ADD5106- Week 2 Discussion 2: Reliability and Validity

CurrentStandardsforValidityChapter4.pdf

Home >Human Resource Management homework help >ADD5106- Week 2 Discussion 2: Reliability and Validity

Objectives 1. Understand the historical context of validity. 2. Understand the current standards for validity. 3. Differentiate the current standards of validity with the triadic model of validity. 4. Understand the methods for demonstrating evidence of validity. 5. Identify whether test developers sufficiently address validity issues in their

assessments.

Current Standards for Validity

Defining Validity: A Brief History In assessment, validity refers to the development, administration, scoring, interpretation, and utilization of an instrument. An instrument is valid if the instrument is an actual mea- sure of a given construct. “Validity refers to the degree to which evidence and theory sup- port the interpretations of test scores entailed by proposed uses of tests . . . [and is] the most fundamental consideration in developing and evaluating tests” (American Educational Research Association [AERA], American Psychological Association (APA), & National Council of Measurement in Education [NCME], 2014, p. 11).

Validity is an evolving process. The manner in which validity was defined in the early 20th century is quite different from the current definition and criteria. Initially, validity was viewed as a fixed, stable attribute of a measure. Guilford (1946) indicated “a test is valid for anything with which it correlates” (p. 429). Thus, validity was evaluated via a cor- relation coefficient, which became known as a validity coefficient. The effect was to rely on statistics to determine evidence of validity. Issues of item content were not even considered (Reynolds & Kamphaus, 2003). Research continued to develop in this area with the defini- tion of validity being extended to criterion evidence (Gulliksen, 1950), which is a relation- ship to a construct or phenomenon, and convergent and discriminant evidence (Campbell & Fiske, 1959), the extent to which items on an instrument for different scales are strongly

Balkin, Richard S., and Gerald A. Juhnke. Assessment in Counseling : Practice and Applications, Oxford University Press, Incorporated, 2018. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/capella/detail.action?docID=5179769. Created from capella on 2023-10-17 21:16:53.

C op

yr ig

ht ©

2 01

8. O

xf or

d U

ni ve

rs ity

P re

ss , I

nc or

po ra

te d.

A ll

rig ht

s re

se rv

ed .

78 | assessment in counsel ing

intercorrelated (i.e., convergent evidence) or show weaker relationships (i.e., discriminant evidence). For example, an individual who endorses the item “I feel sad all of the time” on the Beck Depression Inventory- II (BDI- II) may also endorse other similar items related to depression (convergent evidence), which all could relate to depressive disorder (criterion evidence). On another note, some instruments contain more than one scale. For example, the Substance Abuse Subtle Scale Inventory- 4 has scales that focus on more objective crite- ria related to chemical dependency, such as the amount of alcohol or drugs consumed, but it also has other scales that measure more subtle symptoms, such as feelings of guilt after using. An item measuring guilt may not correlate with objective items but could correlate with items related to feelings that occur as a result of drug or alcohol use. Both convergent and discriminant evidence would show that feelings of guilt correlate with subtle symp- toms of chemical dependency (i.e., convergent evidence) and correlate less with objective symptoms of chemical dependency (i.e., discriminant evidence).

Cronbach and Meehl (1955) identified four types of validity in their seminal article, “Construct Validity in Psychological Tests”: (a) predictive validity, (b) concurrent valid- ity, (c) content validity, and (d) construct validity. The concept of criterion validity was separated into two aspects: (a) predictive validity, which is attained when an instrument is administered and scores are correlated to some criterion that is obtained at a later time, and (b) concurrent validity, which is attained when an instrument is administered while simul- taneously obtaining another score on a criterion. For example, when the Strong Interest Inventory is used to guide a client to a career path, predictive evidence is demonstrated; that the Strong Interest Inventory may correlate with the interest inventory on O*NET demonstrates concurrent evidence. Content validity referred to the acceptance that the items of an instrument measure the intended construct. The process of evaluating content validity was quite ambiguous. According to Cronbach and Meehl (1955), “Content valid- ity is ordinarily to be established deductively, by defining a universe of items and sampling systematically within this universe to establish the test” (p. 282). Cronbach and Meehl fur- ther elaborated that content validity was demonstrated through “acceptance of the . . . con- tent” (p. 282). Thus, content evidence may be demonstrated through a review of previous published research on a phenomenon of interest, expert review, or documentation of prees- tablished acceptance or standards. Cronbach and Meehl suggested construct validation be investigated when a phenomenon of interest lacked an operational definition. “Construct validity must be investigated whenever no criterion or universe of content is accepted as entirely adequate to define the quality to be measured” (p. 282). Construct validity could be established through the investigation of (a) group differences; (b) correlational proce- dures such as factor analysis; (c) evaluations of internal structure, such as reliability and the consistency of responses from a homogeneous sample; (d) evaluations because of changes in conditions, such as the insertion or removal of a test condition; and (e) observations of an individual performance or process of completing an instrument.

In 1966, APA et al. published the Standards for Educational and Psychological Tests and Manuals. Although this was not the first time these organizations had

C op

yr ig

ht ©

2 01

8. O

xf or

d U

ni ve

rs ity

P re

ss , I

nc or

po ra

te d.

A ll

rig ht

s re

se rv

ed .

current standards for Val id ity | 79

published standards for testing and measurement, it was the first joint publication for the three groups. The main change from Cronbach and Meehl’s (1955) conceptualiza- tion of validity was use of the term criterion- related validity, which addressed specific types of predictive and concurrent evidence related to the measure (Hubley & Zumbo, 2001). The 1966 Standards yielded the trinity view (Goodwin & Leech, 2003; Hubley & Zumbo, 2001) of validity. Although the 1966 Standards related three distinct types of validity, the 1985 Standards identified test validity as a single concept and that con- tent, criterion, and construct validity were merely different types of evidence for valid- ity (AERA et al., 1985; Goodwin & Leech, 2003). Thus, a trend evolved to describe validity as a single conceptual element of evaluating instruments, in which different types of evidence may be examined to determine the extent to which an instrument meets standards of validity.

A Present View of Validity In the 1999 Standards for Educational and Psychological Testing (AERA et al., 1999), valid- ity was described as the incorporation of evidence and theory required to support the proposed interpretation and use of test scores. Validity, therefore, included not only the gathering of data to support the interpretation of scores but also how scores were used. This idea was continued in the 2014 Standards. For example, the use of an aptitude test to assess academic achievement and to be used as an exit exam for high school students may be contrary to the explicit definition of validity.

The process of validation involves accumulating relevant evidence to provide a sound scientific basis for the proposed score interpretations. . . . .When test scores are used or interpreted in more than one way (e.g., both to describe a test taker’s current level of the attribute being measured and to make a prediction about a future outcome), each intended interpretation must be validated. (AERA et al., 2014, p. 11)

When evaluating an instrument for use with a client, counselors should focus on the statements related to validity. Essentially, counselors should ask, “Is this the appropriate instrument for this client? Will the interpretation of test scores be used in a manner to which the instrument was designed and interpreted in a manner that is appropriate to the client?” The responsibility to evaluate validity is on both the counselor and the test developer. The developer is responsible for furnishing the evidence related to test inter- pretation and use; the counselor is responsible for evaluating such evidence and using the instrument, scores, and interpretation in an ethical manner (ACA, 2005; AERA et al., 2014). The 2014 Standards for Educational and Psychological Testing emphasized five types of evidences for test validity.

C op

yr ig

ht ©

2 01

8. O

xf or

d U

ni ve

rs ity

P re

ss , I

nc or

po ra

te d.

A ll

rig ht

s re

se rv

ed .

80 | assessment in counsel ing

Evidence Based on Test Content The contents of an instrument (i.e., test items) should represent the intended domain or construct being measured. “Test content refers to themes, wording, and format of the items, tasks, or questions on a test,” as well as the guidelines for procedures regarding adminis- tration and scoring (AERA et al., 2014, p. 14). Evidence based on test content replaced the term content validity but also expanded the definition. Rather than merely focusing on what the instrument contains, evidence based on test content also incorporated pro- cedures related to test use and interpretation. Counselors should be able to identify the extent to which scores and interpretations of an instrument are generated from the items administered and the tasks or processes placed upon the examinee. For example, admin- istering a Wechsler Intelligence Scale for Children- V to a child with deafness may not be valid, as the processes required by the examinee involve verbal explanations of questions asked. If the examinee has difficulty responding to the questions and queries as a result of hearing impairment and not of intellectual impairment, then the measure ceases to be valid because of the inappropriate tasks and processes being placed upon the examinee. Many complications can arise that make a test invalid, such as administering a self- report inven- tory to an individual diagnosed with dyslexia. Counselors who use assessment instruments must consider carefully the ramifications of scoring and interpreting instruments, given the abilities of each individual client.

A term that is often confused with evidence based on test content is face validity, which refers to a superficial evaluation of the instrument according to how the instrument looks. In other words, an instrument would be deemed as valid if it appears valid to the individuals who decide to use it (Anastaci & Urbina, 1997). Face validity is not evidence based and therefore is not considered as sufficient evidence of validity.

How Is Evidence Based on Test Content Evaluated? Analysis and evaluation of evidence based on test content occurs both logically and empir- ically (AERA et al., 2014). Test developers should document how items were derived. Relevant reviews of literature and developed theories should be cited. Often, expert opin- ion is cited and test developers may revise, add, or delete items according to suggestions from reviewers. In creating the Juhnke- Balkin Life Balance Inventory ( JBLI), Davis, Balkin, and Juhnke (2014) engaged in the following processes to demonstrate evidence of test content:

The items on the JBLI were developed by the authors in an attempt to assess the various domains of life balance. The authors of the JBLI consulted with eight experts during the item development phase. . . . The reviewers were asked to rate the relevance (i.e., highly relevant to domain [4] , relevant to this domain [3], neither relevant nor irrelevant to domain [2], irrelevant to domain [1], or highly irrelevant to domain [0]) of each of the 7 question stems within the 12 domains

C op

yr ig

ht ©

2 01

8. O

xf or

d U

ni ve

rs ity

P re

ss , I

nc or

po ra

te d.

A ll

rig ht

s re

se rv

ed .

current standards for Val id ity | 81

included in the JBLI. The domains include Mood, Stress, Physical, Exercise Health & Nutrition, Sleep, Social- Interpersonal, and Marriage- Significant Other, Sexual and Intimacy, Work- Career, Spirituality- Religious, Happiness, Hopelessness- Helplessness, and Substance Abuse- Addictions. . . . The JBLI domains were developed to be inclusive of the areas of work, love, and play or leisure and include assessment of substance abuse. The experts recorded their rat- ings on an expert reviewer form and returned them. The information from the reviewers was utilized to refine the items. (p. 183)

An empirical method for demonstrating evidence based on test content is the index of item- objective congruence, developed by Rovinelli and Hambleton (1977). In this method, a test developer identifies an objective to be measured by each item and expert raters evaluate each item and provide ratings: 1 for an item that clearly measures the objec- tive, – 1 for an item that clearly does not measure the objective, and zero for items in which the measurement of the objective is unclear (Turner & Carlson, 2003). Raters may have dif- ferent ratings for each of the items, depending on their own subjective evaluations. An index for item- objective congruence may be computed to assess the degree to which the raters identified that an item measured a particular objective using the following formula:

I N

nik k= −

− 2 2

( )µ µ

where Iik is the index of item- objective congruence for item I on objective k, N is the num- ber of objectives, µk is judges’ mean rating of item I on objective k, and µ is the judges’ mean rating of item I on all objectives (Crocker & Algina, 1986, p. 221).

A generally accepted value for an index score is .75 (Turner & Carlson, 2003). Although the specific computations of the index of item- objective congruence are outside the scope of this text, readers who have an interest in using the index of item- objective con- gruence may find more information about the measure in the cited materials. The impor- tant element of this discussion is that empirical methods of evaluating evidence based on test content may be used in test development.

Evidence Based on Response Processes Test developers should gather evidence demonstrating that the actual responses of partici- pants on test items is a valid operation for evaluating the construct being investigated (AERA et al., 2014). When an item appears on an instrument, the assumption is that individuals will interpret the item the same way. But this is not always true. For example, consider a common question related to substance abuse assessment, “Has drinking or taking drugs ever caused you any problems?” Such questions can be ambiguous, because individuals who use drugs or alcohol are often in denial about the problematic nature of their use/ abuse. Thus,

C op

yr ig

ht ©

2 01

8. O

xf or

d U

ni ve

rs ity

P re

ss , I

nc or

po ra

te d.

A ll

rig ht

s re

se rv

ed .

82 | assessment in counsel ing

an individual who drinks only on weekends may answer “yes” while another individual who smokes marijuana every day may answer “no.”

How Is Evidence Based on Response Processes Evaluated? When developing an instrument, questioning participants regarding their understanding of the items and strategies in answering the items is important to demonstrating evidence of response processes. By examining the individual responses of examinees and even ques- tioning examinees about their responses or how the responses were derived, test developers may gain insight to the extent to which a desired construct is being measured. There are various types of responses that can be monitored, such as the speed of the response, tasks engaged in developing a response, or physiological responses to an item. When individual differences are noted in terms of a response to an item, the test developer may wish to con- sider alternative formats to an item.

One aspect that is generally acknowledged in standardized testing is the issue of bias. Bias occurs when the interpretation of an instrument is different across various groups. Test developers need to be cautious when subgroups perform differently on an instrument. Investigations into ways items may be interpreted or meaning conveyed is essential so that the instrument remains relevant across a diverse population.

Evidence Based on Internal Structure Previously labeled as construct validity, evidence based on internal structure refers to the interrelationships of the items and the relationships of the items to variables, constructs, or components/ factors being measured. For example, when subscales are developed for an instrument, the expectation is that items on one subscale have higher correlations to each other, as they are measuring the same construct, and have lower correlations to items mea- suring a separate subscale, as the construct may be quite different.

The extent to which items on the same subscale have higher intercorrelations is known as convergent evidence; the extent to which items on one subscale have lower correlations to items on a separate subscale is known as discriminant evidence. Trochim (2000) identified methods for examining item correlations to establish convergent and discriminant evidence. For example, in development of the Crisis Stabilization Scale (Balkin, 2014; formerly the Goal Attainment Scale of Stabilization; Balkin, 2013; Balkin & Roland, 2007), subscales were developed to measure the extent to which adolescent cli- ents in psychiatric hospitalization attained therapeutic goals designed around problem- solving and coping strategies and commitment to a follow- up plan upon discharge from the hospital. In this study, convergent evidence was demonstrated in the problem iden- tification and commitment to follow- up subscales. Sample items were strongly related to items identified as measuring problem identification and less related to items measuring

C op

yr ig

ht ©

2 01

8. O

xf or

d U

ni ve

rs ity

P re

ss , I

nc or

po ra

te d.

A ll

rig ht

s re

se rv

ed .

current standards for Val id ity | 83

commitment to follow- up (see Table 4.1). A similar model appears in commitment to follow- up.

Note that items in bold demonstrate strong intercorrelations. These items demon- strate convergent evidence in that items on the same subscale are more highly correlated to each other. Also note that the items not in bold are weaker correlations. These items demonstrate discriminant evidence, as items on one subscale do not correlate with items on another subscale. Each subscale measures something unique in client therapeutic goal attainment.

How Is Evidence Based on Internal Structure Evaluated? When developing instruments, items should fit the appropriate construct of interest both theoretically (i.e., evidence based on test content) and statistically. Examining intercor- relations of items, as mentioned earlier, is one method. However, often more sophisticated methods are used to determine whether items actually measure a construct in question. Common methods include principal component analysis (PCA), exploratory factor analy- sis (EFA), and confirmatory factor analysis (CFA). The goal of PCA and EFA is to identify items that measure a latent trait and eliminate items that do not contribute to a measure. A latent trait refers to a variable or construct that is not directly observed or measured. For example, in the development of the BDI- II (Beck, Steer, & Brown, 1996), 21 items were used to measure depression. These 21 items may or may not represent the entire pool of items in which data were collected. Items that had low correlations, for instance, may have been removed. Beck et al. conducted an EFA and two latent traits emerged. Items indicat- ing levels of sadness, agitation, loss of interest, and indecisiveness, for example, loaded on a latent trait, which was identified by the authors as Cognitive- Affective dimension. These items were indicative of cognitive or affective symptoms of depression. A second factor

TABLE 4.1 Convergent and Discriminant Evidence

in the GASS

Problem

Identification

Commitment to

Follow- Up

PI1 PI2 PI3 CF1 CF2 CF3

PI1 1.00

PI2 0.85 1.00

PI3 0.80 0.86 1.00

CF1 0.40 0.41 0.37 1.00

CF2 0.41 0.42 0.35 0.81 1.00

CF3 0.43 0.43 0.39 0.68 0.80 1.00

Note: GASS = Goal Attainment Scale of Stabilization; PI = problem identification; CF = commitment to follow- up.

Source: Balkin & Roland, 2007.

C op

yr ig

ht ©

2 01

8. O

xf or

d U

ni ve

rs ity

P re

ss , I

nc or

po ra

te d.

A ll

rig ht

s re

se rv

ed .

84 | assessment in counsel ing

emerged consisting of items such as tiredness, loss of energy, and changes in appetite. These items were somatic in nature and therefore labeled as Somatic dimension. Thus, two latent traits emerged from the EFA.

An important note is that both PCA and EFA are exploratory procedures. Although a theory was in place to create items that measure depression, based on Diagnostic and Statistical Manual of Mental Disorders (fourth edition) criteria, the latent traits were iden- tified through exploratory procedures. The test developers created the items and then studied the factor loadings, which were used to label the latent traits. In exploratory proce- dures, a theory is developed through both item content and statistical analyses. Many test developers do not run additional analyses beyond the exploratory procedures. However, CFA represents a method to statistically test a theory once the factors have been identified. CFA procedures may be more respected, because statistics are used to test a preestablished theory, as opposed to using statistics in theory development. A strong instrument is one in which both exploratory and confirmatory procedures on separate samples are used in test development.

Evidence Based on Relations to Other Variables When developing an instrument, analyses to external variables related to the measure is pertinent to establishing evidence of test validity. Relationship to external variables may be ascertained by examining concurrent evidence and predictive evidence; similar to evaluat- ing evidence of internal structure, convergent and discriminant evidence may also be exam- ined with respect to external variables.

Concurrent evidence refers to an analysis of a relationship between two measures at the same time. For example, Balkin and Roland (2007) administered two instruments, the Goal Attainment Scale of Stabilization (GASS) to measure therapeutic goal attainment for adolescents at the time of discharge from psychiatric hospitalization, and the Clinician Problem Scale– Revised (CPS- R) to measure psychiatric symptoms at the time of discharge. Concurrent evidence was demonstrated on the GASS as a significant relationship was found between the GASS and CPS- R scores, indicating that increases in therapeutic goal attainment (GASS scores) were related to decreased psychiatric symptoms (CPS- R scores).

Predictive evidence is demonstrated when an instrument is related to a specific future outcome. For example, universities often use the ACT or SAT as an admission criterion because they believe that college entrance scores may be predictive of success in college. Predictive evidence, however, can be challenging to evaluate. In developing the Suicide Probability Scale (SPS), Cull and Gill (1982) attempted to demonstrate the extent to which scores on the SPS differentiated between clinical and nonclinical populations. In a clinical population, 70.8% of potential attempters were misclassified as nonsuicidal and 76.9% of nonattempters were misclassified as suicidal. In a nonclinical population, 41.5% of nonattempters were misclassified as suicidal (Golding, 1985).

C op

yr ig

ht ©

2 01

8. O

xf or

d U

ni ve

rs ity

P re

ss , I

nc or

po ra

te d.

A ll

rig ht

s re

se rv

ed .

current standards for Val id ity | 85

Similar to evidence based on internal structure, an evaluation of convergent and dis- criminant evidence to external variables may provide important evidence for validity. For the purposes of evaluating relationships to other variables, stronger correlations between similar measures would be apparent in convergent evidence and weaker correlations would be apparent in discriminant evidence. For example, Beck et al. (1996) demonstrated con- vergent evidence by correlating scores on the BDI- II with the Hamilton Psychiatric Rating Scale for Depression. Because both instruments measure the same construct, depression, the correlation, as expected, between the two instruments was high (r = .71). Discriminant evidence was less evident, as the BDI- II was correlated with the Beck Anxiety Inventory (BAI). The relationship between the BDI- II and the BAI was .60. Beck et al. identified that this finding was not unexpected, as “depression and anxiety have been found to be correlated in clinical evaluations” (p. 27).

How Is Evidence Based on Relations to Other Variables Evaluated? Correlational designs, as noted earlier, tend to be common in identifying evidence of rela- tionships to other variables. Typically, correlations of measures and regression analyses may be used to demonstrate concurrent, predictive, convergent, and discriminant characteris- tics between measures. Occasionally, tests of significance between administrations of two or more measures may be analyzed to demonstrate evidence of validity. Beck et al. (1996), in their revision of the BDI– IA to the BDI- II, conducted t tests to evaluate whether scores were significantly different between the two instruments. Beck et al. identified that more items were endorsed on the BDI- II than on the BDI- IA by a sample of outpatient clients. Despite this difference, which may be used to justify the revision, correlation between the two instruments was strong (.84).

Evidence for Validity and Consequences of Testing In a break from the triadic model, the 1999 and 2014 Standards emphasized the need to identify the benefits, as well as consequences, of using a measure. The benefits of using an instrument should be both stated and implied. When a test is administered, users should be able to glean information and insights that are useful and beneficial to the client and/ or to society. At times, measures may be used in which either the intended construct of interest is lacking evidence of validity or the instruments measure a construct that is unrelated. A clear example of testing consequences can be examined through the advent of high- stakes testing. When No Child Left Behind was passed in 2002, accountability through the examination and demonstrated improvement of student performance was legislated, with the withholding of federal funds as a con- sequence for poor test scores (Thorn & Mulvenon, 2002). Hence, students are now

C op

yr ig

ht ©

2 01

8. O

xf or

d U

ni ve

rs ity

P re

ss , I

nc or

po ra

te d.

A ll

rig ht

s re

se rv

ed .

86 | assessment in counsel ing

placed under enormous pressure to achieve higher test scores. In many states, the extent to which the test measures academic progress, as well as the students’ ability to work under a pressure situation, has come under scrutiny. Although no change in the pres- ent system is likely to occur, the debate about the benefits and the consequences of the test and the practice of high- stakes testing is central to obtaining evidence based on consequences of testing. AERA et al. (2014) cautioned that differentiating between social policy issues versus test validity issues may be difficult. The intent of obtaining evidence of consequences of testing is not to influence social policy but to make sure that interpretations of instruments provide the intended information. As the emphasis on consequences for testing is relatively new, Goodwin and Leech (2003) indicated that few guidelines have been established. Although the intended focus on the benefits and consequences of a test is important, test developers have not identified methods for demonstrating this aspect of validity.

What Are the Implications for Test Validity? The Standards for Educational and Psychological Testing were last revised in 2014; how- ever, neither developers nor reviewers have adhered strongly to the most recent standards. The Minnesota Multiphasic Personality Inventory- II (MMPI- II), in fact, was revised in 1989, and evidence of validity still follow the 1985 Standards, with emphasis on com- paring norms and scales between the MMPI- II and the MMPI. Even instruments that were revised after the 2014 Standards tend to adhere to much older standards of validity. Modern test reviews in Mental Measurements Yearbook still incorporate the terms from the 1985 Standards.

Although the instruments in the case studies are all well established and common to the field, validity studies using the most recent standards are lacking. Counselors should be careful consumers when engaging in standardized assessment practices. Attention to reviews of instruments is essential in identifying the appropriateness of using various assess- ment tools. When evaluating whether the use of an instrument would be valid for a client or group of clients, counselors should consider the following issues:

1. To what extent was the instrument developed under a theoretical framework? This is essential to evidence of test content. What theory or theories were used to develop the instrument? In addition, identification of some type of review by experts is important to providing evidence that the instrument was theoretically derived.

2. To what extent do the processes involved in responding to items provide a mean- ingful measure of the construct in question? For example, if the instrument is a self- report inventory, is this appropriate for the client and the construct being meas- ured? Counselors should consider the nature and type of questions that clients are

C op

yr ig

ht ©

2 01

8. O

xf or

d U

ni ve

rs ity

P re

ss , I

nc or

po ra

te d.

A ll

rig ht

s re

se rv

ed .

current standards for Val id ity | 87

exposed to and identify if the scores obtained are likely to reflect a valid measure for the client.

3. What strategies were used to demonstrate evidence of internal structure? Most tests are validated using sophisticated statistical analyses, such as EFA and CFA. Although many master’s- level counselors will not have course work covering the details of these analyses, being aware of attempts to establish factor structure is important. Counselors should ensure that the subscales in the measure are identi- fied as part of the factor structure. The overall structure of the instrument should account for a large proportion of the variance in the model, but there are no good rules of thumb on this with some instruments as low as 40% (or possibly lower) and others much higher.

4. What measures were used to correlate to the instrument in question? Counselors should look for evidence that the instrument being evaluated was correlated with other instruments that measure a similar construct. If the instrument correlates mod- erately to another existing measure, that serves as evidence that the scores obtained from administration of the instrument may be valid for a particular use.

5. What are the consequences for using a test for a specific client? Counselors should be aware of the issues related to test use and abuse. The scores obtained on a particular assessment measure can help direct or guide treatment or may inadvertently label a client, which may not serve his or her best interest. Counselors should ensure that the scores obtained from a measure are going to be used appropriately.

C op

yr ig

ht ©

2 01

8. O

xf or

d U

ni ve

rs ity

P re

ss , I

nc or

po ra

te d.

A ll

rig ht

s re

se rv

ed .