Developing a Reliable and Valid Process for Estimating Post-Injury Earning Capacity: Part I (the Problem)

Purpose/Introduction

Damages can be specific or general. Specific damages are easily quantifiable while general damages are more difficult to quantify, or they are qualitative. The main primary specific damages are: 1) lost earning capacity, 2) lost household services, and 3) treatment costs. General damages could be: 1) loss of consortium, 2) pain and suffering, and 3) lost value of life.

As one of the primary specific damages, lost earning capacity is likely the most common type an expert witness is asked to evaluate. For purposes of this paper, we are narrowing our analysis to lost earning capacity, and more specifically to post-injury earning capacity. Horner and Slesnick (2016) define earning capacity as “a legal term used by most courts in personal injury cases to designate economic losses due to the impairment of ability to earn money” (p. 1). How often are earning capacity evaluations performed in the United States? When the authors performed a search and reviewing case law on Google Legal between 2013 to 2023, the concept of earning capacity as described by Horner and Slesnick applied to some administrative systems such as workers’ compensation and family law. In the four years before COVID-19 rolled out global lockdowns in early 2020, there were 4,650 cases where earning capacity was an issue. Comparatively, in the four years since the start of the pandemic, there were 3,959 cases (17% drop). These search results are a substantial underestimate of actual cases tried annually involving the concept of earning capacity because only about 11% of cases are appealed and end up in research databases such as Google Legal (Eisenberg, 2004). Furthermore, unpublished opinions often do not end up in these databases at all. In fact, about 87% of federal case opinions in the last few years were unpublished (Brown et al., 2022). Over 95% of cases are settled before they get to trial (Prescott & Spier, 2015; U.S. Department of Justice, Office of Justice Programs, Bureau of Justice Statistics, 2005). Considering all factors, it is likely there are at least 100,000 cases per year involving the concept of earning capacity valued in the billions of dollars – if not in the trillions. The responsibility of the vocational expert (VE) to perform a valid analysis has great ramifications. Consequently, the importance of models to assist experts in arriving at objective conclusions to assist the trier of fact is paramount.

Regardless of the jurisdiction in which the evaluator of earning capacity practices, the professional must consider the primary and secondary evidence relevant to the evaluee and assess it through objective measures (Commission on Rehabilitation Counselor Certification, 2023). However, the law defining earning capacity varies from jurisdiction to jurisdiction and is often not very clear. Moreover, case law is not static, but is dynamic and can change. For example, some states indicate that expected earnings are earnings based on past earnings history while earning capacity is based on expected future earnings. It is incumbent upon the evaluator – typically a VE – to not only have an excellent grasp on the method of earnings capacity analysis, but also its context (e.g., workers’ compensation, personal injury), in the development of their assessment and opinions, and also in the handoff to the forensic economist (FE).

As a team, the VE and FE focus on specific damages. To assist the FE in determining the calculations for earning capacity, the VE must carefully look at what the person “could do” (capacity) rather than what they “choose to do” (choice). The method of estimating pre- and post-injury earning capacity must be consistent with one another. Apples must compare to apples. Otherwise, an evaluator could wind up with an estimate of pre-injury earning capacity that is inconsistent with the post-injury expected earnings analysis; this could bias the award upwards or downward, depending on the contradiction of the analysis.

This paper is different than what currently exists in the literature of how specific damages are estimated by the VE-FE team because we examine not what exists in the damages analysis, but how the analysis is performed. Our effort addresses current methodology from a different angle. First, we focus exclusively on lost earning capacity – the difference in present value between pre-injury earning capacity and post-injury earning capacity. Then, we identify problems with these contemporary models and procedures in this paper, and seek to eventually offer solutions.

Evidence-Based Practice

Codes of ethics constitute a profession’s guidelines of behavior for practitioners. The predominant code of ethics that provides the footprint to adjunct codes upon which VEs rely is the Code of Professional Ethics for Certified Rehabilitation Counselors (2023) from the Commission on Rehabilitation Counselor Certification (CRCC). Specific to forensics and ethical applications of models, theories, and research, the Code outlines the following standards for certified rehabilitation counselors:

• G.1.a. “… produce unbiased, objective opinions and finding that can be substantiated by information and methodologies appropriate to the service being provided … form opinions based on their professional knowledge and expertise, which are supported by the data” (p. 20).

• G.2.d. “… ensure the resources used or accessed in supporting opinions are valid, current, and cited” (p. 21).

• G.2.e. “… use knowledge, consistent with accepted clinical and scientific standards, and accepted data collection methods and procedures for evaluation, treatment, consultation, and scholarly/empirical investigations” (p. 21).

• G.2.e. “… seek to obtain independent verification of data relied upon …” (p. 22).

Consequently, not only are VEs ethically bound to consider the impact of data in the evaluation and apply it objectively, but also VEs must understand and apply a framework, models, and techniques used to arrive at their opinions.

Evidence is primary information collected by the evaluator through two forms – interviews and observations – or through secondary information collected by others but used by the VE that fall into three categories including case-specific documents (e.g., medical records) or related literature/publications, archival records like empirical research databases, and physical artifacts like images (Barros-Bailey, 2018). Critical appraisal literature is when researchers qualitatively analyze all the literature (called a systematic review) or perform a quantitative analysis (called a meta-analysis) of studies on a specific research topic. Because of the expansive nature of this kind of literature, its use is considered the “gold standard.” Of particular importance in the evidence-based analysis is aligning the primary with the secondary evidence specific to the case, or specific to the disability or diagnostic group most pertinent to the evaluee as described in longitudinal or other critical appraisal literature.

Individualized Assessment

At the center of an analysis of loss of earning capacity is the individual plaintiff. Whether evaluating an individual for a clinical or forensic reason, the imperative is to perform an individualized appraisal. The standard for performing an individualized assessment recently celebrated 50 years of legal existence as it was concretized through Section 501 of the Rehabilitation Act of 1973 (U.S. Office of Personnel Management, n.d.). Today, the individualized assessment is common across various government and professional systems involving the evaluation of employment (U.S. Equal Employment Opportunity Commission, 2012) or individuals with disabilities such as the Americans with Disabilities Act (U.S. Equal Employment Opportunity Commission, 2000, 2003). Vocational assessments must be individualized and objective (Commission on Rehabilitation Counselor Certification, 2023). A cookie cutter approach is not the standard across evaluations of a population that is highly heterogenous demographically and functionally. The same expectation of an individualized assessment is foundational to the assessment of earning capacity as found in government (e.g., Aaron v. Saul, 2020), professional (Barros-Bailey, 2021, 2023; Barros-Bailey & Heitzman, 2014), and legal sources.

Therefore, the dilemma arises of how to take general data (e.g., population surveys) and apply them to a plaintiff (N=1). The question arises whether the 20 published models in the VE literature for assessing loss of earning capacity (Barros-Bailey, 2025; Robinson, 2014) sufficiently provide VEs with the tools to address reliability so that the parties and the trier of fact in over 100,000 cases per year in the United States can understand how the facts resulted in the VE’s findings and opinions.

The G2i Dilemma

When scientific information is used in the courtroom, a dilemma often arises. Scientific information is “general” in that it is about populations rather than individuals. The objective in the courtroom, however, usually concerns a determination of facts about individuals rather than populations (Faigman & Fisher, 2015). The application of general scientific information to individuals has been referred to as “going from the General to the individual,” or G2i.

The concept of generalization or generalizability in research can be addressed from quantitative, qualitative, and mixed methods approaches. The concept refers to drawing conclusions or making inferences from findings to a broader population beyond the sample studied. The extent of the generalization impinges on the research design, sampling methods, and the study’s purpose. Often, in the generalizability process, researchers must consider the limitations and potential biases or errors that could affect the generalizability to their findings. The same is true when taking generalized data (e.g., government surveys with broad questions covering all disability groups) and applying it to a plaintiff (N=1). The reverse process of going from population data to the individual can be very problematic as the data are so generalized that they may have little or no bearing on the reality and variability of the individual’s worklife. This is the G2i dilemma in the courtroom—the degree to which general population data (G) can be applied to (2) the individual (i) evaluee being assessed (Faigman & Fisher, 2015). As an example, an epidemiologist may provide research showing whether lead found in the home can cause long-term cognitive difficulties for children living in that home. However, if John Doe sues the landlord of the house he lived in as a child for the effects of lead exposure, Smith must prove not only that lead can cause cognitive difficulties, but also that lead in his house at the time he lived there caused his specific problems that impacted his earning capacity.

An example occurs when estimating pre-injury worklife expectancy as a metric for the duration of someone’s earning capacity. The variables considered in a well-known population-based model of worklife are age, gender, level of education, and whether the individual was active or inactive at the time of the tort (Skoog et al., 2019). Some FEs will go beyond those variables and “tailor” their results (Foster & Rodgers, 2016). If, for example, the plaintiff was a young single female with a strong orientation towards a career, the standard worklife tables may not apply to that particular individual because not all the variables that are important in her case (e.g., marital status, career driven) exist in current worklife expectancy models. A FE might claim that the plaintiff would more likely have the worklife of a male rather than a female to try to capture the variables missing in the tables that would otherwise potentially underestimate the labor force participation of the plaintiff. Several possible solutions to this “missing variable” problem include using the existing loss of earning capacity models despite their apparent drawbacks, adding additional statistically significant variables to the models, using clinical judgment, or ignoring the problem by leaving the decision of estimating the plaintiff’s worklife up to the trier-of-fact without any guidance from the expert. Estimating a specific probability as opposed to referencing a general population statistic such as average age of receiving Social Security benefits would, however, require the expertise of the VE.

Adding more variables might include years of continuous employment, family status (married or not and number of children), career engagement, and whether the person is self-employed. Of course, developing such a model depends upon the availability of sufficient data and, most importantly, having a sound theoretical basis. Refining a well-established statistical model used by an FE can often be an arduous task that leaves the FE vulnerable to defending the approach.

Alternatively, experts can use clinical judgment to adjust based upon their professional experience, education, and expertise, which will be described later. The concept of clinical judgment in forensic rehabilitation counseling enjoys a relatively short history (Barros-Bailey & Neulicht, 2005; A. Choppa et al., 2004; A. J. Choppa et al., 2014; Field et al., 2009; Field & Choppa, 2006) compared to a large body of literature within other clinical and forensic fields (AEgisdottir et al., 2006; Austin & Leahy, 2015; Borum et al., 1993; Buongiorno et al., 2025; Dawes et al., 1989; Dror & Rosenthal, 2008; Gambrill, 2005; Garb, 2005; MacLean et al., 2019; Rosenthal, 2004; Spengler et al., 2009; Strohmer & Leierer, 2000; Tanner, 2006).

Finally, the expert could solve the G2i problem by not providing any opinion concerning worklife. For example, instead of estimating worklife, the expert calculates economic loss year-by-year up through life expectancy. The specific estimate is left to the trier-of-fact based upon other evidence that will be presented at trial. Thus, the expert would not even provide an estimate based upon population averages.

However, it is important to note that the court has generally ruled against deviations from population-based worklife tables. Ireland and Slesnick (2018) examined 16 legal decisions related to worklife expectancy, with the majority concerning pre-injury worklife. The authors stated that:

One very clear message derived from most of these cases is that statistical worklife tables are usually considered a null hypothesis unless there is very good evidence to the contrary … At least based on these court documents, however, it is not entirely clear what specific factors count for an estimate of worklife other than those found in the standard tables such as age, education, and gender (Ireland & Slesnick, 2018, p. 79).

For several decades, well-developed literature has examined the accuracy of clinical judgment in contrast to statistical models. Consequently, it is important to clarify that experts’ clinical judgments, or their observations and inferences made about evaluees, informs their decisions or expert opinions. These studies examine the reliability and validity of decisions made by a variety of professionals including, among others, psychologists and physicians. The studies involve situations where the prediction outcomes made could be tested – a situation that generally does not exist for the expert hired in a court case due to the complexity of a forensic expert changing roles to that of a researcher to follow up on the outcomes of their cases vis-à-vis the opinions they issued. Dawes et al. (1989) provide the following description of the clinical method versus the statistical or actuarial method:

In the clinical method, the decision-maker combines or processes information in his or her head. In the actuarial or statistical method, the human judge is eliminated, and conclusions rest solely on empirically established relations between data and the conditions or events of interest (p. 1668).

This method is valuable to informing judgments in forensic practice if both methods, clinical and statistical, are combined. Statistical methods can include use of psychological or vocational tests or other data sources and/or statistical tools, whereas, clinical methods can include how to interpret and apply these pieces of data to the evaluaee based on their needs and context of the case.

Overall, the authors conclude that based on hundreds of studies, the evidence is overwhelming that statistical models beat clinical judgment in terms of validity and reliability. For example, AEgisdottir et al. (2006) performed a meta-analysis of the research on clinical judgment analyzing clinical versus statistical prediction by mental health professionals covering 67 studies where the effect size for statistical prediction outperformed clinical judgment by 13%. They concluded that “clinical and statistical prediction accuracy varied by type of prediction, the setting in which predictor data were gathered, the type of statistical formula used, and the amount of information available to the clinicians and the formulas” (p. 341). Statistical models generally provide more reliable predictions when properly validated (Collins, 2023; Wee et al., 2018). There are some settings, such as discrimination, where clinical models perform over statistical models that do better with prognostic outcomes (Colunga-Lozano et al., 2024). Specific to expert judgment versus statistical models, Caldeira (2004) suggests that the two models as complementary as the former helps explain complex events and provides context and understanding while the latter provides prediction based on analyzing data and finding patterns.

The late Daniel Kahneman, Nobel Prize Winner in Economics, wrote about similar issues in his New York Times bestseller, Thinking Fast and Slow (2011). Experts will often try to improve the accuracy of their models by increasing the number of variables that are considered. According to Kahneman,

Why are experts inferior to algorithms? One reason … is that experts try to be clever, think outside the box, and consider complex combinations of features in making predictions. Complexity may work in the odd case, but more often than not it reduces validity. Simple combinations of features are better. Several studies have shown that human decision makers are inferior to a prediction formula even when they are given the score suggested by the formula! They feel that they can overrule the formula because they have additional information about the case, but they are wrong more often than not. (p. 224)

However, Kahneman recognizes that there are conditions where intuitive judgments do seem to produce accurate results. What were the characteristics of these groups that were different than other groups in reference to intuitive decision-making? Kahneman (2011) summarizes his conclusions as follows:

If subjective confidence is not to be trusted, how can we evaluate the probable validity of an intuitive judgment? When do judgments reflect true expertise? When do they display an illusion of validity? The answer comes from the two basic conditions for acquiring a skill: an environment that is sufficiently regular to be predictable; an opportunity to learn these regularities through prolonged practice (p. 224).

There is no doubt that many experts see a “sufficient” number of clients to form opinions concerning their economic loss. The problem is that there is little or no feedback to learn whether an analysis is valid or not.

Ogden Tables

It is clear there are major issues involved in estimating special damages. What have some international legal systems done to deal with issues of expert witness lack of reliability and bias? Exploring methods used by other countries found that in some European courts, FEs were entirely removed from the process as a solution to the G2i problem, while VEs were rarely used. The Ogden Tables, for example, are extensively used in the United Kingdom. Other European countries use similar tables. The general goal of courts in those countries is the same as in the United States – to determine the amount of money that will fully compensate a plaintiff for their economic loss. However, there are significant differences in the way courts who use the Ogden Tables handle personal injury and death cases. There are no juries. Rather, the judge makes the final determination of the award, if any. Expert witnesses are not routinely used but may be called upon to render assistance, if needed.

In determining the award in a particular case, the judge is guided by a set of actuarial tables that have been, as indicated in the title page, “Prepared by an Inter-disciplinary Working Party of Actuaries, Lawyers, Insurers, Re-Insurers, Academics, forensic Accountants and other interested parties” (Government Actuary’s Department, 2021, p. 2). The Tables are currently in their eighth edition published in May 2021 (Government Actuary Department). The Tables consider a generic case and arrive at a multiplier that is then used with the base earnings in the case to determine an estimate of economic loss. The multiplier considers the standard economic variables of age, level of education, gender, the discount rate, and other similar factors. The Tables provide what may be viewed as a null hypothesis. It is a starting point which the judge, along with perhaps input from outside experts, may modify based upon specific circumstances.

There is an obvious relationship between the Ogden Tables and the G2i problem. Both are generalized sources that do not provide information at the individual assessment level. One may view the Tables as a general framework, which would then be the basis of most court cases. The problem of moving from G to i is supposedly minimized by reliance on the Tables. Since the Tables are developed by outside experts who are not involved in any particular case and unusual cases are governed by the precedent of similarly situated cases, use of the Tables ideally helps promote both fairness as well as efficient use of the court’s resources. The major question is whether use of the Tables accurately reflect economic damages in a particular case. There is some evidence that the Tables may result in reduced accuracy. One issue is that the Tables are revised infrequently.

Further, the Tables are extremely detailed, which can make it difficult for the court to determine the proper estimate of economic loss. The current edition is 108 pages long and contains extensive explanatory notes and detailed tables which can be used in specific circumstances. It also contains numerous examples of different types of cases. Developing tables for each possible contingency can complicate the entire process given that cases where the Ogden Tables are used do not employ FEs or VEs. Rather, interpretation of the Tables is left to the presiding judge who will likely not have the necessary expertise of the FE or VE who deal with a population that is highly heterogenous in every demographic and geographic variable, and physically, mentally, and cognitively to adequately fit into a single multiplier.

One of the biggest controversies that has arisen concerns the issue of disability. In earlier version of the Tables, disability was not even considered. In its current edition, the Tables view disability as a reduction factor that would normally be applied to post-injury earnings. The definition of disability is like that used in U.S. courts, which focuses on disability having a “substantial adverse effect on the person’s ability to carry out normal day-to-day activities” (Government Actuary Department, 2021, p. 26".)

The Tables recognize the difficulties of accurately considering disability. There have been several well-publicized court cases which highlight these difficulties (Latimer-Sayer & Wass, 2013; Wass, 2015, 2018). At the end of the section discussing disability, the Tables conclude, “All departures will be case-specific. In some cases, it may be difficult to determine the scale of departure, and it may be helpful to consult expert opinion. Expert opinion may also be required to advise upon how the suggested reduction factors should be applied and/or adjusted when the claimant was already disabled at the time of the injury which forms the subject of the claim” (Government Actuary’s Department, 2021, p. 33). The question becomes, then, how often are departures large enough that deviation from the Tables is warranted? If only occasionally, then the Tables are valuable assuming that doing so does not sacrifice accuracy. However, this appears not to be the case in reality. On the other hand, the Tables may serve little function except as a rough starting point if deviations are more common. The dilemma of validity and reliability of an earning capacity analysis continues with this approach.

Reproducible Methods for Assessment, Findings, and Opinions

By definition, validity refers to accuracy while reliability refers to consistency, reproducibility of findings, and conclusions. Earning capacity analyses are subject to scrutiny as to how valid and reliable the process is leading to conclusions and opinions – or, how precise and accurate is the analysis?

Validity and reliability become even more problematic for vocational analyses of earning capacity where evaluators deal with individual and group data constantly. The three types of validity that VEs should hold prominent in their analyses are internal, external, and construct validity when applying group data to the evaluee. Internal validity essentially is how the different components of an analysis hang together (Faigman et al., 2013). The greater the consistency between data points in a study, the greater the internal validity. The same is true of the VE’s evaluation: the greater level of triangulation among data points in the VE’s analysis, the greater the level of internal validity.

External validity is how findings are generalized (Faigman et al., 2013; Faigman & Fisher, 2015), or how the outcome of a study could apply to a population. When dealing with a single individual, external validity is not as apparent as it might be when studying a group and inferring the outcome to a bigger population. Some factors that could be examined to determine the external validity of an earning capacity analysis would be the evaluee’s job acquisition or retention, their consistency in earnings, how similar cases or settings correlate to them (e.g., exploring outcome studies for those with same or similar characteristics), functional capacity alignment, adaptability to market changes, or quality of life indicators. Some of these outcomes may not happen until after the VE and FE are no longer involved in the case.

Lastly, construct validity, ascertains that theoretical constructs are measured for the evaluee (Faigman & Fisher, 2015). Construct validity asks this question: When applying the results of large data sets to the evaluee, how close do the variables in the studied group match to those of the plaintiff? Simply put, it means that the VE considers factors in the individual analysis and uses data that best align with the factors of the population studies.

Although not inherently part of the VE’s evaluation, but a goal for its outcome, is predictive validity – a fourth type of validity. Predictive validity refers to the ability of a forensic assessment model to accurately forecast future outcomes or behaviors relevant to civil legal matters. Predictive validity is predicated upon using validated assessment tools or methods to evaluate the plaintiff, a collection of relevant data, the analysis of the data using established techniques and methodologies, and making prediction about future outcomes or behaviors of the evaluee. Predictive validity is inherently impossible for the VE to do within the forensic context – testing the accuracy (validity) of the predictions to actual outcomes – as the expert does not follow the case longitudinally given their evaluative role. The VE would need to switch to a researcher role, which involves new ethical guidelines and disclosure and informed consent procedures (CRCC, 2023) to follow their cases and report the predictive validity statistics. There is no inherent mechanism within the legal system itself to follow cases longitudinally after conclusion. In the absence of having an integrated long-term procedure to assess predictive validity, expert witnesses must instead rely upon other the preceding components of predictive validity by regularly updating and refining assessment methods and tools for new and changing data, ensuring the appropriateness of the selected methods for the case being evaluated, and considering ethical implications and potential biases in any model used.

Besides the three types of validity inherent to an assessment – as well as keeping in mind the goal of predictive validity – when considering reliability in the application of group data to an evaluee, Faigman and Fisher (2015) posit the importance of consistency in the data through the use of critical appraisal evidence-based literature (meta-analyses, systematic reviews, etc.). Outcome research specific to the disability population, or other demographic or disability groups to which the evaluee belongs, needs to align with the vocational, educational, functional, demographic, or other evaluee variables. This alignment provides the opportunity to refine diagnostic and prognostic conclusions in the earning capacity analysis. If different VEs with the same education, training, and experience arrive at similar conclusions (called inter-rater reliability), then the process has high reliability. Similarly, if a peer review of the VE’s methods using the same assumptions arrives at similar conclusions, the replication of the VE’s analysis (often called a peer or rebuttal review) is considered valid and reliable. Conversely, reliability suggests that different components used in assessing the evaluee correlate back to the group-level research; this is called the internal consistency or reliability of an assessment.

In science, individualized assessments can be complex for the evaluator and are not necessarily parsimonious – they do not necessarily achieve the simplest theory that fits the data equally well. However, parsimonious models – keeping it simple – can improve accuracy and increase efficiency in a vocational analysis. In legal cases, performing an individualized assessment is likewise required. In the courtroom, the evaluator is an educator who must be understood by a jury that can be diverse in levels of education, occupations, and more. Therefore, simplicity or parsimony is the key. What is fundamental to the VE’s evaluation is whether the analysis can be replicated and has a goodness of fit between the group data to the individual and the individual factors back to the group data. That is, the facts of the case and the conclusions reached need to align, fit, and make sense.

Clinical Judgment and Accounting for Cognitive Bias

Clinical judgment can be defined as an expert’s “… observations and inferences about uncertain client [or evaluee] characteristics or events that are systematic, deliberate, and explicit” (Austin & Leahy, 2015, p. 28). Clinical judgments has also been described to"… be predicated on valid, reliable, and relevant foundation information and data … [and] … a methodological approach which has been peer reviewed and generally accepted, coupled with the expert’s specialized knowledge, skill, experience, training and education" (A. J. Choppa et al., 2014, p. 262). Accurate clinical judgments are critically important throughout the individualized assessment process and are needed for VEs to a) make decisions about pertinent evaluee information to attend to and synthesize, b) make shifts in their behavior to establish VE-evaluee rapport, c) select appropriate assessment methods, d) recommend effective vocational rehabilitation interventions, and e) make valid predictions about an evaluee’s post-injury work life earning capacity. Therefore, how VEs observe, process, and decide what evaluee data to collect and attend to, as well as relevant models (or methods) they use, affects their clinical judgments, and ultimately, their expert opinions.

What’s missing that needs to be accounted for is expert cognitive bias which may further influence VEs’ clinical judgments. Cognitive bias is the “… unconscious and systematic errors in thinking that occur when people process and interpret information in their surroundings and influence their decisions and judgments.” (Silva et al., 2023, p. 1). This type of bias therefore can lead to VE judgment errors or inaccurate conclusions or opinions. Certified rehabilitation counselors (CRCs) who are VEs are also ethically required to recognize how such biases may influence their judgments and perhaps contribute to service inequities as indicated in the research (CRCC, 2023). Referral sources who contract with VEs may also apply pressure on VEs to conform favorably to a firm’s case, despite the VE’s ethical obligation to remain neutral (Bloom, 2015). Consequently, it is important to understand how VE cognitive biases may influence their clinical judgments, particularly if they go unchecked.

Although many articles discuss the problem of bias in a court setting, the authors could not find any research that focuses on VE bias in forensic rehabilitation and potential clinical judgment errors that may result from VE cognitive bias when serving in this expert role in hearings, depositions, or when performing vocational evaluations with evaluees. Processing information and making inferences or “a conclusion reached on the basis of evidence and reasoning” (Google’s English Dictionary, n.d.) while applying vocational evaluation models to inform VE opinions when completing an earning capacity evaluation is complex and takes a high level of cognitive ability and skill. A VE’s observations and reasoning skills and strategies to attend to and make sense of pertinent information throughout the evaluation process to make accurate clinical judgments is paramount (Austin et al., 2023; Falvey et al., 2005). In fact, VEs with higher order intellectual skills may be better able to effectively address their own biases, and make more accurate clinical judgments, particularly when working on more complex cases (Holloway & Wolleat, 1980; Spengler & Strohmer, 1994). Fortunately, there have been several studies in related settings that can help VEs better understand how they may make judgments about their evaluees, including how cognitive bias may affect their clinical judgments and how to overcome them (Austin, 2018; Austin et al., 2023; Austin & Leahy, 2015).

From an analogue study, Strohmer and Leierer (2000) found that a random sample of certified rehabilitation counselors primarily used a fully mediated model to make predictive judgments about clients’ likelihood to make rehabilitation progress and ability to maintain competitive employment. Specifically, findings suggest rehabilitation counselor predictions are based on inferences they make about a client’s current functioning and extent condition or “problem” (i.e., disability status, psychosocial adjustment, Intelligence Quotient) is under the client’s control. In this same study, the researchers indicate that such inferential judgments are likely not consistent across counselors. Other research has shown that counselors are susceptible to biases, and therefore possible judgment errors, based on a client’s disability, race, gender, social class, or age (Garner et al., 1994; Lopez, 1989; Rosenthal, 2004; Rybarczyk et al., 2001; Sprong et al., 2017). More experienced rehabilitation counselors have also been found to focus more on negative client information (Strohmer et al., 1995). In other research, counselors have demonstrated a tendency to attend to and gather client data that supports their original assumptions (Strohmer et al., 1990). If this were the case for a VE, this type of skewed attention may limit their ability to attend to positive, strength-oriented, or other relevant evaluee information, and therefore, result in an incomplete evaluee profile (Strohmer et al., 1990, 1995). Given this research presented and importance of clinical judgment in forensic settings (Austin et al., 2023; Beveridge et al., 2019; A. J. Choppa et al., 2014; Wright-McDougal & Toriello, 2013), VEs must be able to decipher evidence and research, as well as make sense of their clinical observations, but also take a stance that actively incorporates techniques intended to counter their cognitive biases to establish a working relationship and make accurate and objective clinical judgments about evaluees.

Conclusions: What is Missing?

Current frameworks or models of earnings capacity analysis regardless of content, process, hybrid, or jurisdiction application seem to be missing guidance to the evaluator or how to analyze the mixed qualitative and quantitative data at the individualized level and from the general to the individual. Further missing is how to move from vocational diagnosis to valid and reliable earnings capacity prognoses. Alternatives such as the Ogden Tables do not contribute to an easier or equitable solution and, in the circumstance of evaluating people with disabilities, may further complicate the trier of fact. Figure 2 provides a visual depiction of a data funnel for the vocational diagnosis and prognosis process:

Existing frameworks, models, and other such processes and procedures provide guidance for the top half of the hourglass figure to determine what kind of data or processes the VE needs to include in the inductive part of an earning capacity or other evaluation. However, these frameworks or models do not take the next step of helping VEs with the how or deductive part of the data analysis in the neck of the hourglass or the deductive bottom part of the figure. The problem is in the bottleneck where models and other such approaches do not provide VEs with sufficient direction for a VE to perform a mixed methods data analysis at the G2i and individualized levels to better ensure biases and errors are mitigated to make valid and reliable conclusions in the bottom half of the hourglass.

What the authors found in examining the present literature and models was the problem presented in this paper:

1. Number and Limitation of Models: There are a large number of loss of earning capacity models available to the VE that speak to the content or process of the analysis (Barros-Bailey, 2025). The number of models itself points towards the need to analyze these models and identify the domains and subdomains that are most generally accepted in the profession.

2. Qualitative/Quantitative Analysis: The VE uses qualitative and quantitative data to arrive at a reasonable estimate of post-injury earning capacity. The G2i process requires the use of both types of data. Attempts, such as the Ogden Tables, to rely solely on quantitative data leave the analysis short as they attempt to apply a cookie cutter to a process that requires an individualized assessment. The attempt to narrow the population data to the individual level in the application of such methods as the Ogden Tables has mainly proven unworkable given the frequent “exceptions” made to the values in such tables for individual case variables.

3. Empirical Support for Objective Data Analysis, Reduction, and Resulting Opinions: According to Robinson (2014), most of the existing models have little or no empirical support. Of those models that are empirically-based, these seem to focus on the steps involved in considering what type of data to gather (e.g., education, vocational, function), gathering the data (e.g., interviews, reviews, case-specific research), and some of the processes for reducing the data (e.g., transferable skills analyses, labor market analysis). What is missing after the data are gathered and processed is a focus on steps to take these data and arrive at conclusion about post-injury worklife and earnings capacity (or other questions that drive the analysis in the first place). I find myself wondering how these evaluation models w/limits we are describing, are informed based on career development theories, and/or other outcome research? Just a thought.

4. Adversarial Legal Context: The adversarial nature of the legal context in which the VE practices bears a burden upon the VE to not let the influence of the side hiring their services to bias the conclusions given all facts in the case but to adhere to the ethical responsibility and guidelines (CRCC, 2023) of providing objective opinions and maintaining competency with new and emerging information in the field.

A goal to solving the dilemma would be to develop a parsimonious model to carefully integrate and synthesize the different data sources to arrive at understandable and defensible opinions (Salmona et al., 2019) in the handoff to the FE.

Future Research

In identifying the shortcomings of present earning capacity models and approaches nationally and internationally, we take on the challenge of developing an adjunct model that complements present analytical models, but at the data analysis (neck of the hourglass figure) and conclusions/opinions deductive levels. In Part II of this series, we hope to provide the first effort to help VEs handoff to FEs findings and opinions that hold greater accuracy and consistency.