Factors Influencing School Social Work Practice: A Latent Profile Analysis

The present study uses a national sample (N = 3769) and a latent profile framework to examine profiles of school social workers who engage in various levels of ecologically oriented practices at school and in the home and community. Three profiles emerged from the data that consisted of school social workers who reported engaging in low, medium and high levels of ecological practice behaviors across school, family, and community domains. Further examination revealed that school social workers fitting into the profile marked by high levels of self-reported practices at school, with families, and to facilitate community-school linkages were more likely to have a graduate degree, work in a state with certification standards, and have 10 or fewer years of experience compared to school social workers in the other two profiles. Additionally, school social workers who reported high levels of ecological practices were more likely to use evidence-based assessments, programs/practices, and engage in universal school-level prevention efforts more frequently compared to those in the low and medium profiles. Practical implications include the need for pre-service and targeted in-service training as well as policies that support minimum—if not lofty—competencies and state or national certification standards for school social work professionals.


Introduction
School-based mental health providers listed in the Every Student Succeeds Act (ESSA), include state-licensed or certified school counselors, psychologists, social workers, or other mental health professionals qualified under state law [ESSA,Title IV,Part A,Subpart 11,Sec. 4102(6)]. These professionals provide critical services to students in school settings. While school psychologists and counselors have more recently integrated multisystem strategies into practice (Perfect & Morris, 2011), school social workers (SSWers) have abided by ecological influences in student support practices for over a century (Allen-Meares, 1994;Costin, 1969).
Distinct from most psychological and biological theories, ecological systems theory (Bronfenbrenner, 1979) directly informs school social work training and practice. SSWers are uniquely trained professionals who provide key services related to a student's adjustment across systems (School Social Work Association of America [SSWA], 2003). Specifically, the microsystem contains the most proximal context in which a student develops (e.g., self, family, peers, school, community, etc.). Exchanges between a student and her immediate microsystems are influential and intimate (Bronfenbrenner, 1979). To promote nurturing exchanges between a student and these systems, SSWers provide direct support services (e.g., crisis response, skills training, parent training, etc.). In addition, the mesosystem describes the context wherein microsystems interact with each other (Bronfenbrenner, 1979). SSWers also intervene at this level in a variety of ways (e.g., wraparound, home-school communication, broker community services) to facilitate successful exchanges between systems that will confer indirect benefits to a student with the goal of assisting her to obtain success at school.
Broadly speaking, results from randomized studies suggest ecologically oriented social work practices such as wraparound are associated with improvements in child level outcomes, including: fewer disruptions in school and residential placements; improved attendance, conduct and functional living outcomes; and fewer school suspensions, rearrests, and incidents of aggressive behavior (Clark & Clarke, 1996). Studies of SSWers ecological practices to link schools, families, and community providers revealed increased involvement of families of high risk children at school-a key mechanism for improving the daily and long term functioning of high risk students (Epstien et al., 2005) and increased agency-school collaboration (i.e., formal collaboration of professional services, attendance at meetings; Eber, Sugai, Smith & Scott, 2002).
Decades of survey efforts have sought to understand the ecological practice choices of SSWers. For example, Costin's (1969Costin's ( , 1983) early surveys of SSWers reflected a case management or targeted student-level approach to practice. Later, Allen-Meares (1994) used a latent factor approach to group responses from a national sample of SSWers (N = 860) into ecologically informed categories. Allen-Meares categories were broadly grouped into: (a) school support practices (e.g., administrative, leadership, policymaking, counseling for children), (b) home support practices (e.g., home visits, liaison between home and school), and (c) community support practices (e.g., brokering community supports for children, families and the school). Although these categories describe practices of SSWers generally, there was no examination of how the frequently SSWers engaged in these practices or whether they related to other characteristics or contextual factors. More recently, results from the 2008 (N = 1639; Kelly et al., 2010a, b) and 2014 national surveys (N = 3769; Kelly et al., 2015) revealed SSWers spend a majority of time engaging in direct student-level practices (e.g., individual therapy, crisis response, wraparound, etc.) and very little time shaping school-level programs and policies. In short, past surveys have generally sought to describe practice choices of SSWers using an ecological orientation as a guiding theory. However, those efforts examine how ecologically oriented practices of SSWers across systems related to other contextual factors.
A number of developments have influenced practice choices of SSWers over the past decade-underlying the need for more recent efforts to determine their daily influence on professional SSWers. First, the Individuals with Disability Education Act (IDEA, 2004) required professionals working in schools to use data driven and evidence-based practices organized within a tiered prevention model (Kelly, 2008;Thompson, Reinke, & Herman, 2014). Tiered prevention models (e.g., Positive behavior interventions and supports, response to intervention), when correctly implemented, utilize school-wide screening data to select, organize, and monitor the impact of a continuum of programs and practices. These efforts include the implementation of universal supports targeting all students and adults in a school setting as well as targeted (e.g., selective and indicated) supports that often extend services across school, home, and the community for students with more intense needs. Second, Frey et al. (2013) created a National School Social Work Practice Model designed to inform professional preparation programs, improve the quality of professional standards, and guide development efforts for pre-service and practicing SSWers (Fig. 1). The model highlights an ecological orientation across school, family, and community; and prioritizes the use of data to organize evidence-based practices within a tiered prevention model. The use of data and the tiered prevention model are also features endorsed in IDEA and No Child Left Behind (NCLB, 2001) and subsequently the Every Student Succeeds Act (ESSA, 2015). However, despite the alignment between the national model and federal legislative efforts to improve schooling, recent national surveys of SSWers indicate many social workers may not be adopting these practices-citing lack of access, training, and time as barriers (Kelly et al., 2016;Phillippo, Kelly, Shayman, & Frey, 2017).
Previous surveys provide some guidance as to what other factors may be associated with the ecologically oriented and evidence-based practices of SSWers. For example, SSWers reported the number of schools served, the length of time in the profession, and the type of school setting were associated with ecological practice choices (Kelly et al., 2015). Specifically, SSWers who served in one school and those with 10 or fewer years of experience engaged more frequently in universal practices compared to those in two or more schools and to their more senior colleagues, respectively. Also, SSWers working in secondary schools were less likely to engage in practices around family or community resources compared to those working in elementary settings (Kelly et al., 2015). Lastly, SSWers with an MSW and in states with certification standards engaged in universal practices and worked across the home and school contexts to support struggling students with greater frequency than those without an MSW or in states without such practice standards, respectively (Kelly et al., 2015(Kelly et al., , 2016. This distinction between newly trained SSWers and their more experienced colleagues has been previously reported (Kelly et al., 2015) and may reflect the intent federal legislation and professional standards captured in the national model.
In summary, although an ecological focus has been a consistent hallmark of the 110 plus year history of school social work practice, legislation over the past few decades calls for some changes in the manner in which SSWers approach practice (e.g., use of evidence-based screening and assessment tools as well as interventions and practices; application of universal and targeted supports at school; and application of supports that cross community, school and family systems). Furthermore, these requirements are reflected in a national model as an effort to promote these practice choices. The ultimate goal of these efforts is to encourage SSWers to adopt effective practices that result in benefits for students. In light of this shift, the present study uses data from the 2014 National School Social Work Survey to ask SSWers about evidence-based practices across ecological domains to examine the following: 1. Do profiles of SSWers use of ecologically oriented practices emerge from the data? 2. Are profiles of SSWers with varying levels of ecological practices associated with other factors previously shown to be related to practices (e.g., degree, licensure, certification, number of schools, years of practice, evidencebased behaviors, and universal practices)?
We hypothesized that at least three profiles will emerge from the data grouping similar SSWers who engage in high, medium, and low levels of ecologically oriented practices across school, families and community-school domains. We further expected the profile marked by SSWers engaging in high levels of self-reported practices at school, with families, and in the community-compared to SSWers in profiles marked by lower degrees of ecological practices-would be associated with education level (e.g., state certification, licensure), be more likely to endorse evidence-based practices (e.g., evidence-based practices and programs), and be more likely to engage in universal or school-wide and systems level practices.

Method
Data were collected between February and April 2014 using an online survey. For a detailed description of the survey development process, see Kelly et al. (2015). Survey participants included 3769 practicing SSWers in the USA. State school social work organizations and three national associations (i.e., National Association of Social Workers, School Social Work Association of America [SSWAA], American Council of School Social Workers) agreed to disseminate the survey. A single participation in the survey was encouraged via two emails: (a) the first embedded within each association's electronic newsletter to all members and (b) the second was a one stand-alone e-mail to each member. According to the membership lists of these groups, the national sampling frame of practicing SSWers in the US consists of approximately 8500 unique professionals-as such, the present study includes approximately 44% of all practicing SSWers in the USA. Table 1 reveals the demographics of the sample is similar to prior surveys in that it is mostly white, female, and largely urban.

Measures
The 2014 national survey built upon prior surveys seeking to understand the practice choices of SSWers (Allen-Meares, 1994;Costin, 1969;Kelly et al., 2015) but also sought to understand the influence of the national model upon the practices of SSWers. Participants provided responses to questions regarding demographics and the frequency of (a) student issues, (b) contextual factors, (c) use of evidencebased practices, (d) practices across school, family, and community, and (e) universal, selective, and targeted practices. Once complete, a panel of researchers and SSWers rated the face validity of the survey items to be both clearly worded and reflective of an ecological orientation (Kelly et al., 2015;N = 20;k = .94).

Ecological Practice Profile Indicators
Participants provided responses to 18 items tapping ecologically oriented practices across home, school and community and were based upon earlier factor analysis results (Allen-Meares, 1994). Next, we applied a stepwise factor analysis (FA) process and relied upon widely agreed upon procedures and statistical indicators to group the 18 ecologically oriented items into related practice domains reflecting practices across school, home, and community. To collapse the items into composites reflective of the practice domains, we used FA with a maximum likelihood (ML) estimator rather than a principal components approach for two reasons. First, the latter is often used as a data reduction method (Costello & Osborne, 2005). By contrast, our goal in using FA with ML was to reveal whether the items reflected factors observed in both prior surveys (Allen-Meares, 1994) and the national model. Second, during FA extraction with ML, the shared variance of an item is partitioned from its unique variance, while principal components analysis does not discriminate between shared and unique variance (Gorsuch, 1997;McArdle, 1990). Also, because the sample size was large and the distribution of the 18 items was normal, we followed convention by selecting ML as an estimator for factor extraction (Fabrigar, Wegener, MacCallum & Strahan, 1999).
To appraise the quality of FA models, we followed common practices related to statistical cutoffs. Factors with eigenvalues equal to or exceeding 1.0 were retained, item loadings were expected to be .35 or greater, and scree results were expected to reveal a clear distinction between unique factors and correlational noise. In addition, we expected the model would have simple structure (i.e., no cross loadings of items on more than one factor) and that overall model tests would indicate statistical fit of the data to the hypothesized factor structure (i.e., Kaiser-Meyer-Olkin [KMO], Bartlett's; Costello & Osborne, 2005;Tabachnick & Fidell, 2001). All FA models were tested using SPSS 21.0. All FA models returned scree plot results suggesting three factors above the breakpoint and that each of the three factors had eigenvalues > 1.0 with loadings .41 or greater. Upon examination, the three factors consisted of items reflective of school, family, and community-school support practices. Lastly, the FA model held simple structure with no cross-loading of items on more than one factor and statistical indicators suggesting the items fit a three-factor structure (i.e., Kaiser-Meyer-Olkin = .86′ Barlett x 2 = 16658.292, p = .001). As such, we had a high degree of confidence the 18 items were reflective of practices across (a) schools, (b) families, and (c) community and therefore the relevant items were collapsed into three, normally distributed composite variables described in greater detail here.
School Support Practices Seven items measuring direct and indirect school support practices were combined into a single factor (M = 2.29, SD = .68; α = .86). The items asked participants to rate the frequency (1 = never to 5 = always) with which they: provided pull-out services to disruptive students, counseled emotionally distraught students, discussed student concerns, provided in-service training, developed individual behavior management plans, provided classroom management consultation, and brokered referrals for related supports.

Family Support Practices
Six items measuring family support practices were combined into a single variable (M = 3.41, SD = .77; α = .80). The items asked SSWers to rate the frequency (1 = never to 5 = always) with which they: made home visits, advocated for parents with administrators, provided parent workshops, provided resources, served on parent advisory boards, and engaged families in needs assessments and wraparound services.

Community-School Support Practices
Five items measuring community-school support practices were combined (M = 2.62, SD = .66; α = .78). The items asked participants to rate the frequency (1 = never to 5 = always) with which they: brokered a continuum of community-based services, established partnerships with service providers, supported community agency involvement at school, and consulted/ collaborated with service providers on student issues.

Contextual Factors
Contextual factors influencing practice included 14 items derived from research on practice barriers for SSWers (Kelly et al., 2010a(Kelly et al., , b, 2014. Items included administrative constraints, availability of resources, lack of cooperation, and language barriers. Participants endorsed how often SSWers experienced barriers using a 5-point scale (1 = never, 5 = always) and the items were combined into a composite with acceptable reliability (a = .74). Other factors previously shown to impact practice included: level of education (Graduate = 1, Bachelors = 0), licensure (LCSW, LPC, etc. = 1, no licensure = 0), whether their job required certification (yes = 1, no = 0), caseload size (number of schools worked in; 1 school = 1, 2 or more = 0), and the number of years of practice (10 or fewer = 1, 11 or more = 0).

Evidence-Based Behaviors
Evidence-based behaviors were appraised using 26 items assessing (a) whether SSWers felt prepared to use evidencebased assessments, (b) how often they used those assessments, (c), how much access they had to evidence-based supports (e.g., programs and practices) and (d) how often they used those evidence-based supports. Again, FA with a ML estimator was used to examine the items. Results (Kaiser-Meyer-Olkin = .88; Barlett's test = x 2 = 1320.112, p = .001) indicated the data fit a multifactor structure and analysis revealed a four-factor solution fit the data as shown in scree plots, eigenvalues (> 1.4), and item factor loadings over .45 with simple structure (Tabachnick & Fidell, 2001) so the 26 items were combined into four normally distributed composites described in greater detail here.

Preparedness to Use Evidence-Based Measures
Preparedness to use evidence-based measures was assessed with 7 items asking participants how prepared (not at all = 1, very = 3) they felt to use direct observation, student/teacher self-report, existing forms of data (i.e., attendance, grades, etc.), standardized scales, universal screening tools, intervention fidelity monitoring tools, or progress monitoring tools (M = 1.7, SD = .40; α = .77).

Use of Evidence-Based Measures
Use of evidence-based measurement tools was assessed by 7 items asking participants how often (never = 1, often = 3) they used direct observation, student/teacher self-report, existing data (i.e., attendance, grades, office referrals), standardized rating scales, universal screening tools, intervention fidelity monitoring tools, or progress monitoring tools (M = 1.82, SD = .36; α = .74).

Access to Evidence-Based Supports
Access to evidencebased resources was assessed with 6 items asking participants how much access (no access = 1, high access = 3) they had to online databases for retrieving research on effective practices, journals or books, online intervention repositories (e.g., What Works Clearinghouse, PBIS.org, etc.), trainings and workshops, regular access to direct practice supervision, and consultation with other school social work peers (M = 1.83, SD = .34; α = .69).

Use of Evidence-Based Supports
Use of evidence-based measurement tools was assessed with 6 items asking participants how often they (never = 1, often = 3): used online databases to access research on effective support, relied on journals or books, drew upon online interven-tion repositories (e.g., What Works Clearinghouse, PBIS. org, Intervention Central, etc.), attended trainings and workshops on practices, relied upon direct practice supervision, and consulted with other peer SSWers (M = 1.89, SD = .33; α = .94).

Universal Practices
To examine how often participants engaged in universal practices, we followed prior surveys asking participants to estimate the proportion of time (0-100%) they engaged in school-wide practices (i.e., school-level program and policy development, providing school-wide interventions; M = 16.49, SD = 18.73; Kelly et al., 2010a, b).

Analysis
We first used a latent profile analysis (LPA) to group respondents with similar practice behaviors across schools, families, and communities. LPA assumes that within each profile the practice behaviors can be explained by an underlying classification of respondents with similar ecological practice patterns (Nylund, Asparouhov, & Muthén, 2007). In short, the goal of LPA is to group SSWers with similar ecologically oriented practices rather than relying on arbitrary variable-centered cutoffs. Second, we proceeded to examine whether certain LPA profiles of SSWers engaging in various ecological practices were also associated with contextual practice factors, evidenced-based beliefs and behaviors, and universal or school-level practices. The LPA was conducted using Mplus 7.0 (Muthén & Muthén, 2013). In LPA, the best fitting model is determined by the alignment of statistical indicators. Successive models are compared using the Akaike Information Criteria (AIC; Akaike, 1981) and the Bayesian Information Criteria (BIC; Schwarz, 1978). In these models, more weight is given to the BIC as recent studies have indicated the BIC provides a reliable indicator of true model fit (Nylund et al., 2007;Swanson, Lindenberg, Bauer, & Crosby, 2012;Vrieze, 2012). Lower BIC estimates are preferred, and we calculated the change in BIC (BICΔ) between successive models (Fraley & Raftery, 1998). We also report the Vuong-Lo-Mendall-Rubin (VLMR; Lo, Mendell, & Rubin, 2001;Vuong, 1989) and a bootstrapped parametric likelihood ratio test (BLRT; McLachlan & Peel, 2000;Muthén & Muthén, 2000). Both the VLMR and the BLRT are Chi-square likelihood ratio tests comparing an LPA model with k profiles against an LPA model with k-1 profiles such that a p value exceeding .05 is indication that the lesser k-profile model is preferred (Lo et al., 2001). Lastly, cohesiveness of each model is assessed using entropy-a statistic with values above .7 indicating persons fit into their identified profile well (Celeux & Soromenho, 1996).
Once the best fitting LPA model was identified using the above indices, the substantive meaning of the profiles were examined using the Mplus Auxiliary function (Muthén & Muthén, 2013). The auxiliary function allows analysts to simultaneously (a) estimate profiles and (b) to conduct statistical tests to determine whether one profile differs from other profiles on key descriptors. The auxiliary function reduces the number of statistical tests and thus the likelihood of a type I error. In the present study, we used the auxiliary function to determine whether the SSWers fitting into a profile could be distinguished from others based on their endorsement of practice and contextual factors (e.g., education, licensure, state certification, number of schools, years of practice) and evidence-based behaviors (i.e., use evidencebased assessments, how often used evidence-based tools, how much access they had to evidence-based resources, and how often they used those resources), and how frequently they participated in universal supports.

Missing Data
Of the 3769 respondents who initiated the survey, 905 (23%) had some missingness, whereas 2521 (67%) provided full responses. Analysis of missing patterns suggested no association between missing patterns and observed characteristics, nor did those with missingness differ from those with full response patterns. Under these conditions, the data meet criteria for the designation of missing at random (MAR; Little & Rubin, 1989). Mplus relies on a full information maximum likelihood (FIML) procedure for MAR data-an extensively tested process utilizing all information to yield parameter estimates similar to imputation (Little & Rubin, 1989Muthén, 2001).

Results
The results are presented in the stepwise fashion in which the analysis proceeded. That is, we first present the results of analyses to identify the best fitting LPA model. Next, we report the differences between profiles on contextual factors, evidence-based behaviors, and universal practices. In the second step, to further reduce the likelihood of a type I error, we present a range of significance values of p < .05, .01 and .001 (McCoach et al., 2010).

Step 1: Estimating LPA of Practice Profiles
First, the three composites representing the ecologically oriented practices at school, home and in the community were used to create the LPA model. Table 2 shows the statistical indices used to determine the quality of fit of each model. We started with 1 profile and successively increased the number of profiles. A three-profile solution fit the data as indicated by alignment of all statistical indicators (i.e., a lower AIC, BIC, a negative BICΔ, significant VLMR and BLRT, and high entropy). Considering the composition of the profiles themselves, Fig. 2 reveals the mean scores of SSWers in each of the profiles. The best fitting solution resulted in three profiles termed here as low (16%; Profile 1), medium (67%; Profile 2), and high (17%; Profile 3) ecologically oriented practice profiles.

Step 2: LPA Profiles, Contextual Factors, Evidence-Based Behaviors, and Universal Practice
Next, we examined differences between the three practice profiles and contextual factors, evidence-based behaviors, and the use of universal school-level practices. The top half of Table 3 includes the proportions of profile participants who endorsed having a graduate degree, being licensed, working in a state with certification standards, working ten or fewer years, or working in a rural setting. Results of overall and specific tests-where overall tests were significant-indicated significant differences between the profiles on education level or degree, state certification, and years of practice. Examining these differences by profile reveal that Profile 3-marked by SSWers with the highest levels of ecological practices across school, home and communitywas more likely to include SSWers with a graduate degree compared to Profile 1 (the low profile), but no significant difference was observed when comparing Profile 3 to Profile 2 (the moderate profile). Next, SSWers in Profile 3 (high) were more likely to consist of persons working in states with certification standards compared to participants in Profiles 1 (low) and 2 (moderate); though here persons in Profile 2 were also more likely to work in states with certification standards compared to those in Profile 1. Lastly, those in Profile 3 (high) were more likely than those in Profiles 1 (low) and 2 (moderate) to consist of persons with 10 or fewer years of experience after graduation; Profile 2 also was more likely to contain SSWers with 10 or fewer years of experience after graduation compared to those in Profile 1. There were no differences between profile membership with regard to licensure or urban versus rural setting.
The bottom half of Table 3 includes participant responses measuring differences in evidence-based behaviors. Here we see that SSWers in Profile 3-the profile with SSWers with high ecological practices across home, school, and community-reported they were significantly more prepared to use evidence-based assessments than those in Profiles 1 (low) and 2 (moderate); persons in Profile 2 (moderate) were also more likely than those in Profile 1 (low) to report being more prepared to use evidence-based assessments. Furthermore, persons in Profile 3 (high) reported using evidence-based assessments with a greater frequency compared to SSWers in Profiles 1 (low) and 2 (moderate); again, those in Profile 2 (moderate) also reported using these measures more often than those in Profile 1 (low). In addition, SSWers in Profile 3 (high) reported significantly greater access to evidencebased resources for directing practice in school settings and reported using those resources with a greater frequency compared to SSWers in Profiles 1 (low) and 2 (moderate). Again, those in Profile 2 (moderate) also differed significantly from those in Profile 1 (low) with regard to increased access to evidence-based resources as well as using evidence-based resources with greater frequency. Lastly, SSWers in Profile 3 (high) engaged in universal or school-wide practices to a significantly greater degree than SSWers in profiles 1 (low) and 2 (moderate). Similar to the other tests, those in Profile 2 (moderate) also engaged in significantly more universal practices compared to those in Profile 1 (low).

Discussion
The present study extends decades of school social work survey research (Kelly et al., 2010a(Kelly et al., , b, 2015. Using an LPA framework, we created profiles of SSWers who self-reported the degree to which they engage in practice behaviors across the ecological domains of school, home and community. The study examines the ecological practices to support students across multiple domains and more contemporary requirements of SSWers to engage in universal practices mandated by legislation (IDEA, 2004;ESSA, 2015) and reflected in the national practice model (Frey et al., 2013(Frey et al., , 2012. The first research question related to the creation of practice profiles based on an ecological orientation. We first identified three distinctly and statistically unique practice profiles similar to prior studies (Allen-Meares, 1994). The three-profile solution provides evidence that a modest percentage (17%, Profile 3) of SSWers reported frequently engaging in support practices at school, with families and to build community-school linkages. The largest profile (Profile 2, 67% of the total sample) reported engaging in practices across those domains to a lesser degree. Lastly, a small percentage (Profile 1, 16%) of SSWers reported engaging in practices across each of the domains to a comparatively low degree. Although it may be viewed as discouraging to note that only 17% of the sample represent the most ecologically focused profile (Profile 3), it is encouraging to note that most SSWers (83%) self-report they engage in ecologically relevant practice behaviors with a moderate to high degree of frequency (Profiles 2 and 3). We also found it concerning that of all the ecologically relevant practice domains (school, family, community), SSWers across all 3 profiles reported engaging in the "school-based support" the least, comparatively speaking. While all SSWers regularly intervene with youth in "pull-out" and crisis situations, relatively few SSWers work to change or alter school-level factors preventing student-level problems (e.g., in-service training, consulting with teachers on behavior problems, and brokering referrals for other school-based supports).
With regard to research question two, we observed SSWers in Profile 3-the profile containing persons who self-reported engaging in ecologically oriented practices across home, school, and community to the highest degreediffered from SSWers in Profiles 1 and 2 with regard to several notable influences. Specifically, SSWers in Profile 3 were significantly more likely to have a graduate level degree and to practice in states with certification standards compared to SSWers in Profiles 1 and 2. Though it has been suggested that SSWers with graduate training who practice in states with certification standards would adhere to national expectations compared to those without graduate degrees (Frey et al., 2012), this study provides some correlational support to suggest that graduate training and state certification standards may improve the quality and effectiveness of SSWers and the services they offer students. This finding alone is not causal, but deserves closer examination given the significant investment needed to provide students with support services at school-an investment that be accompanied by a minimum set of competencies and qualifications.
This study also considers the relationship between services offered by SSWers and years of experience. In contrast to a previous survey of Illinois SSWers (Kelly et al., 2015), the present study identified a relationship between years of experience and self-reported practices. Specifically, Profile 3 contained a larger proportion of SSWers with ten or fewer years of experience compared to Profiles 1 and 2. This association has been observed previously and may reflect shifts over the past decade due to legislative influences (e.g., IDEA, 2004;NCLB, 2001;ESSA, 2015) or professional standards and expectations for school social work practice (e.g., Frey et al., 2013). However, more research would certainly be needed to make any statements of association between these factors. In essence, it is worth noting that SSWers who have been out of graduate training for over 10 years at the time of this survey (2014) may benefit from additional training and coaching support in adopting the latest policy and practice behaviors known to improve student and school outcomes.
It is also worth noting that there were some differences between SSWers captured in Profile 2 (moderate) compared to those in Profile 1 (low). Ultimately, the LPA model reported here represents a complex moderation between practices and factors-that is, those who engage in increasing degrees of ecologically informed practices in their pursuit to support students are also more likely to have better training, work in states with standards, and be more prepared to use evidence-based measurement tools and support practices. The gem in this study may lie in the finding that those with low levels of training (i.e., those with a BS) and who work in states lacking certification standards are not as prepared and use and have less access to evidence-based measures and practices compared to those with graduate degrees and who work in states with higher standards of professional preparation.
In essence, these findings suggest we must continue to train present and future SSWers in the skills needed to be practitioner-scientists. Such training would include skills in data collection, simple analytical methods, and strategies to use data to drive decisional systems surrounding the use of programs and practices. Additional training and coaching for those presently working in schools as well as future SSWers in evidence-based practices across settings (school, home, community) and levels (universal, selective, indicated) is warranted, particularly for seasoned practitioners or those who have been out in the workforce for 10 or more years. It is also important to note that we observed no associations for a variety of contextual factors that prior surveys suggested were barriers to the effectiveness of SSWers. For example, prior studies have noted caseload (i.e., number of schools worked), role restriction, availability of school and community resources, perceived cooperation from teachers/ students/families, and language were barriers. Additionally, licensure (i.e., LCSW, LMSW, LPC, etc.) was not associated with membership in any one of the profiles and nor was setting (urban versus rural, etc.) associated with profile membership. These findings could be regional or response specific findings as prior surveys were smaller or only focused on specific states. By contrast, the present survey is based 1 3 upon a national sampling frame and includes approximately 44% of practicing SSWers.
With regard to the examination of differences between profiles on evidence-based behaviors and engagement in universal practices, SSWers in Profile 3-the most ecologically orientated profile-reported being more prepared to use and reported actually using evidence-based measures and evidence-based practices more often compared to SSWers in Profiles 1 and 2. Similarly, SSWers in Profile 3 also reported engaging in universal or school-based prevention strategies more frequently than their peers in Profiles 1 and 2. We believe this study is the first to consider the relationship between the degree of ecological practices engaged in by SSWers and their engagement in prevention type practices at the school level. On the one hand, it makes sense that those who engage in high degrees of ecological practices at school, home and in the community are systemic action agentsand as such, they are likely forces of change at the systems level within their own schools by advocating for roles at the school policy and universal prevention level. Contrary to prior surveys revealing that most SSWers engage in child level interventions, case management, and crisis intervention (Costin, 1969;Allen-Meares, 1994), this study suggests SSWers who engage in high levels of ecological practices also engage in high levels of school-level prevention. This point is important for several reasons. First, prior surveys of SSWers have suggested these personnel spend an inordinate amount of time focused on targeted intervention for high-risk students. However, as many prior studies suggest, focused school-level prevention practices also confer benefits to all students regardless of risk (Fraser, Thompson, Day, & Macy, 2013). Though we interpret these findings with caution, knowing universal practice behaviors are engaged in by SSWers who simultaneously engage in high levels of ecological practices reveals these effective practices are not mutually exclusive.

Implications for Future Research, Training, and Practice
Perhaps the most important implication of this study is that SSWers engaging in ecologically oriented practices to a high degree are more likely to have a graduate degree and work in a state with certification standards. This may make sense as the coordination between graduate training programs and state departments of education may boost the quality of training as well as act as a gate keeping mechanisms to ensure that SSWers meet minimum standards. While more research could help clarify these correlational findings, this point suggests graduate training may extend skill levels and prepare SSWers to offer more rigorous and evidence-based practices in states with those standards. At this point, there are no national guidelines, but even state requirements for SSWers appear to-at the very least-correlate with ecological practices and the increased likelihood that those SSWers will adopt evidence supported practices and programs which may confer benefits to students and schools.
For states without certification expectations where SSWers are practicing, these findings suggest it is important for state boards of education and state licensure boards to consider adopting such standards as the need for schoolbased mental health professionals becomes increasingly important. States considering standards for school social work certification procedures and standards can refer to NASW's National Standards for School Social Work Services as well as rely on the national model as a guide. In addition, SSWAA (https ://www.sswaa .org/schoo l-socia l-work) offers guidelines on basic competencies for school social workers and frameworks for school-based mental health programs to promote safe and successful schools. Such state level efforts will not only involve the adoption of basic levels of competencies but should also include certification exams based upon those competencies and graduate programs would also need to align curricula to prepare a workforce to meet those standards. At a more local level and in the absence of national and state standards, local district human resource offices and local hiring teams could also adopt basic competencies that include specific ecologically oriented practices, rely on the collection and use of data and evidence-based supports, and include school-wide expectations as important guideposts to consider when designing mental health support services and hiring key personnel.

Limitations
The present study has important limitations. First, there are obvious limitations relating to the sample. Though we relied on the three largest national organizations for SSWers to increase awareness and participation in the survey, and had almost total participation from the extant SSW state associations, there is no comprehensive list available to ensure we included all practicing SSWers, and no way to ensure that all SSWers are part of at least one of the organizations we worked with to get our sample. Additionally, given that the participating organizations chose to send out the link themselves rather than provide us with a master list of e-mail addresses, we know that at least some of the respondents belonged to multiple state and national organizations, and could have conceivably taken the survey multiple times, though we think this is unlikely. Still, these issues reflect the fact that, the sample was ultimately one of convenience and thus suffers from limitations associated with self-selection bias. Despite these limitations, the findings include the voices of about 44% of practicing SSWers based on SSWAA estimates. Future surveys might consider randomly selecting participants from membership lists.
A second limitation relates to an implied assumption between ecological practices across school, home and the community and the quality of those services. We acknowledge it is improper to assume SSWers who self-report engaging in ecologically oriented practices at high rates are providing more effective services to students compared to those in the other two profiles. For starters, the responses used here are self-report-and thus suffer from limitations associated with self-report data. In addition, to test whether an ecological approach to practice has improved outcomes for students, we would need to engage in an entirely different study. Similarly, these analyses do not control for environmental factors such as student symptom severity or other potentially important factors that may drive practice choices of SSWers.
A third limitation is that there may be measurement problems in the manner in which we conceptualized and operationalized the indicators used in this study. As a guide to our process, we relied on (a) the national model, (b) prior surveys, and (c) the guidance of a panel of school social work researchers and practitioners to validate the survey items. Similarly, the items used in the study were only combined following rigorous FA procedures, and analyses suggest the items have acceptable inter-item reliability. However, future surveys might validate the findings of our FA using a confirmatory factor model. Future surveys may also include other validated scales to appraise stress, burnout, and other barriers not addressed in this survey, data that may provide insights into how to support practicing SSWers. Lastly, the survey is cross-sectional and while trends can be observed, there is no way to monitor person level change over time.

Conclusion
Federal policies and the national school social work practice model challenge school-based SSWers to engage in systemic, evidence-based practices to improve student and school outcomes. SSWers are specially trained to have a multifaceted understanding of how to influence students not only through services focusing on the school, but also the home and community. Though more work needs to be done to see if this training translates into better student and school outcomes, it is important to remember that passing policies like ESSA and disseminating the national model which communicates high expectations for SSWers is not enough. These expectations must be accompanied by the contextual supports and competency requirements (e.g., graduate degree, state certifications standards, quality in-service training and coaching for veteran SSWers) that translate into the evidence-based behaviors known to improve outcomes (e.g., increased awareness and use of evidence-based measurement tools and evidence-based practices, use of tiered and ecologically oriented approaches to practice). Lastly, the training and support SSWers receive should impart the benefits of focusing on universal prevention strategies and leadership development needed to meet the policy and program development at the school level. Such practices can improve outcomes for all students-including those at the highest degrees of risk (Fraser et al., 2013). Adopting supports for professionals is the best route to enhance the capacity of SSWers to engage in effective practices to improve safe, welcoming, and civil schools for all students.

Compliance with Ethical Standards
Conflicts of interest The authors declare that they have no conflict of interest.

Ethical Approval
The research performed here was approved and overseen by the Institutional Review Board of Loyola University at Chicago.

Human and Animal Rights
All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
Informed Consent Informed consent was obtained from all individual participants included in the study.