89, p > 0.05. Twenty-nine participants did not answer this question.

The most frequent age category for participants was the 18–25 year range (82%). Each of the other age categories (in 5 year increments) had between 0.1 and 7.2% of the sample. There was no significant difference in the distribution of males and females across the age categories (χ²(11, n = 1058) = 16.2, p > 0.05) nor in the distribution of ethnicity across age categories (χ²(55, n = 1068) = 46.2, p > 0.05). Most subjects reported their highest degree earned was a high school diploma (45.8%), with 16.5% receiving an Associate’s Degree, 15.9% receiving a Bachelor’s degree. Less than 3% reported not finishing high school (1.2%) or attaining either a Master’s degree (2.2%) or doctoral degree (0.9%).

Procedure

Subjects found links to the survey on several research related internet sites including socialpsychology.org, http://psych.hanover.edu/research/exponnet.html, and http://beta.in-mind.org/online-research/, and through facebook and twitter pages of those sites. After clicking through the link to the survey, subjects were taken to a page with the informed consent information and asked to ‘agree and continue’ or ‘disagree and discontinue’. The informed consent described the research and what participants would expect, “You will be asked to fill out a questionnaire which will ask for information related to who you talk to and how often you talk to other people about difficulties in your life, the kinds of experiences you have had in the last year, how many daily activities you are involved in, your gender, and questions about your perceptions of these things.” For those subjects choosing to continue, the next page asked subjects to indicate whether or not they had taken this survey previously. After this, subjects were taken to a page which asked general demographic questions followed by each of the measures including the Perceived Stress Scale-10, the Rhode Island Stress and Coping Scale, and a Social Support Questionnaire-6. At the end of the survey, a page thanking subjects for their participation, information about how to contact the IRB and/or primary researcher, and information about how to enter a raffle for an amazon.com gift card were provided.

Measures

Perceived Stress Scale-10 (PSS-10)

Following demographic information, all subjects were presented with the Perceived Stress Scale-10 (PSS-10) which is a ten item scale (Cohen & Williamson, 1988) used to measure appraised stress. Respondents use a 5-point Likert-type scale (0-never, 1- almost never, 2-sometimes, 3- fairly often, 4- very often) to indicate how often they felt in the way described in the item over the last month (i.e. “felt that you were unable to control the important things in your life”). Higher scores indicate higher perceived stress. The ten-item scale is a shortened version of the larger original 14-item scale (Cohen et al., 1983). Factor analysis of the original scale revealed that the 10-items that were retained loaded on the same factor (Cohen & Williamson, 1988). Additionally, the 10-item version has good reliability (coefficient alpha = 0.78) as well as adequate validity assessed through positive correlations with health behaviors and negative correlations to life satisfaction measures (Cohen & Williamson, 1988).

Social Support Questionnaire (SSQ-6)

Following the coping scale, subjects were asked to “think about people in their environment who provided them with help or support” in specific situations to assess social support. The brief version of the Social Support Questionnaire (SSQ-6) assessed both the perceived quantity and quality of social support (Sarason et al., 1987). For each of six scenerios (i.e. “counting on someone to console you when you are very upset”) subjects listed in open-ended responses the names and relationships of people who provided help or support. The number of individuals listed was tallied to provide the quantity of social support they received (SSQ-N). They also rated the quality of the support or how satisfied (SSQ-Q) they were with the support they received using a 5-point Likert-type scale (very dissatisfied, dissatisfied, slightly dissatisfied, satisfied, very satisfied) for the each scenario. The internal reliability (coefficient alpha) for the SSQ-6 ranged from 0.90 and 0.93 for both the quantity and quality of social support (Sarason et al., 1987). Among several validation studies, the SSQ-6 has been found to be negatively correlated with anxiety and depression (STAI and BDI) and positively correlated with the Perceived Stress Scales for family and friends (PSS) (Sarason et al., 1987).

Results

Data analysis

Both descriptive and inferential statistics were employed to understand the relationship between social support, gender, age, ethnicity, and perceived stress. The descriptive statistics (means and standard deviations) for each of the centered variables and the cronbach’s alpha are presented in Table 1.

Table 1.

Means, standard deviations, and internal reliability of dependent variable and quantitative predictors.

	M	SD	Range	Chronbach’s alpha
PSS-10	42.3	8.3	15–74	.802
SSQ-Quality	24.0	4.4	6–30	.882
SSQ-Number	21.3	12.8	0–48	.927

To test the hypothesis that social support would interact in a non-linear way with gender to predict self-reported stress levels stepwise multiple regression analyses were conducted to determine the variables that predicted perceived stress and the level of variance that was explained by the predictor variables: Gender, Age, Race, linear Gender * Social Support Size and Quantity interaction, curvilinear Gender * Social Support Size² and Quantity², Social Support Size² and Quantity². The linear interaction of Gender*Social support was included as a predictor to determine if the contribution of social support in predicting stress was different for males or females. Because of previous research suggesting a curvilinear relationship between gender and social support (Mallinckrodt et al., 2012; Varvel et al., 2007), the curvilinear predictor was also included; this is included in the model as Gender*Social Support². This would capture not just a gender difference in the contribution of social support (the linear interaction) but also the possibility that the contribution may not be linear (curvilinear interaction). A possible example of this would be that as social support increased for males it had a negative effect on stress but as social support increased for females it had a positive effect. This was done for both measures of social support using the reported quality or satisfaction with social support (SSQ-S) and the number of supportive persons (SSQ-N) with gender.

Preliminary analysis: Data screening, multivariate outliers

Multivariate outliers: The procedures outlined by Tabachnick and Fidell (1996) were used to screen for multivariate outliers. Five variables (quality of social support, quantity of social support, gender, age, and perceived stress) were used to calculate Mahalanobis distances for subjects. Exploration of these cases revealed no problem with data coding or inattentive responding and were retained for subsequent analysis. Multivariate normality and linearity were examined using scatterplots to evaluate the elliptical shapes of the distributions of each variable. Normality plots of all five variables were examined and revealed points close to the straight line indicating normal distributions of variables.

Preliminary data steps: Centering the variables and developing the linear model terms

In order to determine an effect of a curvilinear relationship between social support and gender in predicting perceived stress, a few preliminary steps were required. Each of the variables needed to be centered to reduce the impact of multicolinearity that can occur when two positive variables are multiplied together (as was done for the curvilinear analysis) and to improve the interpretability of the predictor variable in the regression analysis (Aiken & West, 1991; Judd & McClelland, 1989). A new centered variable was calculated by subtracting the mean from the score. For instance, for SSQ-N the mean SSQ-N score was subtracted from each individual SSQ-N score and placed in a new column. The linear interaction term was created by multiplying the centered social support term (SSQ-S and SSQ-N) by the gender variable. The linear interaction was used to determine if there were gender differences in the contribution of social support to the prediction of stress before any curvilinear interaction could be assessed.

Preliminary data steps: Developing the curvilinear model terms

The curvilinear interaction terms were calculated by multiplying the squared social support terms (SSQ-N and SSQ-S) by the gender variable. In this case, the squaring of the variable allowed for the assessment of a nonlinear relationship that would be able to capture the possibility that as social support increases it may have different effects on stress in males and females.

The multiple regression predictor equation

The stepwise multiple regression included the following predictor variables: gender, age, Social Support Quality (SSQ-Q), Social Support Quantity (SSQ-N), the linear effect of gender and social support quantity (SSQ-N*gender), the linear effect of social support quantity and gender (SSQ-Q*gender), the curvilinear effect of social support quantity and gender (SSQ-N²*gender) and the curvilinear effect of social support quality and gender (SSQ-Q²*gender). The predicted, dependent variable was Perceived Stress (PSS-10). The model, utilizing linear and curvilinear predictors, allowed for assessment of additional predictability that the linear social support by gender interaction and that the curvilinear social support by gender interaction provided. The variance influence factor (VIF) of each variable added into the predictor equation was evaluated to assess for potential multicolinearity.

Multiple regression results: Predicting stress

Hypothesis one was supported with evidence from the stepwise multiple regression analysis which indicated that Social Support Quality (SSQ), Gender, Age, and the curvilinear interaction between Social Support Quality and Gender (SSQ-Q²*gender) significantly predicted Perceived Stress (PSS-10), R² = .111, R²_adj = 0.107, F(4, 1079) = 33.4, p < 0.001. The model accounted for 11.1% of variance in Perceived Stress. A summary of the regression model is presented in Table 2. In addition, bivariate and partial correlation coefficients between each predictor and the dependent variable (PSS-10) are presented in Table 3. Ethnicity, SSQ-N, SSQ-N*gender, SSQ-Q*gender, SSQ-N²*gender did not significantly contribute to the predicted stress levels and were excluded from the regression equation. Because measures were taken to reduce or eliminate collinearity (i.e. centering the variables), VIFs for each of the variables entered into the equation were in the range of 1.00 – 1.24 which were well below even the most conservative recommendations that VIF values fall below values of 4 (Pan & Jackson, 2008).

Table 2.

Model summary.

Model step^a	R	R²	R²_adj	Change statistics
Model step^a	R	R²	R²_adj	Δ R²	Δ F	df1	Δ df2	Δ p
1. SSQ-Q	.265	.070	.069	.070	81.270	1	1078	.000
2. Gender	.315	.099	.098	.029	34.759	1	1077	.000
3. Age	.325	.105	.103	.006	7.585	1	1076	.006
4. SSQ²*gender	.333	.111	.107	.005	6.214	1	1075	.013

^aEach variable was added at each subsequent step. For instance, SSQ-Q was added initially, then gender was added additionally into the equation at step 2 such that SSQ-Q and gender were entered into the equation, followed by age, then the curvilinear gender by social support quality interaction (SSQ²*gender).

Table 3.

Coefficients for final model.

	B	β	t	p	Bivariate r	Partial r
Constant	56.76
SSQ-Quality	−.609	−.306	9.840	.000	−.265	−.288
Gender	3.915	.198	6.584	.000	.151	.197
Age	−.492	−.082	2.828	.005	−.069	−.088
SSQ-Q²*Gender	−.023	−.080	2.493	.013	.073	−.076

B indicates beta weights for unstandardized regression coefficients for variables entered into the model including SSQ-Quality, Gender, Age, and the non-linear interaction between SSQ²*Gender. β weights are the standardized coefficients based on z-scores with a mean of 0 and a standard deviation of 1 and can be used to create a prediction equation for standardized variables. The t-values for each variable entered and subsequent p-value are presented. In the last two columns are the bivariate and partial correlations of each entered variable with the PSS-total score.

Taken together these results supported the primary hypothesis that gender, social support quality, and the curvilinear interaction between social support quality and gender would predict stress levels. The results confirmed that there was at least one curvilinear relationship between social support and gender that predicted stress levels. The second hypothesis that age would contribute to the prediction equation was supported as age was a significant predictor.

Gender differences in perceived stress

Consistent with the last hypothesis that females would report more stress than males, analysis of gender differences in reported perceived stress indicated that females (n = 817, Mean PSS-10 = 42.7, SD = 7.9) reported significantly more stress than males (n = 244, Mean PSS-10 = 39.8, SD = 8.8), t(1059) = 5.0, p < 0.001. Additionally, females reported significantly higher social support quality (SSQ-Q M = 24.2, SD = 4.0) than males (M = 23.4, SD = 4.7, t(1059) = 2.4, p < 0.05) and females reported significantly greater levels of social support quantity (M = 22.6, SD = 12.3) than males (M = 17.4, SD =13.8, t(1059) = 5.6, p < 0.001).

Gender differences in social support quality

Additional analysis of gender differences in social support quality (SSQ-Q) were carried out to parse out what might be driving the curvilinear influence in the predictor equation. There was a significant differences between males and females in social support quality (SSQ-Q) with females reporting significantly higher SSQ-Q, M = 24.2, SD = 4.2 compared to males M = 23.3, SD = 5.0, t(932) = 2.4, p = .02.

Age differences in perceived stress

Because age category was an important predictor variable, further analysis was done to understand the relationship between age and perceived stress. There was a significant age group difference in the total perceived stress, F(10, 1167) = 2.2, p < 0.05. Means and standard deviations for each age category and gender on the PSS-10 are reported in Table 4. Interestingly, the 51–55 age group reported significantly lower perceived stress than most of the other age categories (18–25, 26–30, 31–35, 41–45, and 71+) as determined by LSD post-hoc analysis.

Table 4.

Average (and standard deviation) perceived stress scores across age and gender categories.

Age category	Perceived stress scores
	Male		Female		Total
	Mean (SD)	n	Mean (SD)	n	Mean (SD)	n
18–25	39.9 (8.8)	216	43.1 (8.1)	643	42.3 (8.4)^a	859
26–30	37.5 (8.3)	13	42.6 (6.1)	62	41.7 (6.8)^a	75
31–35	38.1 (11.6)	3	42.2 (6.7)	35	41.9 (7.1)^a	38
36–40	33.5 (7.8)	2	39.3 (8.4)	27	38.9 (8.3)^a	29
41–45	45.0 (5.3)	3	41.9 (6.2)	17	42.3 (6.0)^a	20
46–50	38.0 (6.0)	3	38.8 (5.9)	11	38.6 (5.7)	14
51–55	23.0 (0)	1	35.0 (6.6)	3	32.0 (8.0)^a	4
56–60		0	34.0 (11.3)	3	34.0 (11.3)	3
61–65	49.0 (0)	1	30.5 (0.7)	2	36.7 (10.7)	3
66–70		0	42.3 (9.3)	3	42.3 (9.3)	3
71+		0	50.5 (0.7)	2	50.5 (0.7)^a	2
Totals	39.8 (8.8)	244	42.7 (7.9)	808^b	42.3 (8.3)	1050^c

^aPost-hoc analysis with LSD indicated that the age range between 51–56 was significantly lower than the 18–25, 26–30, 31–35, 41–45, and 71+ age ranges.

^bNot all females also provided data about their age, thus the n = 808 is slightly smaller than the total number (n = 817) of females who participated in the study.

^cNot all participants provided age category data, thus the n = 1050 is smaller than the total number (n = 1080) of participants in the overall study.

Discussion

Stress is an inevitable aspect of today’s interactions with the world. However, individual behavioral, cognitive, and social resources can help to reduce the negative impact of those stressors. While there are known gender differences in both perceived levels of stress and social support, this is the first study to examine the possible contribution of a non-linear relationship between gender and social support in predicting perceived stress in the same model. This study confirmed the hypothesis that there was a curvilinear relationship between social support and gender in predicting stress. Specifically, the quality but not the quantity of the social support interacted both linearly and curvilinearly to predict stress. This suggests that the quality of the support given in situations may be beneficial to a certain point and this peak of benefit may be different for males and females. Similar to Asberg et al. (2008), a difference in social support by gender was found to predict stress. Also females reported significantly more stress as well as more quality in their social support networks and more supportive individuals on whom they could rely.

These results support Taylor’s (2006) theory of Tend and Befriend and suggests that males and females are utilizing this important coping mechanism differently to buffer stress. Taylor (2006) has suggested that group affiliation and interaction has evolved as a necessary survival mechanism for our species. Taylor’s Tend and Befriend theory adds to the original conceptions of stress response by including a need for social support during stressful encounters (Taylor, 2006). Specifically, males and females may rely on social support in very different ways for coping with stress. Social support can be seen as an additional resource to Cannon’s (1932) ‘fight or flight’ response whereby an individual either combats a stressor head on or flees from the situation. Taylor et al. (2000) has suggested that while the traditional fight or flight pattern may describe the male approach to stress response, tending and befriending (relying on social support networks) may explain the female approach to stress response. The current findings endorse this theory in findings that females have a larger support network, perceive more quality in their social support, and that social support quality differentially predicts stress levels for males and females.

It was also found that females had significantly higher levels of perceived stress than males. This is consistent with previous findings of a gender difference in perceived stress (e.g. Alfven et al., 2008; Lavoie & Douglas, 2012).

Most importantly was the support for the hypothesis that there would be a curvilinear interaction between gender and social support that predicted perceived stress. In previous research where a curvilinear relationship between social support and other measures were found, researchers were evaluating very specific populations of breast cancer patients (Mallinckrodt et al., 2012) or firefighters (Varvel et al., 2007). The current study was done with a sample of subjects who ranged in age from eighteen years old to ninety-one years old. The current sample did not have limitations of particular disease states or occupation but was open to anyone interested in the topic as reflected in the age range who participated.

Age was also a significant factor in the stress equation. Additional analysis indicated that the 51–55 year old age group reported the lowest perceived stress especially compared to the youngest age groups who reported the highest perceived stress. This is consistent with previous research suggesting that stress varies considerably across the lifespan (Graham et al., 2006).

The predictive model itself was designed to specifically test only for the predictive influences of social support, gender, and age. Yet, there are many factors that are known to influence perceived stress. In our study, the model accounted for around 11% of the variance in predicting perceived stress. According to Lazarus’s theory, it is the combination of all of the behavioral, emotional, cognitive, and social resources that are available to an individual that predicts their reaction to the environmental stressors. Thus, there are a number of other salient factors that may be involved in the relationship between social support, gender, and coping including the number and types of stressors that are currently being faced, the emotional resources that are available and utilized, the psychological resources that are available and utilized, and the cognitive methods employed to combat stressful situations. A brief review of recent research on predicting perceived stress indicated that variables as diverse as adult and god attachment (Reiner et al., 2010), emotionality and sociability (Hintsanen et al., 2011), and mindfulness (Ghorbani et al., 2010) have recently been evaluated for their prediction of stress and coping. Thus coming up with a comprehensive model to accommodate all the possible predictors of perceived stress would be statistically inefficient and theoretically difficult to incorporate all the possible mediators. In addition, not all perceived stress is created equally and differences may separate based on the type of stress that is reported such as predictable or controllable versus uncontrolled stress.

Limitations

This was a convenience sample of individuals who opted to go to a website and participate in the research. There was a significant limitation that only a few websites were utilized and these were websites generally frequented by people interested in psychological research or individuals enrolled in psychology courses who have a course requirement to participate in research studies. While the sample was large and spanned a reasonable age range and representative ethnicities, it had higher representation of younger, Caucasian participants. A broader, more generalized sampling procedure in future studies would provide more information about the relationship of these variables in other populations and be more generalizable. Additionally, the sample is only generalizable to participants in the United States. Broader, more diverse sampling with populations outside of the United States may provide rich information about friendships and stress in other countries and cultures. While the sample size is adequate for power, a larger sample size may provide the opportunity to further explore the impact of variables such as race/ethnicity and age. In addition, a more random sample may provide more diversity and a clearer understanding of the relationships of these variables.

Clinical relevance

The findings of this study suggest that perceived stress is influenced by a number of factors. While the influence is small (11%), it is predicted by age, gender, the quality of social support interactions and the curvilinear interaction between gender and social support and gender meaning that the influence of social support is different for males and females. The results suggest that females and younger participants have higher stress levels. Additionally gender significantly interacts with social support. It is possible that while social support is a known protective factor that ameliorates stress, one also has to be a friend in return and that the stronger the quality of friendships one has can add additional pressures to be a friend in return. In a qualitative analysis of friendships, Moremen (2008) finds that women report balance and reciprocity are important features in friendships they have. Women expect that they will be getting as much as they will be giving in a friend relationship and this might be putting additional strain and stress on women especially if they have large social support networks. It may then be important to discuss prioritizing one’s own needs as well as developing and practicing stress reduction techniques for individuals who are experiencing high levels of stress. While the benefits of a strong social support network have been well documented, if the benefits start to become a source of additional stress, it may be helpful to evaluate potential sources of stress in one’s network.

Future directions

There are a number of future studies which could further our understanding of this work. Future research could replicate this study with a more ethnically diverse sample to understand the relationship of these variables in a sample that could generalize to a larger population. Moreover this study could be expanded to include participants from other countries with the modification of measures to other languages where appropriate to provide further evidence if these relationships generalize to broader populations. Additionally, qualitative interviews with both males and females about the quality of their friendships to understand how those relationships affect them and how that influences both positively and negatively their stress levels is warranted. While much research has been undertaken to understand the positive effects of friendship on stress levels, little has been done to evaluate the potential negative effects that being a friend may have.

Conclusion

Overall the findings support a gender difference in perceived stress and social support quality that warrants further exploration of a broader spectrum of coping variables and age, ethnic, and cultural differences. The research adds to the body of literature on gender differences in stress and social support. Specifically, this research suggests that males and females report different levels of stress and differences in social support. Additionally, the findings confirm that social support is helpful to a point and then may either not add any additional benefit or may become detrimental and that this is different for males and females. This supports Taylor’s theory that ‘befriending’ is an important stress coping mechanism and suggests that males and females differ in how social support mediates stress.

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.

ORCID iD

Meredith Kneavel

Author Biography

Meredith Kneavel: is a Profesor in the Department of Public Health and Nutritionand Associate Dean in the School of Nursing and Health Sciences at La Salle University in Philadelphia, PA.She received her PhD in Psychology with a concentration in Biopsychology from the Graduate School of the City University of New York. Her current research spans three areas: student athlete health, stress, and pedagogy. A main focus of student-athlete health has been understanding concussion and concussion reporting. Her and a colleague developed and assessed a novel peer concussion education program which was supported through grants from the NCAA and Department of Defense. As part of her work with concussions, she also serves as the Director of Assessment and Research for the Center for Concussion Education and Research at Chestnut Hill College. Her pedagogical research focuses on understanding effective techniques in the classroom and particularly ways to enhance learning and understanding in statistics. She has also studied gender differences and the effects of stress on learning and memory.

References

Acitelli

Antonucci

(1994). Gender differences in the link between marital support and satisfaction in older. Journal of Personality and Social Psychology, 67(4), 688–698. https://doi.org/10.1037/0022-3514.67.4.688

Aiken

L. S.

West

S. G.

(1991). Multiple regression: Testing and interpreting interactions. SAGE.

Alfven

Östberg

Hjern

(2008). Stressor, perceived stress and recurrent pain in Swedish schoolchildren. Journal of Psychosomatic Research, 65(4), 381–387. https://doi.org/10.1016/j.jpsychores.2008.07.004

American Psychological Association. (2012). Stress by gender: A stressful imbalance [Press release]. http://www.apa.org/news/press/releases/stress/2012/gender.aspx

Asberg

K. K.

Bowers

Renk

McKinney

(2008). A structural equation modeling approach to the study of stress and psychological adjustment in emerging adults. Child Psychiatry and Human Development, 39(4), 481–501. https://doi.org/10.1007/s10578-008-0102-0

Bellman

Forster

Still

Cooper

(2003). Gender differences in the use of social support as a moderator of occupational stress. Stress and Health, 19(1), 45–58. https://doi.org/10.1002/smi.954

Bowman

Beck

Luine

(2003). Chronic stress effects on memory: Sex differences in performance and monoaminergic activity. Hormones and Behavior, 43(1), 48–59. https://doi.org/10.1016/S0018-506X(02)00022-3

Cohen

Kamarck

Mermelstein

(1983). A global measure of perceived stress. Journal of Health and Social Behavior, 24(4), 385–396. https://doi.org/10.2307/2136404

Cohen

McKay

(1984). Social support, stress, and the buffering hypothesis: An empirical review. In Baum

Singer

Taylor

(Eds.), Handbook of psychology and health (Vol. 4, pp. 253–267). Erlbaum.

10.

Cohen

Williamson

G. M.

(1988). Perceived stress in a probability sample of the United States. In Spaceapan

Oskamp

(Eds.), The social psychology of health (pp. 31–67). SAGE.

11.

Dalgard

O. S.

Dowrick

Lehtinen

Vazquez-Barquero

J. L.

Casey

Wilkinson

Ayuso-Mateos

J. L.

Page

Dunn

, & ODIN Group. (2006). Negative life events, social support and gender difference in depression. Social Psychiatry and Psychiatric Epidemiology, 41(6), 444–451. https://doi.org/10.1007/s00127-006-0051-5

12.

Day

A. L.

Livingstone

H. A.

(2003). Gender differences in perceptions of stressors and utilization of social support among university students. Canadian Journal of Behavioural Science/Revue Canadienne Des Sciences du Comportement, 35(2), 73–83. https://doi.org/10.1037/h0087190

13.

Dickinson

W. J.

Potter

G. G.

Hybels

C. F.

McQuoid

D. R.

Steffens

D. C.

(2011). Change in stress and social support as predictors of cognitive decline in older adults with and without depression. International Journal of Geriatric Psychiatry, 26(12), 1267–1274. https://doi.org/10.1002/gps.2676

14.

Eshbaugh

E. M.

(2008). Brief Report: Gender, social support, and loneliness among residence hall students. The Journal of College and University Student Housing, 35(2), 25–33.

15.

Eurelings-Bontekoe

E. H.

Diekstra

R. F.

Verschuur

(1995). Psychological distress, social support and social support seeking: A prospective study among primary mental health care patients. Social Science & Medicine (1982), 40(8), 1083–1089. https://doi.org/10.1016/0277-9536(94)00182-S

16.

Foorman

Lloyd

(1986). The relationship between social support and psychiatric symptomatology in medical students. The Journal of Nervous and Mental Disease, 174(4), 229–239. https://doi.org/10.1111/1467-8721.00073

17.

Ghorbani

Cunningham

C. J.

Watson

P. J.

(2010). Comparative analysis of integrative self-knowledge, mindfulness, and private self-consciousness in predicting responses to stress in Iran. International Journal of Psychology: Journal International de Psychologie, 45(2), 147–154. https://doi.org/10.1080/00207590903473768

18.

Gitchel

W. D.

Roessler

R. T.

Turner

R. C.

(2011). Gender effect according to item directionality on the perceived stress scale for adults with multiple sclerosis. Rehabilitation Counseling Bulletin, 55(1), 20–25. https://doi.org/10.1177/0034355211404567

19.

Graham

J. E.

Christian

L. M.

Kiecolt-Glaser

J. K.

(2006). Stress, age, and immune function: Toward a lifespan approach. Journal of Behavioral Medicine, 29(4), 389–444. https://doi.org/10.1007/s10865-006-9057-4

20.

Hewitt

P. L.

Flett

G. L.

Mosher

S. W.

(1992). The perceived stress scale: Factor structure and relation to depression symptoms in a psychiatric sample. Journal of Psychopathology and Behavioral Assessment, 14(3), 247–257. https://doi.org/10.1007/BF00962631

21.

Hintsanen

Hintsa

Widell

Kivimäki

Raitakari

O. T.

Keltkangas-Järvinen

(2011). Negative emotionality, activity, and sociability temperaments predicting long-term job strain and effort–reward imbalance: A 15-year prospective follow-up study. Journal of Psychosomatic Research, 71(2), 90–96. https://doi.org/10.1016/j.jpsychores.2011.02.012

22.

House

J. S.

(1987). Social support and social structure. Sociological Forum, 2(1), 135–147. https://doi.org/10.1007/BF01107897

23.

House

J. S.

Umberson

Landis

K. R.

(1988). Structures and processes of social support. Annual Review of Sociology, 14(1), 293–318. https://doi.org/10.1146/annurev.so.14.080188.001453

24.

Judd

C. M.

McClelland

G. H.

(1989). Data analysis: A model comparison approach. Harcourt Brace Jovanovich.

25.

Kahn

Antonucci

(1980). Convoys over the life course: Attachment, roles, and social support. In Baltes

Brim

, Life-Span Development and Behavior (Vol. 3). Academic Press.

26.

Klauer

K. C.

Musch

Naumer

(2000). On belief bias in syllogistic reasoning. Psychological Review, 107(4), 852–884. https://doi.org/10.1037/0033-295X.107.4.852

27.

Konstan

J. A.

Rosser

B. R.

Ross

M. W.

Stanton

Edwards

W. M.

(2005). The story of subject naught: A cautionary but optimistic tale of Internet survey research. Journal of Computer-Mediated Communication, 10(2), article 11. https://doi.org/10.1111/j.1083-6101.2005.tb00248.x

28.

Lavoie

J. A.

Douglas

K. S.

(2012). The perceived stress scale: Evaluating configural, metric and scalar invariance across mental health status and gender. Journal of Psychopathology & Behavioral Assessment, 34(1), 48–57. https://doi.org/10.1007/s10862-011-9266-1

29.

Lee

C. S.

Dik

B. J.

(2017). Associations among stress, gender, sources of social support, and health in emerging adults. Stress and Health: Journal of the International Society for the Investigation of Stress, 33(4), 378–388. https://doi.org/10/1002/smi.2722

30.

Mallinckrodt

Armer

J. M.

Heppner

P. P.

(2012). A threshold model of social support, adjustment, and distress after breast cancer treatment. Journal of Counseling Psychology, 59(1), 150–160. https://doi.org/10.1037/a0026549

31.

Martin

Critelli

Doster

Powers

Purdum

Doster

Lambert

(2013). Cardiovascular risk: Gender differences in lifestyle behaviors and coping strategies. International Journal of Behavioral Medicine, 20(1), 97–105. https://doi.org/10.1037/a0026549

32.

Martin

R. A.

Kazarian

S. S.

Breiter

H. J.

(1995). Perceived stress, life events, dysfunctional attitudes, and depression in adolescent psychiatric inpatients. Journal of Psychopathology and Behavioral Assessment, 17(1), 81–95. https://doi.org/10.1007/BF02229205

33.

Martínez-Hernáez

Carceller-Maicas

DiGiacomo

S. M.

Ariste

(2016). Social support and gender differences in coping with depression among emerging adults: A mixed-methods study. Child and Adolescent Psychiatry and Mental Health, 10(1), 2. https://doi.org/10.1186/s13034-015-0088-x

34.

Matud

M. P.

(2004). Gender differences in stress and coping styles. Personality and Individual Differences, 37(7), 1401–1415. https://doi.org/10.1016/j.paid.2004.01.010

35.

McDonough

Walters

(2001). Gender and health: Reassessing patterns and explanations. Social Science & Medicine (1982), 52(4), 547–560. https://doi.org/10.1016/S0277-9536(00)00159-3

36.

McEwen

(2004). How sex and stress hormones regulate the structural and functional plasticity of the hippocampus. In Gazzaniga

M. S.

(Ed.), The cognitive neurosciences (3rd ed., pp. 171–181). Cambridge

37.

Misra

McKean

(2000). College students’ academic stress and its relation to their anxiety, time management, and leisure satisfaction. American Journal of Health Studies, 16(1), 41–51.

38.

Moremen

R. D.

(2008). The downside of friendship: Sources of strain in older women’s friendships. Journal of Women & Aging, 20(1–2), 169–187. https://doi.org/10.1300/J074v20n01_12

39.

Narayanan

Menon

Spector

(1999). Stress in the workplace: A comparison of gender and occupation. Journal of Organizational Behavior, 20(1), 63–73. https://doi.org/10.1002/(SICI)1099-1379(199901)20:1<63::AID-JOB873>3.0.CO;2-J

40.

Pan

Jackson

R. T.

(2008). Ethnic difference in the relationship between acute inflammation and serum ferritin in US adult males. Epidemiology and Infection, 136(3), 421–431. https://doi.org/10.1017/S095026880700831X

41.

Reiner

S. R.

Anderson

T. L.

Hall

M. E.

Hall

T. W.

(2010). Adult attachment, God attachment and gender in relation to perceived stress. Journal of Psychology and Theology, 38(3), 175–185. https://doi.org/10.1177/009164711003800302

42.

Reips

U.-D.

(2002). Standards for Internet-based experimenting. Experimental Psychology, 49(4), 243–256. https://doi.org/10.1026//1618-3169.49.4.243

43.

Roxburgh

(1996). Gender differences in work and well-being: Effects of exposure and vulnerability. Journal of Health and Social Behavior, 37(3), 265–277. https://doi.org/10.2307/2137296

44.

Sapolsky

(2002). Chicken, egg, and hippocampal atrophy. Nature Neuroscience, 5(11), 1111–1113. https://doi.org/10.1038/nn1102-1111

45.

Sarason

I. G.

Sarason

B. R.

Shearin

E. N.

Pierce

G. R.

(1987). A brief measure of social support: Practical and theoretical implications. Journal of Social and Personal Relationships, 4(4), 497–510. https://doi.org/10.1177/0265407587044007

46.

Selye

(1950b). The physiology and pathology of exposure to stress. Acta, Inc.

47.

Shumaker

S. A.

Hill

D. R.

(1991). Gender differences in social support and physical health. Health Psychology: Official Journal of the Division of Health Psychology, American Psychological Association, 10(2), 102–111. https://doi.org/10.1037/0278-6133.10.2.102

48.

Spielberger

C. D.

Reheiser

E. C.

(1994). The job stress survey: Measuring gender differences in occupation stress. Journal of Social Behavior & Personality, 9(2), 199–218.

49.

Srivastava

S. K.

Barmola

K. C.

(2012). Social support and adjustment of students. Social Science International, 28(2), 303–317.

50.

Tabachnick

Fidell

(1996). Using multivariate statistics (3rd ed.). HarperCollings.

51.

Taylor

S. E.

(2006). Tend and befriend: Biobehavioral bases of affiliation under stress. Current Directions in Psychological Science, 15(6), 273–277. https://doi.org/10.1111/j.1467-8721.2006.00451.x

52.

Taylor

S. E.

Dickerson

S. S.

Klein

L. C.

(2002). Toward a biology of social support. In Snyder

C. R.

Lopez

S. J.

(Eds.), Handbook of positive psychology (pp. 556–562).00 Oxford University Press.

53.

Taylor

S. E.

Klein

L. C.

Lewis

B. P.

Gruenewald

T. L.

Gurung

R. A. R.

Updegraff

J. A.

(2000). Biobehavioral responses to stress in females: Tend-and-befriend, not fight-or-flight. Psychological Review, 107(3), 411–430. https://doi.org/10.1037/0033-295X.107.3.411

54.

Thoits

P. A.

(1984). Explaining distributions of psychological vulnerability: Lack of social support in the face of life stress. Social Forces, 63(2), 453–481. https://doi.org/10.2307/2579057

55.

Thoits

P. A.

(1995). Stress, coping, and social support processes: Where are we next? What next? Journal of Health & Social Behavior, 36, 53–79. https://doi.org/10.2307/2626957

56.

Varvel

S. J.

Shannon

J. K.

Tager

Bledman

R. A.

Chaichanasakul

Mendoza

M. M.

Mallinckrodt

(2007). Multidimensional, threshold effects of social support in firefighters: Is more support invariably better? Journal of Counseling Psychology, 54(4), 458–465. https://doi.org/10.1037/0022-0167.54.4.458

57.

Wallston

B. S.

Alagna

S. W.

DeVellis

B. S.

DeVellis

R. F.

(1983). Social support and physical health. Health Psychology, 4, 367–391. https://doi.org/10.1037/0278-6133.2.4.367

58.

Zhang

Yan

Zhao

Yuan

(2015). The relationship between perceived stress and adolescent depression: The roles of social support and gender. Social Indicators Research, 123(2), 501–518. https://doi.org/10.1007/s11205-014-0739-y

59.

Zhou

Heather

Di Cesare

Ryder

A. G.

(2017). Ask and you might receive: The actor-partner interdependence model approach to estimating cultural and gender variations in social support. European Journal of Social Psychology, 47(4), 412–428. https://doi.org/10.1002/ejsp.2251

60.

Zuckerman

D. M.

(1989). Stress, self-esteem, and mental health: How does gender make a difference? Sex Roles, 20(7–8), 429–444. https://doi.org/10.1007/BF00288001