Abstract
A Canadian context for the diagnosis of students with specific learning disabilities (LD) was investigated in the present literature review. A systematic review of the literature was undertaken to determine the current and best practices in this field. Overall, no agreed upon definition of LD was identified, although core similarities in definitions were noted. Furthermore, recent research shows that many psychological assessments fail to adhere to any one definition when making this diagnosis, and as a result the diagnosis may or may not reflect the presence of a permanent disability that impairs academic functioning at the postsecondary level. There is, therefore, a need to adopt a consistent, evidence-based approach to diagnosis of LD in Canada. Recommendations regarding best practices and appropriate criteria for diagnosis of LD are discussed.
Introduction
From among the ever increasing number of students with disabilities entering postsecondary education in Canada (Blackorby & Wagner, 1996; Canadian Association of Disability Service Providers in Postsecondary Education, 1999; College Committee on Disability Issues, 2005; Harrison & Wolforth, 2007; Ministry of Training, Colleges and Universities, 2008), those with learning disabilities (LD) are the largest segment and receive from 1/3 to 1/2 of all academic accommodations provided (Kerka, 1998; Harrison & Wolforth, 2007; Hubka & Killean, 1999; Ministry of Training, Colleges and Universities, 2008; Vogel, 1998). Learning disabilities are characterized as impairments in neurological processes that interfere with normal learning (Ghelani, Sidhu, Jain, & Tannock, 2004; Kibby, Marks, Morgan, & Long, 2004; Learning Disabilities Association of Ontario [LDAO], 2003). Too often, assessment documents received by Disability Services Offices (DSOs) at Canadian postsecondary institutions have used a wide variety of diagnostic criteria and are of varying quality. Hence, proof of the existence of a LD is not only difficult to verify but it is also not easy to determine whether the student is experiencing an inequity of opportunity that would require provision of academic accommodations. There is therefore a pressing need for a consistent, evidence-based approach to the diagnosis of LD. From this will flow accommodation requests that are supported by empirical evidence.
Need for an Agreed Upon Definition of LD
A definition delineates the nature and limits of a disorder (Kavale, Forness, & Lorsbach, 1991; Siegel & Smythe, 2005). In order to provide appropriate services and supports to persons with a particular diagnosis, there must first exist an agreed upon definition for such a diagnosis. Most physical disabilities (e.g., blindness, deafness, paraplegia, etc.) have specified, agreed upon criteria for diagnosis. Despite almost half a century of research, however, there is no consistency in assessing and diagnosing LD. Those individuals being labelled as LD are quite a heterogenous mix (Harrison, Nichols, & Larochette, 2008; Sparks & Lovett, 2009a, 2009b). This leads to confusing research regarding the causes and best treatments for this disorder (Flanagan et al., 2006a; Kavale & Forness, 2000). Recent research further suggests that many clinicians fail to consistently follow any one diagnostic model when making the diagnosis (Harrison et al., 2008; Rosenblum, Larochette, Harrison, & Armstrong, 2009), or do not employ any accepted definition of LD when making this diagnosis (Rosenblum et al., 2009; Sparks & Lovett, 2009b). The result of this inconsistency is confusion regarding who is or is not actually LD, and more importantly, whether a student truly demonstrates an impairment in functioning that causes an inequality of participation at the postsecondary level. There is no duty to accommodate if the condition does not compromise the student’s equal opportunity to participate in education (i.e., there is no functional limitation). Hence, documentation must also demonstrate that specific impairments are causing a functional impact in the educational or work environments that creates an unequal opportunity for participation relative to most other students.
Options for Diagnosis
Currently, there exist a number of published options for diagnosis of LD in young adults. For example, the Diagnostic and Statistical Manual of Mental Disorders (4th ed., text revision [DSM-IV-TR]; American Psychiatric Association, 2000) offers a consensus-based set of criteria for various learning disorders (i.e., reading disorder; disorder of written expression; math disorder; learning disorder not otherwise specified). Researchers in the United States and Canada have also championed their own definitions of the concept of LD (e.g., Fletcher, Coulter, Reschly, & Vaughn, 2004; Flanagan, Ortiz, & Alfonso, 2006a; Mather & Gregg, 2006; Dombrowksi, Kamphaus, & Reynolds, 2004; Kavale & Forness, 2000; Lyon, Shaywitz, & Shaywitz, 2003; Siegel, 2003). To further complicate things, provincial and federal advocacy groups promote their own definitions (developed through consensus), as do individual ministries of education within each province (Kozey & Siegel, 2008). It is therefore not surprising that considerable variability exists in the methods used across Canada for diagnosing LD. Two common methods for diagnosing LD are IQ-achievement discrepancy and a diagnosis based on academic impairment.
The IQ-Achievement Discrepancy Model
An exceedingly popular but now largely discredited approach to diagnosing LD is the discrepancy model wherein the presence of a marked disparity between an individual’s IQ and performance on academic tasks is taken to signify the presence of LD (Willis & Dumont, 2006). This method of diagnosis has persisted, despite widespread criticism based on psychometric flaws in the concept (Fletcher, Francis, Rourke, Shaywitz, & Shaywitz, 1992; Fletcher, Morris, & Lyon, 2003; Meyer, 2000; Restori, Katz, & Lee, 2009; Siegel, 1988, 1992, 2003; Spear-Swerling & Sternberg, 1998; Stage, Abbott, Jenkins, & Berninger, 2003; Stanovich, 1991; Sternberg & Grigorenko, 2002; Vellutino, Scanlon, & Lyon, 2000) and demonstrated misuse of discrepancy formulas within various states in the United States (Fuchs, Mock, Morgan, & Young, 2003; Lester & Kelman, 1997; MacMillan, Gresham, & Bocian, 1998; Peterson & Shinn, 2002; see also Cotton, Crewther, & Crewther, 2005; for an extended discussion regarding measurement errors that occur when using the discrepancy definition).
The IQ-achievement discrepancy model has several limitations. First, while intelligence is a relatively good predictor of academic achievement (effect sizes in the r = .3 to .5 range; Sternberg, Grigorenko, & Bundy, 2001), there is a great deal of variance (75% to 90%) unaccounted for by IQ (Lyon, 1995; Siegel, 1988, 2003; Spear-Swerling & Sternberg, 1998; Stanovich, Cunningham, & Freeman, 1984; Sternberg et al., 2001). Simply performing below one’s measured IQ is not a strong predictor of impairment (Lovett, Gordon, & Lewandowski, 2009; Sparks & Lovett, 2009b). As a result, many researchers such as Stanovich (1991, 1993, 1999, 2005) and Siegel (1988, 1992, 2003) have advocated for a definition of LD wherein intelligence has no role. By contrast, others (e.g., Berninger, Abbott, Thomson, & Raskind, 2001; Kavale & Flanagan, 2007; Reynolds & Shaywitz, 2009; Swanson, 2008; as cited in Reynolds & Shaywitz, 2009) have advocated for the inclusion of some measures of intellectual ability in the conceptualization of and treatment for LD. Most experts in the field of LD (Hale et al., 2010) agree that simply performing below one’s measured IQ is not a sufficient basis for diagnosis of this disorder.
Second, even though low academic achievement is integral to the concept of LD (Speece, Case, & Molloy, 2003), the discrepancy model fails to clearly define what it means to be achieving below average. High-IQ students identified by this method often achieve at or above grade level (Fletcher, Denton, & Francis, 2005). By contrast, low-achieving children commonly show achievement greater than their measured IQ and thus don’t produce an IQ-achievement discrepancy and as such are not identified, even though it is apparent that they are struggling academically. Furthermore, there is no evidence to support the concept that students with IQ-achievement discrepancies learn differently or benefit differently from remedial programs compared with students who are simply achieving below average (Fletcher et al., 2002; Francis, Shaywitz, Stuebing, Shaywitz, & Fletcher, 1996; Hoskyn & Swanson, 2000; Siegel, 1992; Stage et al., 2003; Vellutino et al., 2000). This finding has held even in studies that used a regression-based discrepancy formula for identification (e.g., Fletcher et al, 2003; Stanovich, 2005).
A third and significant limitation of the discrepancy model results from the statistical phenomenon of regression to the mean. Generally, achievement scores will fall somewhere between a person’s IQ and the mean. Those with higher than average IQs will typically produce scores that are somewhat closer to average (but below their IQ) on tests of achievement, whereas those with lower than average IQs tend to achieve at levels higher than their IQ (Cone & Wilson, 1981; Evans, 1992). This means that discrepancy formulae consistently overidentify those with higher IQs and underidentify those with lower IQs (Fletcher, Lyon, Fuchs, & Barnes, 2007). This could be interpreted as discriminatory against those with more modest IQ scores and not a fair or equitable way to identify the presence of a true disability (see Fletcher et al., 2007; Spear-Swerling & Sternberg, 1998 for a review of these issues). In addition, Kavale and Flanagan (2007) discuss the mistaken assumption adopted by many clinicians that a student with superior ability in one area should have superior ability in all areas. They point to research identifying just how common is significant variability in an individual’s cognitive ability profile and conclude that simply having a few average scores in an otherwise above average profile is not sufficient to demonstrate the presence of a disability.
Overall, there is much objective evidence to question the validity of discrepancy formulae as a means of diagnosing LD. Indeed, after reviewing all of the research on the validity of discrepancy formulae in the diagnosis of LD, Ysseldyke (2005) concluded, “there is little empirical support for test-based discrepancy models in identification of students as LD” (p. 125).
The Academic Impairment Model
The majority of current experts in the field of LD have championed diagnosis of LD based on academic impairment relative to most other individuals. Whereas discrepancy definitions do not require academic impairment in the absolute sense, (as a student may have only average achievement in the presence of above average IQ), most contemporary scholars in this field currently emphasize that academic impairment in the absolute sense is essential to the definition of LD. For instance, Dombrowski, Kamphaus, and Reynolds (2004) proposed a definition that would require academic impairment as the main criterion for diagnosis of LD. In their model, students who achieve below average both in the classroom and on standardized tests of achievement (one standard deviation (SD) or more below the mean), who evidenced such problems well before age 18 and for whom other causes of learning failure had been ruled out would be diagnosed as LD. Numerous other experts in the field (e.g., Fletcher et al., 2007; Lyon et al., 2003; Siegel, 1988, 1999, 2003; Stanovich, 1991, 2005) have all put forward definitions of LD that emphasize the core deficit of impaired academic achievement. The Learning Disabilities Association of Ontario (LDAO) definition makes it clear that the first necessary condition for diagnosis of a LD of any sort is that there be evidence of unexpectedly low academic achievement (or in-class achievement attained only with much support).
The Response to Intervention (RTI) model is based firmly on the notion of LD as an academic impairment relative to most other students (see Fletcher et al., 2005, for an extended discussion of RTI). In essence, this approach does not employ standardized tests or measures of aptitude for the establishment of a diagnosis of LD. Typically, RTI programs consist of continuously monitoring students’ progress in basic academic skills through brief curriculum-based probes (Madaus & Shaw, 2006). Students not progressing at an acceptable rate in the early grades are then exposed to additional, intensive, and varied means of scientifically validated instruction in basic academic skills for a set period of time, with further evaluation and intensive training being provided for those who fail to respond. If a student drawing toward the end of Grade 2 has not progressed despite these various interventions, he or she would then be eligible for identification as LD. In essence, RTI is being purported as an alternative means to diagnosing LD—those children who respond to intervention do not have LD while those who do not respond are likely LD and eligible for special education services and supports. There have been numerous criticisms of this approach to diagnosis (e.g., Hale et al., 2010). Some have noted that it is difficult to define what counts as adequate progress in academic skill development, and measures of progress adequacy often disagree with each other (see Barth et al., 2008).
No longitudinal studies have been conducted to confirm that RTI accurately identifies all children at risk for long-term academic failure. Nevertheless, the focus in the RTI approach on consistently below average academic achievement despite appropriate teaching and intervention underscores the central importance of this concept in the diagnosis of LD.
The Three Component Combined Model
Many experts in the field (see Hale et al., 2010) currently endorse a definition of LD that includes three components: below average academic achievement relative to most other individuals; impairments in the cognitive processes responsible for normal development of the deficient academic abilities; and ruling out other reasonable causes for the academic deficits, including academic difficulties due to generally lower abilities required for reasoning and learning. In the United States, the main proponents of this type of model have been Dawn Flanagan and colleagues (e.g., Flanagan et al., 2006a, 2006b) and Kenneth Kavale and colleagues (e.g., Kavale & Forness, 2000).
This approach requires not only the presence of academic underachievement for diagnosis of LD but also evidence of deficits in specific processing functions shown to be empirically linked to adequate or inadequate development of a particular academic skill (Flanagan, Fiorello, & Ortiz, 2010). Proponents point to empirical research linking specific processing deficits with specific types of learning disabilities, such as research demonstrating the causal relationship between phonological processing deficits and the development of reading disabilities (c.f. Vellutino, Fletcher, Snowling, & Scanlon, 2004). Thus, diagnosis of a reading disability requires identification, for instance, of both impaired reading ability and impairment in some aspect of phonological coding skills. Similarly, specific neurological impairments caused by white matter disease or dysfunction contribute to the development of a nonverbal LD (Rourke, 1995), and methods to measure such impairments have been described elsewhere in this issue (Casey, 2012).
In the combined approach, impairments in the cognitive processes essential for academic achievement are necessary for diagnosis, but on their own they do not constitute a LD. The reason is that many nondisabled individuals produce a few subtest scores on commonly administered psychological tests that fall within the impaired range (e.g., Binder, Iverson, & Brooks, 2009; Brooks & Iverson, 2010; Brooks, Iverson, Sherman, & Holdnack, 2009; Heaton, Grant, & Matthews, 1991). For instance, Binder et al. (2009) recently demonstrated that the majority of the nondisabled WAIS-III and WAIS-IV normative sample participants had at least one subtest score that fell within the impaired range, and over half had two subtests scores that fell within this range. As level of intelligence increased, so too did the frequency of large discrepancies between various IQ indices, attesting to the nondiagnostic nature of such intra-IQ discrepancies in those with above average IQ’s. Note that this is consistent with what would be expected due to regression to the mean. Furthermore, as the number of tests in a flexible battery increases, so too does the likelihood of obtaining more scores that fall below average (e.g., Binder et al., 2009; Brooks & Iverson, 2010; Brooks, Iverson, et al., 2009; Evans, 1992; Heaton et al., 1991; Russell, Russell, & Hill, 2005). Hence, having a few subtest scores on any flexible test battery that fall within the below average or impaired range is a normal, expected finding and not, in and of itself, diagnostic of any type of disability. Indeed, an impairment in a processing ability only becomes disabling when it interferes substantially with an individual’s ability to carry out a regular or routine task that relies on the use of skills or knowledge in that area (Barnartt & Scotch, 2001; Brant & Pope, 1997; U.S. Equal Employment Opportunity Commission, 2009, footnote 6).
Recommended Models for LD Assessment and Diagnosis in Canada
In Canada, the working definitions of LD proposed by the Learning Disabilities Association of Canada (LDAC, 2002) and by Harrison (2005) in reference to the LDAO definition are based on the three component combined model. The LDAC definition essentially defines LD as lifelong, central to the individual and possibly resulting from genetic factors, illness or injury during or before birth, and incidents after birth, nutritional deprivation and exposure to toxic substances (LDAC, 2002), and has been accepted by several of Canada’s provinces (Kozey & Siegel, 2008). More specifically, the definition references the following elements as key features of the disability:
may affect verbal or nonverbal information processing,
at least average abilities essential for thinking and/or reasoning,
distinct from global intellectual deficiency,
impairments in various psychological processes,
unexpected academic underachievement or school achievement obtained through very high levels of effort,
may coexist with other conditions, and
not caused by extrinsic factors.
These features as stated lend form to a feasible assessment process whereby standardized tests thought to capture the listed traits may be selected and administered to the individual in question. Each criterion in this definition is necessary, but not sufficient on its own, to make the diagnosis. Explicit guidelines around recommended tests, discrepancy sizes, measurement of effort, and extrinsic factors, however, are not provided or even suggested by LDAC.
Harrison (2005) and the LDAO (LDAO, 2001) undertook a more specific operationalization of the definition of LD (see Harrison, 2005 for an extended discussion of this definition). These provide details of skills to measure, tests that may be used, what to cover during an interview, and what to include in a report. They then offer a clear set of diagnostic criteria that must be met in order for an individual to be diagnosed with LD. In this definition, impaired academic functioning is also a necessary but not sufficient criterion for diagnosis. In addition, one must also demonstrate that failure in an academic area is the result of an underlying impairment in the cognitive processes necessary for adequate learning or production of that academic skill and not due to other cause (Harrison, 2005).
The LDAO definition makes the first necessary condition for diagnosis be evidence of unexpectedly low academic achievement (or in-class achievement attained only with much support). The next necessary condition is that the academic achievement deficit is associated with impairment in one or more of the underlying processes necessary for learning or production of that skill. This is necessary because, especially in adolescents or adults, there can be many non-LD reasons for an individual performing poorly on tests of academic achievement. Hence, having evidence of “unexpectedly low” academic achievement is essential, and one must further demonstrate that the academic achievement deficits are not remediable (i.e., they are caused by a processing impairment in a skill that is integral to that academic function). Note that this means that there is no such thing as a learning disability in processing speed or working memory, per se, as such processing weaknesses must be paired with functional impairment in an academic skill. An analogy here would be that to diagnose a bacterial infection it is not sufficient to simply identify the presence of bacteria on the skin of an individual. One must also confirm that there is active destruction of tissue (impairment) and that the bacteria in question are known to produce this kind of infection (causal relationship). In isolation, the presence of bacteria on the skin is neither unusual nor diagnostic of any disease process.
Overall, these Canadian definitions echo the themes endorsed by the three-component model and would therefore meet the requirements set out by experts in LD.
Problems With Current Practice
Failure to Adhere to Evidence-Based Diagnostic Practices
Not only is there variability in the choice of definitions or diagnostic criteria employed, but recent research suggests that the majority of clinicians eschew using any published or agreed upon criteria when diagnosing LD, despite the choice of definitions and diagnostic criteria in play. Indeed, Sparks & Lovett (2009b) examined disability documentation provided by students attending postsecondary institutions in the United States to see if their documentation actually demonstrated an impairment sufficient to be considered disabling. They used five different methods of classifying individuals as LD: Three were variations on the discrepancy model requiring an IQ-achievement discrepancy of 2 standard deviations (SD), 1.5 SDs or 1 SD as the only criterion for diagnosis. The fourth classification scheme was based on impaired academic achievement relative to most people demonstrated by scoring below average in at least one academic area. The fifth classification method required that the student’s school-based impairment (described in the fourth classification scheme) had also been lifelong. These authors found that only 54% of the assessment reports provided by these students met the most lenient of all the definitions used (1 SD discrepancy). Fewer than 10% showed evidence of academic impairment relative to the average person. In addition, less than 7% had evidence of lifelong impairment in academic functioning. These authors concluded that there need to be more objective, agreed upon ways to determine who is actually LD.
Similar concerns have been voiced by others with respect to lack of adherence to the DSM-IV diagnostic criteria by practicing clinicians. Ferrari (2009) reported on the outcome of updated assessments for 176 vocational rehabilitation clients with a previous diagnosis of LD. In total, only 22% met “the general spirit” of the DSM criteria for either a reading or math disorder (defined as at least a 1.5 SD difference between one measure of intelligence and one measure of achievement), and only 57.9% showed a discrepancy of at least 1 SD. Furthermore, almost half of those previously diagnosed as LD actually had overall intelligence scores below 80. Ferrari concluded that a previous diagnosis of LD is not sufficient to ensure that the individual would currently meet the diagnostic criteria for such a disorder and appealed to clinicians to adhere to accepted criteria when first making this diagnosis.
Purpose of Accommodations
Referrals for psychoeducational assessments in a school setting are often for the purpose of determining the student’s eligibility for program accommodations and access to assistive technology. Problems in making an accurate diagnosis of LD can arise when the psychological service providers feel that their role when conducting a psychoeducational assessment is to act as advocates for students. In a North American survey of psychological service providers who conducted assessments for those applying to take a high-stakes entrance examination, Gordon, Lewandowski, Murphy, and Dempsey (2002) found that nearly one third of these clinicians believed that the purpose of conducting an assessment was to make a case for accommodations, rather than to gather information sufficient for an objective diagnosis. These authors conclude that assessors may be more concerned with justifying an identification rather than being impartial reviewers of data. Their findings also raise concern regarding the objectivity and validity of diagnoses being made by professionals (Gordon et al., 2002).
Lovett, Gordon, and Lewandowski (2009) remind clinicians to remember the criterion of significant impairment in functioning when making a diagnosis of LD. Human rights legislation, both in the United States and in Canada, is not meant to guarantee academic success, as it is outcome neutral. It only requires that one be allowed equal participation—it guarantees access not success. Many students and their advocates find this hard to understand. A good reason for the confusion is that legislation governing the elementary and secondary schools differs from that governing postsecondary education. The goal of accommodations at the elementary and high school level is to ensure student success. In contrast, at the postsecondary level disability legislation mandates equal opportunity to participate (National Joint Committee on Learning Disabilities [NJCLD], 2007), but does not require success as the outcome of accommodation.
Why Different Methods of Diagnosing LD Cause Problems at the Postsecondary Level
Several studies in Canada and the United States have found that the disability documentation provided at the postsecondary level is deficient, with many students providing inadequate documentation or failing to provide evidence of a past diagnosis at all. For instance, Harrison et al. (2008) reviewed disability documentation provided at three different Canadian postsecondary institutions by 247 students requesting accommodations for a LD. In total, only half of these students provided a psychological report, with the remainder providing either no documentation or school-based identification protocols alone. Of the 133 students who submitted a psychological assessment, 44% included a clear diagnostic statement (though not always a diagnosis of LD). Twenty-seven percent of the LD diagnoses made were based solely on an IQ test score, and 45% were made based on an IQ-achievement discrepancy. Similar concerns about documentation were reported by Gregg and Hoy (1990), who noted that only 35 of 110 students requesting accommodations for LD provided assessment documentation, and of the documentation received less than half contained an actual LD diagnosis. Problems with documentation quality were also reported by McGuire, Madaus, Litt, and Ramirez (1996), who noted that the majority of assessment reports diagnosed LD based simply on IQ test scores. These concerns mirror the sentiments offered by Siegel (1999) where she noted the extent to which provided documentation at Boston University failed to support the existence of a LD. Disagreement over ways to define LD and use of inappropriate tests thus appear to be the main issues that prevent postsecondary DSOs from determining if a disability is present.
The comprehensiveness of the initial assessment on which the LD diagnosis was made can also influence whether or not the problems identified are permanent in nature. In a Canadian-based study, Rosenblum et al. (2009) compared the previous and current assessment/identification of 100 Grade 7 children all of whom had previously been diagnosed as having a LD. These authors examined the consistency of the findings from the first to second assessment and also whether the children continued to meet diagnostic criteria for a LD.
Employing a definition based on the LDAO criteria, these authors found that 25% of the students previously diagnosed as LD no longer demonstrated symptoms or test scores associated with such a diagnosis. Furthermore, the best predictor of whether or not a diagnosis would change was the degree to which the initial assessment had been comprehensive and had ruled out other causes for reported learning problems. In other words, children whose initial identification of LD was based on less comprehensive assessment criteria were less likely to continue to qualify for a diagnosis of LD when seen in Grade 7. These results suggest that a previous diagnosis is not always an indication of a permanent disability, and the cognitive and academic profile of a student can genuinely change over time. These data point to the need for students to obtain an updated, comprehensive, diagnostic assessment to demonstrate an ongoing (i.e., permanent) impairment in academic functioning when preparing for postsecondary life as the initial diagnosis may not have been accurate.
Simply having the diagnosis of LD in a psychoeducational assessment report does not guarantee that the student meets real criteria for a permanent disability (Ferrari, 2009; Siegel, 1999; Sparks & Lovett, 2009b). A diagnostic statement alone does not ensure that the student suffers from a permanent disability that would cause an inequity of opportunity relative to most other people. Indeed, evidence suggests that many assessment reports utilize inconsistent criteria for diagnosis, often employing IQ subtest scatter alone or IQ-achievement discrepancy as the sole measure of LD (Harrison et al., 2008; McGuire et al., 1996). Furthermore, the less comprehensive the initial assessment, the more likely that any previously identified learning problems may have disappeared when reassessed at a later date. This finding has important implications for how one determines who at the postsecondary level has a LD, and for the recency of documentation on which accommodation decisions are made.
Finding Common Ground
There is no consensus in Canada, or indeed anywhere in North America, on the best way to diagnose LD or the specific criteria to use when making this diagnosis. The lack of consistency in definition used across provinces and even between school boards within the same province can generate significant difficulties for students moving from secondary to postsecondary educational settings. Indeed, the NJCLD (2007) identified concerns about this documentation disconnect between high school and postsecondary settings, outlining the legal and philosophical differences that exist regarding accommodation and required documentation at these two levels of education. They identified the lack of consistency that currently exists in the documentation provided by students making the transition to postsecondary settings, as well as differing laws that govern provision of services and supports to those with disabilities at each level of education. In addition, the NJCLD (2010) presented another review in which they acknowledge that many assessments provided by students to support their LD are inappropriate in nature and that such inconsistency interferes with determination of actual prevalence rates and consistent research into the nature of LDs.
We need to draw research and clinical practices closer together, perhaps through increased and united attempts at operationalization. An additional finding is that the majority of researchers and clinicians roundly discourage reliance on a sole criterion for measuring LD (e.g., Gregg, Coleman, Davis, Lindstrom, & Hartwig, 2006; NJCLD, 2010). LD continues to be viewed as a multifaceted concept and its definition and process for diagnosis must reflect this by utilizing a blend of appropriate measurement techniques, examination of learning history and consideration of factors reasonably likely to contribute to academic underachievement. A definition that includes the key components drawn from the convergence of modern conceptualizations of LD is therefore recommended for use in assessment of LD at the postsecondary level.
The key components of the definitions of LD described in the current literature show that many commonalities do exist (see Table 1). This table adapts and expands upon the summary published by Flanagan et al. (2006a) after their empirical review of current American definitions of LD. As shown in Table 1, the columns on the left reflect the greatest consensus as to the elements thought to be central to the definition of LD. They are as follows: (a) a history of academic difficulties; (b) academic deficits (i.e., functioning below average, typically the 16th percentile or lower); (c) impaired underlying processing abilities as contributing to the noted academic deficits; (d) exclusion factors; and (e) functional impairment (i.e., interfering substantially with acquisition or use of a skill).
Current Conceptualizations of Learning Disabilities Definition
Note: RTI = response to intervention; ? = unclear; LDAC = Learning Disabilities Association of Canada; LDAO = Learning Disabilities Association of Ontario. DSM-IV-TR = Diagnostic and Statistical Manual of Mental Disorders (4th ed., text revision).
Areas not strongly related to academic deficit.
Discrepancy between ability and achievement.
We suggest that a consensus definition of LD include all of these components, and could be operationalized across Canada. An operationalized definition of LD is essential to allow students with LD to receive accommodations throughout their educational years. Currently, documentation for these students is marked by its lack of consistency across the years. Our review suggests that there are a number of agreed upon components of a definition of LD that should be included in an operational definition of LD. As such, we recommend that a diagnosis of LD, given by a qualified professional 1 should include the following components (Table 2) in order for accommodations and supports to be provided at the postsecondary level. Each of these components is necessary, but not sufficient by itself, for the diagnosis of LD.
Recommended Operational Criteria for Diagnosis of LD in Canada
Note: LD = learning disabilities.
A diagnosis should include objective evidence of a consistent history of impairment, supported by behavioural observation, collateral report, school records, and/or actual test scores that all demonstrate a pattern consistent with diagnosis of a lifelong, neurologically based disability (Dombrowski et al., 2004; Gregg et al., 2006).
To diagnose LD one must do more than simply identify a few unrelated scores that fall below average, or scores that, while average, are below a student’s measured IQ. There must be a demonstration of academic underachievement (one standard deviation or more below average) relative to the average student so as to identify a potential inequity of opportunity to participate. There is the possibility that in-class achievement is no longer impaired due to high levels of support or academic accommodations; however, when such high levels of support or accommodation are removed (such as during a standardized assessment) the impairments in academic or life functioning should reappear. Some research has demonstrated that underlying deficits in phonological processing skills may improve to normal levels in individuals who were identified at an early age and received intensive remediation (see Shaywitz, 2003) but such well remediated individuals may continue to lack fluency and automaticity in reading (Lefly & Pennington, 1991; Shaywitz, 2003). As such, they may still require some academic accommodations at the postsecondary level; however, slow processing or reading speed alone without a history of remediated phonological skills and in the absence of any other academic impairment, is not sufficient for a diagnosis of a LD in a young adult.
Often students with reading disabilities (one type of LD) have difficulty with fluent word recognition and poor spelling and decoding abilities (Berninger et al., 2006; Lyon et al., 2003); thus, assessment of reading and decoding skills is essential to a diagnosis. Siegel (1999) suggests that standardized tests be used to evaluate reading comparing the student to other children of the same age. A score below the 16th percentile is low enough to demonstrate the academic underachievement component of an LD (e.g., Berninger, 2006; Flanagan & Kaufman, 2004; Flanagan, Keiser, Bernier, & Ortiz, 2003; Rourke & Finlayson, 1978; Swanson & Saez, 2005; Vellutino et al., 2000).
Evidence that academic impairments are related to meaningful and empirically supported deficits in specific psychological processes (see Flanagan et al., 2010 for specific examples of such empirically supported academic-processing deficit associations). This would require administration of tests measuring specific processing skills associated with learning such as working memory, phonological awareness and decoding, processing speed, rapid naming, attention, and executive functions. Note that a low score on one such processing test in the absence of a functional impact on academic performance is not sufficient to support the diagnosis of LD.
Evidence that the identified deficit causes functional impairments. In order to be considered a disability that legally requires the provision of accommodations and supports at the postsecondary level, the disturbance in academic functioning (e.g., reading, math, writing, etc.) must significantly interfere with academic achievement or activities of daily living that require reading, writing or math skills. This reflects the purpose of human rights legislation, which is to ensure equal access and equal opportunity to participate for all individuals regardless of disability.
Exclusion clause: The majority of LD definitions include a clause stipulating that other potential causes for the observed academic difficulties be explored and ruled out prior to making the diagnosis of LD. Over and above the errors in diagnosis caused by improper interpretation of one or two abnormal test scores in a flexible battery, Fletcher and his colleagues (Fletcher et al., 2007) allege that much of the increase in diagnosis of LD is actually due to false positive diagnostic errors when clinicians fail to rule out alternate causes for poor achievement. This is why the concept of RTI has been so appealing, in that it has helped to ensure that most common causes of reversible underachievement can be addressed early in a child’s school career. Especially when undertaking an initial assessment at the adolescent or adult level, it is important for clinicians to investigate causes other than LD that may lead to cognitive and information processing difficulties that, in turn, can affect academic functioning (see Harrison, 2005 for an extended discussion of many potential causes for academic underachievement).
Apart from the typical factors such as lack of appropriate education, English as a second language, and psychological/emotional causes for depressed academic achievement (although individuals so labeled could also have a LD), one must also rule out other noncognitive factors such as lack of motivation (Adelman, Lauber, Nelson, & Smith, 1989; Marinak & Gambrell, 2008) lack of engagement (Guthrie, Wigfield, & Von Secker, 2000) and symptom exaggeration (Alfono & Boone, 2007; Sullivan, May, & Galbally, 2007), as these factors have been shown to interfere with accurate assessment of learning skills.
Updating Diagnostic Data
Ideally, individuals with LD would not have to continue to obtain costly updated assessments to verify the presence of a permanent disability and their need for accommodation. If early screening and empirically supported intervention were offered to all young children at risk for academic failure, and if initial diagnoses of LD were made based on universally agreed upon criteria, no questions would arise at the postsecondary level regarding the legitimacy of this diagnosis. Assessment would occur at transition points only (e.g., from high school to postsecondary) to document the current level of functioning and evaluate areas where equal participation may be impaired even with adaptive technology and learned compensatory strategies (Harrison, 2005). Given, however, that no agreed upon definition or diagnostic criteria are currently used when making this diagnosis, that some clinicians continue to make this diagnosis based on IQ-achievement discrepancies or without employing any agreed upon criteria, and that students with this label may not demonstrate current impairments in functioning, we recommend that the documentation demonstrating a LD be current. Until clinicians can employ agreed upon criteria for making this diagnosis, it seems prudent to recommend that the documentation on which postsecondary accommodation requests are being made be no more than 3 years old (unless a comprehensive assessment and diagnosis was rendered after the person was 18 years of age or older). Human Resources and Skills Development Canada (2011) provides a grant to students in financial need to assist them in paying for such updated assessments, and in some provinces (e.g., Ontario), provincially funded assessment centres exist that can provide such assessments based on a sliding fee scale.
Conclusion
This evaluation and synthesis of the current literature has identified problems in the application of published criteria when making the diagnosis of LD. This has lead to confusion not only regarding who suffers from this condition but also by interfering with research into this disorder. Although discrepancy definitions were once a popular way to make this diagnosis, they have not demonstrated empirical validity as a method of identifying individuals who are struggling academically. Diagnostic methods based on academic underachievement and associated impairments in real-world functioning, along with a consistent history of impairment in the absence of other reasonable explanations, is offered as an operational definition of LD for use across ages and across provinces. We believe that this method of diagnosis will improve research into the best methods for treating or accommodating individuals with LD and will improve the portability of documentation as students progress through the educational system.
Footnotes
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Ministry of Training, Colleges and Universities of Ontario.