Association Between Addition of Learning Support Professionals and School Performance

Abstract

This study capitalizes on the gradual introduction of learning support professionals (LSPs; e.g., school social workers) into an urban school district’s elementary schools (n = 71) between 2001–2002 and 2008–2009. The time series allowed for a school-level investigation of associations between the presence of LSPs and aggregate school achievement utilizing fixed-effects methodology. Schools with an LSP in a given year, as well as the cumulative years in which a school had an LSP, were both positively associated with the percentage of students who scored at or above proficient in reading achievement on the California Standards Test and were negatively associated with the cumulative number of years a school was in program improvement status, but not associated with mathematics achievement. Findings suggest that school social workers function as a school resource and indicate the utility of fixed-effects methodology in assessing the impact of social work services in schools.

Keywords

school social work student academic achievement fixed-effects methodology

Recent models of contemporary school social work practice suggest that school social workers may impact student academically related outcomes via several potential mechanisms (Early & Vonk, 2001; Frey & Dupper, 2005) . One mechanism is through direct interventions with students and/or their caregivers, such as case management or counseling. A second mechanism, which is often indirect, occurs through intervention strategies that alter the school settings in which students are embedded. Key examples of these school social work strategies include teacher consultation, as well as classroom-, peer-, and school-level programming (e.g., bullying prevention and intervention). A final set of mechanisms occurs via larger organizational structure and processes. School effects-oriented educational literature, for example, discusses the role of school resources (e.g., financial and human) on both student- and school-level academic outcomes (Cohen, Raudenbush, & Ball, 2003; Grubb, 2008). This literature suggests that, one or more social workers, as members of the school staff, can function as a school-level human resource, whose skills both individually and in combination with other school resources, may exert direct and indirect influences on student- and school-level achievement outcomes.

Both a narrative review of the literature (Early & Vonk, 2001) and a recent meta-analysis evaluating school social work interventions (Franklin, Kim, & Tripodi, 2009) indicate that a majority of studies reflect the first, direct mechanism of school social work practice effects. For example, 14 of the 21 studies included in the meta-analysis were in individual or small group formats targeting individual student change. The remaining seven studies reported on outcomes that can best be characterized as related to the second mechanism of effects such as classroom-level strategies, including a teacher consultation intervention, a conflict mediation program, and a pregnancy prevention curriculum. Overall intervention effect sizes generated from the meta-analysis were .23 for student externalizing symptoms and .40 for student internalizing symptoms.

While there is no doubt that the distribution of research on intervention strategies represents the emerging nature of research studies of sufficient methodological quality within this practice knowledge domain (Franklin, Kim, & Tripodi, 2009), the lack of research on the third potential mechanism of school social work effects represents a critical gap in the knowledge base. This gap may be, in part, explained by a lack of a school-effects perspective in the school social work literature (Phillippo & Stone, 2011) . “School effects” perspectives represent an important conceptual and methodological paradigm in educational research (e.g., Grubb, 2008; Monk, 1989.; Raudenbush & Bryk, 2002; Rutter & Maughan, 2002). This body of research typically investigates the association between aggregate school characteristics and individual student- and school-level achievement-related outcomes and aggregate school characteristics. These aggregate school characteristics (also sometimes termed resources; see Grubb, 2008) include (a) monetary factors, such as per pupil spending, (b) student body and teacher characteristics, (c) structural factors such as size and sector, and (d) specific school processes and practices (e.g., principal leadership).

In the school social work literature, this gap is manifest not only in terms of the outcome domains typically considered in school social work intervention research but also in terms of the overall conceptualization of potential school effects on school social workers and their practices and, conversely, in terms of school social worker effects on schools. School social work intervention research has not consistently considered student academic outcomes. Franklin, Kim, and Tripodi’s (2009) meta-analysis contained relatively few studies that considered effects on student grades and attendance and none considered standardized achievement scores. The relative paucity of school-based intervention effects on indicators of student academic performance in general also characterizes the larger school-based psychosocial intervention literature (Atkins, Hoagwood, Kutash, & Seidman, 2010; Hoagwood et al., 2007).

Few studies have directly considered the influence of school characteristics on school social workers and their practices, though there is some evidence that school characteristics relate to school social worker reports of their practice strategies, including whether they most often engage in individual case work with students versus other practice modalities (Kelly & Stone, 2009). Other work also demonstrates that school-level compositional and organizational features shape the delivery of psychosocial services in schools. The effects of evidence-based, cognitive–behavioral treatment on students, for example, are attenuated in schools characterized by high levels of disorganization (Gottfredson, Jones, & Gore, 2002). School features such as rates of concentrated poverty and teacher turnover, urbanicity, size, and principal support relate to the quality of implementation of school-based prevention programs (Payne, 2009).

Finally, recent research on school resources, though often not explicitly considered when discussion the roles and practices of school social workers, is increasingly seen as highly relevant to understanding the contribution of school social workers to the school as a whole (Phillippo & Stone, 2011). Grubb (2008) defines four categories of school resources that potential account for school effects on student academic outcomes. Simple resources reflect school “inputs” and are typically conceptualized as financial and human (e.g., per pupil revenues, student background characteristics, and teacher educational background). Compound resources are combinations of two or more resources (e.g., relatively high per pupil revenues and high percentage of well-qualified teachers). Complex resources refer to less tangible phenomena such as the overall instructional approach of the school staff, which often must be created through investments in professional development. Finally, abstract resources refer to the “web of relationships and practices” among school staff (Grubb, 2008, p. 108).

While Grubb’s (2008) work provides an important conceptual rationale that the addition of a school social worker to a school community could potentially shape school academic outcomes, there is also more direct evidence that suggests the plausibility of such an effect. Particular tasks and associated roles performed by school social workers strongly suggest that their efforts could function as a school-level resource. Surveys show that school social workers often perform administrative duties, coordinate community resources, and engage in leadership and policy-making activities in schools (Allen-Meares, 1977, 1993, 1994).

Additionally, research on urban school improvement provides evidence of the potential benefits of school features reflecting complex and abstract schools resources. For example, strong principal leadership, high levels of trust and strong working relationships among school staff, the extent to which staff hold high expectations and show strong support for students, the extent to which staff are oriented toward learning and innovation, and the extent to which staff are engaged with parents and the school community are associated with growth in student achievement (Bryk, Sebring, Allensworth, Luppescu, & Easton, 2009; Cosner, 2009; Hoy, Tarter, & Hoy, 2006). Importantly, school social workers have been argued to strongly contribute to school reform efforts, particularly in terms of their critical role in building such school resources (e.g., in strengthening parent, school, and community ties; Lewis, 1998; Teasley, 2004).

We could locate only one study that was directly relevant to understanding how school social workers functioned as a school resource. Bagley and Pritchard (1998) followed one elementary and one secondary school over 3 years wherein a teacher and social worker (a full-time position in the elementary school and a half-time position in the secondary school) were introduced in each school. Compared to a demographically similar elementary and secondary schools, schools with an added social worker had comparatively lower rates of student reported theft, truancy, bullying, substance use, and suspensions/expulsions than schools without these additional staffing resources. While the design-related limitations of this study are significant, including the small total number of schools studied, conflation of social worker and teacher effects, limited control of confounding variables, and reliance on student self-reports of outcomes, it provides preliminary evidence that the addition of a school social worker—as a staff resource—may have important influences on student academically related outcomes and, thus, at the very least, justifies empirical replication.

We suspect that replication of this work has not systemically occurred to date, in part, due to several design-related challenges. As noted above, given that a social worker in a school can function as a school resource in multiple ways that are related to the extent of other available resources in a given school, the nature of effects on indicators of student achievement are likely to be complex and difficult to detect. Scholars may be dissuaded from investigating such effects given their so-called modest entitivity (Cook, 2007, p. 137), although research investigating the relationship between school support services and student achievement is urgently needed (Hoagwood et al., 2007) . Furthermore, in addition to considering measures of academic achievement, successfully building upon prior research requires data from a sufficient number of schools over time, access to plausible control variables, and ability to reliably distinguish levels of the intervention (i.e., addition of a social worker vs. not).

The current study takes advantage of a natural opportunity to assess school social service provider effects in the San Francisco Unified School District (SFUSD). Between the school years 2001–2002 and 2008–2009, the district gradually added at least one half-time support staff (referred to as Learning Support Professionals [LSPs] by the district) to almost all each elementary schools in the district, ultimately allowing for a time series of support staff presence at schools. This relatively lengthy time series allows for the implementation of fixed-effects methodology to robustly estimate the association between presence of staff and school aggregated student achievement measures. Based on prior research, we hypothesized that addition of support staff would be associated with higher levels of reading and mathematics achievement. In addition, we hypothesized that schools with LSPs would accumulate fewer years of “Program Improvement” status. Federal education statutes, referred to as No Child Left Behind compel states to implement statewide accountability systems standards in reading and mathematics, annual testing for all students in Grades 3–8, and annual statewide progress goals ensuring that all groups of students reach proficiency within 12 years. Schools and districts that fail to make “adequate yearly progress (AYP)” toward statewide goals are subject to corrective measures. In California, Program Improvement is the designation for Title I-funded schools and local education agencies (e.g., districts) that fail to make AYP for two consecutive years (California Department of Education, 2011).

Method

Study Context and Timeframe

The study, which draws on school-level observational data, considers the 71 traditional elementary schools within the SFUSD that remained open between 2001–2002 and 2008–2009 (two closed or merged over the time period). Table 1 shows descriptive, school demographic, resource, and performance characteristics over the study period. Of note, SFUSD elementary schools are racially and ethnically diverse, including particularly substantial percentages of Asian and Latino children, as well as English language learners. Over the study period, the average proportion of children in schools receiving free and/or reduced lunch was 63%.

Table 1.

Sample Characteristics (n = 71 Schools)

	2002	2003	2004	2005	2006	2007	2008	2009
M (SD) or Percentage
Percentage at or above proficient, reading	37.13	37.02	37.62	41.59	44.99	45.33	47.24	51.56
Percentage at or above proficient, reading	(18.20)	(18.29)	(18.10)	(18.50)	(18.88)	(19.46)	(20.90)	(20.79)
Percentage at or above proficient, math	45.38	45.34	45.94	53.52	55.87	57.07	58.15	59.77
Percentage at or above proficient, math	(19.45)	(19.46)	(19.57)	(19.01)	(19.03)	(19.52)	(21.30)	(21.33)
Cumulative years in Program Improvement	0.06	0.21	0.41	0.65	0.77	0.94	1.13	1.21
Cumulative years in Program Improvement	(0.23)	(0.53)	(0.89)	(1.22)	(1.49)	(1.71)	(1.94)	(1.99)
School had an LSP (%)	4	31	28	35	37	45	97	97
Cumulative years with LSP	0.04	0.35	0.63	0.99	1.35	1.80	2.77	3.73
	(0.20)	(0.56)	(0.94)	(1.35)	(1.57)	(1.92)	(1.95)	(2.01)
	372.00	363.83	354.23	351.76	349.49	352.56	353.45	358.79
Enrollment	(135.12)	(136.47)	(139.16)	(142.30)	(143.68)	(142.90)	(144.73)	(140.80)
Enrollment	19.96	19.58	19.94	20.56	20.60	20.66	19.88	20.16
Average class size	(1.89)	2.12	2.33	2.29	2.39	2.56	2.95	2.87
Average class size	18.19	17.27	17.99	17.38	17.91	17.81	17.23	17.54
Student–teacher ratio	(2.15)	2.60	2.73	2.56	2.63	2.73	2.59	2.57
Student–teacher ratio	16.45	15.91	15.99	16.33	15.87	15.06	15.64	14.96
African American (%)	(17.18)	(17.49)	(18.27)	(19.10)	(18.92)	(17.86)	(18.91)	(17.84)
African American (%)	24.71	24.98	24.70	25.17	25.50	25.18	25.95	26.40
Latino (%)	(20.85)	(21.90)	(22.71)	(23.13)	(24.05)	(23.76)	(24.65)	(24.88)
Latino (%)	38.91	38.58	38.48	36.37	35.62	33.95	33.11	32.35
Asian (%)	(24.40)	(25.02)	(25.84)	(26.30)	(26.38)	(26.51)	(27.41)	(27.49)
Asian (%)	9.75	9.37	8.90	8.91	9.23	9.60	11.50	11.65
White (%)	(10.70)	(10.31)	(10.06)	(9.52)	(9.85)	(10.04)	(11.45)	(11.41)
White (%)	41.29	40.77	41.55	42.95	41.74	38.96	41.10	43.98
English learners (%)	(27.99)	(25.06)	(26.69)	(28.32)	(25.06)	(25.33)	(26.45)	(26.77)
English learners (%)	52.59	67.62	70.58	63.00	63.16	65.22	62.43	62.69
Free/reduced lunch (%)	(23.46)	(21.27)	(21.71)	(19.96)	(18.36)	(20.09)	(20.74)	(22.38)
Title 1 site (%)	80	82	82	80	80	80	73	73
Fully credentialed teachers (%)	98.44	94.21	96.88	97.23	98.12	98.22	98.81	99.36
Fully credentialed teachers (%)	(3.00)	(7.44)	(5.25)	(6.70)	(4.64)	(4.92)	(3.27)	(1.58)
New principal (%)	0	1	31	18	25	25	13	8
Star site (%)	37	37	37	37	37	32	32	32

Note. LSP = learning support professional.

In 2001, SFUSD initially matched funds with a private donor whose goal was to ensure that every school in the district ultimately had access to a social worker or, in some cases, another mental health service provider (e.g., marital and family therapists and nurses) referred to by the district as LSPs. Initially targeted to the lowest performing schools in the district, LSPs were gradually infused into a majority of elementary schools in the district. LSPs were directed by their district administrator to utilize a generalist model of psychosocial provision in schools. That is, based on their assessment of the school environment and student functioning, they were directed to intervene either indirectly or directly with individual students, given the unique characteristics of the school. This approach is also referred to in other literatures as a school public health model (e.g., Adelman & Taylor, 2006).

Measures

Measures of outcome, independent, and control variables in the study are all time varying and are summarized in Table 1. There are three outcome variables considered. The first two are the percentages of students in a school who have scores falling at or above proficiency levels in reading and mathematics achievement on the California Standards Test. The third outcome represents, in a given year, the number of years a school has been designated as being in “Program Improvement.”

The two key independent variables of the study, which are estimated separately, are (1) whether the school in a given year had an LSP and (2) in a given year, the cumulative total of years a school had an LSP.

Given the observational nature of these data, the current study also includes several measures of school structural, compositional, and resource characteristics that emerge directly from prior school effects research as related to student achievement outcomes (Grubb, 2008; Monk, 1989; Raudenbush & Bryk, 2002; Rutter & Maughan, 2002). Structural characteristics include total school enrollment, average class size, and student–teacher ratios. Student demographic compositional characteristics include the percentages of African American, Latino, Asian, and White students, the percentage of students who were English language learners, the percentage of students receiving free or reduced lunch, and a dichotomous indicator indicating whether the school was Title 1 (i.e., served a critical density of low-income students). Two proxies of resources are also utilized. The first is the percentage of fully credentialed teachers as an indicator of overall teacher quality. In addition, given the salience of principal support, we also include a dichotomous indicator indicating whether, in a given year, the principal was new. We control for an additional salient confound. Given that LSPs were introduced into schools when there were other ongoing accountability-related intiatives related to No Child Left Behind, we control for whether the school, in a given year, was designated a “STAR” school. The STAR initiative, a district-led school reform initiative, provided additional supports, such as additional teaching personnel and summer school programming, in chronically high-need schools.

Analytic Approach

Given the observational nature of our data, we utilize fixed effects methodology to adjust for bias emanating from time-invariant unobservable variables. Fixed-effects analytic methods require either panel or other forms of nested data (see Allison, 2009 for a general discussion). They represent a flexible method to control for omitted variable bias and, for example, are extensively used in the econometric analysis of educational outcomes (Babcock & Betts, 2009; Clotfelter, Ladd, & Vigdor, 2007; Hanushek & Rivkin, 2009; Lipscomb, 2007). This powerful method accounts for all time-invariant omitted variables (Wooldridge, 2009) by exploiting within-unit variation. For example, in our case, a model with school-level fixed effects would control for such factors as past history and the physical environment. In summary, the overarching goal of this approach is to remove all bias due to the correlation of included variables with omitted time invariant variables. Importantly, these methods only adjust for the time invariant effects of these variables. For these reasons, as noted above, we include measures of important time-varying confounds in our models implicated in the school effects literature. The general form of our models is as follows:

y_{i t} = μ_{t} + β_{x n i t} + α_{i} + ϵ_{i t}

where y _it is the outcome of interest for school i at time t, μ _t represents intercepts that differ for each time point (i.e., dummy variables representing each school year). The term β _xn _it denotes a vector of time-varying school characteristics, including LSP-related indicators and other school characteristics. For models estimating LSP effects on reading and mathematics achievement, we also control for the number of years a school has been in program improvement. When we estimate the number of years in program improvement, we control for reading achievement (mathematics achievement is highly related to and thus collinear with reading achievement, and reading achievement is more strongly related to the number of years in program improvement). Given that schools are nested in time in this data set, we utilize robust standard errors to adjust for clustering.

We also conducted several tests to confirm the robustness of our results to different specifications of the trend of the time series (e.g., using a simple linear trend). Additionally, given that virtually all schools had access to an LSP provider by the 2007–2008 school year and thus decreased variability in this respect, we also modeled results using shorter time series (i.e., using a shorter time series up to the 2007–2008 school year). Because results were identical, we present the longer time series, as this is recommended (Allison, 2009). Finally, we also confirmed that the LSP effect was not only confined to latter school years wherein relatively higher performing schools were provided with LSPs.

Findings

Table 2 shows results from three fixed-effects models estimating the relationship between key LSP effects and the percentage of students at or above proficient in reading and mathematics on the California Standards Test and the cumulative number of years a school was in program improvement. All models adjust for variables that have show relationships to student achievement in prior school effects research. Even after controlling for these potential confounders, we find that both having an LSP in a given year and the cumulative years in which a school had an LSP in a given year were both positively associated with the percentage of students who scored at or above proficient in reading on the California Standards Test. The Cohen’s d effect sizes, respectively, for each effect are .07 and .04, respectively, representing weak overall sizes. The same relationship was not observed for mathematics achievement. We also find that both having an LSP in a given year and the cumulative years in which a school had an LSP in a given year were both associated with fewer accumulated years in program improvement. The Cohen’s d effect sizes, respectively, for each effect are −.18 and −.20, respectively, representing small overall sizes.

Table 2.

Fixed-Effects Models Estimating LSP Associations (n = 71 Schools)

	Percentage at or Above Proficient in Reading		Percentage at or Above Proficient in Math		Years in Program Improvement
Had an LSP	1.36*		0.92		−0.25*
Had an LSP	(0.64)		(0.86)		(0.11)
Cumulative Years with LSP		0.75**		0.51		−0.29***
Cumulative Years with LSP		(0.29)		(0.39)		(0.05)
School year 02–03	1.98*	1.97*	0.72	0.71	0.33*	0.18
School year 02–03	(0.92)	(0.92)	(1.24)	(1.24)	(0.16)	(0.15)
School year 03–04	3.43***	3.32***	2.49⁺	2.42⁺	0.57**	0.39*
School year 03–04	(0.96)	(0.96)	(1.29)	(1.29)	(0.16)	(0.16)
School year 04–05	7.32***	7.12***	10.76***	10.63***	0.87***	0.58***
School year 04–05	(0.92)	(0.93)	(1.24)	(1.25)	(0.16)	(0.16)
School year 05–06	10.72***	10.37***	13.67***	13.44***	1.11***	0.74***
School year 05–06	(0.95)	(0.98)	(1.28)	(1.32)	(0.18)	(0.18)
School year 06–07	11.44***	10.88***	15.51***	15.13***	1.39***	0.88***
School year 06–07	(1.11)	(1.16)	(1.49)	(1.56)	(0.20)	(0.20)
School year 07–08	11.70***	11.24***	14.60***	14.30***	1.66***	0.75***
School year 07–08	(1.15)	(1.18)	(1.54)	(1.59)	(0.20)	(0.21)
School year 08–09	15.84***	14.72***	16.37***	15.62***	1.88***	0.71**
School year 08–09	(1.19)	(1.38)	(1.60)	(1.86)	(0.22)	(0.25)
Enrollment	0.00	0.01	0.00	0.01	0.00	0.00
Enrollment	(0.01)	(0.01)	(0.01)	(0.01)	(0.00)	(0.00)
Average class size	0.14	0.17	−0.39	−0.38	−0.02	−0.02
Average class size	(0.18)	(0.18)	(0.25)	(0.25)	(0.03)	(0.03)
Student–teacher ratio	0.08	0.00	−0.37	−0.42	−0.02	−0.03
Student–teacher ratio	(0.19)	(0.19)	(0.26)	(0.26)	(0.03)	(0.03)
African American (%)	0.01	0.02	0.10***	0.10	0.03	0.02
African American (%)	(0.11)	(0.11)	(0.15)	(0.15)	(0.02)	(0.02)
Latino (%)	0.31**	0.29**	0.21	0.20	0.04*	0.04*
Latino (%)	(0.11)	(0.11)	(0.15)	(0.15)	(0.02)	(0.02)
Asian (%)	0.36***	0.35***	0.50	0.50***	0.01	0.01
Asian (%)	(0.10)	(0.10)	(0.13)	(0.13)	(0.02)	(0.02)
White (%)	0.43***	0.40***	0.45**	0.43**	0.03	0.01
White (%)	(0.12)	(0.12)	(0.16)	(0.16)	(0.02)	(0.02)
English learners (%)	−0.08⁺	−0.09⁺	0.00	0.00	0.00	0.00
English learners (%)	(0.05)	(0.05)	(0.06)	(0.06)	(0.01)	(0.01)
Free or reduced lunch (%)	−0.08**	−0.08**	−0.03	−0.02	−0.01	−0.01
Free or reduced lunch (%)	(0.03)	(0.03)	(0.04)	(0.04)	(0.00)	(0.00)
Title 1	−3.74**	−3.74**	−4.17*	−4.17*	0.61*	0.45⁺
Title 1	(1.42)	(1.42)	(1.91)	(1.91)	(0.24)	(0.24)
Fully credentialed teachers (%)	0.15**	0.14*	0.11	0.10	0.01	0.01
Fully credentialed teachers (%)	(0.06)	(0.06)	(0.08)	(0.08)	(0.01)	(0.01)
New principal	−1.49*	−1.48**	−1.22	−1.22	−0.02	−0.06
New principal	(0.60)	(0.59)	(0.80)	(0.80)	(0.10)	(0.10)
Star site	−4.07*	−3.04	−7.33**	−6.64*	0.56	0.74*
Star site	(2.00)	(2.01)	(2.68)	(2.70)	(0.34)	(0.33)
Years in program improvement	−1.07***	−1.32***	−1.12**	−1.29***
Years in program improvement	(0.26)	(0.27)	(0.36)	(0.36)
Percentage at or above proficient in reading					−0.03***	−0.04***
Percentage at or above proficient in reading					(0.01)	(0.01)
Intercept	2.89	3.27	24.53	24.78	−0.58	−0.53
Intercept	(11.69)	(11.66)	(15.70)	(15.69)	(1.99)	(1.93)
Adjusted r ²	.95	.95	0.72	0.71	0.72	0.74

Note. LSP = learning support professional.

***p < .001. **p < .01. *p < .05. ⁺ p < .10

Discussion and Applications to Social Work

The current study attempted to fill important gaps in the literature related to school social service providers as school resources as well to address the paucity of achievement-related effects in prior school social work research. We found that both having an LSP in a given year and the cumulative number of years in which a school had an LSP were positively, albeit modestly, related to greater percentages of students reading at or above proficiency levels on the California Standards Tests as well as fewer years in program improvement over the study time period. It is important to note that these relationships were detectable over and above other variables previously associated with student achievement in school effects research. Similar relationships, however, were not found with mathematics proficiency levels. In summary, two of our three hypotheses were confirmed.

That relationships between presence of an LSP in a school and/or cumulative years a school had an LSP were not found is interesting. Overall, data from the National Assessment of Educational Progress suggest that mathematics achievement has grown more steadily than reading achievement over time (a trend that is also reflected in these data), suggesting that different combinations of instructional and school resources may account for gains (Rampey, Dion, & Donahue, 2009). Therefore, addition of social service providers may not add value over other school resources and characteristics in this particular academic outcome domain.

That LSPs were associated with reading achievement and the years accumulated in program improvement does support the notion that providers can add value in these areas and thus function as a core school resource. While it is important to note that the overall provider effects are quite modest, that these potentially complex and “modestly entitive” effects were observed using strongly controlled analytic methods suggests that they are important. Moreover, they are consistent with prior evidence suggesting that addition of social service providers is an important component of schools’ strategies to improve student achievement over time (Bagley & Pritchard, 1998). The similarity in absolute effect sizes for the presence of and LSP versus the cumulative years in which a school had an LSP suggests that LSPs may have an important initial value-added effect. Yet, these findings also point to the need for future research to understand exactly how the addition of social service providers may function as school resources. The rough indicators of LSP presence are not able to distinguish whether LSPs operate as simple, compound, complex, or abstract resources. Indeed, more fine grained analyses of exactly what roles school social workers played and how they distributed their time would be a logical next step for future analyses.

We also note three additional limitations. First, although one strength of this study is its relatively lengthy time series and a near population-level sample of schools in this district, it is important to remember the relatively small overall sample size of schools in these analyses (n = 71). Second, there is considerable variation by school, district, and state in terms of how (and whether) school social workers are employed and utilized in school settings (Kelly et al., 2010). Per pupil ratios of school social workers in California are among the lowest in the nation (AB 722 Study Work Group, 2003). Given this context, findings generated from the SFUSD may not be easily generalized to other districts and states. Finally, although the current study represents an important attempt to rigorously adjust for bias emanating from omitted variables using fixed-effects methodology, it is important to note that there are likely other time-varying omitted variables that are not accounted for in the current study. Although we control for a set of plausible confounds, we cannot rule out that the LSP effects we estimated may be biased. Nevertheless, we provide additional evidence that suggests the potential fruitfulness of our conceptual and analytic approach in future work on school social work effects, particular in terms of how school social workers may function as school-level resources.

Footnotes

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 received no financial support for the research, authorship, and/or publication of this article.

References

AB 722 Study Work Group. (2003). Study of pupil personnel ratios, services and programs. Sacramento: California Department of Education.

Adelman

Taylor

(2006). Mental health in schools and public health. Public Health Reports, 121, 294–298.

Allen-Meares

(1977). Analysis of tasks in school social work. Social Work, 22, 197–201.

Allen-Meares

(1993). Pull and push: Clinical or macro interventions in schools. Social Work in Education, 15, 3–5.

Allen-Meares

(1994). Social work services in schools: A national study of entry-level tasks. Social Work, 39, 560–565.

Allison

P. D.

(2009). Fixed effects regression models. Series: Quantitative applications in the social sciences. Thousand Oaks, CA: Sage

Atkins

Hoagwood

Kutash

Seidman

(2010). Toward the integration of education and mental health in schools. Administration and Policy in Mental Health and Mental Health Services Research, 37, 40–47.

Babcock

Betts

J. R.

(2009). Reduced-class distinctions: Effort, ability, and the education production function. Journal of Urban Economics, 65, 314–22.

Bagley

Pritchard

(1998). The reduction of problem behaviours and school exclusion in at-risk youth: An experimental study of school social work with cost-benefit analyses. Child and Family Social Work, 3, 219–26.

10.

Bryk

A. S.

Sebring

P. B.

Allensworth

Luppescu

Easton

J. Q.

(2009). Organizing schools for improvement: Lessons from Chicago. Chicago, IL: University of Chicago Press.

11.

California Department of Education (2011). 2011 Adequate Yearly Progress Report Guide. Sacramento, CA: Author.

12.

Clotfelter

C. T.

Ladd

H. F.

Vigdor

(2007). Teacher credentials and student achievement: Longitudinal analysis with student fixed effects. Economics of Education Review, 26, 673–682.

13.

Cohen

D. K.

Raudenbush

S. W.

Ball

D. L.

(2003). Resources, instruction and research. Educational Evaluation and Policy Analysis, 25, 1–24.

14.

Cook

T. J.

(2007). School based management: A concept of modest entitivity with modest results. Journal of Personnel Evaluation in Education, 20, 129–145.

15.

Cosner

(2009). Building organizational capacity through trust. Educational Administration Quarterly, 45, 248–291.

16.

Early

T. J.

Vonk

M. E.

(2001). Effectiveness of school social work from a risk and resilience perspective. Children & Schools, 23, 9–32.

17.

Franklin

Kim

J. S.

Tripodi

S. J.

(2009). A meta-analysis of published school social work practice studies. Research on Social Work Practice, 19, 667–677.

18.

Frey

Dupper

(2005). A broader conceptual approach to clinical practice for the 21st century. Children and Schools, 27(1), 33–44.

19.

Gottfredson

G. D.

Jones

E. M.

Gore

T. W

(2002). Implementation and evaluation of a cognitive–behavioral intervention to prevent problem behavior in a disorganized school. Prevention Science, 3, 43–52.

20.

Grubb

W. N.

(2008). Multiple resources, multiple outcomes: Testing the “improved” school finance with NELS88. American Educational Research Journal, 45, 104–144.

21.

Hanushek

E. A.

Rivkin

S. G.

(2009), Harming the best: How schools affect the black-white achievement gap. Journal of Policy Analysis and Management, 28, 366–393.

22.

Hoagwood

K. E.

Olin

S. S.

Kerker

B. D.

Kratochwill

T. R.

Crowe

Saka

(2007). Empirically based school interventions targeted at academic and mental health functioning. Journal of Emotional and Behavioral Disorders, 15, 66–92.

23.

Hoy

Tarter

Hoy

(2006). Academic optimism of schools: A force for student achievement. American Educational Research Journal, 43, 425–446.

24.

Kelly

M. S.

Berzin

S. C.

Frey

Alvarez

Shaffer

O’Brien

(2010). The state of school social work: Findings from the National School Social Work Survey. School Mental Health, 2, 132–141.

25.

Kelly

Stone

(2009). An analysis of factors shaping interventions used by school social workers. Children in Schools, 31, 163–176.

26.

Lewis

M. R.

(1998). The many faces of school social work practice: Results from a research partnership. Social Work in Education, 20, 177–190.

27.

Lipscomb

(2007). Secondary school extracurricular involvement and academic achievement: A fixed effects approach. Economics of Education Review, 26, 463–472.

28.

Monk

(1989). The education production function: Its evolving role in policy analysis Educational Evaluation and Policy Analysis, 11, 31–45.

29.

Payne

A. A.

(2009). Do predictors of the implementation quality of school-based prevention programs differ by program type? Prevention Science, 10, 151–167.

30.

Phillippo

Stone

(2011). Toward a broader view: A call to integrate knowledge about schools into school social work research. Children and Schools, 33(20), 71–81.

31.

Rampey

B. D.

Dion

G. S.

Donahue

P. L.

(2009). NAEP 2008 trends in academic progress (NCES 2009–479). Washington, D.C.: National Center for Education Statistics, Institute of Education Sciences, U.S. Department of Education.

32.

Raudenbush

S.W.

Bryk

A.S.

(2002). Hierarchical Linear Models: Applications and Data Analysis Methods. Thousand Oaks, CA: Sage.

33.

Rutter

Maughan

(2002). School effectiveness findings, 1979-2002. Journal of School Psychology, 40(6), 451–475.

34.

Teasley

(2004). School social workers and urban education reform with African American children and youth: Realities, advocacy, and strategies for change. School Community Journal, 14, 19–38.

35.

Wooldridge

J. M.

(2009). Introductory econometrics: A modern approach. Mason, OH: Southwestern-Cengage Learning.