Child protection decision making: a factorial analysis using case vignettes.

Stokes, Jacqueline ; Schmidt, Glen

The death of a child due to mistreatment and abuse is tragic and evokes both private distress and public outcry. A public inquiry into such a death is a common response, and the inquiry may lead to a major overhaul of the child protection system responsible for protecting vulnerable children. The Gove report (1995), which followed the death of five-year-old Matthew Vaudreuil in British Columbia, Canada, and the Laming report (2003), which followed the death of Victoria Climbie in the United Kingdom, are two examples of this process. Reviews of this nature examine the circumstances of the child's life and death and pay particular attention to the decisions made by social workers, supervisors, and managers. Social workers are asked to defend their procedural practice. A limited amount of attention is paid to the complexity of information that social workers face in their daily decision making. Despite the history of numerous reviews conducted from the perspective of hindsight and the adoption of various technocratic policies and procedures that occur after such inquiries, child protection social workers cannot be reduced to automatons who follow rigidly prescribed blueprints when working with children and families. Although there is a plethora of literature on child protection practice, only a minimal amount of research has been devoted to understanding the decision-making process or how the professional thinks in the context of everyday practice (Kondrat, 1992; Walmsley, 2004). This research examines decision making in child protection through the use of case vignettes.

DECISION-MAKING PROCESS

Decision making in social work, and in particular in child protection, is difficult, challenging, and fraught with risk and a degree of uncertainty. Decisions are often made with insufficient, unreliable, conflicting, or missing information within a stressful and pressured organizational and political context (Budd, 2005; Gambrill, 2005a; Webb, 2002). Despite this uncertainty, social workers must decide whether a child is safe, whether a child should be removed from his or her home, whether to provide additional supports to a family, and how to engage with a family to facilitate positive change for a child. These everyday decisions can dramatically affect the lives of children and their families. Increasingly, child protection practice has turned to objective or evidence-based knowledge to augment the effectiveness, accountability, and transparency of these decisions. The development of risk assessment is a primary example of the application of research-based knowledge to a practice and procedural tool. Risk items are included when there is empirical research showing a statistical link between the items or factors and the circumstances of the case being assessed (Cash, 2001). These risk assessment tools standardize service interventions and provide some degree of predictability. In many jurisdictions the risk assessment tool has become the central decision-making mechanism in child welfare (DeRoma, Kessler, McDaniel & Soto, 2006).

However, the practice of applying knowledge to a specific situation is more than an exercise in technical rationality; it requires social workers to translate theoretical knowledge into skills and know-how for practice (Kondrat, 1992; Parton, 2003). In the real world, it requires what Squires (2005) referred to as "repetition-with-variation." The art or practice of social work is to interpret the individual client situation and take actions that are consistent with the organizational and societal context. This implicit store of knowledge is sometimes referred to as intuition, common sense, or tacit knowledge (Zeira & Rosen, 2000). Although there is some dispute in the literature about the denotative meanings of these concepts, the literature agrees that there is more to professional practice than technical know-how. Whereas empirical or scientific knowledge allows for the application of systemic theory and procedural rules, secondary forms of experiential knowledge take into account the professional's store of cultural, personal, and practice knowledge. It is this personal store of knowledge that becomes internalized and allows the professional to filter a situation through his or her own thinking process, to decide which information is relevant, to discover patterns of meaning, and to value an individual's unique experience of their everyday world.

It is not surprising that there are differing views among social workers and experienced practitioners about the process of decision making (Gambrill, 2005b; Reamer, 1993; Sicoly, 1989). Initial judgments have been shown to be important in decision making. Gambrill (2005a), in her work on critical thinking in social work, suggested that "we have a tendency to believe in initial judgments, even when we are informed that the knowledge on which we based our judgments was arbitrarily selected" (p. 19). Furthermore, she contended that these initial beliefs are resistant to change, even when new evidence is provided. In a study of reasoning in child protection in Britain, Munro (1999) developed similar conclusions. She found that most determinations of risk were based on a limited range of data and subsequently, even with evidence contrary to the worker's initial case disposition decision, revision of judgment about cases was slow or nonexistent.

Although measuring risk is a predominant focus of a child protection worker's practice, it is not the only one. Other areas of attention include kinship care, family collaboration, and family mediation, For these practices, social workers are required to develop an interpersonal rapport that facilitates positive change for clients (Kondrat, 1995).

Much has been written about the necessity of a good, trusting therapeutic relationship in clinical practice. Building relationships with non-voluntary clients has not been explored as extensively; however, practitioners in child welfare regard relationship building as extremely important (Yatchmenoff, 2005). Kondrat (1995) argues that "in order to form the relationships necessary for change, the most essential issue for the practitioner is that of informing his or her judgment with reliable understanding--understanding of the other, of the self, and of the emerging relationship" (p. 409).

RESEARCH CONTEXT

Recognition of the rapidity with which social workers have to make decisions and the reluctance of social workers to modify those judgments heightens the need to understand decision making from first contact. In an era of risk management it seems that individual or professional judgment has been largely ignored. The question arises as to how social workers in child protection make decisions about children's safety. Our research on this question was conducted in the province of British Columbia (BC), Canada. In BC, child welfare protection services are primarily delivered through the Government of BC's Ministry of Children and Family Development (MCFD).

METHOD

To examine how social workers make everyday decisions, it is important to use a research process that replicates as much as possible the circumstances of daily decision making. The factorial survey method of research, first developed by Rossi and Nock (1982), is a hybrid technique that studies people's perceptions, beliefs, judgments, and decisions associated with complex multidimensional phenomena (Jasso, 2006; Ludwick et al., 2004; Shlay, Tran, Weinraub, & Harmon, 2005). The factorial survey technique serves to bridge "two research paradigms by combining elements of experimental designs and probability sampling, with the inductive, exploratory approach of qualitative research" (Ganong & Coleman, 2006, p. 455). Although this method has limited exposure in social work research, Taylor (2006) argued that "factorial survey has potential as a method for rigorous study of the impact of client, family, and context factors on decisions by social work and social care staff" (p. 1187). In the factorial survey method, the respondents are presented with contrived hypothetical situations, or vignettes.

The factorial survey method possesses important strengths as a research design, including internal and external validity and robustness. Internal validity is high due to the random combination of factors within vignettes and the random allocation of vignettes to participants (Ganong & Coleman, 2006; Landsman & Copps Hartley, 2007; Taylor, 2006). External validity is high because the decisions closely resemble those made in normal work situations. External validity is also enhanced through consistent wording of the same factors, which reduces the potential for bias in interpretation of the statements (Landsman & Copps Hartley, 2007).

The robustness is increased because the vignette is the unit of analysis rather than the respondent. Because the factors within the vignette have an orthogonal relationship to one another, each vignette is considered independent, and this provides the required randomness. Another advantage of this method is how the sample size is defined. "In the conventional survey design, the sample size is determined by the number of participants in the study, and researchers are interested in variation across subjects and subject groups" (Shlay, Tran, Weinraub, & Harmon, 2005, p. 403). However, because the vignette is the unit of analysis, the number of variations on the vignette forms the sample size. This method maximizes the statistical power (Landsman & Copps Hartley, 2007, p. 458). Although the vignette is the unit of analysis, demographics and other factors relating to the respondent can be considered for their effect on the decisions (Taylor, 2006).

The development of the vignette is critical to the factorial survey method. The vignette is designed to tell a logical story that simulates real-life practice; therefore, the independent variables within the vignette are important. In this research, consideration was given to all of the dimensions currently identified in the BC Risk Assessment Model as potential independent variables. The factors included as independent variables were those that either dominate the literature or are present in a substantial number of children or families that are investigated by child protection agencies. Eight factors were chosen as independent variables in decision making: (1) harm to child, (2) income, (3) housing, (4) culture, (5) parental substance use, (6) family violence, (7) resources and support, and (8) cooperation. A vignette framework was then developed to include these factors.

Vignette Framework

Vignettes were structured using the following basic format:

   The following situation has been presented at
   a team meeting: Susan, age 7 is--harm to
   child--. Susan lives with her mother and
   father who--income--. They are living
   in--housing--. Susan and both her
   parents are--race--. The parents
   --substance use--, and Susan reports that
   --family violence--. The prior contact
   record shows that the family--resources
   and supports--. The parents have
   --cooperation--.

Each of the independent variables, referred to as dimensions, is broken into levels. For example, harm to child had four levels: (1) neglect, (2) physical harm, (3) emotional harm, and (4) sexual abuse. Income also had four levels: (1) no known income, (2) income assistance, (3) one minimum wage, and (4) two jobs. With the exception of culture (which had two levels: aboriginal and Caucasian), all other dimensions had four levels. A computer-generated survey tool was developed that would allow for the random assignment of levels for each dimension, thereby ensuring that any non-orthogonality of the independent variables was due to random error only. To ensure the realism of the vignettes, three expert consultants with experience in child protection decision making reviewed the vignette construction and the list of dimensions and levels for content validity and clarity. Changes were made to the vignette to ensure that the vignettes had internal believability. With seven dimensions that have four levels and one dimension that has two levels, 32,798 possible vignettes exist in this study.

Sample Vignette

The following populated vignette represents an example that a respondent might receive.

   The following situation has been presented at
   a team meeting: Susan, age 7 has been attending
   school with no lunch, without breakfast
   and often seems tired and lethargic; she has
   few clothes and no winter coat and is often
   cold. Susan lives with her mother and father
   who are relying on one minimum wage job.
   They are living in a house that is poorly maintained
   with numerous broken windows, open
   electrical outlets, and evidence of animal
   waste inside the house. Susan, and both her
   parents, are Aboriginal people. The parents
   are known to have a serious problem with
   drug abuse, and Susan reports that she has
   seen her father hit and shove her mother. The
   prior contact record shows that the family has
   little consistent, or reliable, support from
   friends or family members. The parents have
   had ambivalence about change and often miss
   appointments but have some follow through
   to services offered.

Responding to the Vignettes

Four questions about four dependent variables were developed. Questions one and two dealt with the objective and technical aspects of the risk assessment tool, whereas questions three and four related to the relationship development factors important to the contextual understanding of the client's situation.

Question 1 asked, "On a scale of risk, what is your initial impression of the level of risk?" Respondents were asked to circle one number on a five-point scale on which 1 = no risk and 5 = extreme risk.

Question 2 asked, "Based on the information you have so far, what is likely to be your placement decision throughout the investigation process?" Respondents were given five possible answers and asked to choose one:

(1) Close file, no further service required.

(2) Provide a referral to a community service provider.

(3) Provide intensive family support services with MCFD case management.

(4) Arrange an informal placement with a family support network (e.g. kith and kin in which MCFD does not have temporary custody).

(5) Develop a formal in-care arrangement (e.g. family foster care, or a group home in which MCFD has temporary or flail custody).

Question 3 asked, "In this situation, how important is it that you visit the family home in determining the degree of risk and placement decision?" Respondents were asked to circle one number on a five-point scale on which 1 = not very important and 5 = extremely important.

Question 4 asked, "How many hours would you plan to spend getting to know the parent(s) over the next four weeks?" Respondents were asked to write any number of hours into a box.

Sampling

A request to participate in the survey was sent via e-mail to all social workers in the BC MCFD. The social workers who agreed to participate were provided with up to three unique vignettes. Most participants responded to all three vignettes, but some addressed only one or two. In total, 118 respondents responded to 327 unique vignettes. Because the unit of analysis is the vignette, the sample size (n) is 327. The participants included 95 women (80.5 percent) and 23 men (19,5 percent). Thirty percent of all participants were between the ages of 35 and 44, and all held a minimum of a bachelor's degree in either social work or child and youth care. Almost 70 percent of the participants had six or more years experience in child protection; 88 percent of all participants had received specialized training in child abuse, and 94 percent of all participants had received training in risk assessment.

Analysis and Findings

Multiple regression was the primary statistical method used to measure the relationship between each factor and the decision made. Analysis of variance was used to establish statistical reliability. Taylor (2006) suggested that by using multiple regression "it is possible to infer a causal explanation (that is, that the factors actually cause the change in the decision, rather than merely being associated with it by 'accident') because the factors in the vignette are virtually independent" (p. 1196).

The initial analysis provided the frequency, mean, median, and standard deviation (SD) of the responses on each of the dependent variables. The median response for all vignettes in terms of risk was 4 (on a five-point scale in which 1 = no risk and 5 = extreme risk) (SD = 0.77). For the service provision decision, the median response was to provide intensive family support services (SD = 0.90). The median response for visit importance was 4 (on a five-point scale in which 0 = not very important and 5 = extremely important) (SD = 0.78). For the number of contact hours, the median response was 7.25 hours (SD = 4.91). For all six possible combinations of dependent variables, there was a positive correlation. That is, a higher risk evaluation positively correlated with a more intensive intervention, higher importance placed on a home visit, and an increase in the number of hours to be spent with the family. The highest positive correlation was for the interaction between risk level and service provision ([R.sup.2] = .45, r = .673). For the other five comparisons, there was a positive but weak correlation.

A multiple regression analysis was used to examine the effect of the independent variables on the assessment of risk, the service provision, the visit importance, and the amount of contact hours (see Table 1). Using the conventional probability level of .05, the independent variables that had at least one level with a statistically reliable effect on the decision of risk from the vignette were harm to child, housing, substance use, spousal violence, and cooperation. The independent variables of harm to child and housing had at least one level that had a statistically reliable effect on the decision to provide service. The factors housing and substance use had an impact on visit importance, and only harm to child had a statistically reliable effect on the number of contact hours. No factor had a statistically reliable effect on all decisions. Income, culture, and resources and support did not have a statistically reliable effect on any decision.

Although the vignette is the unit of analysis and not the respondent, in factorial survey design the factors about the respondents are treated as independent variables. Respondents were asked questions pertaining to ten independent variables: gender, age, job satisfaction, type of degree qualification, the amount of child protection experience, the amount of experience in the community, delegation, whether they practiced in an urban or rural location, satisfaction with supervision, and the specific role they played in protection work (for example, intake, resource, family service) (see Table 2). Only degree and supervision satisfaction had a statistically reliable effect on the decision about risk level. Degree, child protection experience, and community experience had at least one level that had a statistically reliable effect on service provision. Gender, age, job satisfaction, degree, and current social work role had at least one level that had a signficant effect on visit importance. The number of contact hours decision was affected at statistically reliable rates on eight of the 10 variables; the only variables that showed no statistically reliable effect were job satisfaction and supervision satisfaction. Of the independent variables associated with the respondent, all had at least one level that had a statistically significant effect on at least one dependent variable. Degree was the only variable that had a statistically reliable effect on all four of the dependent variables.

DISCUSSION

Decision making in child protection work is complex, and workers are influenced by factors from multiple sources. Our research does not demonstrate a clear causal relationship between the decisions made and the source of knowledge on which those decisions rest. However, it does appear that social workers are more likely to use technocratic, evidence-based knowledge derived from the case situation when making decisions about risk level or service provision. This contrasts with decisions that are related to developing a relationship with the client. Variables about the social workers themselves affect factors such as time spent with the family and the importance of a home visit.

Although the debate about whether social work is a science or an art is a long-standing one, this research indicates that despite the emphasis on the quantitative and procedural mechanism of the risk assessment tool, some social work decisions are affected by the social worker's own experiences and who he or she is as a person. This is consistent with the findings of Parada, Barnoff, and Coleman (2007), who, in their ethnographic study on social work decision making in Ontario, concluded that there is an interaction between the structured system of standardized procedures and the use of intuition that was likely grounded in the social workers' internalized knowledge. Similarly, Blom, Nygren, Nyman, and Scheid (2007) found that child protection workers used multiple forms of knowledge.

In BC, Judge Gove (1995) recommended that the Risk Assessment Tool be used but argued that no formal risk assessment system can replace good social worker judgment (Callahan & Swift, 2007). However, an unintended consequence of the focus on the so-called objectivity and proceduralization of a device, such as the risk assessment tool, has been a neglect of the development of further understanding of how professional judgment and personal subjectivity affect the decision-making process. Current practices have devalued the importance of the skills it takes to develop and sustain an effective relationship in child protection. Our research found that child protection workers are more likely to pay attention to objective facts about a case situation when making decisions about risk and service provision, but more subjective factors are likely to affect decisions about home visits and contact hours.

This is an important point, as it shows a shift away from investigative work to work that emphasizes family support. Researchers such as Yatchmenoff (2005), who identified how positive client engagement is a beneficial component of child welfare, and Lee and Ayon (2004), who argued that the nature of the client relationship is a primary prerequisite for effective assessment and intervention in child welfare, have highlighted the positive effect of a good helping relationship with involuntary and sometimes resistant clients. Furthermore, Lee and Ayon (2004) found a significant correlation between a good worker-client relationship and improvements in children's physical care, discipline, emotional care, and parental coping. The necessity of a good helping relationship has been further argued by Platt (2008), who concluded that in the current context of increasing proceduralization in statutory social work changing the structures and techniques may be helpful, "but the roles of skilled workers are at least equally important" (p. 314). Our research indicates that when the importance of home visits and the amount of client contact are considered critical to the development of the relationship, the characteristics of the social worker are paramount in decision making.

This indicates that more attention should be paid to relationship building skills that continue past the formal academic environment into the practice setting. Although schools of social work provide foundational relationship building skills in their curriculums, more advanced in-house instruction and training should be provided in order to make certain that these skills remain viable and integrated into practice.

The role of relationship development seems to he critical to client change in child protection, but it requires more than technical instrumentality; social workers must make effective use of self to exercise professional judgment. In contemporary society professional judgment has been relegated to a secondary form of knowledge widely perceived as being untrustworthy. In complex situations in which there is no simple agreed-on solution, workers need to be encouraged by their supervisors to reflect on the complexities of the case. This kind of reflective practice involves talking and sharing on both rational and affective levels (Ruch, 2002) and requires time for social workers to describe and interpret their experiences, ideas, thoughts, and feelings. This can occur in self-reflective ways through internal dialogue, dialogue with oneself through writing or audio recording, or through dialogue with colleagues and supervisors. This iterative process of questioning and reflecting leads to a conscious and more objective use of self (Mandell, 2007).

Child protection work is complex, and the environment in which it occurs can be challenging. Mistakes and tragedies become the subjects of intense media scrutiny, which leads to political pressure to change procedures and social work practice. Social workers may be vilified in the media, and the credibility of the profession is hurt. When changes m practice are implemented in reaction to public scrutiny and media and political pressure, the resulting swings or shifts can be dramatic. In the case of BC, the media attention and the subsequent Gove Inquiry surrounding the death of Mathew Vaudreuil saw the child protection system shift its emphasis from significant use of family support services involving professional judgment to a primary focus on technocratic risk assessment. Neither of these approaches is inherently good or bad, and perhaps the best child welfare practice uses both in a balanced and thoughtful manner.

Our research indicates that despite pressures to adopt a technical, rational approach to decision making, social workers are holding on to the humanistic and artistic components of social work practice. They are using self and their experiential knowledge to make decisions about developing client relationships while paying attention to empirical research linked to evidence within case situations to make decisions about risk and safety.

POINTS & VIEWPOINTS

If you would like to make a substantive response to the premises, studies, and intellectual positions in an article you read in Social Work, this is your forum! This column seeks to stimulate dialogue that will help the profession respond to changing conditions. Send your manuscript (seven double-spaced pages or fewer) as a Word document through the online portal at http://swj.msubmit.net (initial, one-time registration is required).

doi: 10.1093/sw/swr007

Original manuscript received February 19, 2010

Final revision received August 5, 2010

Accepted August 17, 2010

Advance Access Publication June 4, 2012

REFERENCES

Blom, B., Nygren, L., Nyman, C., & Scheid, C., (2007). Social work students' use of knowledge in direct practice--Reasons, strategies and effects: Social Work & Society, 5(1). Retrieved from http://www.socwork.net/sws/article/view/120.

Budd, K. S. (2005). Assessing parenting capacity in a child welfare context. Children and Youth Services Review, 27, 429-444.

Callahan, M., & Swift, K. (2007). Great expectations an unintended consequences: Risk assessment in child welfare in British Columbia. In L. T. Foster, & B. Wharf (Eds.), People, politics, and child welfare in British Columbia (pp. 158-183). Vancouver, Canada: UBC Press.

Cash, S.J. (2001). Risk assessment in child welfare: The art and science. Children and Youth Services Review, 23, 811-830.

DeRoma, V. M., Kessler, M. L., McDaniel, R., & Soto, C. M. (2006). Important risk factors in home-removal decisions: Social caseworker perceptions. Child and Adolescent Social Work Journal, 23, 263-277.

Gambrill, E. (2005a). Critical thinking in clinical practice: Improving the quality of judgments and decisions (2nd ed.). New York: John Wiley & Sons.

Gambrill, E. (2005b). Decision making in child welfare: Errors and their context. Children and Youth Services Review, 27, 347-351.

Ganong, L. H., & Coleman, M. (2006). Multiple segment factorial vignette designs. Journal of Marriage and Family, 68, 455-468.

Gove, T. (1995). Report of the Cove inquiry into child protection. Victoria, Canada: Ministry of Social Services.

Jasso, G. (2006). Factorial survey methods for studying beliefs and judgments. Sociological Methods & Research, 34, 334-423.

Kondrat, M. E. (1992). Reclaiming the practical: Formal and substantive rationality in social work practice. Social Service Review, 66, 237-255.

Kondrat, M. E. (1995). Concept, act, and interest in professional practice: Implications of an empowerment perspective. Social Service Review, 69, 405-427.

Laming Lord. (2003). The Victoria Climbie inquiry: Report of an inquiry by Lord Laming. Norwich, UK: The Stationery Office.

Landsman, M., & Copps Hartley, C. (2007). Attributing responsibility for child maltreatment when domestic violence is present. Child Abuse & Neglect, 31, 445-461.

Lee, C. D., & Ayon, C. (2004). Is the client-worker relationship associated with better outcomes in mandated child abuse cases? Research on Social Work Practice, 14, 351-357.

Ludwick, R., Wright, M. E., Zeller, R. A., Dowding, D. W., Lauder, W., & Winchell, J. (2004). An improved methodology for advancing nursing research:

Factorial surveys. Advances in Nursing Science, 27, 224-238.

Mandell, D. (2007). Use of self." Contexts and dimensions. In D. Mandell (Ed.), Revisiting the use of self" Questioning professional identities (pp. 1-19). Toronto: Canadian Scholars' Press.

Munro, E. (1999). Common errors of reasoning in child protection work. Child Abuse & Neglect, 23, 745-758.

Parada, H., Barnoff, L., & Coleman, B. (2007). Negotiating "professional agency": Social work and decision-making within the Ontario child welfare system. Journal of Sociology & Social Welfare, 34(4), 35-56.

Parton, N. (2003). Rethinking professional practice: The contributions of social constructionism and the feminist "ethics of care." British Journal of Social Work, 33, 1-16.

Platt, D. (2008). Care or control? The effects of investigations and initial assessments on the social worker-parent relationship. Journal of Social Work Practice, 22, 301-315.

Reamer, F. G. (1993). The philosophical foundations of social work. New York: Columbia University Press.

Rossi, Pi H., & Nock, S. L. (Eds.). (1982). Measuring social judgments: The factorial survey approach. Beverly Hills, CA: Sage Publications.

Ruch, G. (2002). From triangle to spiral: Reflective practice in social work education, practice and research. Social Work Education, 21, 199-216.

Shlay, A. B., Tran, H., Weinraub, M., & Harmon, M. (2005). Teasing apart the child care conundrum: A factorial survey analysis of perceptions of child care quality, fair market price and willingness to pay by low-income African American patents. Early Childhood Research Quarterly, 20, 393-416.

Sicoly, F. (1989). Prediction and decision making in child welfare. Computers in Human Services, 5(3), 43-56.

Squires, G. (2005). Art, science, and the professions. Studies in Higher Education, 30, 127-136.

Taylor, B. J. (2006). Factorial surveys: Using vignettes to study professional judgement. British Journal of Social Work, 36, 1187-1207.

Walmsley, C. (2004). Social representations and the study of professional practice. International Journal of Qualitative Methods, 3(4), 1-18.

Webb, S. (2002). Evidence-based practice and decision analysis in social work: An implementation model. Journal of Social Work, 2, 45-63.

Yatchmenoff, D. K. (2005). Measuring client engagement from the client's perspective in nonvoluntary child protective services. Research on Social Work Practice, 15, 84-96.

Zeira, A., & Rosen, A. (2000). Unraveling "'tacit knowledge": What social workers do and why they do it. Social Service Review, 74, 103-123.

Jacqueline Stokes, EdD, RSW, is dean, School of Academic Foundation and Academic Services, College of New Caledonia, Prince George, Canada. Glen Schmidt, PhD, RCSW, is associate professor, School of Social Work, University of Northern British Columbia, Prince George, Canada. Address correspondence to Glen Schmidt, School of Social Work, University of Northern British Columbia, Prince George, B.C. Canada V2N 4Z9; e-mail: schmidt@unbc.ca.

Table 1: Risk Assessment (n = 327)

                                  Risk Level

Variable                df   [chi square]   P

Harm to child           3    104.98         <.0001
Income                  3      0.96          .8100
Housing                 3      7.84          .0494
Culture                 1      0.55          .4569
Substance use           3     15.89          .0012
Spousal violence        3     11.62          .0088
Resources and support   3      5.25          .1546
Cooperation             3     13.86          .0035

                               Service                 Visit
                             Provision               Importance

Variable                [chi square]   P        [chi square]   P

Harm to child             102.79       <.0001       4.08       .2529
Income                      1.01        .7996       2.46       .4823
Housing                    15.31        .0016      11.81       .0081
Culture                     1.22        .2690       1.52       .2173
Substance use               1.57        .1347      10.33       .0167
Spousal violence            3.16        .3679       0.40       .94505
Resources and support       0.47        .9256       4.37       .2241
Cooperation                 7.60        .0550       3.75       .2903

                          Contact
                           Hours

Variable                F      P
Harm to child           4.63   3
Income                  0.94   .4234
Housing                 2.10   .1011
Culture                 0.04   .8405
Substance use           2.00   .1147
Spousal violence        0.85   .4657
Resources and support   0.37   .7763
Cooperation             1.55   .2019

Table 2: Respondent and Organizational Variables (n =118)

                                 Risk Level

Variable                     df  [chi square]   P

Gender                       1    3.29          .697
Age group                    3    2.28          .5171
Job satisfaction             3    2.10          .5510
Degree                       4   11.29          .0235
Protection experience        4    8.96          .0621
Community experience         5    7.26          .2024
Delegation experience        1    0.08          .7772
Practice in urban location   2    0.88          .6430
Supervision satisfaction     6   15.99          .0138
Current social work role     1    5.40          .7138

                                 Service            Visit
                                 Provision          Importance

Variable                     [chi square]   P       [chi square]

Gender                        0.74          .3892    6.55
Age group                     4.87          .1815    8.26
Job satisfaction              3.21          .3602   24.29
Degree                       17.31          .0017   20.69
Protection experience        17.52          .0015    4.61
Community experience          9.90          .0780    6.40
Delegation experience         0.16          .6882    1.61
Practice in urban location    3.52          .1716    1.40
Supervision satisfaction     10.30          .1125    6.17
Current social work role      7.91          .4426   74.79

                             Visit
                             Importance   Contact Hours

Variable                     P            F       P

Gender                        .0105        3.96    .0475
Age group                     .0409        9.06   <.0001
Job satisfaction             <.0001        1.93    .125
Degree                        .0004        7.76   <.0001
Protection experience         .3294        0.73    .5719
Community experience          .2691        3.61    .0035
Delegation experience         .2039        6.59    .0108
Practice in urban location    .4965       12.64   <.0001
Supervision satisfaction      .4042        1.85    .0906
Current social work role      .0017        2.08    .383