There are two major approaches to classification: categorical and dimensional. The categorical approach is based on the assumption that differences between normal and abnormal behavior or between different types of abnormal behavior are qualitative. The dimensional approach, on the other hand, is based on the assumption that these differences are quantitative, since behavior can be seen as distributed on a continuum from normal to abnormal. The threshold model of behavior is a compromise between these two approaches. This model basically states that an individual may exhibit characteristics of a disorder without experiencing any adverse effects until a certain point, or threshold. As they cross that threshold, the quantitative element, however, there is an increase in the number of problems experienced; therefore, qualitatively, the individual can be evaluated as having a disorder.

In diagnosing mental disorders, physicians seek to diagnose syndromes. Syndromes are a group or pattern of symptoms that occur together in a persistent fashion, potentially constituting evidence of a mental disorder. A symptom is any characteristic of a person's actions, thoughts, or feeling that could serve as a potential indicator of the presence of a mental disorder. Symptoms are self-reported and generally not observable while signs, such as affect, are observable features of the person's mental state. By using both signs and symptoms, a diagnosis can be met. The ability to define syndromes and to observe how they appear in the general population, or epidemiology, is very important in understanding the etiology of mental disorders and in finding the most effective form of treatment. This is the reason that descriptive statistics such as age of onset, prevalence, and gender differences are important. If, for example, we are able to understand such things as why major depression is more prevalent among women--whether it is because women are more likely to report such distress or because greater societal acceptance of women being depressed has led to biases in the diagnostic criteria--the entire system as a whole becomes more systematic, valid, and reliable.

Diagnostic approaches should always attempt to be high in validity and reliability. Another goal of classification systems is to be high in both specificity and sensitivity. Specificity refers to the extent to which any definition excludes invalid cases. If a definition has poor specificity, it is high in false positives. This means that it labels individuals as having a disorder when there is really no disorder present. Sensitivity refers to the extent to which any definition includes all valid cases. If a definition has poor sensitivity, it is high in false negatives (individuals who have a disorder present are falsely being diagnosed as not having one).

Two diagnostic approaches, which highlight the cost and benefits of poor specificity and sensitivity are the monothetic and polythetic classes. The monothetic category is defined as features that are both necessary and sufficient in order to identify members of that class. This category leads to more homogeneity across members, which results in greater specificity but poor sensitivity. The benefits of this diagnostic approach are that it leads to improved research and treatment findings and decreases marginalization of individuals having a mental disorder. Yet the costs, as with any system that is poor in sensitivity, are that the validity of the research findings decreases and the number of people who need treatment also decreases, since the potential for false negatives increases. The polythetic category presents a broad set of characteristics that are neither necessary nor sufficient. Instead, each member of the category must possess a certain minimal number of defining characteristics, but none of the features have to be found in each member of the group. The heterogeneous quality of this category increases sensitivity but leads to a decrease in specificity. Benefits of this approach are that the amount of individuals who need help receive treatment, yet at the same time, this may lead to punishing and labeling of individuals who are normal as "sick," which may, in turn, have an iatrogenic effect of perpetuating the disorder and maintaining functional impairment. It is also dangerous, as any medical field, to medicate and treat individuals for diseases and disorders that they do not truly have, thus leading to a decrease in reliability of treatment methods.

There are three types of validity: etiological, concurrent, and predictive. Etiological validity looks at the extent to which we can identify distinct factors or characteristics that contribute to the onset of a disorder. Concurrent validity is concerned with factors currently associated with the presence of a disorder. And predictive validity assesses the outcome of individuals affected with a disorder. By looking at these three components of validity, advancements in research concerning the causes, classification, and treatment of disorders can be made. The criteria for "vulnerability markers" for developing a disorder, for instance, states that the marker must distinguish between people who have the disorder and those who do not, and should identify more people among the relatives of the individual with the disorder than relatives of people in the general population, both of these being characteristics of concurrent validity. Furthermore, the marker should also be transmitted genetically, etiologically valid, and a stable characteristic over time, and should predict the future development of the disorder, qualities that are inherent in the definition of predictive validity.