Helping may be Harming: unintended negative consequences of providing social support

Abstract

While social support is generally considered a helpful resource for employees, it can also serve as a job stressor. Unhelpful workplace social support (UWSS) is any action taken by a supervisor and/or colleague that the recipient believes was intended to benefit him or her but is perceived as unhelpful or harmful. Two studies, one qualitative and one quantitative, identified types of UWSS and demonstrated that unhelpful support can operate as a job stressor in relating to strains. In Study 1, critical incidents were collected from 116 employees, and a content analysis revealed 11 distinct categories of UWSS. In Study 2, the taxonomy of UWSS was further refined using quantitative methods. Results of two samples (176 diverse employees and 496 registered nurses) demonstrate that UWSS is associated with higher job-related negative affect, lower competence-based self-esteem, lower coworker satisfaction, higher work-related burnout, higher organisational frustration, and more physical symptoms (e.g. headache, nausea, and fatigue) among recipients. Together, the studies demonstrate that unhelpful workplace social support is a meaningful job stressor worthy of further investigation.

Introduction

Work stress imposes an estimated $187 billion societal burden in the United States (Hassard, Teoh, Visockaite, Dewe, & Cox, 2018). Fortunately, research suggests the detrimental effects of job stress can be at least partially buffered by the receipt of social support (e.g. Cohen & Wills, 1985). While many researchers and organisations advocate increasing social support in the workplace to reduce job stress and/or its negative impact, social support does not always alleviate stress. The purpose of this research is to gain a heightened understanding of a counterintuitive role of social support, that is, social support as a job stressor.

The overall goal of this research is to provide meaningful insights into failures of workplace social support. Using a mixed method approach, the research aims to provide an overarching categorisation scheme to organise research on unhelpful workplace social support. Researchers have long recognised that “[i]n order for any field of science to advance, it is necessary to have an accepted classification scheme for accumulating and categorizing empirical findings,” (Barrick & Mount, 1991, p. 23). Additionally, the research establishes that unhelpful support does indeed act as a stressor that is associated with strains.

Unhelpful workplace social support

Social support refers to “an exchange of resources between two individuals perceived by the provider or the recipient to be intended to enhance the wellbeing of the recipient,” (Shumaker & Brownell, 1984, p. 11). Social support can take the form of instrumental support (e.g. physical assistance, tangible materials), emotional support (e.g. empathic understanding, concern), informational support (e.g. knowledge, information), or appraisal support (e.g. affirmation, evaluative feedback; House, 1981). This manuscript examines social support based on this holistic conceptualisation of received (rather than available) support.

Notably, the definition of social support does not specify the outcome of social support for the recipient. While social support is typically considered a beneficial resource, there are numerous instances in which social support does not enhance the wellbeing of the recipient as intended. Unhelpful workplace social support refers to any action taken by a supervisor and/or colleague that the recipient believes was intended to benefit him or her but is perceived as unhelpful or harmful.

Unhelpful workplace social support was initially recognised when researchers were studying the beneficial effects of social support at work. Researchers have found that social support can help protect workers from experiencing negative psychological and physiological responses to workplace stressors (i.e. strains; Cohen & Wills, 1985; House, 1981). Meta-analytic evidence suggests that social support reduces strains, mitigates perceived work stressors, and buffers against strains (Viswesvaran, Sanchez, & Fisher, 1999). However, not all studies have found beneficial effects of workplace social support, and some studies have found that workplace social support can exacerbate strain outcomes (i.e. reverse-buffering effects; Glaser, Tatum, Nebeker, Sorenson, & Aiello, 1999; Jenkins & Elliott, 2004; Kaufmann & Beehr, 1986; Kickul & Posig, 2001). A recent meta-analysis demonstrates that reverse-buffering effects are common, and the authors recommend examining contextual factors to determine when support may exacerbate negative impacts of workplace stressors (Mathieu, Eschleman, & Cheng, 2019).

A number of factors have been examined to explain counterintuitive findings. These include measurement issues such as unmeasured third variables (e.g. stressor severity, employee adjustment; Barrera, 1986; Gleason, Iida, Bolger, & Shrout, 2003; Seidman, Shrout, & Bolger, 2006), characteristics of the support recipient (e.g. self-esteem, social anxiety, emotional state; Bolger & Eckenrode, 1991; Gino & Schweitzer, 2008), when support is provided by the same person who is the source of the recipient’s stress (Beehr, Farmer, Glazer, Gudanowski, & Nair, 2003; Blau, 1981; Mayo, Sanchez, Pastor, & Rodriguez, 2012), and when receipt of support leads to feelings of indebtedness (Gleason et al., 2003; Walster, Berscheid, & Walster, 1973). While all of the explanations are likely relevant to different observed cases of counterintuitive findings, another explanation is also worthy of investigation: some supportive actions may be unhelpful due to the nature of the support provided.

Organisational literature on unhelpful workplace social support

There have been only a handful of organisational studies of unhelpful workplace support. Three forms have been linked to strains in work samples–imposed support, which is support that recipients do not want (Beehr, Bowling, & Bennett, 2010; Deelstra et al., 2003; Song & Chen, 2014), inadequacy-implied support, which is help that makes the recipient feel inadequate or incompetent (Beehr et al., 2010), and stress-focused support, which is support that causes the recipient to focus on their stress (Beehr et al., 2010).

The negative impacts of imposed support and inadequacy-implied support have been explained using the threat-to-self-esteem model (Deelstra et al., 2003). The model posits that supportive actions contain elements of both support and threat, and recipient reactions are negative when supportive acts are perceived as primarily threatening (Fisher, Nadler, & Whitcher-Alagna, 1982). Researchers have hypothesised that imposed social support is perceived as threatening because it threatens individuals’ autonomy and competence while inadequacy-inducing support is perceived as threatening because it directly threatens individuals’ positive self-image. The negative impact of stress magnifying support has been explained with social information processing theory (Beehr et al., 2010), which suggests social interactions provide a means of gaining information and forming beliefs (Salancik & Pfeffer, 1978). If a supervisor or colleague emphasizes a workplace stressor while providing support, the recipient may form more negative views of the stressor. Thus, the support provider may legitimize a recipient’s initial worries and concerns regarding a stressor, causing the stressor to become more salient and concrete to the recipient.

Study 1

To date, only three forms of unhelpful social support have been studied in a work context, and additional forms of unhelpful workplace social support may remain unexamined. To provide an in-depth snapshot of unhelpful workplace support, we conducted a qualitative study of workers’ experiences with unhelpful workplace social support. The study was conducted using an open-ended, structured questionnaire that asked employees to describe an incident in which they received unhelpful social support at work. Similar research approaches have been used in a variety of qualitative studies to gain insights from critical incidents (e.g. Eby, McManus, Simon, & Russell, 2000; Gottlieb, 1978; Motowidlo, Packard, & Manning, 1986; Narayanan, Menon, & Spector, 1999; Newton & Keenan, 1985; Parkes, 1984; 1985). The open-ended methodology is especially appropriate for identifying incidents that have not been theoretically or empirically recognised in previous literature. Although the research was largely exploratory in nature, an overarching research question was addressed.

Research Question: What are categories that best describe workers’ experiences with unhelpful workplace social support?

Method

Participants

Responses from 116 individuals with experience working at least 20 h per week in a current or prior job were analyzed for the qualitative survey study. Participants were recruited from two sources: an American university participant pool (94 participants) and a social media website (22 participants). Participants recruited from the American university participant pool (14 male, 76 female, 1 other, 3 did not report) ranged in age from 18 to 57 (M = 20.97, SD = 4.78). The majority of the university participants reported working in a service (44.68%), sales or office (18.09%) position. Participants recruited from the social media website (7 male, 13 female, 2 did not report) ranged in age from 21 to 53 (M = 26.80, SD = 6.68). The most frequently reported occupations held by social media participants were in management, professional, or related occupations (39.66%). The use of two samples was appropriate and desirable because the goal was to sample participants with a range of experiences with unhelpful workplace social support.

Initially, 250 responses were collected, but responses from 91 participants were excluded from the analysis because they did not address the prompt (58 from the university participant pool, three from the social media website), or they did not contain enough description to be analyzed (27 from the university participant pool, three from the social media website). Another 43 responses were excluded from the analysis because the participants could not recall a relevant experience (38 from the university participant pool, five from the social media website).

Measures

Unhelpful workplace social support

Participants were asked two open-ended questions regarding unhelpful workplace social support. First participants were asked to “Recall a time when you received ineffective/unhelpful support at work (i.e. a time someone tried to help you at work, but the support was not actually helpful). Please describe the incident.” Then the participants were asked, “Why do you think the support you received was ineffective/unhelpful? (i.e. What about the support made it unhelpful?)” Participants were given text entry boxes to provide their responses, and responses to both questions were combined for analysis.

Procedure

The data were analyzed using a content analysis approach informed by literature on content analyses (Weber, 1990) as well as research utilising the analysis (Eby et al., 2000; Motowidlo et al., 1986; Narayanan et al., 1999; Newton & Keenan, 1985; Parkes, 1984, 1985). To guide the development of a high-quality taxonomy, the researchers followed four guidelines. First, the category definitions should be conceptually clear and precise. Second, the categories should be mutually exclusive so that any one unsupportive action should primarily fit into one and only one category. Third, the categories should be exhaustive so that every experience of unhelpful social support reported by participants should be classifiable. Fourth, the categories should be generalisable to instances other than those specifically reported by the participants.

Two researchers were involved in creating the taxonomy. They read the first 20 qualitative responses and independently created possible categories and definitions to capture the data. They also independently categorised each response to one and only one of their categories. Then the researchers met to discuss the categories they independently developed, and after a thorough discussion including a third researcher, they reached consensus on the most effective categorisation scheme for the initial data. The researchers went through an iterative process until consensus was achieved on an overarching taxonomy for the full dataset as well as the categorisation of every response. As a final test of the taxonomy, a fourth researcher independently matched all of the responses to the developed categories, and inter-rater agreement on the full dataset was calculated. The overall percentage of agreement was 86.21%, and 100% agreement was reached among all four of the researchers after a final consensus discussion.

Study 1 results and discussion

The final taxonomy consists of 11 types of unhelpful workplace social support displayed in Table 1 with their data-driven definitions, relative frequencies, and example responses.

Table 1.

Taxonomy of unhelpful workplace social support.

Category	Frequency	Definition	Example Response
Conflicting social support	2% (2)	Social support in which multiple providers offer differing advice or instructions	“[…] other employees will often give me confusing and conflicting advice on how to attack problems or approach my boss […]”
Critical social support	9% (11)	Social support that directly leads the recipient to feel insulted, criticised, and/or attacked	“[I] was given advice on how to perform better. I was already doing the things that were mentioned, and it seemed insulting to be told to do what I was already doing.”
Imposing social support	6% (7)	Social support that is unwanted and forced on the recipient in a noncritical manner	“I was given a chance to show initiative, but my supervisor acted ahead of me when it was not in her job description. It was stepping over bounds rather than allowing me to exhibit initiative.”
Impractical social support	24% (28)	Social support that is unreasonable, misinforming, and/or leads the recipient to stray from company policy or general practices	“I had an incident with a supervisor, and while the supervisor tried to listen to my concerns, the response was not a reasonable solution. [His response] was unhelpful because while my supervisor thought the solution would work [… , it] neglected the hierarchy of the organisation and therefore wasn’t feasible”
Incompatible social support	7% (8)	Social support in which the provider attempts to work with the recipient to help complete a task, but the provider and recipient work differently and struggle to work cohesively	“[Help] was ineffective because we both were getting confused since we were placing and doing things differently.”
Partial social support	23% (27)	Social support that does not benefit the recipient because it is incomplete, imprecise, or unclear	“The instructions were vague.”
Poorly assigned social support	3% (3)	Social support in which a supervisor assigns an employee to help the recipient complete a task, but the assignment was untimely, unneeded, and/or low-quality	“[My manager assigned] coworkers not fully prepared for a heavy shift to work with me when we needed two more fully experienced workers.”
Shortsighted social support	3% (4)	Social support in which the provider takes over a task without teaching the recipient the skills to complete the task on his/her own in the future	“I was working, and I could not figure something out […] Instead of helping me figure it out, someone just took over for me. I didn’t find it helpful because I would have rather learned and figured it out with their help […]”
Stress magnifying social support	3% (3)	Social support that causes the recipient to focus more on the initial stressor in a way that exacerbates the recipient’s stress	“[A] colleague asked to help me on a project […] when I wasn’t prepared. He wanted to drill a given topic with me [… , but] I was overwhelmed and reminded by the fact that I was behind and unprepared.”
Uncomforting social support	1% (1)	Social support in which the provider tries to give emotional support (not advice or tangible assistance), but the recipient does not feel adequately comforted or validated	“[I receive unhelpful support] whenever my coworker comments on a tough situation, and the comment is annoying and useless to the situation. [He] is just really bad at comforting others, and I feel like I have to give him a pity laugh. Basically, [it’s] more trouble than if he just didn’t say anything.”
Undependable social support	19% (22)	Social support in which the provider promises and/or attempts to complete a recipient’s task, but the provider does it in an unreliable, delayed, or low-quality manner	“A coworker attempted to help me answer a phone call while I was starting to walk away from my desk. Although their intentions were good, they ended up being very awkward on the phone as this was not within their daily duties.”

Note: Taken from Gray (2018).

The results demonstrate that workers can recall a variety of experiences receiving unhelpful workplace support, and the categories provide a holistic, standardised categorisation scheme for studying unhelpful workplace social support. In addition to informing a taxonomy, the results suggest that unhelpful workplace social support may act as a job stressor that relates to strains. For example, one respondent stated that well-intended feedback “made me feel ambushed.” Another recipient of well-intended support said that the help “made me even more upset.” Yet another claimed that well-intended, low-quality advice “was frustrating.” Thus, the study also suggests strain outcomes of unhelpful workplace social support.

Study 2

Study 1 used a qualitative approach to identify forms of unhelpful workplace social support. The purpose of Study 2 was to demonstrate, using a quantitative methodology, that unhelpful workplace social support types, identified in Study 1, would act as stressors that would positively relate to strains. This study focussed specifically on coworker social support because workers typically have many interactions with their coworkers. A survey method was used with two samples that both allowed for cross-validation to mitigate the possibility that findings were due to Type 1 error, and to extend generalizability by using dissimilar samples.

Hypothesised strain outcomes

The threat-to-self-esteem model has been the leading explanation for reactions to unhelpful workplace social support (e.g. Deelstra et al., 2003). The model posits that support that is perceived by the recipient as more threatening than supportive leads to negative affect and unfavorable self-evaluations (Fisher et al., 1982). In line with the threat-to-self-esteem model, the following hypotheses were proposed:

Hypothesis 1:

Unhelpful workplace social support is positively associated with recipient jobrelated negative affect.

Hypothesis 2:

Unhelpful workplace social support is negatively associated with recipient competence-based self-esteem.

Hypothesis 3:

Unhelpful workplace social support is negatively associated with coworker satisfaction.

In addition to strains suggested by the threat-to-self-esteem model, unhelpful workplace social support has been associated with work-related emotional exhaustion and physical symptoms (Beehr et al., 2010). Replicating previous findings, the researchers of this study hypothesised that:

Hypothesis 4:

Unhelpful workplace social support is positively associated with work-related burnout.

Hypothesis 5:

Unhelpful workplace social support is positively associated with physical symptoms.

Additionally, some respondents in Study 1 mentioned that unhelpful workplace social support impeded their ability to accomplish work tasks. For example, one respondent in the service industry said that help he received from a coworker “made [his job] more difficult,” and he had to “remake […] food.” Previous research has identified that constraints are sources of frustration for employees (O’connor et al., 1984). Therefore, it was hypothesised that:

Hypothesis 6:

Unhelpful workplace social support is positively associated with organizational frustration.

Alternative explanations for UWSS-strain relationships

The relationship between unhelpful workplace social support and helpful workplace social support was examined to see whether or not unhelpful workplace social support is simply a lack of helpful workplace social support. If that is the case, we would expect the two forms of support to correlate negatively, and when placed in the same analysis, to explain the same variance in strains. In order to test for discriminability between the two forms of support, it was hypothesised that:

Hypothesis 7:

Unhelpful workplace social support is a unique predictor of recipient strains above and beyond helpful workplace social support.

In an effort to examine a potential third variable explanation, a final hypothesis was proposed to examine whether or not the associations between UWSS and strain outcomes are attributable to the current mood of the participants taking the survey.

Hypothesis 8:

Unhelpful workplace social support is a unique predictor of recipient outcomes above and beyond mood.

Examination of reverse-buffering effects

As discussed in the introduction, moderating effects of social support are inconsistent. While some studies find that social support buffers against negative outcomes of work-place stressors, other studies find the opposite (i.e. reverse-buffering effects; Mathieu et al., 2019). Given the prevalence of findings that workplace social support can exacerbate strain outcomes, we wanted to examine if unhelpful support may provide an explanation. Perhaps workplace social support makes the experience of stressors worse when the support falls into our definition of unhelpful support.

Hypothesis 9:

Unhelpful support will moderate the relationships of stressors (a) workload, (b) organizational constraints, and (c) interpersonal conflict with strains, such that the relationships are stronger with more unhelpful support.

Study 2 method

Participants

Sample 1

Data were collected from 176 full-time employees (71 female, 105 male) working in a variety of occupations recruited from Amazon Mechanical Turk (MTurk; Buhrmester, Kwang, & Gosling, 2011). Participants ranged in age from 20 to 63 (M = 34.27, SD = 8.62), and the majority of participants were white (142 participants). Participants held a wide variety of occupations, including engineers, teachers, nurses, servers, and sales representatives. The yearly salaries of participants ranged from less than $25,000 to over $100,000, with a median income between $25,000 and $49,999. Participants were compensated $2.80 for their participation. To help ensure that we received high quality responses, data were only analyzed from participants who responded appropriately to an attention check item. The item stated, “Please select somewhat agree to demonstrate that you are reading the items.” Five participants failed to respond appropriately to the item.

Sample 2

Participants consisted of 496 registered nurses (41 male, 452 female, 3 non-binary) working at least 30 h per week in the United States. Nursing is an especially high stress occupation (e.g. Duquette, Kérowc, Sandhu, & Beaudet, 1994), and nurses frequently interact and assist each other. Therefore, nurses may serve as an especially relevant occupation to examine unhelpful workplace social support. A survey was sent to potential participants through their email addresses, which were obtained from a publicly available list of Florida licensed healthcare providers. Approximately 100,000 emails were sent requesting voluntary participation, and 496 participants completed the entire survey. The high nonresponse rate is likely a result of inaccurate or incorrect email addresses, unseen/unopened emails, occupation changes, and nurse retirement. Many undeliverable email notices were received (approximately 1,100), and many former nurses sent reply emails informing researchers of their occupation changes and/or retirement (approximately 100). According to the email distribution platform (Qualtrics), the completion rate was 45 percent of those who opened the survey.

Participants ranged in age from 22 to 78 (M = 50, SD = 11.42), and the majority of participants were white (418 participants). Participants held a wide variety of nursing positions, including bedside registered nurses, inpatient ARNPs, charge nurses, Chief CRNAs, and directors of nursing. The yearly salaries of participants ranged from less than $25,000 to over $200,000, with a median income between $75,000 and $99,999. To help ensure that we received high quality responses, data were only analyzed from participants who responded appropriately to an attention check item. The item asked participants to “Please select somewhat agree to demonstrate that you are reading the items.” Fifty-eight additional nurses failed to respond to the item appropriately.

Scales

Unhelpful workplace social support

Unhelpful workplace social support was measured using the Unhelpful Workplace Social Support Scale (UWSSS) created in this study. The resulting scale contains 28 items measuring seven subscales and overall unhelpful workplace social support. The subscales demonstrated internal consistency reliability well beyond the generally accepted .70 minimum standard (Sample 1 αs = .88 – .94, see Table 3; Sample 2 αs = .88–.97, see Table 4). The subscale scores were averaged to create a general unhelpful workplace social support score to examine the study hypotheses. The overall scale demonstrated high internal consistency reliability in both samples (Sample 1 and Sample 2 α = .91).

Table 3.

Descriptive statistics and correlations among study 2 variables using sample 1 data (N = 176).

	M	SD	1	1a	1b	1c	1d	1e	1f	1g	2	3	4	5	6	7	8	9	10	11	12
1. Unhelpful Support	2.13	0.89	(.91)
1a. Critical	1.50	0.91	.71	(.91)
1b. Imposing	2.37	1.10	.79	.50	(.88)
1c. Partial	2.29	1.14	.86	.56	.60	(.90)
1d. Undependable	2.26	1.10	.82	.45	.54	.69	(.89)
1e. Shortsighted	2.22	1.14	.80	.51	.66	.61	.53	(.92)
1f. Uncomforting	1.93	1.10	.83	.61	.57	.64	.67	.57	(.90)
1g. Conflicting	2.32	1.17	.86	.46	.60	.73	.75	.62	.66	(.94)
2. Job-related NA	1.95	0.76	.63	.49	.36	.62	.52	.42	.60	.57	(.91)
3. Mood	1.28	0.60	.56	.56	.38	.46	.37	.43	.55	.44	.59	(.93)
4. Self-esteem	5.50	1.00	−.57	−.53	−.39	−.48	−.41	−.48	−.52	−.42	−.59	−.60	(.80)
5. Coworker Satisfaction	4.93	0.96	−.61	−.50	−.50	−.46	−.48	−.44	−.54	−.55	−.56	−.37	.50	(.80)
6. Burnout	2.46	0.85	.52	.32	.38	.48	.47	.30	.45	.56	.75	.39	−.56	−.59	(.91)
7. Physical Symptoms	1.70	0.67	.57	.34	.35	.53	.43	.47	.57	.55	.65	.63	−.47	−.39	.53	(.91)
8. Frustration	3.12	1.52	.47	.28	.32	.43	.39	.29	.44	.48	.68	.31	−.47	−.60	.76	.43	(.82)
9. Helpful Support	3.13	0.59	−.45	−.36	−.29	−.43	−.39	−.32	−.43	−.37	−.50	−.35	.47	.67	−.49	−.30	−.42	(.78)
10. Workload	3.29	0.92	.20	.02	.12	.25	.23	.09	.16	.24	.45	.13	−.23	−.32	.58	.36	.49	−.18	(.85)
11. Constraints	1.84	0.74	.71	.50	.50	.68	.58	.49	.58	.69	.70	.41	−.48	−.63	.68	.56	.57	−.50	.46	(.91)
12. Conflict	1.65	0.76	.65	.64	.48	.52	.46	.53	.58	.48	.56	.43	−.46	−.55	.42	.46	.36	−.35	.26	.54	(.88)

Note: Correlations above .13 are significant at p < .05; Cronbach’s alpha in parentheses.

Table 4.

Descriptive statistics and correlations among study 2 variables using sample 2 data (N = 496).

	M	SD	1	1a	1b	1c	1d	1e	1f	1g	2	3	4	5	6	7	8	9	10	11	12
1. Unhelpful Support	2.13	0.91	(.91)
1a. Critical	1.57	0.91	.72	(.90)
1b. Imposing	2.19	1.06	.73	.45	(.88)
1c. Partial	2.40	1.19	.88	.60	.56	(.93)
1d. Undependable	2.51	1.18	.83	.47	.48	.72	(.93)
1e. Shortsighted	1.98	1.09	.73	.43	.57	.54	.49	(.95)
1f. Uncomforting	2.02	1.18	.84	.59	.49	.68	.71	.51	(.92)
1g. Conflicting	2.27	1.25	.88	.58	.56	.77	.71	.57	.71	(.97)
2. Job-related NA	2.22	0.75	.54	.44	.29	.52	.47	.37	.49	.46	(.90)
3. Mood	1.28	0.51	.46	.41	.23	.41	.39	.38	.42	.34	.61	(.88)
4. Self-esteem	5.56	0.86	−.33	−.30	−.16	−.30	−.18	−.37	−.30	−.24	−.52	−.47	(.71)
5. Coworker Satisfaction	4.70	1.16	−.64	−.53	−.38	−.57	−.61	−.32	−.59	−.57	−.53	−.40	.18	(.81)
6. Burnout	2.99	0.81	.50	.38	.30	.45	.46	.34	.47	.40	.75	.49	−.47	−.50	(.91)
7. Physical Symptoms	1.96	0.66	.42	.35	.25	.37	.36	.33	.38	.33	.57	.58	−.30	−.38	.59	(.86)
8. Frustration	3.91	1.68	.51	.38	.28	.50	.48	.30	.47	.45	.62	.42	−.46	−.51	.73	.43	(.84)
9. Helpful Support	3.20	0.71	−.60	−.53	−.28	−.54	−.57	−.32	−.58	−.53	−.45	−.37	.22	.74	−.44	−.36	−.46	(.85)
10. Workload	3.70	1.04	.29	.17	.14	.24	.32	.25	.27	.24	.43	.27	−.25	−.28	.56	.43	.49	−.20	(.87)
11. Constraints	2.27	0.96	.57	.41	.33	.51	.53	.40	.50	.50	.63	.48	−.37	−.58	.68	.56	.61	−.50	.59	(.91)
12. Conflict	1.77	0.65	.59	.52	.38	.51	.53	.35	.53	.50	.59	.45	−.25	−.58	.51	.47	.43	−.50	.33	.56	(.79)

Note: All correlations are significant at p < .01; Cronbach’s alpha in parentheses.

An initial pool of items was generated for the UWSSS with the intent of reducing it down to a shorter scale after further development. The taxonomy of unhelpful workplace social support developed in Study 1 provided a framework for item generation, and the original narrative responses from Study 1 were used to help generate scale items. Generating the items from qualitative responses ensured that the items reflect actual experiences reported by a diverse sample of employees. Other items were drawn from the Unsupportive Social Interactions Inventory, which is intended to measure unsupportive actions provided by others in response to a stressful event such as cancer or death of a loved one (USII; Ingram, Betz, Mindes, Schmitt, & Smith, 2001).

Four to six items were developed to measure each category of unhelpful workplace social support, resulting in an initial 47-item scale with 10 subscales. An example critical support item is “My coworkers criticize me while trying to help me tackle work problems.” Imposing support was measured with items such as “My coworkers provide unwanted guidance when I don’t ask for it.” Similar items were developed to measure impractical support, incompatible support, partial support, shortsighted support, stress-focused support, uncomforting support, undependable support, and conflicting support. The poorly assigned subscale was not included because it only applies to supervisor support rather than coworker support. See Appendix A for more example scale items.

To examine the initial set of 47 scale items, exploratory factor analyses (EFA) were conducted using the common factor model in SPSS 25. The principle components method was used to analyze the correlation matrix, and missing values were excluded listwise. The scree plot showed points of inflection at two and five factors, and Eigenvalues were greater than one with up to six factors. The two, three, four, five, and six factor solutions were extracted using principal axis factoring (PAF) and an oblique rotation (promax) to allow the factors to correlate. The rotated pattern matrix suggested that none of the extracted solutions was a robust, stable factor structure for the data. Because the initial number of dimensions suggested by the scree plot did not lead to a simple solution, and the scale was formed based on 10 dimensions identified in Study 1, the researchers decided to work backwards to obtain a robust, interpretable solution.

A 10-factor solution was rotated next, and factors were interpreted with the 10 a priori dimensions in mind. The rotated pattern matrix largely supported the a priori factor structure, but the undependable and incompatible items merged into a single factor, leaving the tenth factor with only one item that loaded over .30, and it had a cross-loading of .48. The tenth factor was removed, and a 9-factor solution was rotated next. Most of the items loaded well on their respective factors, but the six impractical support items had factor loadings below 0.55. Those six items were removed, and an 8-factor solution was attempted next. The undependable and incompatible items loaded on the same factor, but all but one of the incompatible items had factor loadings below 0.6. They were removed, and the analysis was re-run. At that stage, the factor structure was relatively clean, but a 7-factor solution was performed to mitigate over factoring. The stress-focused items merged with the conflicting items, but a couple of them had cross loadings over .3, and the other stress-focused items did not conceptually align with the conflicting items. They were removed. An additional five items with relatively low loadings and/or cross-loadings above 0.3 were removed to develop a clean 7-factor solution, with each of the factors corresponding to one of the dimensions from Study 1. All of the remaining 28 items had loadings of at least 0.5 on their respective factors, and no items had cross-loadings above 0.3. The seven factors represent critical, imposing, partial, shortsighted, uncomforting, undependable, and conflicting social support from Study 1. The factor loadings and communalities based on principal axis factoring analysis with an oblique rotation are presented in Table 2.

Table 2.

Principal axis factoring and item reliability of UWSS scale (UWSSS; N = 176).

Scale Item	Factor Loading	Communality	Item-Total Correlation	Cronbach’s Alpha with Item Removed	Cronbach’s Alpha
Critical					.91
1	.90	.80	.83	.88
2	.78	.71	.78	.90
3	.91	.84	.86	.87
4	.89	.71	.78	.90
Imposing					.88
1	.63	.54	.66	.89
2	.88	.79	.79	.84
3	.80	.76	.78	.84
4	.70	.75	.77	.84
Partial					.90
1	.52	.67	.75	.89
2	.95	.79	.79	.87
3	.82	.77	.82	.86
4	.57	.69	.76	.88
Undependable					.89
1	.65	.54	.67	.89
2	.62	.64	.75	.86
3	.72	.73	.79	.85
4	1.03	.90	.86	.83
Shortsighted					.92
1	.87	.74	.79	.91
2	.91	.82	.84	.89
3	.67	.79	.81	.90
4	.82	.80	.84	.89
Uncomforting					.90
1	.69	.75	.80	.85
2	.87	.83	.83	.83
3	.58	.73	.77	.88
Conflicting					.94
1	.77	.76	.83	.93
2	.75	.75	.80	.93
3	.83	.77	.84	.92
4	.72	.70	.81	.93
5	.95	.88	.89	.91

Once the dimensionality was determined through the exploratory factor analyses, an item reliability analysis was conducted. All of the subscales demonstrated high internal consistency reliability (α = .88–.94). Out of the 28 items, only one item would increase the internal consistency reliability of its respective subscale if deleted. The item was retained because the cost of losing breadth seemed greater than the benefit of gaining 0.01 internal consistency reliability. The item reliability analysis is depicted in Table 2.

Given the iterative exploratory nature of the factor analysis in Sample 1, it was important to verify the structure on a new sample. Using sample 2 data, a confirmatory factor analysis (CFA) was conducted in SAS 9.4 using proc calis to test the hypothesised measurement model of the seven unhelpful workplace social support dimensions. A first-order model was examined with seven correlated factors. Although the x² measure of fit was statistically significant [x²(329) = 885.02, p < .05], the descriptive measures indicated good model fit. The Non-Normed Fit Index (NNFI) and the Comparative Fit Index (CFI) were higher than .95 as recommended by Hu and Bentler (1998; NNFI = .96, CFI = .96). The root mean square error of approximation (RMSEA) and the standardised root mean residual (SRMR) were at or lower than the .06 and .08 cutoffs recommended by Hu and Bentler (1998; RMSEA = .06, SRMR = .04). All of the measured variable indicators had standardised loadings of at least 0.6 on their corresponding factors (.61–.96). An alternative 1-factor model was also examined to compare with the a priori 7-factor model. The alternative model fit the data significantly worse than the 7-factor model [Δx²(21) = 5456.90, p < .05] providing additional construct validity support for the measure and taxonomy.

Helpful workplace social support

Helpful workplace social support was measured with a four-item social support scale created by Caplan, Cobb, French, Van Harrison, and Pinneau (1975). The scale can be used to measure social support provided by an immediate supervisor, other people at work, or friends/family. For this study, social support from coworkers was measured. An example item is “How much do your coworkers go out of their way to do things to make your work life easier for you?” Participants responded on a four-point scale (1 = not at all, 4 = very much). The scale demonstrated adequate internal consistency reliability (Sample 1 α = .78; Sample 2 α = .85).

Job-related negative affect

Job-related negative affect was measured using the 10-item negative emotion subscale of the Job-related Affective Well-being Scale (JAWS; Van Katwyk, Fox, Spector, & Kelloway, 2000). Participants were asked to indicate the extent to which their job generally makes them feel emotions such as angry, anxious, and frightened. They responded on a five-point scale (1 = never, 5 = extremely often). Based on the generally accepted level of .70 (Nunnally & Bernstein, 1994), the scale demonstrated adequate internal consistency reliability (Sample 1 α = .91; Sample 2 α = .90).

Competence-based self-esteem

Competence-based self-esteem was measured using six items measuring job competence developed by Warr (1990). An example item is “I can do my job well.” Participants responded on a seven-point scale (1 = strongly disagree, 7 = strongly agree). The scale demonstrated adequate internal consistency reliability (Sample 1 α = .80; Sample 2 α = .71).

Coworker satisfaction

Coworker satisfaction was measured using the four-item coworker subscale of the Job Satisfaction Survey (JSS; Spector, 1994). An example item is “I like the people that I work with.” Participants responded on a six-point scale (1 = disagree very much, 6 = agree very much). The scale demonstrated adequate internal consistency reliability (Sample 1 α = .80; Sample 2 α = .81).

Burnout

Participants completed the seven-item work-related burnout subscale of the Copenhagen Burnout Inventory (CBI; Kristensen, Borritz, Villadsen, & Christensen, 2005). An example item is “Do you feel burnt out because of your work?” Participants responded on a fivepoint scale (1 = never, 5 = always). The scale demonstrated adequate internal consistency reliability (Sample 1 and Sample 2 α = .91).

Physical symptoms

Physical symptoms were measured using the 13-item Physical Symptoms Inventory (PSI; Spector & Jex, 1998). Participants were asked to report the frequency with which they experience a variety of physical symptoms such as an upset stomach or nausea, a backache, or trouble sleeping. They responded on a five-point scale (1 = not at all, 5 = several times per day). The scale demonstrated adequate internal consistency reliability (Sample 1 α = .91; Sample 2 α = .86).

Organisational frustration

Participants completed a three-item organisational frustration scale (Peters, O’Connor, & Rudolf, 1980). The items were slightly modified to reflect frustration associated with one’s job as opposed to a particular task. An example item is “Trying to get my job done is a very frustrating experience.” Participants responded on a seven-point scale (1 = strongly disagree, 7 = strongly agree). The scale demonstrated adequate internal consistency reliability (Sample 1 α = .82; Sample 2 α = .84).

Mood

Mood was measured using eight negative mood items taken from Mohr et al. (2005). Participants were asked to indicate the extent to which they feel negative emotions such as angry and sad at the present moment. They responded on a five-point scale (1 = very slightly or not at all, 5 = extremely). The scale demonstrated adequate internal consistency reliability (Sample 1 α = .93; Sample 2 α = .88).

Workload

Workload was measured with the 5-item quantitative workload inventory (QWI; Spector & Jex, 1998). A sample item is “How often does your job require you to work very fast?” Participants responded on a 5-point scale (1 = less than once per month or never; 5 = several times per day). The scale had acceptable internal consistency reliability (Sample 1 α = .85; Sample 2 α = .87).

Organisational constraints

Participants responded to the 11-item organisational constraints scale (OCS; Spector & Jex, 1998). They reported the frequency with which it is difficult or impossible to perform their jobs due to constraints such as poor equipment and inadequate training on a 5-point scale (1 = less than once per month or never; 5 = several times per day). The measure demonstrated adequate internal consistency reliability (Sample 1 and Sample 2 α = .91).

Interpersonal conflict

Interpersonal conflict was measured with a 4-item interpersonal conflict at work scale (ICAWS; Spector & Jex, 1998). An example item is “How often are people rude to you at work?” Participants responded on a 5-point frequency scale (1 = never; 5 = very often). The measure had sufficient internal consistency reliability (Sample 1 α = .88; Sample 2 α = .79).

Study 2 results and discussion

Hypotheses 1 – 6 concerned the relationships between overall unhelpful workplace social support and strains. Supporting those hypotheses, general unhelpful workplace social support was significantly associated with higher job-related negative affect, lower competence-based self-esteem, lower coworker satisfaction, higher work-related burnout, higher organisational frustration, and higher physical symptoms in both samples (Sample 1 rs = |.47–.63|, ps < .01; Sample 2 rs = |.33–.64|, ps < .01). Correlation matrices including all study variables for both samples are presented in Tables 3 and 4, for Samples 1 and 2, respectively.

Correlations between each of the seven social support dimensions and the strains are also shown in Tables 3 and 4. As can be seen, every dimension of social support was significantly related to every strain in both samples, although there were some differences in magnitude of correlations. To investigate these relationships further, we conducted a series of multiple regression analyses in which each strain was separately regressed on all seven UWSSS subscales in order to determine if they each explained unique variance in stains. These analyses showed that in no case were all seven subscales significant in the regression analysis, as different types of UWSS predicted unique variance in different strain outcomes. For example, partial support and uncomforting support were significant predictors of job-related negative affect across both samples (Sample 1: β = .34, β = .28, ps < .05; Sample 2: β = .23, β = .17, ps < .05; betas reported are standardised coefficients) whereas critical support, shortsighted support, and uncomforting support were significant predictors of competence-based self-esteem across both samples (Sample 1: β = −.26, β = −.21, β = −.20, ps < .05; Sample 2: β = −.14, β = −.35, β = −.20, ps < .05; betas reported are standardised coefficients). Results of each of the regression analyses are displayed in Table 5.

Table 5.

Multiple regressions of UWSS subscales predicting study 2 criterion variables.

DV	Critical Sample 1 β	Imposing Sample 2 β	Partial Sample 1 β	Undependable Sample 2 β	Shortsighted Sample 1 β	Uncomforting Sample 2 β	Conflicting Sample 1 β	Sample 2 β	Sample 1 β	Sample 2 β	Sample 1 β	Sample 2 β	Sample 1 β	Sample 2 β
NA	.13	.15*	−.18*	−.10	.34*	.23*	.01	.11	−.02	.09	.28*	.17*	.19	<.01
Multiple R2													.48	.33
F													22.07*	33.89*
Self-Esteem	−.26*	−.14*	.05	.17*	−.13	−.21*	−.02	.20*	−.21*	−.35*	−.20*	−.20*	.04	.10
Multiple R2													.38	.21
F													14.45*	18.41*
Satisfaction	−.23*	−.23*	−.18*	−.02	.10	−.07	−.03	−.32*	.04	.14	−.16	−.16*	−.31*	−.11
Multiple R2													.41	.48
F													16.94*	64.27*
Burnout	.02	.10*	.08	−.02	.13	.16*	.03	.21*	−.19*	.08	.13	.20*	.43*	−.10
Multiple R2													.35	.28
F													12.79*	26.80*
Symptoms	−.11	.16*	−.16	−.04	.24*	.11	−.16	.12	.17	.15*	.42*	.13	.26*	−.09
Multiple R2													.42	.20
F													17.56*	17.30*
Frustration	−.01	.08	.01	−.07	.13	.25*	−.04	.18*	−.11	<.01	.22*	.15*	.35*	.02
Multiple R2													.27	.30
F													8.88*	29.70*

Note: βs are standardised coefficients; NA = job-related negative affect;

^*=p < .05; Satisfaction = coworker satisfaction.

Alternative explanations for results

Tables 3 and 4 show that the helpful and unhelpful social support scales were significantly correlated. Helpful social support correlated with overall unhelpful support −.45 in Study 1 and −.60 in Study 2. Correlations between helpful support and the seven dimensions of unhelpful support ranged from −.43 to −.29 in Study 1 and −.58 to −.28 in Study 2. Clearly helpful and unhelpful support are not independent, leading to the concern that both scales reflect much the same construct, and would explain the same variance in strains. Multiple regression analyses were conducted to examine the hypotheses that the relationships between unhelpful workplace social support and strain outcomes cannot be attributed to helpful social support, and thus the two scales reflect different, albeit related, constructs. Each of the strain outcomes were separately regressed on overall unhelpful workplace social support and helpful workplace social support. Unhelpful workplace social support remained a significant predictor of each of the outcomes in both samples (Sample 1 βs = |.35–.55|, ps < .05; Sample 2 βs = |.30–.44|, ps < .05). The findings support hypothesis seven (see Table 6).

Table 6.

Multiple regressions of UWSS and control variables predicting criterion variables.

	Helpful Workplace Social Support		Mood
	Sample 1 β	Sample 2 β	Sample 1 β	Sample 2 β
DV: Job-related Negative Affect
Unhelpful Workplace Social Support	.51*	.44*	.44*	.33*
Control (Helpful support or mood)	−.26*	−.18*	.35*	.46*
R2	.45	.32	.48	.46
DV: Competence-based Self-esteem
Unhelpful Workplace Social Support	−.44*	−.30*	−.34*	−.15*
Control (Helpful support or mood)	.27*	.04	−.41*	−.40*
R2	.38	.11	.44	.24
DV: Coworker Satisfaction
Unhelpful Workplace Social Support	−.38*	−.31*	−.59*	−.58*
Control (Helpful support or mood)	.50*	.55*	−.04	−.14*
R2	.57	.60	.37	.42
DV: Work-related Burnout
Unhelpful Workplace Social Support	.38*	.38*	.45*	.35*
Control (Helpful support or mood)	−.31*	−.21*	.14	.33*
R2	.35	.28	.29	.34
DV: Physical Symptoms
Unhelpful Workplace Social Support	.55*	.32*	.32*	.20*
Control (Helpful support or mood)	−.05	−.17*	.45*	.49*
R2	.33	.20	.47	.37
DV: Organisational Frustration
Unhelpful Workplace Social Support	.35*	.37*	.42*	.41*
Control (Helpful support or mood)	−.26*	−.24*	.08	.23*
R2	.27	.30	.22	.31

Note: βs are standardised coefficients;

^*p < .05.

Additional multiple regression analyses were conducted to examine the hypotheses that the relationships between unhelpful workplace social support and strain outcomes cannot be attributed to the mood of the participants. Perhaps participants report high levels of UWSS and strains because they are upset, or they report low levels of UWSS and strains because they are content. To help rule out mood as a possible third variable explanation, each of the strain outcomes were separately regressed on UWSS and mood. Unhelpful workplace social support remained a significant predictor of each of the outcomes in both samples (Sample 1 βs = |.32–.59|, ps < .05; Sample 2 βs = |.15–.58|, ps < .05). The findings support hypothesis eight (see Table 6).

Examination of buffering effects

Moderated multiple regression analyses were conducted to examine the moderating effects of unhelpful workplace social support. After computing unhelpful support by job stressor interaction terms, unhelpful support and job stressors were entered in Step 1. The interaction terms were entered in Step 2. Eighteen regression analyses were run in each sample, one for each paired combination of stressors (workload, organisational constraints, and interpersonal conflict) and strains (job-related negative affect, competence-based self-esteem, coworker satisfaction, work-related burnout, organisational frustration, and physical symptoms). Tables 7 and 8.

Table 7.

Examining Main Effects and Buffering Effects of Unhelpful Workplace Social Support with Sample 1 Data.

	Job-related NA			Competence-based Self-esteem			Coworker Satisfaction			Work-related Burnout			Physical Symptoms			Organisational Frustration
	B	β	R 2	B	β	R 2	B	β	R 2	B	β	R 2	B	β	R 2	B	β	R 2
Step 1			.43			.34			.41			.29			.34			.22
A Interpersonal Conflict	.26*	.26*		−.21	−.16		−.33*	−.26*		.16	.14		.14	.16		.21	.10
B UWSS	.40*	.46*		−.52*	−.46*		−.48*	−.44*		.41*	.43*		.36*	.47*		.69*	.40*
Step 2			.43			.34			.43			.33			.35			.25
Intercept	.62			6.87			7.37			.11			.33			−.32
A Interpersonal Conflict	.29	.29		−.15	−.12		−.91*	−.72*		.95*	.85*		.39*	.44*		1.29*	.65*
B UWSS	.42*	.49*		−.49*	−.43*		−.83*	−.77*		.89*	.94*		.51*	.67*		1.34*	.78*
A x B	−.01	−.05		−.02	−.07		.21*	.73*		−.29*	−1.13*		−.09	−.45		−.39*	−.86*
Step 1			.50			.33			.41			.51			.39			.38
A Workload	.28*	.33*		−.13	−.12		−.21*	−.21*		.45*	.49*		.19*	.26*		.69*	.42*
B UWSS	.49*	.56*		−.61*	−.54*		−.61*	−.57*		.41*	.43*		.39*	.52*		.66*	.38*
Step 2			.50			.35			.41			.51			.39			.39
Intercept	.23			8.52			6.87			−.51			.31			−1.00
A Workload	.21	.25		−.53*	−.49*		−.19	−.19		.64*	.70*		.17	.23		.83*	.50*
B UWSS	.37	.43		−1.27*	−1.12*		−.58*	−.54*		.72*	.75*		.36	.48		.89*	.52*
A x B	.03	.17		.20*	.76*		−.01	−.04		−.09	−.43		.01	.06		−.07	−.18
Step 1			.52			.34			.45			.47			.37			.33
A Constraints	.52*	.50*		−.22	−.16		−.52*	−.41*		.71*	.62*		.27*	.31*		.98*	.48*
B UWSS	.24*	.27*		−.51*	−.45*		−.35*	−.32*		.08	.08		.27*	.36*		.21	.12
Step 2			.52			.34			.46			.52			.39			.35
Intercept	.17			7.09			7.26			−.40			.14			−.53
A Constraints	.70*	.68*		−.28	−.21		−.87*	−.68*		1.49*	1.31*		.55*	.61*		1.77*	.87*
B UWSS	.38*	.44*		−.56*	−.49*		−.63*	−.58*		.70	.73		.49*	.65*		.84*	.49*
A x B	−.07	−.33		.02	.08		.14	.50		−.31*	−1.26*		−.11	−.56		−.32*	−.71*

^*p < .05.

Table 8.

Examining Main Effects and Buffering Effects of Unhelpful Workplace Social Support with Sample 2 Data.

	Job-related NA			Competence-based Self-esteem			Coworker Satisfaction			Work-related Burnout			Physical Symptoms			Organisational Frustration
	B	β	R 2	B	β	R 2	B	β	R 2	B	β	R 2	B	β	R 2	B	β	R 2
Step 1			.41			.11			.47			.32			.26			.29
A Interpersonal	.48*	.42*	−.12		−.09		−.55*	−.31*		.40*	.32*		.36*	.35*		.51*	.20*
Conflict
B UWSS	.25*	.30*		−.26*	−.28*		−.58*	−.45*		.28*	.31*		.16*	.22*		.74*	.40*
Step 2			.41			.12			.47			.32			.26			.30
Intercept	.54			6.93			6.68			1.27			1.17			.48
A Interpersonal Conflict	.65*	.56*		−.44*	−.33*		−.43*	−.24*		.63*	.51*		.26*	.25*		1.04*	.40*
B UWSS	.38*	.45*		−.52*	−.54*		−.48*	−.38*		.46*	.52*		.08	.11		1.14*	.61*
A x B	−.06	−.27		.13*	.47*		−.05	−.14		−.09*	−.36*		.04	.18		−.20*	−.39*
Step 1			.38			.14			.42			.45			.28			.39
A Workload	.22*	.31*		−.15*	−.18*		−.13*	−.11*		.37*	.47*		.22*	.34*		.61*	.37*
B UWSS	.38*	.45*		−.27*	−.28*		−.77*	−.60*		.32*	.36*		.24*	.32*		.75*	.41*
Step 2			.38			.14			.42			.45			.29			.40
Intercept	.79			6.88			6.19			.99			1.26			−1.04
A Workload	.17*	.23*		−.20*	−.24*		.04	.04		.35*	.45*		.06	.09		.90*	.55*
B UWSS	.27*	.33*		−.37*	−.39*		−.45*	−.35*		.30*	.33*		−.08	−.11		1.32*	.71*
A x B	.03	.17		.03	.14		−.08	−.34		.01	.04		.08*	.57*		−.15*	−.40*
Step 1			.45			.16			.47			.48			.33			.41
A Constraints	.37*	.47*		−.24*	−.27*		−.39*	−.33*		.50*	.59*		.33*	.48*		.83*	.47*
B UWSS	.23*	.27*		−.17*	−.18*		−.57*	−.45*		.14*	.16*		.11*	.15*		.46*	.25*
Step 2			.45			.16			.48			.49			.34			.43
Intercept	.84			6.64			6.55			1.19			1.38			−.23
A Constraints	.39*	.50*		−.31*	−.34*		−.28*	−.24*		.64*	.77*		.17*	.25*		1.34*	.77*
B UWSS	.25*	.31*		−.25*	−.26*		−.44*	−.35*		.32*	.36*		−.09	−.13		1.09*	.59*
A x B	−.01	−.06		.03	.14		−.05	−.17		−.07*	−.34*		.07*	.46*		−.23*	−.58*

^*p < .05.

Hypothesis 9 was partially supported, although mostly unsupported. Unhelpful support demonstrated reverse-buffering effects in two cases in Sample 2. The positive relationship between workload and physical symptoms was stronger with more unhelpful support, and the relationship between organisational constraints and physical symptoms was stronger with more unhelpful support (βs = .57 and .46, ps < .05). Twelve moderation effects occurred in the opposite direction. Although unhelpful support was consistently associated with higher strains, the relationships between stressors and strains were sometimes weaker with more unhelpful support. One potential explanation is a ceiling effect; when strains are high, adding another stressor in addition to unhelpful support may not make a significant difference.

General discussion

While workplace social support is typically considered a beneficial job resource, workplace social support can also serve as a job stressor. Unhelpful workplace social support (UWSS) is defined as any action taken by a supervisor and/or colleague that the recipient believes was intended to benefit him or her but is perceived as unhelpful or harmful. Two studies, one qualitative and one quantitative, identified 11 dimensions of UWSS, developed a measure of UWSS, and established that at least seven of those UWSS dimensions relate to a series of strains. Together, the studies demonstrate that unhelpful workplace social support is a meaningful job stressor worthy of further investigation.

In Study 1, responses from 116 employees provided real-life examples of unhelpful workplace social support. The responses demonstrated that workers encounter instances receiving unhelpful workplace social support, and a content analysis revealed 11 distinct forms of unhelpful workplace social support. Study 2 confirmed that UWSS from a coworker is associated with numerous strain outcomes, including higher job-related negative affect, lower competence-based self-esteem, lower coworker satisfaction, higher work-related burnout, higher organisational frustration, and more physical symptoms. The associations remained significant even after controlling for helpful workplace social support and mood.

This is the first known research to create a holistic categorisation scheme of unhelpful workplace social support. This research paves the way for other researchers to study the construct using an accepted classification scheme for accumulating and categorising empirical findings. The research also provides an internally consistent measure of unhelpful workplace social support with initial evidence for construct validity that might be of use in future research.

One noteworthy takeaway from this research is that unhelpful workplace social support can be more detrimental than a lack of social support. Some of the qualitative responses help to clarify why that seems to be the case. One participant said,

[I receive unhelpful support] whenever my coworker comments on a tough situation. [… He] is just really bad at comforting others, and I feel like I have to give him a pity laugh. Basically, [it’s] more trouble than if he just didn’t say anything.

In a more extreme example, a participant recalled a time in which,

Another server once greeted my table at work because I was busy […] It was helpful because the table was greeted. However, it turned out the table was of secret shoppers (people evaluating the quality of service). She did not follow the steps of service (having a name tag, introducing herself, suggesting specific dinks and appetizers, being in uniform, etc.), so I failed the shop.

Both cases (and many others) demonstrate that unhelpful support can have worse implications than no support at all.

A related finding is that unhelpful workplace social support often appears to be more strongly related to strain outcomes than helpful workplace social support. This can be seen in the correlations; for example, the relationship between overall unhelpful workplace social support and job-related negative affect (r_{sample 1} = .63, r_{sample 2} = .54) was significantly stronger than the relationship between helpful workplace social support and job-related negative affect (r_{sample 1} = −.50, r_{sample 2} = −.45) in both sample 1 and sample 2, t(173) = 2.15, p < .05, t(493) = 2.68, p < .05, respectively. This pattern of relationships held across two different samples and five strain outcomes: job-related negative affect, competence-based self-esteem, work-related burnout, physical symptoms, and organisational frustration. The correlation differences were significant in five of 12 cases (see Table 9). Helpful workplace social support was only more strongly related to coworker satisfaction. The findings support the merits of studying unhelpful workplace social support in addition to helpful workplace social support. Additional evidence is provided by the regression analyses.

Table 9.

Differences in correlation magnitude by support type (helpful versus unhelpful).

	rUnhelpful	rHelpful	rUnhelpfulxHelpful	N	t
DV: Job-related Negative Affect
Sample 1	.63	−.50	−.45	176	2.15*
Sample 2	.54	−.45	−.60	496	2.68*
DV: Self-esteem
Sample 1	−.57	.47	−.45	176	1.56
Sample 2	−.33	.22	−.60	496	2.89*
DV: Coworker Satisfaction
Sample 1	−.61	.67	−.45	176	−1.09
Sample 2	−.64	.74	−.60	496	−3.85*
DV: Work-related Burnout
Sample 1	.52	−.49	−.45	176	0.46
Sample 2	.50	−.44	−.60	496	1.74
DV: Physical Symptoms
Sample 1	.57	−.30	−.45	176	4.07*
Sample 2	.42	−.36	−.60	496	1.65
DV: Organisational Frustration
Sample 1	.47	−.42	−.45	176	0.73
Sample 2	.51	−.46	−.60	496	1.47

Note: t test for dependent correlations was conducted; Unhelpful = unhelpful workplace social support; Helpful = helpful workplace social support;

^*p < .05.

Research commonly finds that support can exacerbate negative outcomes of workplace stressors rather than buffer against them (Mathieu et al., 2019). This research provides a potential explanation for some instances of reverse-buffering effects: some of the social support provided might be unhelpful (rather than helpful) support. Findings of this study suggest that unhelpful workplace social support may sometimes result in reversebuffering effects (see Table 9).

A potentially important issue in this domain is the attributions people make toward social support providers. The authors tried to limit this research to instances in which the recipients believed the support provider was trying to be helpful. To illustrate well-intended attributions, one participant in Study 1 said,

a coworker attempted to help me answer a phone call while I was starting to walk away from my desk. Although their intentions were good, they ended up being very awkward on the phone as this was not within their daily duties.

Another said,

One day at work I was having a really bad day. I was in pain from coming back to work after a car accident. My boss noticed this, and in an attempt to make me feel better, sent me to a different station. However, this new station was even worse on my injured back and made me even more upset […]

Future studies could examine whether different attributions are associated with different outcomes. This research suggests that support can be harmful even when recipients believe the support provider was trying to be helpful.

From a practical standpoint, this research helps to identify ways in which people can better provide support. For example, support providers may consider asking employees if help is wanted before providing support to mitigate the provision of imposing support. Support providers may also phrase feedback, advice, and suggestions tactfully to reduce the likelihood of inadvertently criticising an employee while trying to help. More research should be conducted that proposes and evaluates strategies for reducing unhelpful workplace support.

Limitations

This research has several strengths and limitations. Perhaps the greatest study limitation is that the measure of helpful support selected for this research does not structurally and conceptually align with the measure of unhelpful support. The helpful support measure contains four items, and it is not explicitly behavioural. In contrast, the unhelpful support measure contains twenty-eight items, and it is a behavioural measure. Furthermore, the helpful support measure includes items pertaining to both available and received support whereas the unhelpful measure only captures received support. The helpful support measure also has more emotional support content than the unhelpful support measure. Future research could attempt to replicate the findings using a more analogous measure of helpful social support. Such research could help to rule out the possibility that unhelpful workplace support explained variance in strains above and beyond helpful workplace support because the unhelpful support measure is more comprehensive.

Many of the participants in Study 1 were employees taking classes at an American university, and one sample of participants in Study 2 consisted of employees recruited from Amazon Mechanical Turk. Some researchers have expressed concern that findings utilising such samples may not generalise to other samples of employees (Elsevier, 2014, Introduction section, para. 6; Highhouse & Gillespie, 2009). However, the results of this research support a general research question: Is unhelpful workplace social support a meaningful workplace stressor? Any sample of workers for which the research question is intended to generalise is appropriate for examining the question, including employees taking classes and Amazon Mechanical Turk workers (Landers & Behrend, 2015). In a chapter on sampling, Highhouse (2009) claims, “a theory about occupational satisfaction and commitment might apply to nurses, coaches, priests, or professional skateboarders. Any one of these samples is appropriate for testing the theory” (p. 264). Other researchers concur, stating that “any currently or potentially employed person falls within the population of interest to most I-O psychologists.” (Landers & Behrend, 2015, p. 3). By utilising three very different samples of employees in this study that yielded similar results, researchers have greater confidence in the generalizability of implications of unhelpful workplace social support.

On the topic of samples, the authors acknowledge growing skepticism of data collected from Amazon Mechanical Turk workers (Moss & Litman, 2018). The authors sought to demonstrate the merits of their data by collecting data on the same constructs from a sample of Amazon Mechanical Turk workers and a sample of nurses. The patterns of variable relationships were extremely similar across the samples. The replication suggests acceptable data quality from the Amazon Mechanical Turk workers and the generalizability of the research findings.

The studies relied on self-report measures. Though objective measures are sometimes preferred over self-report measures, the variables of interest in this research are best measured with self-reports. Unhelpful workplace social support is subjective by definition because the support recipient must perceive the support as unhelpful and/or harmful. In favour of a subjective support measure, previous research has found that recipient perceptions of support are more relevant to recipient outcomes than objective evaluations of received support (Wethington & Kessler, 1986). The strain measures included in the study are also likely best measured with self-reports. Job-related negative affect, self-esteem, coworker satisfaction, burnout, and organisational frustration are personal experiences that likely cannot be well-assessed observationally or physiologically. While some physical symptoms can be assessed with physiological measures, a holistic assessment of physical symptoms, including headaches, fatigue, and nausea would be difficult to gauge with physiological measures, especially over an extended period. Some previous research has found that the receipt of imposing social support is associated with a subsequent increase in heartrate and a decrease in respiratory sinus arrhythmia (Deelstra et al., 2003). Taken together, the findings of Deelstra et al. (2003) and this research suggest that unhelpful workplace social support may have physical implications.

Although the use of self-report measures is likely merited, there is threat of common method bias. Recent literature on common method bias suggests that researchers should be concerned with “extraneous and unintended systematic influences on a measured variable, some of which might be shared with other measured variables (CMV) and some of which is not (UMV)” (Spector, Rosen, Richardson, Williams, & Johnson, 2019, p. 856). One potential common method variance source applicable to this study was mood. Perhaps participants report receiving unhelpful workplace social support and report experiencing strain outcomes because they are angry or otherwise upset. Examining the support-strain relationships after controlling for mood helped to rule out mood as a common method variance source in this research. However, further research on other sources of method variance that may impact the measurement of variables used in this study would greatly inform this and other research. Further, although common method variance has the potential to affect bi-variable correlations, they are unlikely to affect results of complex multiple regression analyses with several predictors (Siemsen, Roth, & Oliveira, 2010), as we have in this paper. All this being said, common method variance should not be assumed simply because the research used a cross-sectional self-report design (Conway & Lance, 2010; Spector, 2006).

Conclusion

Overall, two research studies using very different methods provide evidence for social support as a meaningful job stressor. Future research should continue to examine the effects of unhelpful workplace social support as well as identify practical solutions to combat the stressor. Such knowledge would have important implications for employees who desire to be helpful while providing support and employees who desire to be helped while receiving support.

Appendix A

Example Unhelpful Workplace Social Support Scale Items

My coworkers …

1. Try to help by completing tasks for me that I want to do myself.

2. Provide unwanted guidance when I don’t ask for it.

3. Give me imprecise suggestions at work.

4. Don’t give me enough information when trying to help me.

5. Provide vague solutions to my work problems.

6. Criticise me while trying to help me tackle work problems.

7. Insult me when trying to help me improve my work.

8. Do my tasks for me rather than showing me how to do them.

9. Are not helpful when trying to comfort me.

10. Do not follow through after offering to complete a task for me.

11. Do things wrong when completing a work task for me.

12. Offer advice that isn’t helpful because it clashes with other advice I have received at work.

13. Advise courses of action that aren’t helpful because they conflict with previous advice I’ve received.