A Psychometric Evaluation of the STAI-Y,BDI-II,and PAI Using Single and Multifactorial Models in Young Adults Seeking Psychoeducational Evaluation

Abstract

A psychometric evaluation on the measurement of self-report anxiety and depression using the Beck Depression Inventory (BDI-II), State Trait Anxiety Inventory, Form-Y (STAI-Y), and the Personality Assessment Inventory (PAI) was performed using a sample of 534 generally young adults seeking psychoeducational evaluation at a university-based clinic. Confirmatory factor analysis was used to evaluate single-factor and multifactorial models (including hierarchical and higher-order models). Fit indices indicated superiority of the hierarchical model where the BDI-II and PAI depression subscales loaded onto a depression factor, the PAI anxiety subscales loaded onto an anxiety factor, and the STAI-Y State and Trait scale scores loaded onto a separate factor that indexed variance associated with both depression and anxiety. Findings are discussed in regards to the construct validity of the BDI-II, STAI-Y, and PAI in young adults seeking psychoeducational evaluation and relations among these measures.

Keywords

depression anxiety psychometrics STAI-Y BDI-II PAI

The relationship between anxiety and depression has long been unclear (Clark & Watson, 1991; Watson, 2005), particularly from a psychometric standpoint. Some authors suggest that anxiety and depression should be considered a single construct due to the high degree of overlap between them (Feldman, 1993; Wolpe, 1971). However, a number of researchers (Endler, Cox, Parker, & Bagby, 1992; Endler, Macrodimitris, & Kocovsk, 2003) as well as the Diagnostic and Statistical Manual of Mental Disorders (4th ed., text rev.; DSM-IV-TR; American Psychiatric Association (APA), 2000), conceptualize depression and anxiety as two distinct constructs. In addition, Clark and Watson (1991) have proposed the tripartite model which suggests that positive correlations between the constructs of anxiety and depression result from one overarching factor of general psychological distress or negative affectivity (Clark, Steer, & Beck, 1994; Watson, Clark, Weber, & Assenheimer, 1995; Watson, 2005). Factors such as symptom overlap (Segerstrom, Tsao, Alden, & Craske, 2000), disorder comorbidity (Kessler et al., 2008), and etiological confluence continue to fuel the debate. Related to theoretical arguments about the structure of anxiety and mood, research has also examined the instruments used to assess anxiety and depressive symptoms as their psychometric properties have been shown to significantly influence their ability to differentiate between anxiety and depression (Endler et al., 1992). Exploration of the psychometric properties of these measures may enable a better understanding of the theories explaining anxiety and depressive symptomology and also aid clinicians who routinely use self-report instruments to measure these constructs.

Psychometric Assessment of Depression and Anxiety Using the BDI-II and STAI-Y

The Beck Depression Inventory, Second Edition (BDI-II; Beck, Steer, & Brown, 1996) is a measure of depression as it asks individuals to report the degree to which they have felt depressive symptoms for the past 2 weeks. It is a revised version of the Beck Depression Inventory (BDI; Beck, Ward, Mendelson, Mock, & Erbaugh, 1961). The State-Trait Anxiety Inventory, Form-Y (STAI-Y; Spielberger, Gorsuch, Lushene, Vagg, & Jacobs 1983) was developed to measure acute versus chronic aspects of anxiety. State anxiety is considered a transient emotional condition during which a person experiences tension and apprehension where trait anxiety is typically thought of as a stable, individual vulnerability to experience anxious symptoms.

Karagozoglu, Masten, and Baloglu (2005) examined this issue in 443 Turkish college students and reported significant correlations between the BDI-II and STAI State (r = 0.52) and Trait (r = 0.64) scales. The authors used two exploratory factor analyses (EFAs) to examine how items from the BDI-II related to items from the (1) STAI State Anxiety scale and (2) STAI Trait Anxiety scale. The first EFA yielded a three-factor model with the STAI State items loading onto two factors and the BDI-II items loading onto one factor. In the second EFA resulted in a four-factor model where items of the STAI Trait scale loaded onto two factors and items of the BDI-II loaded onto two factors. Karagozoglu et al. (2005) suggested that although anxiety and depression are significantly correlated, distinct mood factors may be differentiated by the STAI and BDI-II.

Despite some evidence that the BDI-II and STAI may adequately differentiate between anxiety and depression, a number of item and factor analyses have found that anxiety, as measured by the STAI, also measures aspects of depression and well-being (Caci et al., 2003; Grös, Antony, Simms, & McCabe, 2007). Some authors have suggested that the STAI is indexing general psychopathology (Kennedy, Schwab, Morris, & Beldia, 2001), general negative affect (Andrade et al., 2001), or general distress (Endler et al., 1992) as opposed to being a “pure” measure of anxiety. Support for these hypotheses has been provided through factor analysis. Bieling, Antony, and Swinson (1998) found that the STAI Trait scale tends to produce a hierarchical structure of a general negative affect factor with two lower-order anxiety and depression factors, labeled STAI-A and STAI-D, respectively. They also reported that the STAI Trait scale correlated with the BDI at r = 0.72. Another study divided the items of the BDI-II into cognitive-affective scores and somatic-depressive scores and found that both of these scores, along with the BDI-II total score, correlated significantly with the STAI Anxiety (STAI-A) and STAI Depression (STAI-D) scales (Storch, Roberti, & Roth, 2004). For the BDI-II total score specifically, STAI-A correlated at r = 0.69 and STAI-D correlated at r = 0.76. Endler et al. (1992) examined the psychometric properties of the BDI, STAI, and Endler Multidimensional Anxiety Scales (EMAS; Endler, Edwards, & Vitelli, 1991) using EFAs. He found that the EMAS and BDI were able to adequately differentiate between anxiety and depression while the STAI was not useful in discriminating between the two constructs. They reported that BDI score correlated with STAI State from r = 0.35–0.55 (males–females) and STAI Trait from r = 0.41–0.66 (males–females). These findings provide further evidence that the STAI may not adequately measure anxiety as a distinct construct but may be measuring general negative affect and depressive symptoms as well.

Present Study

This study specifically sought to investigate the psychometric relationships between three instruments commonly used to assess anxiety and depression in psychoeducational evaluations: the STAI-Y, the BDI-II, and the Personality Assessment Inventory (PAI; Morey, 1991). The PAI has validated subscales that measure components of anxiety and depression (i.e., affective, cognitive, and physiological). Despite the extensive clinical use of the STAI, BDI-II, and PAI, we were unable to locate any studies that examined the construct independence of anxiety and depression using all three of these well-known instruments in a sample of college students. Four models were investigated. First, a single-factor, “negative mood” model consistent with Feldman (1993) and a two-factor model similar to the DSM-IV-TR’s current conceptualization of anxiety and depression as distinct but correlated constructs. Two three-factor models were investigated: a higher-order model with the STAI State and Trait scales loading on a general negative affect factor that causes anxiety and depression factors and a hierarchical model where all measures are indicators of a general negative affect factor, the PAI anxiety subscale are indicators of an anxiety factor, and the PAI depression subscale and BDI-II are indicators of a depression factor. All three factors were orthogonal in both models. It should be noted that neither three-factor model is an exact representation of the tripartite model as both lack specific hyperarousal and low positive affect factors. However, our three-factor models and the tripartite model are conceptually related and were proposed to give us a potentially more in-depth understanding of the psychometric relations between the three instruments utilized in this study.

A secondary goal of the current study was to investigate the construct validity of the STAI-Y, BDI-II, and PAI as they relate to one another. Previous studies provide conflicting data about the usefulness of the STAI-Y as a measure of pure anxiety. Some have shown it to be highly, positively correlated with measures of depression (Grös et al., 2007; Kennedy et al., 2001) and some items of the STAI-Y may measure constructs other than anxiety (Andrade et al., 2001; Bieling et al., 1998; Caci et al., 2003; Endler et al., 1992). We are aware of no studies to date that have used the BDI-II and STAI-Y in conjunction to validate the anxiety and depression subscales of the PAI, though the PAI manual does list some relevant validation data.

Method

Participants

This sample initially consisted of 675 individuals who completed comprehensive psychoeducational evaluations at an outpatient university-based clinical psychology training clinic. Individuals were evaluated from 2001 to 2011 and were mainly students seeking evaluation for attentional and learning difficulties. No individuals with IQ < 70 or a diagnosis of either mental retardation or severe mental illness (i.e., schizophrenia, bipolar disorder) were utilized. Individuals whose PAI profiles were invalid (i.e., inconsistency T score >73, infrequency T score >75, positive impression management T score >68, or negative impression management T score >92) were also excluded from further analyses. Ninety individuals were excluded based on these criteria or were judged to be outliers leaving 545 individuals for the present analysis. An additional 11 observations were lost to missing or unscorable scales, yielding a usable N of 534.

The mean age of the remaining 534 individuals was 23.3 years old (SD = 6.6) and mean years of education was 14.0 (SD = 2.1). Of these individuals, 50.7% were female, 83.7% Caucasian, 9.4% African American, 3.4% Latino, and 3.5% of other ethnicity. The mean WAIS-III Full Scale IQ was 104.3 (SD = 12.7).

Measures

Beck Depression Inventory (BDI-II)

The BDI-II (Beck et al., 1996) contains 21 items designed to measure cognitive, affective, and somatic symptoms associated with depression. The BDI-II was designed to correspond closely with DSM-IV-TR diagnostic criteria for major depressive disorder. There are four possible choices for each question with answers receiving either 0, 1, 2, or 3 points. Higher scores are indicative of more self-reported depressive symptoms. The test–retest reliability is reported to range from 0.72 to 0.93 (Beck et al., 1996; Osman, Kopper, Barrios, Gutierrez, & Bagge, 2004). The Cronbach’s α of the BDI-II for a subsample of 287 individuals from the current study was 0.92. The Buros Mental Measurement Yearbook (Arbisi & Farmer, 2001) described the BDI-II as having good psychometric properties and reported the following: α coefficient r = 0.92 for outpatient sample, α coefficient r = 0.93 for nonclinical samples, and test–retest reliability r = 0.93 at 1 week.

State-Trait Anxiety Inventory (STAI)

The STAI was designed to measure severity of anxiety symptoms and differentiate acute (state) from chronic (trait) anxiety (Spielberger, Gorsuch, & Vagg, 1970). The STAI was revised in 1983 (Spielberger et al., 1983) and the revised version was used for the current study. The STAI-Y is composed of 40 questions that are answered using a 4-point Likert-type scale. Multiple items are reverse scored, and scores range from 20 to 80 for both the state and trait forms. The STAI state test–retest reliability has been reported to range from 0.40 to 0.54 and the trait test–retest reliability has been reported as 0.86 in two separate studies (Spielberger et al., 1970; Rule & Tarver, 1983). The Cronbach’s αs of STAI state and trait subscales were both 0.92 for a subsample of 287 individuals from the current study. The Buros Mental Measurement Yearbook (Dreger & Katkin, 1985) described the STAI as a good measure of anxiety and reported that α coefficients ranged from 0.83 to 0.92 for State scores and 0.86 to 0.92 for Trait scores.

Personality Assessment Inventory (PAI)

The PAI (Morey, 1991) is a self-report measure composed of 344 questions. It yields 4 validity scales, 11 clinical scales, 5 treatment scales, and 2 interpersonal scales. The clinical scales are consistent with categories used in the DSM-IV (APA, 2000). Individuals are asked to choose one answer from four choices (false, slightly true, mainly true, and very true). It is emphasized that individuals provide their own opinion of themselves. The anxiety and depression scales of the PAI are each divided into three subscales: affective, cognitive, and physiological. The test–retest reliability of the individual PAI scales is reported as ranging from 0.85 to 0.94 for adults and 0.66 to 0.90 for college students (Morey, 1991). The Cronbach’s αs were calculated for a subsample of 287 individuals in the current study for the PAI affective anxiety (0.77), cognitive anxiety (0.87), and physiological anxiety (0.75) subscales as well as the PAI affective depression (0.82), cognitive depression (0.76), and physiological depression (0.71) subscales. The Buros Mental Measurement Yearbook (Andrew, Thorpe, & Dawson, 2010) described the PAI as a psychometrically sound clinical and research instrument and reported the following: internal consistency coefficients ranging from 0.70 to 0.86. and test–retest reliability studies with a mean correlation of r = 0.76.

Procedures

All ethical guidelines for conducting human research were followed. This study was approved by the university’s Institutional Review Board. Informed consent for use of the data in research was obtained for all participants. All individuals completed the BDI-II (Beck et al., 1996), STAI-Y (Spielberger et al., 1983), and PAI (Morey, 1991) as part of a larger psychoeducational evaluation. Each measure was administered according to its standardized protocol. For the purposes of this study, raw scores were used for the BDI-II, STAI-Y, and PAI. Higher scores on each measure indicate increased endorsement of symptom severity on the construct in question.

Data Analysis

A multivariate analysis of variance (MANOVA) was performed with gender as the independent variable and STAI-Y State and Trait raw scores, BDI-II total raw score, and PAI anxiety and depression subscale raw scores as dependent variables. Pearson product moment correlations were used to examine the impact of age and education on the study variables. Pearson product moment correlation coefficients were calculated for the raw scores of the PAI anxiety and depression subscales, the BDI-II raw score, and the raw scores of the STAI-Y State and Trait scales as a preliminary examination of the relationship between these items. Next, four confirmatory factors analyses were run using Amos 7 (Arbuckle, 2006). The four models examined included a single-factor (negative mood) model, a two-factor (anxiety and depression) model, and two distinct three-factor (anxiety, depression, and general negative affect) models: a higher order and a hierarchical model. Model fit was assessed first using the χ² statistic. χ² analyses are sensitive to sample size and may indicate statistical significance with relatively minimal differences in model fit in large samples so several goodness-of-fit indices were also examined. The fit indices used were the normed fit index (NFI; Bentler & Bonnett, 1980), the comparative fit index (CFI; Bentler, 1988), and the root mean square error of approximation (RMSEA; Browne & Cudeck, 1993). Values for NFI and CFI range from 0 to 1 and high values (>0.90) indicate adequate fitting models, and values >0.95 indicate good model fit. Values larger than 0.10 for RMSEA indicate poor fitting models, between 0.08 and 0.10 indicate adequate fit, between 0.05 and 0.08 indicate reasonable fit, and less than 0.05 indicate good fit. All analyses used raw scores of the STAI-Y state subscale, the STAI-Y trait subscale, the BDI-II raw score, and the PAI affective, cognitive, and physiological anxiety subscales and the affective, cognitive, and physiological depression subscales.

Results

Means and standard deviations calculated for the BDI-II, the STAI-Y State and Trait scores, and the PAI anxiety and depression subscale scores are presented in Table 1. The MANOVA revealed a significant main effect for gender. Females scored significantly (p < .05) higher than males on all the variables except the STAI Trait and the PAI affective depression subscale. With the large sample size, many of the correlations of age and education with the study variables were significant (p < .05), but very small (the largest was STAI Trait with age of r = 0.138). Pearson’s product moment correlation coefficients among the scales and subscales scores ranged from 0.35 to 0.83 and were all significant (Table 2; p < .01).

Table 1.

Means and Standard Deviations for BDI-II Total Raw Score, STAI-Y State and Trait T Scores, and PAI Anxiety and Depression Subscale Raw Scores by Gender.

	Gender	Mean	SD	N
BDI-II*	Male	11.4	8.8	265
	Female	13.9	9.6	273
PAI cognitive anxiety*	Male	9.0	5.5	269
	Female	11.4	5.7	276
PAI affective anxiety*	Male	8.2	4.5	269
	Female	10.6	4.6	276
PAI physiological anxiety*	Male	6.8	4.0	269
	Female	8.2	4.7	276
PAI cognitive depression*	Male	7.5	4.0	269
	Female	8.2	4.3	276
PAI affective depression	Male	6.1	4.5	269
	Female	5.8	4.2	276
PAI physiological depression*	Male	7.1	4.1	269
	Female	8.3	4.8	276
STAI State*	Male	41.4	11.1	266
	Female	43.9	11.0	274
STAI Trait	Male	43.8	11.3	265
	Female	45.6	10.3	272

p < .05.

Table 2.

Descriptive Statistics and Correlations Between Raw Scores of the BDI-II, STAI-Y State, STAI-Y Trait, and PAI Subscales.

	1	2	3	4	5	6	7	8	9
1. PAI cognitive anxiety	0.87
2. PAI affective anxiety	0.82	0.77
3. PAI physiological anxiety	0.68	0.72	0.75
4. STAI State	0.52	0.56	0.47	0.92
5. STAI Trait	0.64	0.64	0.56	0.70	0.92
6. BDI-II	0.53	0.54	0.54	0.55	0.72	0.92
7. PAI cognitive depression	0.44	0.50	0.45	0.45	0.63	0.65	0.76
8. PAI affective depression	0.50	0.52	0.50	0.43	0.68	0.71	0.70	0.82
9. PAI physiological depression	0.43	0.48	0.56	0.34	0.48	0.55	0.47	0.54	0.71
Mean	10.2	9.4	7.5	42.7	44.7	12.7	7.9	6.0	7.7
SD	5.7	4.6	4.4	11.1	10.8	9.3	4.2	4.3	4.4
Skewness	0.33	0.42	0.76	0.21	0.14	0.92	0.69	1.03	0.43
Kurtosis	−0.77	−0.31	0.28	−0.22	−0.45	0.66	0.22	1.03	−0.20

Note. Cronbach’s αs are presented in the diagonal in boldface.

Confirmatory Factor Analysis

Four models were assessed using CFA and are presented in Figures 1 to 4. Mardia’s (normalized) coefficient (7.99) for the data indicated that the data were significantly kurtotic; therefore, robust Sattora–Bentler χ² and robust fit indices were reported. First, a single-factor model (Model 1) was evaluated with all STAI-Y, BDI-II, and PAI scores as indicators of a single negative affect construct, consistent with Feldman (1993). The hypothesis for this model was that the STAI-Y, BDI-II, and PAI anxiety and depression subscales data would best fit a single construct. However, the hypothesized single-factor model was a poor fit (χ²[27] = 503.59, p < .001; CFI = 0.80; NFI = 0.88; RMSEA = 0.18; 90% CI [.17, .20]).

Figure 1.

Model 1: Single general distress or general negative affect factor.

Figure 2.

Model 2: Depression and anxiety as two correlated factors.

Figure 3.

Model 3: Higher order three-factor model of general negative affect, anxiety, and depression.

Figure 4.

Model 4: Hierarchical model of general negative affect, anxiety, and depression.

Second, a two-factor model (Model 2) composed of discrete anxiety and depression constructs, consistent with the relationship between anxiety and depression as typically clinically conceptualized, was tested. The hypothesis for this model was that affective, cognitive, and physiological anxiety as measured by the PAI and state and trait anxiety as measured by the STAI-Y would comprise an anxiety construct that would be correlated with a depression construct composed of scores from the BDI-II as well as the affective, cognitive, and physiological subscales of the PAI Depression scale. The hypothesized two-factor solution was a marginal fit (χ²[26] = 336.83, p < .001; CFI = 0.87; NFI = 0.86; RMSEA = 0.15; 90% CI [.14, .16]).

Finally, two different three-factor models were examined. It was hypothesized that allowing the STAI-Y State and Trait raw scores to load onto a general negative affect factor (consistent with a tripartite model conceptualization) may improve model fit as previous studies found that it may be contaminated with items that measure constructs other than anxiety. This was examined using a higher order three-factor model (Model 3) that had a general negative affect construct and separate anxiety and depression constructs. The hypothesized three-factor solution allowed STAI-Y State and Trait scores to load onto a general negative affect factor, PAI anxiety subscale scores to load onto an anxiety factor, and PAI depression subscale scores and the BDI-II scale score to load onto a depression factor. The higher order three-factor model yielded better a better fit than the single- and two-factor models (χ²[25] = 130.24, p < .001, CFI = 0.96, NFI = 0.95; RMSEA = 0.09; 90% CI [.07, .10]).

The first-order factors were highly correlated in the two-factor model and all the loadings were very high in the single-factor model, raising the issue of whether a hierarchical three-factor model might not be more appropriate. It was hypothesized that allowing all of the scales to load onto a general negative affect factor may improve model fit as all of the measures examined are significantly correlated. The hierarchical three-factor model (Model 4) had a general negative affect factor upon which all of the subscales loaded. It also had a separate anxiety construct on which the PAI anxiety subscales loaded, and a separate depression construct on which the PAI depression subscales and the BDI-II loaded. The hypothesized hierarchical three-factor solution allowed STAI-Y State and Trait scores to load only onto the general negative affect factor. The hierarchical three-factor model was the best fitting model (χ²[20] = 113.93, p < .001, CFI = 0.96, NFI = 0.95; RMSEA = 0.09; 90% CI [.08, .11]). As the two three-factor models are nested they were compared using a χ² difference test (Yung, Thissen, & McLeod, 1999). The χ² difference indicated that the hierarchical model fit the data significantly better than the higher-order model (χ²[5] = 16.31, p < .01).

Discussion

The present study sought to investigate the relation between common measures of anxiety and depression in a sample of young adults seeking psychoeducational assessment using the STAI-Y, BDI-II, and PAI. All three instruments are commonly used in psychoeducational evaluations of college-aged individuals but the psychometric relationships between them have not previously been reported in this specialized population.

The results of confirmatory factor analyses performed in this study indicate that the hierarchical three-factor model provided the best fit for our data. In this model, while all measures indexed general negative affect, a significant portion of the variance unaccounted for by general negative affect for the PAI anxiety subscales can be accounted for by an orthogonal anxiety factor; an orthogonal depression factor also accounted for significant variance in the PAI depression subscales and the BDI-II. We have labeled this separate factor general negative affect, consistent with tripartite model terminology. However, it should be noted that the three-factor model examined in this study, while conceptually similar, is not wholly consistent with the tripartite model.

There are several implications for this finding. First, all of the measures examined share significant overlap as indicated by high loadings on the general negative affect factor. No scale or subscale proved orthogonal. However, the PAI anxiety and depression subscales do appear to index anxiety and depression, respectively. The results suggest that the STAI-Y state and trait subscales measure nonspecific, negative mood rather than pure anxiety, and when using the State and Trait scale scores in psychometric assessment, it may be best to conceptualize the results as such. Both of our three-factor models were supported over the single-factor model where all of the above measures represent a single negative affect construct (Feldman, 1993) and the two-factor model (APA, 2000) with distinct anxiety and depression factors. Our findings are at odds with previous research in college samples that, using only the BDI-II and STAI, found these instruments measure separate depression and anxiety factors, respectively (Alzeghoul et al., 2001; Karagozoglu et al., 2005). However, our findings are consistent with research in nonpsychoeducational samples that found the STAI-Y is not a specific measure of anxiety but instead indexes general distress or negative affect (Andrade et al., 2001; Kennedy et al., 2001; Grös et al., 2007; Bados, Gómez-Benito, & Balaguer, 2010). Endler et al. (1992) found that the EMAS and BDI-II adequately differentiated anxiety and depression while the STAI did not, and he concluded that the STAI is not an effective measure for distinguishing between anxiety and depression. Studies that have examined individual items of the STAI using factor analyses have found that both anxiety and depression factors emerge (Bieling et al., 1998; Storch et al., 2004). This is not to say that the STAI is insensitive to measuring anxiety symptoms, but rather that the STAI appears to measure both anxiety and depressive symptoms.

An additional finding of the present study was that the BDI-II and PAI depression subscales do appear to index a depression construct once general negative affect is accounted for. The depression factor is orthogonal to both general negative affect and anxiety. Given the considerable construct validity demonstrated for the BDI-II, this indicates that the PAI depression subscales validly measure the depression construct. Moreover, the current results also indicate that the BDI-II scale score and the PAI depression subscales are separate from, but highly correlated with, the anxiety construct. This finding further supports their use as measures of a specific depression construct that is distinct from anxiety and general negative affect.

In terms of the anxiety factor, given that the PAI anxiety subscales loaded significantly onto the anxiety factor without the STAI state or trait subscales, the interpretation of significance of that result is somewhat limited. The PAI anxiety subscales appear to be measuring both general negative affect and a construct that is separate from depression, which we conceptualized as anxiety. However, it is difficult to say conclusively that the construct in question is actually anxiety as no other well-validated measure was available in the dataset to load with the PAI anxiety subscales and support that assumption. Still, the currently presented data do quantitatively define the relationship between the PAI anxiety subscales and the other measures utilized in this study.

There are a number of limitations to the present study. First, in Clark and Watson’s (1991) tripartite model, low positive affect is unique to depression and hyperarousal is specific to anxiety. It should be noted that the sample utilized in this study did not complete any measures of positive affectivity or discrete measures of hyperarousal yielding limited theoretical applicability to that model. However, the primary focus of this study was to examine the relationship between the BDI-II, STAI-Y, and PAI and how those measures may apply to a tripartite conceptualization of mood and anxiety in that population in a young adult sample seeking psychoeducational evaluation. Second, while we believe that this is a largely psychiatrically normal sample, the current findings may not apply to nonclinical young adult samples as most individuals used in this study presented for assessment of issues related to academic difficulties. However, this makes our findings of particular utility to clinicians performing psychoeducational assessments in similarly aged samples. Third, Watson and Tellegen (1985) found that adults had lower levels of reported depressive and anxiety symptoms than both student and patient samples and the present results may not generalize to older individuals. In addition, the presented factor loadings and model fit statistics may vary considerably in more psychiatrically severe samples, thus requiring cross-validation.

The current study is relevant to a large number of clinicians given the burgeoning demand for psychoeducational evaluative services and the direct applicability of our findings to that group. We also believe these results are relevant to the many researchers examining anxiety and depressive symptoms in college students given that our sample represents a significant subgroup in that population. The results suggest that while the BDI-II and PAI subscales index general negative affect, they also measure distinct depression and anxiety constructs and shed some light on the relation between these measures. One implication of the current study is that the STAI-Y may be best conceptualized as a measure of general negative affect rather than as a measure of anxiety. The results of the current study do appear to support that (1) the PAI anxiety and depression subscales are measuring both general negative affect and distinct anxiety and depression constructs, (2) some of the variance of the PAI anxiety subscales is measuring a construct that is separate from negative affect and the construct indexed by the STAI-Y, and (3) the PAI depression subscale is measuring a construct similar to what is assessed by the BDI-II.

We view the current study as a psychometrically focused evaluation of the relation between three common clinical measures of anxiety and depression. From a clinical interpretation standpoint, our results suggest a number of possibilities. First, all of the measures may be elevated in an individual experiencing either anxiety or depression as general negative affect may be elevated in both subgroups. Second, interpreting the STAI-Y as a standalone instrument for assessing anxiety symptoms runs the risk of possible false positive diagnoses of anxiety disorders in young adults seeking psychoeducational evaluation because it is also sensitive to negative affect/general distress symptoms. Third, the PAI anxiety subscales and the STAI-Y State and Trait scales would likely both be elevated in an individual with clinically significant anxiety but may diverge in an individual with predominantly depressive symptoms, where the STAI-Y scales may still be elevated even though the PAI anxiety subscales are not. Finally, the PAI depressive subscales and the BDI-II should likely both be elevated in individuals with clinically significant depression.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

Alzeghoul

E. A.

Masten

W. G.

Caldwell-Colbert

A. T.

Williams

Gadzella

Helton

J. R.

Ball

S. E.

(2001). Empirical evidence for discriminating anxiety from depression in college students. The Irish Journal of Psychology, 22(1), 73-78.

American Psychiatric Association (APA). (2000). Diagnostic and statistical manual of mental disorders: DSM-IV-TR. Washington, DC.

Andrade

Gorenstein

Vieira Filho

A. H.

Tung

T. C.

Artes

(2001). Psychometric properties of the Portuguese version of the State-Trait Anxiety Inventory applied to college students: factor analysis and relation to the Beck Depression Inventory. Brazilian Journal of Medical and Biological Research, 34(3), 367-374.

Andrew

Thorpe

G. L.

Dawson

R. F. S.

(2010). Test review of the PAI. In: Spies

R. A.

Carlson

J. F.

Geisinger

K. F.

(Eds.), The eighteenth mental measurements yearbook [Electronic version]. Retrieved from http://www.unl.edu/buros

Arbisi

P. A.

Farmer

R. F.

(2001). Review of the BDI-II. In: Plake

B. S.

Impara

J. C.

(Eds.), The fourteenth mental measurements yearbook [Electronic version]. Retrieved from http://www.unl.edu/buros

Arbuckle

J. L.

(2006). Amos (version 7.0) [Computer Program]. Chicago, IL: SPSS.

Bados

Gómez-Benito

Balaguer

(2010). The State-trait anxiety inventory, trait version: Does it really measure anxiety? Journal of Personality Assessment, 92, 560-567.

Beck

A. T.

Steer

R. A.

Brown

G. K.

(1996). Beck depression inventory (2nd ed.). San Antonio, TX: The Psychological Corporation.

Beck

A. T.

Ward

C. H.

Mendelson

Mock

Erbaugh

(1961). An inventory for measuring depression. Archives of General Psychiatry, 4, 561-571.

10.

Bentler

P. M.

(1988). Comparative fit indexes in structural models. Psychological Bulletin, 107, 238-246.

11.

Benlter

P. M.

Bonett

D. G.

(1980). Significance tests and goodness of fit in the analysis of covariance structures. Psychological Bulletin, 88, 588-606.

12.

Bieling

P. J.

Antony

M. M.

Swinson

R. P.

(1998). The state-trait anxiety inventory, Trait version: Structure and content re-examined. Behaviour Research and Therapy, 36, 777-788.

13.

Browne

M. W.

Cudeck

(1993). Alternative ways of assessing model fit. In Bollen

K. A.

Long

J. S.

(Eds.), Testing structural models. Newbury Park, CA: SAGE.

14.

Caci

Bayle

Dossios

Robert

Boyer

(2003). The Spielberger trait anxiety inventory measures more than anxiety. European Psychiatry, 18, 394-400.

15.

Clark

D. A.

Steer

R. A.

Beck

A. T.

(1994). Common and specific dimensions of self-reported anxiety and depression: Implications for the cognitive and tripartite models. Journal of Abnormal Psychology, 103, 645-654.

16.

Clark

D. A.

Watson

(1991). Tripartite model of anxiety and depression: Psychometric evidence and taxonomic implications. Journal of Abnormal Psychology, 100, 316-336.

17.

Dreger

R. M.

Katkin

E. S.

(1985). Review of the State-Trait Anxiety Inventory. In: Mitchell

J. V.

Jr. (Ed.), The ninth mental measurements yearbook [Electronic version]. Retrieved from http://www.unl.edu/buros

18.

Endler

N. S.

Cox

B. J.

Parker

J. D. A.

Bagby

R. M.

(1992). Self-reports of depression and State-Trait Anxiety: Evidence for differential assessment. Journal of Personality and Social Psychology, 63, 832-838.

19.

Endler

N. S.

Edwards

J. M.

Vitelli

(1991). Endler multidimensional anxiety scales (EMAS): Manual. Los Angeles, CA: Western Psychological Services.

20.

Endler

N. S.

Macrodimitris

S. D.

Kocovski

R. P.

(2003). Anxiety and depression: Congruent, separate or both? Journal of Applied Biobehavioral Research, 8(1), 42-60.

21.

Feldman

(1993). Distinguishing depression and anxiety in self-report: Evidence from confirmatory factor analysis on nonclinical and clinical samples. Journal of Consulting and Clinical Psychology, 6, 631-638.

22.

Grös

D. F.

Simms

L. J.

Antony

M. M.

McCabe

R. E.

(2007). Psychometric properties of the State-Trait Inventory for Cognitive and Somatic Anxiety (STICSA): Comparison to the State-Trait Anxiety Inventory (STAI). Psychological Assessment, 19, 369-381.

23.

Karagozoglu

Masten

W. G.

Baloglu

(2005). Evidence for differentiating between anxiety and depression in Turkish college students. Social Behavior and Personality, 33, 579-586.

24.

Kennedy

B. L.

Schwab

J. J.

Morris

R. L.

Beldia

(2001). Assessment of state and trait anxiety in subjects with anxiety and depressive disorders. Psychiatric Quarterly, 72, 263-276.

25.

Kessler

R. C.

Gruber

Hettema

J. M.

Hwang

Sampson

Yonkers

K. A.

(2008). Co-morbid major depression and generalized anxiety disorders in the National Comorbidity Survey follow-up. Psychological Medicine, 38, 365-374.

26.

Morey

L. C.

(1991). Personality assessment inventory professional manual. Odessa, FL: Psychological Assessment Resources.

27.

Osman

Kopper

B. A.

Barrios

Gutierrez

P. M.

Bagge

C. L.

(2004). Reliability and validity of the Beck Depression Inventory-II with adolescent psychiatric inpatients. Psychological Assessment, 16(2), 120-132.

28.

Rule

W. R.

Traver

M. D.

(1983). Test-retest reliabilities of State-Trait Anxiety Inventory in a stressful social analogue situation. Journal of Personality Assessment, 47, 276-277.

29.

Segerstrom

S. C.

Tsao

J. C. I.

Alden

L. E.

Craske

M. G.

(2000). Worry and rumination: Repetitive thoughts as a concomitant and predictor of negative mood. Cognitive Therapy & Research, 24, 671-688.

30.

Spielberger

C. D.

Gorsuch

R. L.

Lushene

Vagg

P. R.

Jacobs

G. A.

(1983). Manual for the State-Trait Anxiety Inventory (Form Y). Palo Alto, CA: Consulting Psychologist Press.

31.

Storch

E. A.

Roberti

J. W.

Roth

D. A.

(2004). Factor structure, concurrent validity, and internal consistency of the Beck Depression Inventory—Second Edition in a sample of college students. Depression and Anxiety, 19, 187-189.

32.

Watson

(2005). Rethinking the mood and anxiety disorders: A quantitative hierarchical model for DSM-V. Journal of Abnormal Psychology, 114, 522-536.

33.

Watson

Clark

L. A.

Weber

Assenheimer

J. S.

(1995). Testing a tripartite model: II. Exploring the symptom structure of anxiety and depression in student, adult, and patient samples. Journal of Abnormal Psychology, 104, 15-25.

34.

Watson

Tellegen

(1985). Toward a consensual structure of mood. Psychological Bulletin, 98, 219-235.

35.

Wolpe

(1971). Neuroticdepression: experimental analog, clinical syndromes and treatment. American Journal of Psychotherapy, 25, 362-368.

36.

Yung

Y. F.

Thissen

McLeod

L. D.

(1999). On the relationship between the higher-order factor model and the hierarchical factor model. Psychometrika, 64(2), 113-128.