Predictors of Time Demand were investigated as a means of estimating case and caseload difficulty in Rehabilitation Counseling. Differences in the time investment of 100 rehabilitation counselors were investigated among a statewide sample of 1330 cases. Criterion time variables were comprised of various groupings of the federal status codes prescribed by the United States Rehabilitation Services Administration (e.g., planning, counseling, restoration, training, and placement). Time was thereby partitioned to represent the roles and functions of rehabilitation counselors. Using the data reduction facilities of the SPSS AnswerTree program (SPSS, 2001), the rehabilitation planning, training and placement variables were categorically significant, parametrically confirmed and cross-validated. Four case types of optimal statistical and practical significance resulted. The rehabilitation process was reduced to new and existing cases, which were each split into high and low service complexity groups. The results of the study have implications for quality improvement and organizational development in mental health and rehabilitation program management.
Keywords: Rehabilitation, caseload difficulty, time demand, burnout, workload.
Workload measurement was designed to address problems caused by inequities in time demand. In the workload measurement literature it is widely accepted that systemic workload inequity has a negative impact upon organizational effectiveness and efficiency, and upon workforce morale (Kern, 1987, 1988, 1993; Peat, Marwick, & Mitchell, 1978a, 1978b; USDHEW, 1972, 1978; Walls & Moriarty, 1977). The underlying assumption is that in controlling these imbalances, objective workload measurement techniques are superior to subjective workload estimation.Workload management problems develop when factors that impact time demand differently are not adequately understood or accounted for (Ebb, 1994; Tarvydas and Peterson, 1999). Problems associated with unbalanced workloads include: low morale due to overwork, confusion as to the nature and extent of the work expected, arbitrary and subjective case assignments, increased need for detailed supervisory intervention, and agency difficulty in justifying the need for additional staff and funding (Haring, 1974). Exploring the nature of time demand in the rehabilitation process is needed.
the accurate partitioning of time demand in this study include several
and external threats to validity. Internally, selection and
require consideration (Campbell & Stanley, 1963; Cook &
Gall, Borg & Gall, 1996). Although the sample was not randomly
the systematic selection of every 12th case on each caseload
approximated random selection, and sample size (N=1330) was adequate.
instrumentation, the accuracy of the time log used for data collection
through multiple significance testing and it was cross-validated (SPSS,
The time log was self-report, however, and susceptible to data
The purpose of the study was to identify the strongest available predictor of time demand and use it to partition the dependent time variable into comparative and optimally cross-validated models. The criterion time variable was comprised of counselor time investment in minutes reported within 14 functional service categories. The 14 service categories represented the essential roles and functions of the rehabilitation process as formerly developed and mandated through the Rehabilitation Services Administration’s Federal Status Codes. The Federal Status Codes were used to operationalize the first two of six predictor variables (i.e., rehabilitation process levels, case movement, service complexity, caseload size, percent of active cases, and disability type) selected according to their potential for predicting time demand. The null hypothesis was used to test each model.Literature Review
Time demand is inferred in the rehabilitation role and function literature as well. The Federal Status Codes, developed and initially mandated by the Rehabilitation Services Administration, strongly reflect both time demand and the roles and functions of the profession. These groupings, in order, are: referral, intake, eligibility determination, plan development, counseling services, restoration services, training services, job placement, employment, successful closure, unsuccessful closure after plan development, unsuccessful closure before plan development, and post-employment. Although counselor role and function has been as much a matter of legislated philosophy as of research, some functions are more preeminent and have endured both inquiry and legislative change. Planning, for example, may be the most integral. Although the name of the Individualized Written Rehabilitation Program was recently changed to the Individualized Plan for Employment in the Workforce Investment Act of 1998 (National Governor’s Association, 1998), the planning function remains largely unchanged as the centerpiece of the vocational rehabilitation process.
Recent developments in the rehabilitation role and function literature have been centered in the emergence of counselor specializations (Goodwin, 1992). Counselor specialties often effect changes in the structure of work in the organization. Due to the complexity of the field, counselors are increasingly working with specialty caseloads rooted in either the counselor functions themselves (e.g., job placement services) or in serving specific disability groups (Roessler & Rubin, 1995).
Support in the rehabilitation literature for time demand as an aspect of the role and function of rehabilitation counseling is admittedly indirect and scant (Goodwin, 1992; National Governor’s Association, 1998; Roessler & Rubin, 1995). This report on the Main (2002) doctoral dissertation is the first study to directly link the two. When references to time or time demand are found in rehabilitation research, it is most often in the context of caseload difficulty.Caseload Difficulty
Due to the difficulty and expense involved in implementing weighted closures, Walls and Moriarty (1977) developed an alternative in which a flexible variety of performance measures could be profiled as percentages or stanines in comparison to national, state or area norms. The average time that clients were staying in a particular status (e.g., 02-10) was suggested as an example of possible time measures that could be included in counselor outcome profiles.
Zadny and James (1977) asked 319 counselors in seven states to estimate their time spent in travel, placement-related travel, and nine case process measures. These were counseling, paperwork, coordinating services, placement, job development, public relations, planning, and professional growth. Time estimates were correlated with two sets of factor analyzed measures representing six outcome dimensions. Total rehabilitations, rehabilitations of persons with severe disabilities, and the percentage of cases closed as not rehabilitated comprised the number of successful closures factor. The nature of successful closures factor was comprised of the percent of all closures that were competitive, the percent of closures in sheltered workshops, and average earnings at closure.
There were no significant correlations in the nature of closures factor, but several relationships in the number of closures factor were significant at the .01 level. The percent of cases closed not rehabilitated was inversely related to time spent in placement (r = -.22), job development (r = -.27), hours spent out of the office (r = -.31), miles traveled per week (r = -.31), travel time spent in job development (r = -.38), and travel time spent in placement (r = -.39). Rehabilitations for clients with severe disabilities were related to time spent in job development (r = .20) and planning (r = .22). Total rehabilitations were inversely related to time spent in coordinating services (r = -.20) and planning (r = -.23). The investment of time in placement related activities was recommended. (Zadny & James, 1977)
Chan, Rosen, Wong and Kaplan (1993) used discriminant analysis to differentiate a group of 18 students from 18 experienced counselors on the basis of caseload process measures (p<. 001). A computer-based multiple case management simulation was developed to measure six task groupings. Three caseload measures were significant (total number of status changes, index of preferred actions, index of inappropriate status changes) and three measures were not significant (money spent on clients, number of actions taken, appropriate closures). Using the classification function, 89% of the participants were accurately classified. This study approximates the use of time standards because participants were given one hour to process eight case types.
Related to caseload difficulty, concern over stress in rehabilitation counseling has also been more recently voiced. Professional disengagement, compassion fatigue, and burnout, have been acknowledged (Greenwood, 1995; Roessler & Rubin, 1995), and counselors have been cited as experiencing disempowerment in the effort to empower persons with disabilities (Emener, 1993). Most importantly, organizational stressors such as role overload (e.g., caseloads that are too large or too active) have been clearly outlined in rehabilitation stress management models (Wood, 1999), and the impact of the organizational system on counselor-client ratios has been given ethical consideration. Tarvydas & Peterson (1999) directly cite the demoralization that occurs for some counselors when counselor-client ratios are not kept consistent with the complexity of the caseload management system and the difficulty of the client population served. Clearly, workload management has potential for addressing these issues.Interpretive Summary and Variables
that is not directly operationalized in the present study is severity
disability. The dichotomous designation of severity of disability
the Rehabilitation Services Administration did not evidence enough
on the criterion to be of use in this study. This omission requires
deliberation because relationships between severity of disability and
factors have been previously demonstrated (General Accounting Office,
Zadny & James, 1977). Upon first consideration it appears that this
does not contain a measure of severity of disability with an adequate
scores or categories to demonstrate that time demand is related to the
or magnitude of functional limitations created by the disability. State
agencies were first mandated to focus their efforts on serving persons
disabilities in the Rehabilitation Act of 1973. The presumption within
field then and now appears to be that that time demand is unrelated to
of disability and that caseload difficulty is a function of severity of
disability and disability type (
The possibility that time demand is independently or negatively related to severity of disability must be considered, however. Specialty services may have become so well organized around disability groups with most severe disabilities that the actual time needed to serve a person in this category is less than might be expected. Similarly, persons with severe disabilities who are fairly well established in their careers may need only one or two VR services. In essence, time demand itself is the primary mediator of severity of disability to the counselor. On this basis the criterion variable is taken also as an indication of severity of disability.
were used as predictors of time demand (i.e., rehabilitation process
movement, service complexity, caseload size, percent active, and
type). Case movement is informed by the work of Chan et al. (1993) who
an indication that when time is held constant, different counselor
and rates of task processing (e.g., number of status changes, number of
preferred actions, and number of inappropriate status changes) emerge.
independent variables service complexity, caseload size and percent
appear to be substantiated by the professional judgment of the
counselors surveyed by McAnally and
The long history of support for accounting for disability type in rehabilitation research is particularly clear from the studies on caseload difficulty in rehabilitation (Cooper & Pearce, 1980; Miller & Barillas, 1967; Miller et al., 1965; Noble, 1973; Sermon, 1972; Silver, 1969). There is definitely an awareness reflected in the literature that different case types vary dramatically in their difficulty level. There is some indication (Wallis & Bozarth, 1971; Walls & Moriarty, 1977) that successful and unsuccessful clients do not necessarily differ on several matched factors (i.e., age, gender, education, referral source and disability type). This leaves open the possibility that factors related to time demand might account for some of this difference.
that no case
differences existed between weighted and unweighted cases (Silver,
& Bozarth, 1971; Walls & Moriarty, 1977) suggests that case
or workload may be similar when time demand is not accounted for. The
recommendation for the use of time-in-status as a performance measure
and Moriarty (1977) raises two issues. The recommendation directly
calls for a
real time measure, and it substantiates the need to control for the
time demand in performance appraisal to avoid biased penalties and
This appears to apply to estimated time spent serving persons of a
disability type (McAnally & Linz, 1988), and to production goals
set on the
basis of successful closure rates (Cooper & Pearce, 1980). Efforts
added weight to more difficult closures based on disability type were
lived, possibly due to the fact that time demand was not taken into
Time logs were used to record time investments in each of the process categories for one month. Counselors self-reported their time investments for each case in minutes. Demographic information used to determine client types was also recorded on the time log form. Data was taken directly from the computer information system of the state agency or from the case file. Of 15 variables for which data were collected by the state agency, five were selected and modified as independent variables for the study. No validity or reliability studies of the time log instrument were conducted, but the reliability of the log was largely confirmed through the parametric cross-validation in the analysis.
The analysis began with the evaluation and implementation of the recommendations suggested by Peat et al. (1978b) and others to enhance accuracy and reduce the impact of skewed time distributions. Peat recommended the use of the mode for work unit standards and the removal of extreme scores and case types that do not apply to the whole sample. Adjustments were also made for missing values.
Secondly, the strongest predictor of time demand was identified through the categorical data reduction of the six predictor variables on the criterion time variable. Categorical data reduction is the term used in the AnswerTree program (SPSS, 2001) to describe the process of running multiple significance tests among competing predictor variables, identifying the most significant predictors and splits in each target variable. The process results in a diagram called a tree in which each category is called a node, and each split is called a branch. When the data is reduced until no further splits are statistically significant, terminal nodes are attained.
Third, the strongest predictor was used to partition the time variable from comparative theoretical and empirical standpoints. The most viable models, finally, were cross-validated against a random sample of approximately 50 withheld cases.Measures
The second predictor variable was the dichotomous number of federal status code changes, per case, in the month of data collection, taken as a measure of case movement (one or more than one). This variable gave an indication of the relative speed at which different case types move through the rehabilitation process. The third predictor was the number of specific service categories from the month prior representing service complexity (one or more than one). Fourth, caseload size was dichotomized as small or large (below the median; at or above the median). Fifth, the proportion of cases that demanded time for each counselor was taken as a measure of the percent of active cases (below 50% and 50% and up). Finally, disability type included visual, hearing, orthopedic, mental illness, developmental disabilities, learning disabilities, and all other conditions.Data Analysis
Furthermore, no outlying cases or caseloads were removed from the analysis. In assessing the impact of outlying scores, the 10 individual cases with the most total minutes reported ranged from 895 to 525. These cases appeared to be valid and proportionately distributed among the disability types. The largest and smallest workloads reported were from blind services caseloads, which limited their extraneous impact. The largest number of minutes reported for a caseload was 2985, or an average of just under four hours spent on each of the 14 cases sampled over the month of data collection. The smallest number of minutes reported for a caseload was zero. This caseload had been recently vacated and assigned to another counselor for triage. None of the extreme cases or caseloads seemed unreasonable and none were adjusted or removed.
Several adjustments were made for missing values, however. Missing values on the total time variable (N = 41) were computed as the sum of all process minutes reported. Missing values on the service complexity predictor demanded only slightly less time than the time demanded by cases having one service category and were combined (i.e., coded less than two, or two or more). Three of 100 caseloads missing the caseload size value were coded at the mean caseload size for their respective caseload types (i.e., blind services, 56; general caseload, 90). Missing disability type values were assigned separate codes within their respective caseload types (i.e., visual, hearing, mental health, general). The adjusted data set included an alternative coding system that made use of available information about each missing value. Potentially, the full data set (N = 1330) could be used for some analyses, and missing values could be selectively removed.The full data set was used to identify the strongest predictor of time demand categorically, as recommended by Peat et al. (1978b). The model was nonparametric and automatic, meaning that the insignificant levels of the variables could all be merged or reduced.
The rehabilitation process levels entered the model first (Chi-square = 215.3811; p = .0000) and the levels were reduced to three groups. Planning was merged with new cases. Service categories 14, 16, and 18 (i.e., counseling, restoration, training) were merged with existing cases, and the placement category remained a single group. The strongest predictors, in order, were process levels, percent active, service complexity, and case movement (Chi-square > 4.8190; p < 0.0282).
Cross-validation, however, fell short of the parametric capacity of the AnswerTree program. Parametric confirmation had to be customized. In other words, the parametric version of the AnswerTree program identified more accurate group splits that were not used in this reduced, categorical model. Two terminal case movement nodes on the high complexity, new cases branch were not confirmed (F = .6011; p = .2750), and multiple interactions were found in the cross-validation sample (N = 68) that did not exist in the confirmation sample (N = 1262). Also, four of the nine, terminal cross-validation sample nodes had less than five cases. Parametric confirmation of the process levels predictor was strong. The null hypothesis was rejected, but a parametric model with stronger splits and branches was clearly indicated for future workload measurement purposes.
Several theoretical models were tested, meaning that different partitions of the process levels predictor were forced into the analysis and could not be reduced. One of the strongest of these was a symmetrical five group model (i.e., planning, counseling, restoration, training and placement). The theoretical rehabilitation process groups better approximate the role and function literature. Cases with missing federal status codes were removed from this model to avoid their extraneous mergers with the theoretical groupings (N = 1175). The five theoretical process categories were each significant when tested using the parametric confirmation facility of the CHAID procedure (p < .0211, F > 5.4135). The null hypothesis was again rejected. Service complexity was the strongest nonparametric predictor of rehabilitation planning, counseling and restoration service groups (Chi-square > 8.4063, p < .0038), and the percent active variable was the strongest predictor of training and placement groups (Chi-square > 17.7934, p < .0001). Six terminal nodes did not cross-validate, however, not improving on the reduced model.
Therefore, an automatic, parametric data reduction was used and cross-validated to more fully tap the practical utility of the AnswerTree program. For this automated model, the modified means were used as the process variable and the process groups were scaled ordinally. Although the nominal scaling of the process categories produced more nodes with more exacting mergers, the ordinal scaling of the groups was simpler, more logically understandable, and more consistent with the theoretical process categories.
Two process levels resulted (F = 202.4067; p = .0000). Cases in status codes 00, 01 and 02 were merged into a new-cases node (i.e., referral, new and applicant statuses, respectively). Cases above status 02 comprised an existing-cases node (i.e., eligible, plan completed, counseling, restoration, training, existing, and placement statuses, respectively).
Service complexity was the strongest predictor for both new and existing cases (F >14.6111; p < .0003). The base standards in Figure 1 are based on the process and complexity nodes of the second level, before the third variable entered the model. In the existing-cases node, percent active was the third and final variable to enter the model (F >19.0323; p = 0.0000).
Each of the
nodes cross-validated in the predicted direction. No interactions were
in the cross-validation sample (N = 53), and only one node contained
five cases (N = 3). The optimal splitting function of the AnswerTree
was strongly implemented, and no reasons were discovered for
model. Therefore, the case type means from this automatic model
appeared to be
optimal for the further development of workload standards. Figure 1
D. T. & Stanley, J. C. (1963). Experimental and
quasi-experimental designs for research.
Chan, F., Rosen, A. J., Wong, D. W., & Kaplan, S. (1993). Evaluating rehabilitation caseload management skills through computer simulations. Journal of Counseling and Development, 71 (5), 493-498.
T. & Campbell, D. (1979). Quasi-experimentation: Design
and analysis issues for field settings.
Cooper, P. G., & Pearce, R. L. (1980). A model for the estimation of caseload potential. Evaluation Review, 4 (6), 789-801.
N. (1994). Child support reform: A state checklist for change.
W. G. (1993). Empowerment in rehabilitation: An empowerment
philosophy for rehabilitation in the 29th Century. In M. Nagler (Ed.), Perspectives
on Disability (pp. 297-305).
M. D., Borg, W. R. & Gall, J. C. (1996). Educational Research:
Introduction (6th ed.).
Accounting Office. (1991). Vocational rehabilitation:
Clearer guidance could help focus services on those with severe
Report to the Chairman, Subcommittee on Select Education, Committee on
Education and Labor, House of Representatives (GAO Publication No.
Glaser, M. (1993). Reconciliation of total quality management and traditional performance improvement tools. Public Productivity and Management Review, 16, 379-386.
Goodwin, L. (1992). Rehabilitation counselor specialization: The promise and the challenge. Journal of Applied Rehabilitation Counseling, 23 (2), 5-11.
B. (1974). Workload measurements in child welfare.
J. C. (1997, June). Report on the goals of the Client
First Hearing before the United States Senate on the Reauthorization of the Rehabilitation Act of 1973, 108th Cong., 1t Sess. (1997) (testimony of Bobby Simpson, representing the Council of State Administrators of Vocational Rehabilitation).
Kern, H. D. (1987). Child protective services workload analysis and management. Protecting Children, 4, 4-6.
H. D. (1988). Workload standards in child protective services. In
Guidelines for a Model System of Protective Services for Abused and
Neglected Children and Their Families.
H. D. (1993). Workload analysis and management methodology: How
to develop and use a workload system. In J. Fluke (Ed.), Child
Workload Analysis and Resource Management Resource Manual (pp. V-1
R. (1995). Management.
J. H., & Schumacher, S. (1989). Research
in education (2nd ed.).
P. L., &
Miller, L. A., & Barillas, M. G. (1967). Using weighted 26-closures as a more adequate measure of counselor and agency effort in rehabilitation. Rehabilitation Counseling Bulletin, 11 (2), 117-121.
Miller, L. A., Muthard, J. E., & Barillas, M. G. (1965). A time-study of vocational rehabilitation counselors. Rehabilitation Counseling Bulletin, 9, 53-60.
National Governor’s Association, Employment and Training Program, Center for Best Practices (1998, August 3). Workforce Investment Act of 1998 (H.R. 1385): Summary and final description of final compromise. Legisline, 1-2.
Noble, J. H., Jr. (1973). Actuarial system for weighting case closures. Rehabilitation Record, 14 (5), 34-37.
Peat, Marwick, Mitchell & Co. in association with the Child Welfare League of America (1978a). System of Social Services for Children and their Families (DHEW Publication No. OHDS 78-30136). Washington, DC: National Center for Child Advocacy, Children’s Bureau, Administration for Children, Youth and Families, Office of Human Development Services, United States Department of Health, Education and Welfare.
Marwick, Mitchell & Co. in association with the Child Welfare
League of America (1978b). Developing workload standards for
family social services (DHEW Publication No. OHDS 78-30131).
Randolph, A. H. (1975). The Rehabilitation Act of 1973: Implementation and implications. Rehabilitation Counseling Bulletin, 18 (4), 200-204.
Rehabilitation Act of 1973, Pub. L. No. 93-112, § 701, 87 Stat. 355 (1973).
Services Administration (1998). United States Department
of Education, Office of Special Education and Rehabilitative Services,
Rehabilitation Services Administration. Vocational Rehabilitation
Program: Fiscal Year 1998 Data Analysis Guide and National Data Tables.
R. T. & Rubin, S. E. (1992). Case Management and
Rehabilitation Counseling (2nd ed.).
Inc. (1998). Answer Tree 2.0 Users Guide.
Sermon, D. T. (1972). The difficulty index: An expanded measure of counselor performance. Minnesota Department of Vocational Rehabilitation (Available from the National Clearinghouse of Rehabilitation Training Materials, 5202 North Richmond Hill Drive, Stillwater, OK, 74078-4080).
Silver, D. L. (1969). A look at evaluation of VR counselor performance. Journal of Rehabilitation, (November-December), 13-14.
Inc. (1998). Answer Tree
2.0 Users Guide.
Information Network of
V. M., & Peterson, D. B. (1999). Ethical issues in case
management. In F. Chan & M. Leahy (Eds.), Health care and
case management (pp. 317-354).
States Department of Health, Education and Welfare, Office of
Human Development Services, Office of Child Development, Children’s
(1978). Child Welfare in 25 States - An Overview (DHEW
States Department of Health, Education and Welfare, Social and
Rehabilitation Service, Assistance Payments Administration (1972). Work
Measurement and Work Simplification (DHEW Publication No. SRS
Wallis, J. H., & Bozarth, J. D. (1971). The development and evaluation of weighted DVR case closures. Rehabilitation Research and Practice Review, 2 (3), 55-60.
Walls, R. T., & Moriarty, J. B. (1977). The caseload profile: An alternative to weighted closure. Rehabilitation Literature, 38 (9), 285-91.
C. (1999). Self-management for case managers. In F. Chan & M.
Leahy (Eds.), Health care and disability case management (pp.
Zadny, J. J., & James, L. F. (1977). Time Spent on Placement. Rehabilitation Counseling Bulletin, 21, 31-35.