A psychological test refers to a systematic procedure for comparing the behaviour of two or more people. Standardization refers to collecting a sample for the purpose of norm-referencing and it refers to the administration of a measure according to a consistent set of rules. A standardized administration is necessary to produce reliable and valid measurement.

A diagnostic schedule (e.g. rating scale) refers to a specialized psychometric method that provides a structured procedure for collecting and categorizing behavioural data that corresponds to diagnostic categories or systems. It is used to diagnose a syndrome. The goal of an instrument determines whether it is a diagnostic schedule (e.g. diagnose or not). Rating scales allow for the rapid and accurate identification of domains of behaviour that may require diagnosis or intervention.

There are several definitions used when testing children:

1. Raw score
   This refers to the sum of the item scores on a certain measure and does not give any information of performance compared to a norm-group.
2. Standard score
   This refers to a raw score that is converted to a distribution that reflects the degree to which the individual has scored below or above the sample mean.
3. T-score
   This is a type of standard score with a mean of 50 and a standard deviation of 10.
4. Linear T-scores
   This is a type of score which maintains the skewed shape of the raw score distribution, meaning that the relationship of percentile ranks to T-scores is unique for each scale.
5. Uniform T-score (UT)
   This is a type of standard score which maintains the skewness of the original raw score distribution to ensure that the relationship between percentile ranks and T-scores is constant across scales.
6. Scaled score
   This is a type of standard score with a mean of 10 and a standard deviation of 3.
7. Percentile rank
   This refers to a person’s individual relative position within a norm group but has unequal units along their scale (e.g. the difference between the first and fifth percentile rank is larger than the difference between the 40^th and the 50^th).
8. Norm-referenced interpretation (i.e. information on age-typicalness)
   This refers to the comparison of children’s scores to some standard or norm.
9. Norm-referenced achievement tests
   This refers to tests that compare children’s scores to others in the same grade.
10. Local norms (e.g. gender-referenced norm; clinical norm).
    This refers to norms based on a specific population in a specific setting or location. This is used when the clinician wants to limit comparisons to a certain group.
11. National norms
    This refers to norms based on the population as a whole.

Standard scores allow for comparisons across tests because they have equal units along the scale. Gender norm-referencing could erase the gender effects in groups and subsequently reduce gender biases in diagnosing. However, the DSM-5 often does not have a different threshold for boys and girls.

A negatively skewed distribution has the tail on the left and is the opposite of a positively skewed distribution. Normal distribution cannot be assumed for personality tests or behaviour ratings. Forcing normality on a non-normal distribution could make interpretation of the scores easier but could lead to more false positives.

Reliability refers to the degree to which test scores are free from measurement error and includes the presumed stability (1), consistency (2), and repeatability (3) of scores for a given individual. Reliability coefficients represent the amount of reliable variance associated with a test or the degree to which observed scores on the test reflect true scores on the construct. The error variance is 1 minus the reliability coefficient.

The test-retest method refers to retesting the same group in a brief period of time. The internal consistency assesses the average correlation among items in a test or scale and assesses the homogeneity of the test item pool (e.g. split-half reliability). Subtest specificity refers to the amount of reliable variance that can be attributed to a single subtest or scale. This can be used to assess how much confidence a clinician should have in conclusions that are based on a single subtest or scale. The standard error of measurement (SEM) refers to the standard deviation of the error distribution of scores and gives an indication of the amount of error associated with test scores.

There are several factors that influence reliability (HERRR):

• Homogeneity of the sample.
• Error may be introduced if a poor translation is used.
• Reliability can suffer when there is a longer interval between assessment.
• Reliability can be affected by rater or child characteristics (e.g. fatigue; age).
• Reliability can differ for different score levels (e.g. test can be reliable for depressed students but not for non-depressed students).

Validity refers to the degree to which evidence and theory support the interpretation of test scores for the proposed uses of a test (i.e. can the test score be used for what it was supposed to be used). A test is validated or invalidated for specific uses or circumstances meaning that a test, in general, cannot be valid or invalid. There are different types of validity:

1. Construct validity
   This refers to the degree to which a test measures a construct and is obtained as a result of long-term accumulation of evidence.
2. Content validity
   This refers to the appropriate sampling of a particular content domain.
3. Concurrent validity (i.e. predict outcomes on similar measure)
   This refers to the test showing correlations with other measures to which it is theoretically related.
4. Predictive validity
   This refers to the ability of a test to predict some later criterion.
5. Convergent validity (i.e. correlation between two tests)
   This refers to when scores on a scale correlate with other measures with which it is hypothesized to have an association.
6. Discriminant validity
   This refers to when scores on a scale do not correlate with other measures with which it is hypothesized to not have an association.

Factor analysis refers to a data reduction technique that attempts to explain common variance by reducing it to several ‘overlapping factors’. These factors are then used as latent variables. The factor loading refers to the correlation between a scale and a larger factor and range from -1 to +1.

Confirmatory factor analysis uses a set number of factors based on theory and uses statistical techniques to determine whether this is the case. Cluster analysis (e.g. latent class analysis) attempts to reduce the complexity of a dataset by grouping individuals with common characteristics.

A response set refers to a tendency to answer questions in a biased way. There are several response sets;

1. Social desirability response set
   This refers to the tendency for a person to respond in ways that seem socially appealing.
2. Acquiescence response set
   This refers to the tendency to answer ‘true’ or ‘yes’ to the majority of the items.
3. Deviation
   This refers to the tendency to give unusual or uncommon responses to items.

Clinical utility refers to being able to make a meaningful difference in relation to diagnostic accuracy, case formulation and treatment outcomes.