Independent Children’s Social Work Practice Pilots

Abstract

Objectives:

To investigate whether a new model that delegates some out-of-home care services from the public to the private and not-for-profit sectors in England enhances practitioners’ job control and stress levels.

Methods:

A 3-year longitudinal matched-control evaluation examined changes in Karasek demand-control model and Maslach burnout levels of 2,050 staff working in five social work practice (SWP) pilots, their host local authorities and comparable sites.

Results:

Mixed-effect models indicated no significant difference in main outcomes among SWP staff when compared to staff in host and comparison local authorities. There were notable differences in relation to job insecurity and social support.

Conclusion:

The minimal effects observed may relate to the diverse nature of SWPs with no specific work model predominating.

Keywords

Karasek Maslach Burnout Inventory social services mixed-effect models matched-control evaluation

Introduction

Social work is characterized by high levels of emotional involvement and is considered one of the “emotionally taxing” professions (Brotheridge & Grandey, 2002). Similarly, the social work profession has constantly featured as one with higher stress levels than average (Lloyd, King, & Chenoweth, 2002; Tracy, Bean, Gwatkin, & Hill, 1992). Some social work stressors are inherent in nature relating to the burden and responsibilities associated with the job; a poor public image is also an important factor (LeCroy & Rank, 1986; Zapf, Knorz, & Kulla, 1996). Other factors contributing to burnout are organizational in nature and may reflect the structure and dynamics of social work environments. Several features may enhance job stress, for example, high levels of administration, lack of role clarity and conflict, and hierarchical decision-making processes. On the other hand, studies have found that supportive work environments, involvement in decision making, and workplace strategic planning (Mackie, Holahan, & Gottlieb, 2011; Shier & Graham, 2011) are associated with reduced levels of stress and burnout. Research also indicates associations between environmental workplace factors and child protection practitioners’ burnout levels (Savicki & Cooley, 1994). In this article, we investigate the impact of a new model of social work that was envisaged to enhance environmental workplace factors contributing to the diminishing of practitioners’ job stress and enhancing their autonomy.

The social work practice (SWP) model was promoted in the United Kingdom amid long-standing concerns over the recruitment and retention of social workers working with children and families, specifically in child protection services (Tham, 2007). Overall, the profession has experienced difficulties in recruitment of students (Orme et al., 2009), with endemic staff shortages in some areas leading to “role slumping” where managers have to offer intensive support to inexperienced staff (Searle & Patent, 2012). These challenges have fueled considerable debate on how to facilitate better practices while facing harsh economic realities.

In England, there has been a clear direction from the government to move social work responsibilities (and other public services) from the larger more complex organizational structure of local authorities to smaller and more autonomous environments (Mutuals Taskforce, 2012). The intention to pilot SWPs with children in out-of-home care (SWPs) emerged in England in a government consultative Green Paper Care Matters: Time to deliver for children in care (Department of Children, Schools and Families [DCSF], 2006) produced by the New Labour government. The policy aspirations were formulated in developing smaller social worker-led organizations, independent of local authorities. These were envisaged as improving the morale and retention of children’s social workers, reducing bureaucracy, and facilitating professional decision making. Movement toward “privatization” directions can also be seen in other developed countries including the United States: However, there are key differences in relation to its history and success and these are discussed further in Stanley et al. (2012a).

The model hypothesized that better outcomes for children and young people would result through greater consistency and improved stability of care for children in out-of-home care and care leavers, which can be facilitated by this new model of working (Le Grand, 2007). The development of SWPs was encouraged by those originating the idea in the New Labour government and then subsequently in the Coalition government. It is in this context, and as part of the Children’s Workforce Strategy (DCSF, 2008) that the SWPs were established as autonomous organizations, which would contract with local authorities to provide children out-of-home care services. The ambition was for smaller, social worker-led organizations that are free from local authorities’ restrictions to bring the decision-making process closer to frontline practice, improving outcomes for both practitioners and children in out-of-home care (Le Grand, 2007). This model was thus hoped to alter the culture and climate construct of social work, which in turn would improve staff job satisfaction and commitment (Glisson & James, 2002).

These pilots, however, were not intended to replace all the services provided by local authorities for children in out-of-home care; the pilots therefore displayed varying degrees of dependency on their host local authorities. Five SWP pilots were included in this study, each operating a different model: One in-house pilot that has remained within the local authority as a separate and discrete unit, two professional practices, one run by a private company and the other run as a social enterprise or mutual (a business that helps the community but owned directly by those involved in its operation), and two practices run by charities. The sixth planned pilot failed to start as the host local authority was diverted by a negative inspection report. The smaller and more autonomous structure of the pilots, compared to the more conventional local government social work setting, was intended to free social workers’ time for more direct work with children and their families and to enhance their decision autonomy within a reduced hierarchy (for full discussion of SWP structure and how it differs from the traditional working model, see Stanley et al., 2013).

The U.K. Government commissioned an independent longitudinal matched-control process and outcome evaluation of the pilots (2009–2012, for details of the study, see Stanley et al., 2012b). This article reports on part of this evaluation focusing on changes in practitioners’ job satisfaction, autonomy, and decision authority that could be attributed to working in the pilots for a year. The matched-control design allowed controlling for confounding factors by comparing the experience of practitioners in the SWP with two other groups: those working in the “traditional” environment who remained in their “host” local authorities and a comparison group working in matched sites. The analysis reports on findings from the surveys of children and family social care practitioners at two time points (prior to SWP start-up and 12 months after their onset) to examine the hypothesis that this new model of working positively influenced staff autonomy and morale, and reduced burnout levels.

Job Control and Burnout

To enable us to measure and compare these outcomes systematically between the three groups (SWP, host, and comparison), we used the validated scales of Karasek’s Job Content Questionnaire (JCQ; Karasek, 1979; Karasek, Brisson, et al., 1998), the Maslach Burnout Inventory (MBI), and job satisfaction scales. JCQ is a self-completed instrument designed to identify two crucial aspects: job demands—the stressors existing in the work environment—and job decision latitude—the extent to which employees have the potential to control their tasks and conduct throughout the working day (Karasek, Brisson, et al., 1998). The control-demand (CD) model postulates that job strain is the result of an interaction between demand and control, meaning that a job with high demand and low control would be labeled as “high strain” while a job with low demand and high control would be viewed as “low strain.” The JCQ social support scale combines both coworkers’ and supervisory support scales. Such support is theorized to moderate or buffer the impact of job-related stress (Karasek, Triantis, & Chaudhry, 1998); in particular, individuals in high stressor jobs will have lower psychological strain in the presence of social support. Karasek’s job strain model has been examined in relation to several different health outcomes, most notably cardiovascular disease (Hallqvist, Diderichsen, Theorell, Reuterwall, & Ahlbom, 1998; Landsbergis & Theorell, 1999). It was also used in studies of risk factors for work-related musculoskeletal disorders (Hoogendoorn, van Poppel, Bongers, Koes, & Bouter, 2000; Toomingas, Theorell, Michelsen, & Nordemar, 1997).

One definition of burnout is the exhaustion resulting from excessive demands on energy and resources (Lee & Ashforth, 1990). The MBI (Maslach & Jackson, 1984) is the most widely used measure of burnout. This 22-item, 6-point anchored Likert-type scale has three components: “emotional exhaustion” (EE; feelings of being overextended and depleted of emotional and physical resources); “depersonalization” (Dp; negative or excessively detached responses to various aspects of the job); and reduced “personal accomplishment” (PA; feelings of incompetence and a lack of achievement at work). A high degree of burnout is reflected in high scores of EE and Dp subscales and a low score on the PA subscale. Scales are computed separately and cannot be combined to form a single scale so the following sections are presented consecutively (Maslach, Jackson, & Leiter, 1996).

The three elements of MBI represent distinct, separate constructs of burnout. However, empirical research has shown that workers who are more emotionally exhausted will become more depersonalized and are more likely to leave their jobs (Koeske & Koeske, 1989; Lee & Ashforth, 1990). Although scores for PA of the MBI are “independent of the other two subscales and cannot be assumed to be the opposite of EE or DP” (Maslach & Jackson, 1984); they indicate a subjective feeling of professional success and job satisfaction.

Method

Participants

Practitioners were recruited through their employers who were invited to take part in the research. Employers produced a list of electronic contact details for the research team. Practitioners were contacted directly by the team with a request to complete an electronic survey with the option to opt out from participation. Data were collected from practitioners in the host and comparison authorities (see below for details on selecting the comparison group) at the onset of the SWPs (T1), then from practitioners in the pilot SWPs, host and comparison authorities a year later (T2: 12 months after the pilots’ onset). Practitioners completed the survey anonymously with a chance to win a prize draw of gift vouchers.

Response Rate

T1 survey achieved 50% response rate (n = 1,101 responses) and the response rate at T2 was 43% (n = 949 responses). To establish the representativeness of respondents, the research team requested aggregate information on key characteristics of children’s services staff regarding age, gender and, in some cases, ethnicity (although limited information was available in respect of this). These distributions were compared to those of respondents to the online survey and were found to match well, with a range of ±2%; thus, the representativeness of respondents in terms of age and gender was assured and no weighting was applied.

Sample

A total of 2,050 practitioners working with children in out-of-home care completed the survey at the two time points; Table 1 presents participants’ distribution by key characteristics. Two of the main differences between the pilots, host, and comparison sites were that SWP staff were less likely to be qualified social workers and more likely to be male. SWP staff were also younger at T2 with a mean age of 42 years (standard deviation [SD] = 10.2) compared to 46 and 44.5 years in the host and comparison sites (SD = 17.2 and 10.3, respectively).

Table 1.

Respondents’ Characteristics by Type of Site and Time.

Characteristics	Host Local Authorities		Comparison Sites		SWP Pilots
Characteristics	Time 1	Time 2	Time 1	Time 2	Time 2
Gender: women (%)	83.6	86.4	84.2	85.5	75.9
Valid total	389	491	355	365	58
Mean age	43.12	46.02	43.02	44.64	41.98
Standard deviation (SD)	10.09	17.19	10.56	10.3	10.15
Valid total	386	489	353	374	57
Ethnicity
White British (%)	74.2	77.9	82.9	80.8	78.4
White other (%)	5.2	7.2	5.6	6.4	3.9
Mixed (%)	6.3	3.7	2.2	1.0	3.9
Asian (%)	5.7	3.7	2.8	5.1	3.9
Black (%)	7.0	6.3	5.6	6.4	9.8
Any other (%)	1.6	1.2	0.8	0.3	0
Valid total (%)	384	430	357	312	51
Qualified social workers (%)	59.1	59.7	70.9	86.7	59.7
Valid total	536	495	492	375	57
Mean number of years in sector	11.69	13.74	13.26	14.49	11.22
SD	8.67	9.23	10.12	10.22	8.04
Valid total	537	497	492	375	58
Mean satisfaction with job (max 7)	4.74	4.73	4.71	4.78	5.15
SD	1.33	1.24	1.2	1.19	1.27
Valid total	398	383	365	312	55
Mean satisfaction with employer (max 7)	4.49	4.19	3.75	4.14	5.15
SD	1.54	1.48	1.56	1.48	1.38
Valid total	398	383	365	312	55
Feel children enter the care system at the right time (%)	47.1	49.7	41.7	43.4	23.1
Valid total	446	435	403	339	52
Perceived to spend the right amount of time in direct work with children (%)	26.9	22.9	23.1	26.2	48.2
Valid total	472	454	438	351	58

Note. SWP = social work practice.

Material and Procedure

Study Design

The evaluation took place from December 2009 to March 2012, starting at the onset of the pilots with a matched-control longitudinal design to capture changes that could be attributed to the new SWP. A cohort study approach thus allowed us to examine the relationship between a postulated “cause,” in this case, the organizational change to social work pilots, and the “effect,” achieving some or all of desired outcomes, over a period of time. The experience of practitioners in the SWP was compared to that of their peers who remained working in the more conventional environment within the same locality: host local authority. However, to reduce possible “contamination” effect between the SWP and host group due to geographical proximity, we introduced a third comparison group: practitioners working with the similar groups of clients and providing similar services in other sites (for full details of the evaluation, see Stanley et al., 2012b). The comparison group allowed a rigorous program evaluation by providing an estimate of the desired effect without the exposure to the new work model (Howell & Yemane, 2006; Schlesselman, 1982).

Comparison Sites’ Selection

Six local authorities¹ were selected as comparators for the local authorities hosting the pilots according to a set of essential criteria supplemented by other desirable “key” criteria. Essential criteria included (1) General demographic criteria—the ratings accorded to local authorities in the Children’s Services Statistical Neighbour Benchmarking Tool (http://www.dcsf.gov.uk/rsgateway/DB/STA/t000712/index.shtml) and (2) key characteristics of the local looked after populations—the two DCSF indicators used for this were Local Area Agreement 1 (children in out-of-home care rate per 10,000 children aged under 18 years) and NI 62 (the percentage of children in out-of-home care with three or more placements. Desirable criteria included (1) Workforce criteria—the total social services social work vacancy rate as provided by a 2009 survey of local authorities carried out by Community Care magazine and (2) safeguarding criteria—to capture the wider activity of children’s services in relation to children in out-of-home care and to offer an indication of an authority’s threshold for entry into the care system. Each pilot site was initially matched to two or three possible comparison sites, these were approached and the first authority that agreed to participate in the evaluation was selected.

The evaluation applied a mixed-methods approach to capture and understand various perspectives including the views of children and young people, families, carers, local authority social workers, and SWP staff as well as professionals working in other agencies (see Stanley et al., 2012b, for details). Ethical approval for the evaluation was provided by the Institute of Education’s Research Ethics Committee and from research governance committees in participating local authorities.

Measures

The practitioners’ survey was piloted with practitioners working with children in out-of-home care services and the final version was distributed electronically to all staff identified by local authorities as working with children and families. The survey design was based on relevant literature and research on the social care workforce, and collected information on practitioners’ demographics, job specifics, workload, time allocation, and perceptions of autonomy and decision making as well as the two validated scales: Karasek’s “JCQ” and “MBI.” Participants were asked to score their satisfaction with their job and their employer separately on scales from 1 (low) to 7 (highest). The survey also collected precoded and free-text responses on practitioners’ perceptions of SWPs as organizations and their aspirations. The survey was designed and managed using the “Survey Monkey” online tool.

Analysis

To examine possible factors influencing burnout and job demand control, we imputed baseline data for the pilot responses at T1 based on matched responses from the host local authorities (matched on key demographics and job roles of respondents²). We used mixed-effect models³ to account for multilevel error structure (to separate the effect of cluster from that of individual participants) and for the effect of time; the models also accounted for covariance at baseline and different personal and organizational characteristics. We constructed a series of mixed-effect models using a framework of analysis to incorporate the importance of individual and organizational characteristics and individual participants’ beliefs and perceptions of different elements of care and job dynamics. Analyses were conducted using R software version 2.1 (R Development Core Team, 2007).

Results

Job Satisfaction

The average level of satisfaction with current jobs was 4.72 (σ = 1.27) for T1 and 4.78 (σ = 1.24) at T2. The differences between pilot, host, and comparison site participants and across time were not significant. In terms of satisfaction with employer, this was slightly lower at 4.14 (σ = 1.59) for T1 and 4.24 (σ = 1.50) at T2. At T1, satisfaction with employer was significantly higher among comparison than host sites; by T2, host and comparison sites’ levels of satisfaction were almost identical at 4.1 and 4.2. At T2, SWP staff’s level of satisfaction with their employers was relatively higher at an average of 5.2 (σ = 1.4) out of 7; these are significantly higher scores than those observed in the host and comparison groups, as well as higher than those found in the host sites at baseline.

JCQ: Decision Latitude

In the JCQ, decision latitude is the summation of two main elements: decision authority and skills’ discretion. Table 2 shows positive change in overall decision latitude among participants from the comparison sites (significant change $f r o m \overset{ˉ}{X} = 68 . 37 t o 69 .0 7$ ). Scores for SWP participants were higher than those for host and comparison T2 groups. SWP T2 decision latitude score (71.13) was the closest to the norm for social workers in the United States as calculated by Karasek (1985): $\overset{ˉ}{X} = 71.9$ .

Table 2.

Karasek Subscale Scores for Pilot, Host, and Comparison Groups at Time 1 and Time 2.

JCQ Subscale	Comparison Site		Host Local Authority		Pilot SWP	All
JCQ Subscale	T1	T2	T1	T2	T2	T1	T2
Decision latitude scores
Mean ( $\overset{ˉ}{X}$ )	68.37	69.07	66.87	67.2	71.13	67.59	68.27
SD (σ)	8.67	9.06	9.68	9.36	9.24	9.23	9.3
N	365	312	397	382	55	762	749
CI upper	69.26	70.08	67.82	68.14	73.57	68.25	68.94
CI lower	67.48	68.06	65.92	66.26	68.69	66.93	67.6
Psychological job demand
Mean ( $\overset{ˉ}{X}$ )	39.24	38.08	38.55	37.69	36.69	38.88	37.78
SD (σ)	5.85	5.63	5.91	5.87	5.55	5.89	5.76
N	365	312	397	382	55	762	749
CI upper	39.84	38.7	39.13	38.28	38.16	39.3	38.19
CI lower	38.64	37.46	37.97	37.1	35.22	38.46	37.37
Job insecurity
Mean ( $\overset{ˉ}{X}$ )	5.59	5.69	5.43	5.94	5.98	5.51	5.84
SD (σ)	2.18	2.17	2.41	2.37	1.39	2.3	2.23
N	364	313	398	382	55	762	750
CI upper	5.81	5.93	5.67	6.18	6.34	5.67	6
CI lower	5.37	5.45	5.19	5.7	5.62	5.35	5.68

Note. CI = confidence interval; JCQ = Job Content Questionnaire; SD = standard deviation; SWP = social work practice.

The full results of the final mixed-effect model examining JCQ decision latitude are presented in Table 3. Model 1 in Table 3 indicates no significant differences in decision latitude scores between SWP pilot and host sites (β = 1.46; p = .133); there was also negligible effect over time. The factors that significantly improved the decision latitude scores across the sample were as follows: belonging to “White” ethnicity (β = 2.74; p < .001); having higher job satisfaction scores (β = 2.64; p < .001); and to a lesser extent higher scores of employer satisfaction (β = 0.50; p = .011). Older participants reported significantly lower scores of decision latitude (β = −0.06; p = .006).

Table 3.

Results of Final Mixed-Effect Models for Karasek Decision Latitude, Psychological Job Demand and Maslach Emotion Exhaustion, Depersonalization, and Personal Accomplishment.

	JCQ				MBI
	Model 1 Decision Latitude		Model 2 Job Demand		Model 3 Emotional Exhaustion		Model 4 Depersonalization		Model 5 Personal Accomplishment
	SD	% var	SD	% var	SD	% var	SD	% var	SD	% var
Random effects
Groups
Site	0.589	0.51	0.269	0.25	0.155	2.26	0.197	0.26	0.001	<0.001
Residual	8.223	99.49	5.354	99.75	1.019	97.74	3.893	99.74	5.112	99.99
Variables	β	p Value	β	p Value	β	p Value	β	p Value	β	p Value
Fixed effects
Time	0.489	.309	−0.957	.002*	0.004	.951	−0.074	.816	0.012	.921
Age^a	−0.063	.006**	−0.005	.789	0.009	.002*	−0.047	.009**	0.055	<.001*
Comparison vs. host	1.389	.053	0.285	.463	0.186	.064	0.014	.815	0.026	.934
Pilot vs. host	1.462	.133	−0.944	.141	−0.018	.886	0.379	.156	0.895	.141
White ethnicity vs. BME	2.735	<0.001*	−0.658	.133	−0.069	.416	0.785	.014**	−0.925	.028**
Men vs. women	0.657	.296	0.015	.971	−0.060	.446	0.894	.003*	−0.374	.340
Length of work in SC^a	0.017	.513	−0.043	.009**	0.004	.917	−0.032	.047**	0.042	.044**
Social worker vs. other roles	−0.806	.136	2.163	<.001*	2.303	<.001*	0.271	.285	−1.120	<.001*
Satisfaction with Job^a	2.638	<.001*	−1.094	<.001*	−4.335	<.001*	−1.199	<.001*	1.785	<.001*
Satisfaction with employer^a	0.496	.011**	−0.134	.294	−0.612	.001*	0.048	.604	0.016	.897
Perception that children do not enter the system at the right time vs. other	−0.034	.945	0.561	.079	0.865	.057	0.891	<.001*	−0.310	.307
Perception of spending the right amount of time in direct work with children vs. other	0.741	.180	−2.032	<.001	−1.387	.007	−0.245	.351	0.655	.058

Note. BME = Black and minority ethnic; JCQ = Job Content Questionnaire; MBI = Maslach Burnout Inventory; SC = social care; SD = standard deviation.

^aContinuous variables; a total of 1,229 cases were used in each model after crosswise deletions.

*Significant at p < .005.

**Significant p < .05.

JCQ: Psychological Job Demand

Findings presented in Table 2 indicate some reductions in psychological job demand scores from T1 to T2 for both the comparison and the host groups, though these changes did not appear to be significant. The overall scores of SWP participants were lower than others; however, all confidence intervals (CIs) overlap, so there were no clear significant differences. For all groups, including the SWPs, job demand scores appeared considerably higher than those observed for social workers by Karasek in the U.S. context some decades previously (32; σ = 8.4; Karasek, 1985).

Factors that reduced psychological job demand included time (β = −0.96; p = .002); experience (measured as length in social care sector; β = −0.04; p = .009); job satisfaction (β = −1.09; p < .001); and participants’ perception that they spent the right amount of time in direct work with children and young people (β = −2.03; p < .001; see Model 2 in Table 3). On the other hand, social workers scored significantly higher than nonsocial work staff (such as support workers or nonqualified practitioners) with regard to psychological job demand (β = 2.16; p < .001); and participants who felt that children did not enter the public child care system at the right time also had higher scores of borderline significance (β = 0.56; p = .079).

There were no strong correlations between job control and demand among participants, while scatter plots of both scores indicated that level of decision latitude (control) remained constant while job demand increased. Among SWP staff there was a negative (but not significant) correlation between control and demand, indicating that, as job demand increased, their job control decreased.

JCQ: Job Insecurity

Table 2 shows that overall levels of job insecurity over time remained almost identical among comparison site participants; however, they were slightly higher among SWP staff, especially when compared to host T1 results. Overall, all job insecurity scores were high. Examining CIs of these scores at T2, job insecurity mean scores for SWP staff (M _js = 5.98; CI [5.62, 6.34]) were not significantly higher than those observed in the host (M _js = 5.94; CI [5.70, 6.18]) and comparison sites (M _js = 5.69; CI [5.45, 5.93]). However, they were significantly higher than those found at baseline in the host sites.

JCQ: Social Support

Figure 1 indicates that social support scores were not significantly different between host and comparison groups and changed very little within each group from T1 to T2. However, social support scores were significantly higher among SWP staff. The overall social support mean score for all survey participants at T2 was 24.9, which is slightly lower than the norm for other social workers in the Karasek model; but in the SWPs, it was identical to that norm (26.3; Karasek, 1985).

Figure 1.

Karasek scores and confidence intervals of social support by site groups and time.

MBI: EE

Table 4 shows that levels of EE for all groups of participants and over time were within the “average” levels of burnout (the middle range for social services was 17–27; Maslach et al., 1996, p. 6). The results of the mixed-effect models for different elements of MBI listed in Table 3 (Models 3-5) indicate that negligible proportions of variance in EE scores were attributable to site effects (2%), with no significant differences between pilot, host, and comparison groups. However, higher job and employer satisfaction and feelings of spending the right amount of time in direct work significantly reduced participants’ levels of EE (β = −4.34, p < .001; β = −0.61, p = .001; and β = −1.39, p = .007, respectively). On the other hand, being a qualified social worker (as opposed to other roles such as support workers) and the perception that children in care do not enter the system at the right time significantly increased levels of EE (β = 2.30, p < .001 and β = 0.86, p = .057, respectively).

Table 4.

Maslach Scores Statistics by Site Group and Time.

MBI Subscales	Comparison		Host		Pilot	All
MBI Subscales	T1	T2	T1	T2	T2	T1	T2
EE scores
Mean ( $\overset{ˉ}{X}$ )	24.25	22.47	23.31	22.58	20.05	23.76	22.35
SD (σ)	9.7	8.93	10.66	9.99	10.26	10.22	9.59
N	365	312	397	382	55	762	749
CI upper	25.25	23.46	24.36	23.58	22.76	25.25	23.46
CI lower	23.25	21.48	22.26	21.58	17.34	23.25	21.48
Dp scores
Mean ( $\overset{ˉ}{X}$ )	7.22	7.16	6.83	6.66	5.6	7.01	6.79
SD (σ)	4.42	4.19	4.48	4.1	4.24	4.45	4.16
N	365	312	397	382	55	762	749
CI upper	7.67	7.62	7.27	7.07	6.72	7.67	7.62
CI lower	6.77	6.70	6.39	6.25	4.48	6.77	6.70
PA scores
Mean ( $\overset{ˉ}{X}$ )	30.76	31.04	30.84	31.6	32.8	30.8	31.46
SD (σ)	5.49	5.68	6.23	5.4	6.23	5.88	5.59
N	365	312	397	382	55	762	749
CI upper	31.32	31.67	31.45	32.14	34.45	31.32	31.67
CI lower	30.20	30.41	30.23	31.06	31.15	30.20	30.41

Note. CI = confidence interval; Dp = depersonalization; EE = emotional exhaustion; MBI = Maslach Burnout Inventory; PA = personal accomplishment; SD = standard deviation.

MBI: Dp

The analysis showed almost no change in Dp mean scores for participants from comparison and host groups, with levels of Dp significantly higher among the comparison group (Table 4). What is clearly different is the score of SWP staff, which was significantly lower at 5.6. The results of mixed-effect models (see Model 4 in Table 3) confirm that SWP participants had significantly lower Dp scores (β = −1.29; p = .006); however, the results also showed that over a quarter of observed variance in Dp scores was attributable to unmeasured individual site effects highlighting the importance of overall structure and dynamics of the workplace. Other measured variables that significantly improved Dp scores included participants’ age (β = −0.05; p = .009); experience (β = −0.03; p = .047); and job satisfaction (β = −1.20; p < .001). On the other hand, men and participants of White ethnicity had significantly higher levels of Dp (β = 0.894, p = .003 and β = 0.78, p = .014, respectively). Staff who felt that children do not enter the system at the right time had significantly higher levels of Dp (β = 0.891; p < .001).

MBI: PA

Table 4 indicates that levels of PA were almost unchanged among participants from both the comparison and the host groups from T1 to T2, albeit with small upward movements. A sense of PA among SWP participants was greater than the others at T2 but the differences were not significant. All scores of PA were, however, on the “middle range of social services” norm (30–36; Maslach et al., 1996). The results of the mixed-effect models (Model 5 in Table 3) show that older participants, those with higher job satisfaction, and those who feel that they spend the right amount of time in direct work with children in out-of-home care all have significantly higher PA scores (β = 0.05, p = .001; β = 1.78, p ≤ .001 and β = 0.67, p = .058, respectively). Being of White ethnicity or being a social worker significantly reduced PA scores (β = −0.93, p = .028 and β = −1.12, p < .001, respectively).

Discussion and Application to Practice

The study is limited in a number of ways: First, the follow-up period of 12 months is relatively short to fully capture possible changes and lagged effects of job climate and cultural changes. Second, the evaluation design did not follow the same individuals over time; rather it collected information from the same “unit” of work overtime, which may result in some possible averaging errors. Both these elements were beyond the control of the evaluation as the first related to the funding period of the research and the second was associated with feasibility of work, particularly in the social work profession where staff turnover is relatively high.

Additionally, better scores of SWP practitioners in relation to job satisfaction and social support may be attributed to a “Hawthorne effect” (McCarney et al., 2007) which entails improved performance in groups that are the objects of study; the shorter life span of the pilots, which limited exposure to workplace problems, may also explain such differences. Other research also shows the importance of social embeddedness in the workplace and work community in perceptions of high level of support (Hayton, Camabuci, & Eisenberger, 2011), and SWP staff may have experienced this as a sense of embodying a new model of working that had received government support; however, only one of the SWPs can be considered as a true mutual model owned by those directly involved in its operation (Mutuals Taskforce, 2012; Stanley et al., 2013).

The survey response rate of 43–50%, while within the expected range for this type of survey (Nulty, 2008), may pose potential selection bias where certain groups of participants may be more likely to participate in the survey. However, participants’ distribution on key characteristics (such as age and ethnicity) was similar to that of the overall sites; thus, we can expect responses are representative of the population of practitioners working with children in out-of-home care in the selected sites. Another possible reporting bias may relate to the overrepresentation of men within SWPs.

This study attempted to measure the impact of a new model of working with children in out-of-home care that aimed to improve the culture and climate of social work. It specifically examined the hypothesis that the new SWP model of working would improve practitioners’ job control and reduce their burnout level over a period of a year due to associated organizational changes. Using two comparison groups (those remaining within the host local authorities and those working in other comparable sites and performing similar roles), the results indicated that working in the new SWP model did not have any significant effect on practitioners’ job demand or their decision latitude, while having the perception of spending the right amount of direct work with children significantly reduced workers’ psychological job demand as measured by Karasek JCQ.

The results indicate that high level of job satisfaction significantly improved decision latitude and reduced workers’ psychological job demand; however, when controlling for other factors, the effect of working in the SWP was not significantly associated with either. On the other hand, being a social worker, compared to other practitioners working with children in out-of-home care, significantly increased workers’ psychological job demand, perhaps related to the high levels of responsibility that social workers assume. The finding that men scored significantly higher on the Maslach Dp scale may highlight the effect of their overall minority status in this feminized profession and warrants further research.

The positive relation with time in reducing levels of job demand across all workers is likely to be related to the changing image of social work in England from 2009 to 2012. While in 2009 there was great deal of negative media coverage following the death of 17-month-old Peter Connelly from multiple injuries (Stevenson, 2009), by 2011, social work in England was projecting a more positive public image. This was evident in the establishment by government of The College of Social Work and television documentaries such as Kids in care and Protecting our children, which portrayed social workers sympathetically (British Broadcasting Corporation [BBC], 2010, 2011). Shortages of social workers were also being resolved as noted above, not least by the major investment in the social work degree in England and by overseas recruitment. These contextual factors may help explain why the demand-control balance seemed to improve among workers who reported higher levels of job satisfaction regardless of levels of satisfaction with their particular employer (the latter was not significantly associated with any of CD elements).

Indicative findings of negative correlation between Karasek control and demand among SWP practitioners with the possibility of them moving to a state of “unresolved stress” (Karasek, 1985) are consistent with the findings from staff interviews, which showed that, in some SWP sites, control (e.g., over budgets) was reduced over time as numbers of children and young people in individual workers’ caseloads increased (Stanley et al., 2012b).

One of the main differences observed among SWP pilot staff was their high level of perceived social support when compared to their counterparts in host and comparison sites. SWP staff were more likely than the others surveyed to report that both coworker and supervisor support were higher and it is this that contributed to feelings of higher social support. These findings are consistent with the significantly higher levels of agreement about the priority given to supervision among SWP pilot staff when compared to both host and comparison groups (see Stanley et al., 2012b, for details).

Very little U.K. research has used MBI to measure burnout among social service workers although, in related fields, Evans et al. (2006) surveyed mental health social workers; and Pajak, Mears, Kendall, Katona, and Medina (2003) surveyed psychiatrists in England and Wales. Participants in the SWP evaluation scored lower in EE and Dp, but these scores were not significantly different from those participating in these two studies of related professionals. For example, at T2 EE scores were 22.5 (comparison); 22.58 (host); and 20.05 (pilot) compared to 26.3 (Evans et al., 2006) and 24.2 (Pajak, Mears, Kendall, Katona, & Medina, 2003) with overlapping CIs. However, our participants from all sites scored significantly lower in the PA element than in the other two studies of different professional groups (31; 31.6 and 32.8, respectively, compared to 33.9 in Evans et al., 2006 and 36 in Pajak et al., 2003). The latter concurs with findings from the United States that children’s service workers are one of the staff groups most prone to burnout (Drake & Yadama, 1996). Here, the PA element is the most affected.

The findings indicated that while SWP staff appeared to score better than host and comparison staff on all elements of MBI, when controlling for other factors, these differences were not significant. For example, the overall lower mean score of Dp among SWP practitioners may reflect the novelty of their work and enthusiasm about employing a “new” model. Lower Dp levels indicate a lower level of burnout—the standard “lower third” for social services staff is ≤5 (Maslach et al., 1996), meaning that all scores from different groups including the pilots (M _Dp = 5.6) were at the average score, while the SWP staff scored on the lower band.

However, as with the case of CD, the mixed-effect models indicate that individual site effects are almost negligible in explaining the variation in burnout scores, highlighting the importance of the inherently stressful nature of working in children’s social services. Other important factors such as age, experience, and degree of satisfaction with the job were more important than whether participants worked in the pilots, hosts, or comparison sites. Workers’ own perceptions of whether they spend the right amount of time in direct work with children in out-of-home care and how early their local authorities identified and responded to the needs of children in out-of-home care also played an important role, albeit to a lesser extent. The findings highlight the importance of job stress management strategies in social work with children, such as emotion-focused coping when stress sources are external and outside the person’s control (Anderson, 2000).

There may be other influences at play; findings from other elements of this evaluation indicated that the SWPs received substantial free external support with their professional practice and training, as well as help with setup (Stanley et al., 2012a, 2012b); this needs to be acknowledged, and successor SWPs may not be so advantaged.

Some of the findings reported in this article appear consistent with changes to the image, status and perception of children’s social work in England from 2009 to 2012. The surveys revealed differences between participants but also similarities. Chief among these is the finding that decision latitude did not greatly differ between SWP staff and others. However, notable differences emerged related to job insecurity and social support. By their nature, SWPs and other outsourced local authority services (in the commercial and third sectors) are likely to give rise to some insecurity around contracts, which involve negotiations and the risks of nonrenewal.

One of the main messages from this element of the SWP evaluation reported and discussed in this article is the similarity between the responses of SWP staff and their comparators on most of the JCQ and MBI subscales, with the exception of social support, where SWP staff scored relatively well and, to a lesser extent, job insecurity, where they scored less well than host authority staff had done at the evaluation outset. The findings highlight the importance of key work practices in influencing practitioners’ job control and levels of burnout, for example, “spending the right amount of time in direct work with children” and the perception that “children are entering the system at the right time.” The findings also highlight the significant effect of workers’ satisfaction with their job and perceptions of how well they and their establishments met the needs of children in out-of-home care. This latter set of elements appeared to be achievable in different work environments and was not restricted to certain models of organization or ownership. Further research may focus on identifying key organizational factors that enhance practitioners’ autonomy and reduce burnout that are replicable within different work structures, including social enterprises.

Footnotes

Acknowledgment

We are most grateful to all those who participated in the study and assisted with data collection.

Authors’ Note

The views expressed here are those of the authors alone and should not be attributed to the Department of Education.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was funded and supported by the Department for Education, UK.

Notes

References

Anderson

D. G.

(2000). Coping strategies and burnout among veteran child protection workers. Child Abuse and Neglect, 24, 839–848. doi:10.1016/S0145-2134(00)00143-5

British Broadcasting Corporation. (2010). Kids in care. Panorama. Retrieved October 14, 2010, from http://www.bbc.co.uk/programmes/b00v77vn

British Broadcasting Corporation. (2011). Protecting our children. BBC2 January–February 2012. Retrieved from http://www.bbc.co.uk/programmes/b01bskrq/episodes/guide

Brotheridge

Grandey

(2002). Emotional labor and burnout: Comparing two perspectives of “People Work.” Journal of Vocational Behavior, 60, 17–39. doi:10.1006/jvbe.2001.1815

Department of Children, Schools and Families. (2006). Care matters: Transforming the lives of children and young people in care (Cm 6932). Norwich: TSO. Retrieved from https://www.education.gov.uk/publications/standard/publicationDetail/Page1/CM%206932

Department of Children, Schools and Families. (2008). 2020 children and young people’s workforce strategy: Evidence base (DCSF-01054-2008). London, England: Crown Copyright. Retrieved from https://www.education.gov.uk/publications/standard/_arc_Childrensworkforce/Page1/DCSF-01054-2008

Drake

Yadama

(1996). A structural equation model of burnout and job exit among child protective services workers. Social Work Research, 20, 179–187.

Evans

Huxley

Gately

Webber

Mears

Pajak

… Katona

(2006). Mental health, burnout and job satisfaction among mental health social workers in England and Wales. British Journal of Psychiatry, 188, 75–80.

Glisson

James

(2002). The cross-level effects of culture and climate in human service teams. Journal of Organizational Behavior, 23, 767–794.

10.

Hallqvist

Diderichsen

Theorell

Reuterwall

Ahlbom

(1998). Is the effect of job strain on myocardial infarction risk due to interaction between high psychosocial demands and low decision latitude? Results from Stockholm Heart Epidemiology Program (SHEEP). Social Science Medicine, 46, 1405–1415.

11.

Hayton

Camabuci

Eisenberger

(2011). With a little help from my colleagues: A social embeddedness approach to perceived organizational support. Journal of Organizational Behavior, 33, 235–249.

12.

Hoogendoorn

van Poppel

Bongers

Koes

Bouter

(2000). Systematic review of psychosocial factors at work and private life as risk factors for back pain. Spine, 25, 2114–2125.

13.

Howell

Yemane

(2006). An assessment of evaluation designs: Case studies of 12 large federal evaluations. American Journal of Evaluation, 27, 219–236.

14.

Karasek

(1979). Job demands, job decision latitude, and mental strain: Implications for job redesign. Administrative Science Quarterly, 24, 285–308.

15.

Karasek

(1985). Job Content Questionnaire and user’s guide. Lowell, England: University of Massachusetts Lowell.

16.

Karasek

Brisson

Kawakami

Houtman

Bongers

Amick

(1998). The Job Content Questionnaire (JCQ): An instrument for international comparative assessments of psychosocial job characteristics. Journal of Occupational Health Psychology, 3, 322–355.

17.

Karasek

Triantis

Chaudhry

(1998). Coworker and supervisor support as moderators of associations between task characteristics and mental strain. Journal of Occupational Behaviour, 3, 181–200.

18.

Koeske

(1989). Work load and burnout: Can social support and perceived accomplishment help? Social Work, 34, 243–248.

19.

Landsbergis

Theorell

(1999). Measurement of psychosocial workplace exposure variables. Self-report questionnaires. Occupational Medicine, 15, 163–171.

20.

LeCroy

Rank

(1986). Factors associated with burnout in the social services: An exploratory study. In Gillespie

(Ed.), Burnout among social workers (pp. 23–40). London, England: Haworth.

21.

Lee

Ashforth

(1990). On the meaning of Maslach’s three dimensions of burnout. Journal of Applied Psychology, 75, 743–747.

22.

Le Grand

(2007). Consistent care matters: Exploring the potential of social work practices. London, England: Department for Education and Skills.

23.

Lloyd

King

Chenoweth

(2002). Social work, stress and burnout: A review. Journal of Mental Health, 11, 255–265.

24.

Mackie

K. S.

Holahan

C. K.

Gottlieb

N. H.

(2001). Employee involvement management practices, work stress, and depression in employees of a human services residential care facility. Human Relations, 54, 1065–1092.

25.

Maslach

Jackson

(1984). Patterns of burnout among a national sample of public contact workers. Journal of Health and Human Resource Administration, 7, 189–212.

26.

Maslach

Jackson

Leiter

(1996). Maskach burnout inventory manual (3rd ed.). Mountain View, CA: Consulting Psychologists Press.

27.

McCarney

Warner

Iliffe

van Haselen

Griffin

Fisher

(2007). The Hawthorne effect: A randomized, controlled trial. BMC Medical Research Methodology, 7, 30. doi:10.1186/1471-2288-7-30

28.

Mutuals Taskforce. (2012). Public service mutuals: The next steps. London, England: Cabinet Office.

29.

Nulty

D. D.

(2008). The adequacy of response rates to online and paper surveys: What can be done? Assessment & Evaluation in Higher Education, 33, 301–314. doi:https://dx-doi-org.web.bisu.edu.cn/10.1080/02602930701293231

30.

Orme

McIntyre

Lister

P. G.

Cavanagh

Crisp

Hussein

… Stevens

(2009). What (a) difference a degree makes: The evaluation of the new social work degree in England. British Journal of Social Work, 39, 161–178.

31.

Pajak

Mears

Kendall

Katona

Medina

(2003). Workload and working patterns of consultant psychiatrists: An investigation into occupational pressures and burdens. London, England: Royal College of Psychiatrists’ Research Unit.

32.

R Development Core Team. (2007). R Data Import/Export v 2.1. Retrieved from cran.r-project.org

33.

Savicki

Cooley

(1994). Burnout in child protective service workers: A longitudinal study. Journal of Organizational Behavior, 15, 655–666.

34.

Schlesselman

(1982). Case-control studies, design, conduct, analysis. New York, NY: Oxford University Press.

35.

Searle

Patent

(2012). Recruitment, retention and role slumping in child protection: The evaluation of in-service training initiatives. British Journal of Social Work. doi:10.1093/bjsw/bcs043

36.

Shier

Graham

(2011). Work-related factors that impact social work practitioners’ subjective well-being: Well-being in the workplace. Journal of Social Work, 11, 402–421.

37.

Stanley

Austerberry

Bilson

Farrelly

Hargreaves

Hollingworth

… Strange

(2012b). Social work practices: Report of the national evaluation (Research Report DFE-RR233). London, England: Department for Education.

38.

Stanley

Austerberry

Bilson

Farrelly

Hargreaves

Hussein

… Ridley

(2012a). Establishing social work practices in England: The early evidence. British Journal of Social Work, 37, 57–79. doi:10.1093/bjsw/bcs101

39.

Stanley

Austerberry

Bilson

Farrelly

Hussein

Larkins

… Ridley

(2013). Turning away from the public sector in children’s out-of-home care: An english experiment. Children and Youth Services Review, 35, 33–39.

40.

Stevenson

(2009). Baby P review calls for overhaul of social services. Politics.co.uk. Retrieved from http://www.politics.co.uk/news/2009/03/12/baby-p-review-calls-for-overhaul-of-social-se

41.

Tham

(2007). Why are they leaving? Factors affecting intention to leave among social workers in child welfare. British Journal of Social Work, 37, 1225–1246.

42.

Toomingas

Theorell

Michelsen

Nordemar

(1997). Associations between self-rated psychosocial work conditions and musculoskeletal symptoms and signs. Stockholm MUSIC 1 Study Group. Scandinavian Journal of Work Environment and Health, 23, 130–139.

43.

Tracy

Bean

Gwatkin

Hill

(1992). Family preservation workers: Sources of job satisfaction and job stress. Research on Social Work Practice, 2, 465–478.

44.

Zapf

Knorz

Kulla

(1996). On the relationship between mobbing factors, and job content, social work environment, and health outcomes. European Journal of Work and Organizational Psychology, 5, 215–223.