Psychometric perspectives on diagnostic systems
D. Borsboom (2008)
Journal of clinical psychology

Abstract

Four conceptualizations of the relation between symptoms and disorders as utilized in diagnostic systems are: 1) A constructivist perspective, disorders are conveniently grouped sets of symptoms. 2) A diagnostic perspective, disorders are latent classes underlying the symptoms. 3) A dimensional perspective, symptoms measure latent continua. 4) A causal systems perspective, disorders are causal networks consisting of symptoms and direct causal relations between them

Introduction

The movement standardization has not been paralleled by theoretical advances in understanding the conceptual and psychometric underpinnings of diagnostic systems in general.

The central question in this article is: What is it that a researcher, who uses the DSM classification, really does?

The constructivist view

The researchers that uses the DSM for classification constructs classes of people based on a convenient grouping of symptoms into syndromes. The classification system of the DSM is seen as relatively arbitrary, which renders the resulting classes of people socially constructed kinds rather than naturally existing ones. The concept of a disorder is a socially constructed kind in the sense that it is implicitly defined by a convenient grouping of key attributes. The concept that describes the group does not identify a homogenous group of people. The label is merely useful to delineate a group of people who share some key attributes,  but does not ‘cut nature at its joints’.

Constructivist conceptualizations does not imply that the whole process of diagnosis and the results of scientific research on mental does orders, are also arbitrary. For instance, the symptoms of depressing hang together reliably, in the sense that they are moderately positively correlated, so the syndromes constructed out of them have a sense of reliability as well. The higher the intercorrelations between a set of measures, the higher internal consistency will be. People may respond to treatment with a reliable change of symptoms while they suffer from very different conditions.

Constructivist deny that a group of symptoms is anything more than just that, a group of symptoms. A constructivist accepts that a set of symptoms may have high internal consistency, but denies that they all measure the same latent variable (unidimensionality). Internal consistency is nothing more than a summary statistic of the intercorrelations between a set of variables, and these correlations may come from everywhere and nowhere. Any set of positively correlated variables will show high internal consistency if run through the relevant statistical analysis, even though they do not measure the same latent variable.

The inference that a set of indicators are affected by the same latent variable requires an abductive step. The acceptance of the latent variable hypothesis is not mandated by the data alone, but requires an appeal to the explanatory merits of this hypothesis.

Formative modelling consider a theoretical term to be a function of the observable symptoms rather than a common cause of them.

The diagnostic view

A researcher who uses the diagnostic view is involved in the determination of latent class membership, on the basis of manifest responses to diagnostic questions. Symptoms in the DSM are indicators of some underlying condition that, although we may not have direct observational access to it, does exist as a phenomenon independent of any diagnostic activities.

To admit the possibility of an erroneous diagnosis implies the acceptance of the hypothesis that the condition itself exists.

Because conditions like being depressed are not directly observable, the distinction between people who do and people who do not suffer from a mental disorder necessarily has hypothetical elements. The hypothesis that there is such a distinction is not inconsequential. The symptoms should be statistically independent conditional on class membership.

This viewpoint suggests future courses of research. There must be something deeper than the mere symptoms, and this homogenizes the people suffering from a given mental disorder. Those who suffer from a mental disorder form an equivalence class.

The dimensional view

In the dimensional view, clinical research involves the determination of persons’ positions on a latent continuum on the basis of their manifest responses to diagnostic questions. This differs from the diagnostic view mainly because its proponents conceptualize disorders as continua rather than discrete classes. In this view, continua are real, but the cut points that define disorders may be arbitrary. Proponents of the dimensional view commonly see patient populations as extremes on a continuum that may extend into the normal population, so that people who suffer from a mental disorder need not to be homogeneous in the sense that they share something that normal people do not have.

Symptoms are considered to have a location on the continuum, called a threshold, which determines how easily they develop. The lower the threshold, the easier one will develop the symptom. The liability-threshold model hypothesizes a trade-off between symptom properties (thresholds) and person properties (liabilities).

Because the dimensional view conceptualizes mental disorders as continuous attributes, it naturally allows for greater heterogeneity at the level of such attributes. The assumption that a latent continuum underlies the symptoms, has consequences for the structure of the probability distribution over the item responses. Given that latent variables are latent, we have no direct way of knowing what their structure is.

The latent continuum hypothesis has testable consequences. We may expect to see heterogeneity in item responses, but with strict ordering. This makes items exchangeable.

Although dimensional models allow for some heterogeneity, this may not be the kind of heterogeneity that we would expect for psychopathology symptoms. Such symptoms are likely to be heterogeneous in a way that uni-dimensional latent variables do not allow for, in the sense that symptoms measure different things.

The most important hypothesis in these models is that the different symptoms measure the same attribute.

Measurement of causality

The intuitions that render the diagnostic and dimensional views attractive are that symptoms do not correlate by accident, and symptoms within a disorder correlate more strongly than symptoms between disorders.

Measurement requires more than model fit. There must be a plausible account of how the attributes to be measured are causally connected to a set of indicators. For a set of indicators to measure an attribute it is required that differences in position on the attribute structure cause differences in the symptoms. When such causal relevance of the attribute for the indicators is absent, it is hard to defend the supposition that the indicators are valid measures of the attribute, because in this case they are not measures of the attribute at all.

There are several problems with the requirement of causal connection as applied to the relation between symptoms and disorders. 1)There exists no substantively motivated account of what this causal relation should be, and 2) there are problems with the causal ontogenesis of symptom patterns. A latent variable model views correlations between indicator variables as spurious, they do not reflect direct causal relations between the indicators, but arise as a result of the fact that the indicators measure the same attribute.

One consequence of a common cause relation is that conditioning on the levels of the common cause screens off correlations between its effects. Local independence means that one position on the attribute is considered at a time, and the indicators are statistically independent in the subpopulation of people who occupy this position. This is difficult if the indicators do not share a common dependence on this latent structure.

There is a lack of correspondence between models for interindividual differences and intraindividual processes. The structure of a model, as derived from interindividual differences research has no discernible implication for the structure of the processes that go on within an individual. The processes that generate data at the individual level may have a completely different structure from that present in a model for the differences between people. This does not mean that we cannot investigate what the relation between intra-individual and inter-individual structures look like. The causal relation between latent variables and indicators is not straightforward. It requires theoretical and empirical justification that goes beyond fitting a latent variable model to individual differences.

If we cannot routinely assume that the latent variables model is valid for describing the causal ontogenesis of symptom groups within individual people, then we are forced to devote considerable attention to a theoretical analysis of the processes that may generate data on which we execute analysis. At this level, the diagnostic and dimensional view run into problems. Symptoms are not effects of a common cause, they stand in direct causal relations to each other.

Disorders as causal systems

Suppose that the relations between symptoms as captured in diagnostic systems are direct causal relations. We could no longer say that the indicators measure the construct, because the causal structure of measurement is not satisfied. The relation between indicators and constructs is a mereological (part-whole) relation rather than a causal one. The symptoms are part of a larger system of symptoms and causal connections that we refer to. We do not measure the construct, but constitute it in a non-arbitrary way.

The cut-off scores for diagnoses convey non-arbitrary information about whether the causal system of symptoms is at all activated in a person, and where in the causal system of symptoms a person is located. Central symptoms are symptoms with many incoming and outgoing relations from and to other symptoms that make up the disorder. The converse of centrality occurs when variables are on the border of causal system.