Tripartite Model of Anxiety and Depression: Psychometric Evidence and Taxonomic Implications
三元模型的焦慮與憂鬱：心理測量證據與分類學意涵

The puzzle of the relation between anxiety and depression is as old as the study of the syndromes themselves. In recent times, they have been viewed as: (a) different points along the same continuum, (b) alternative manifestations of a common underlying diathesis, © heterogeneous syndromes that are associated because of shared subtypes, (d) separate phenomena, each of which may develop into the other over time, and (e) conceptually and empirically distinct phenomena (L. A. Clark, 1989). Whereas each of these viewpoints is supported by some research, the current nomenclature, the Diagnostic and Statistical Manualof Mental Disorders (rev. 3rd ed.; DSM-III-R; American Psychiatric Association, 1987) primarily reflects the categorical view (e), although certain aspects are also compatible with views © and (d). However, many researchers today feel that the evidence supporting the more dimensional views (a) and (b) is sufficiently strong that the inclusion of one or more mixed anxiety-depression diagnoses in the nomenclature must be considered. Some investigators have been most concerned with mild levels of mixed affective symptomatology-which are especially common in general medical populations (e.g., Katon & Roy-Byrne, 1989; Klerman, 1989)-whereas others have been concerned with the overlap at severe levels of psychopathology (e. g., Akiskal, 1990; Blazer et al., 1988; Blazer et al., 1989; Leckman, Merikangas, Pauls, Prusoff, & Weissman, 1983). Those espousing view (b), in particular, have noted that these disorders show longitudinal as well as cross-sectional comorbidity. That is, some patients exhibit both an anxious and a depressive syndrome but at different points in time, and various hypotheses have been offered to explain this phenomenon (e.g., Breier, Charney, & Heninger, 1984; Maser & Cloninger, 1990).
焦慮與憂鬱之間關係的謎題與這些綜合症的研究一樣古老。近來，它們被視為：(a) 同一連續體上的不同點，(b) 共同潛在易感性的一種替代表現，(c) 由於共享亞型而相關的異質綜合症，(d) 各自獨立的現象，隨著時間的推移可能相互發展，以及 (e) 概念上和實證上明顯不同的現象（L. A. Clark, 1989）。雖然這些觀點都有一些研究支持，但目前的命名法，即《精神疾病診斷與統計手冊》（修訂版第三版；DSM-III-R；美國精神醫學會，1987 年）主要反映了類別觀點 (e)，儘管某些方面也與觀點 (c) 和 (d) 兼容。然而，許多當今的研究者認為，支持更具維度觀點 (a) 和 (b) 的證據相當強大，因此必須考慮在命名法中納入一個或多個混合焦慮-憂鬱診斷。 一些研究者對輕度混合情感症狀特別關注，這在一般醫療人群中尤其常見（例如，Katon & Roy-Byrne, 1989；Klerman, 1989），而其他研究者則關注於嚴重心理病理學的重疊（例如，Akiskal, 1990；Blazer 等, 1988；Blazer 等, 1989；Leckman, Merikangas, Pauls, Prusoff, & Weissman, 1983）。特別是持有觀點 (b) 的人指出，這些疾病顯示出縱向和橫斷面的共病性。也就是說，一些患者在不同的時間點上同時表現出焦慮和抑鬱的症狀，並且提出了各種假設來解釋這一現象（例如，Breier, Charney, & Heninger, 1984；Maser & Cloninger, 1990）。

The views expressed in this article are those of the authors and do not represent the official positions of the DSM-IV Task Force of the American Psychiatric Association.
本文中表達的觀點僅代表作者個人意見，並不代表美國精神醫學會 DSM-IV 工作組的官方立場。

Correspondence concerning this article should be addressed to Lee Anna Clark or to David Watson, Department of Psychology, Southern Methodist University, Dallas, Texas 75275-0442.
有關本文的通信應寄給李安娜·克拉克或大衛·沃森，南方衛理公會大學心理學系，德克薩斯州達拉斯 75275-0442。

Therefore, we ask: To what extent do empirical research findings support the existence of one or more mixed mood disorders for inclusion in DSM-IV? We began by reviewing the psychometric data relevant to this issue, focusing on important properties of measures of anxiety and depression in both patient and nonpatient samples, including the convergent and discriminant validity of self- and clinical ratings and the interrater reliability of clinical ratings. Although most of the available rating data were static in nature (i.e., anxious and depressive phenomena were assessed at a single point in time), we considered longitudinal phenomena when possible. This review led us, in turn, to analyses of how context and scale content influence ratings, to factor-analytic data, and to an examination of the role of (low) positive affect ( PA ) in depression.
因此，我們詢問：實證研究結果在多大程度上支持將一種或多種混合情緒障礙納入 DSM-IV？我們首先回顧了與此問題相關的心理測量數據，重點關注患者和非患者樣本中焦慮和抑鬱測量的重要特性，包括自我評估和臨床評估的聚合效度和區別效度，以及臨床評估的評分者間信度。儘管大多數可用的評分數據本質上是靜態的（即焦慮和抑鬱現象在單一時間點進行評估），但我們在可能的情況下考慮了縱向現象。這一回顧使我們進一步分析了上下文和量表內容如何影響評分，進行了因素分析數據的研究，並檢視了（低）正性情感（PA）在抑鬱中的作用。

Gradually, it became clear that the data were best captured by a tripartite structure of a general distress factor and specific factors for anxiety and depression, respectively. Jointly these factors provide a framework for the development of a more satisfactory diagnostic scheme for the anxiety and depressive disorders and suggest the need for a new diagnosis of mixed anxiety-depression. Furthermore, the structure helps to explain why the various views of anxiety and depression mentioned earlier have developed and represents a framework for their synthesis. That is, studies that have focused on the shared general distress factor have tended to view anxiety and depression as points on a continuum or as having a common diathesis (views a and b), whereas those that have focused on the specific factors have concluded that they are distinct phenomena (views d and e). Obviously, if both general and specific factors exist, then a complete characterization of anxiety and depression must incorporate each of these views; it will also subsume view ©, with the substitution of “a common component” for “shared subtypes.”
逐漸地，數據顯示最佳的結構是由一個一般困擾因素以及分別針對焦慮和抑鬱的特定因素組成的三元結構。這些因素共同提供了一個框架，以發展出更令人滿意的焦慮和抑鬱症的診斷方案，並暗示需要一個新的混合焦慮-抑鬱診斷。此外，這一結構有助於解釋為何之前提到的各種焦慮和抑鬱的觀點會發展出來，並代表了它們綜合的框架。也就是說，專注於共享的一般困擾因素的研究往往將焦慮和抑鬱視為連續體上的點或具有共同的易感性（觀點 a 和 b），而專注於特定因素的研究則得出它們是不同現象的結論（觀點 d 和 e）。顯然，如果一般因素和特定因素同時存在，那麼對焦慮和抑鬱的完整描述必須納入這些觀點中的每一個；它還將包含觀點 c，將“共同成分”替換為“共享亞型”。

In this article we present the results of our review of the psychometric and related literatures, the conclusions-including a description of the tripartite structure-that we derived from this review, and some implications we see for the diagnosis of anxiety and depressive disorders.
在本文中，我們呈現了對心理測量及相關文獻的回顧結果，從這次回顧中得出的結論——包括三元結構的描述，以及我們對焦慮和抑鬱症診斷的一些看法。

One factor that contributes to the confusion in the vast literature on anxiety and depression is the multiple ways in which the terms are used. The several differentiable meanings of anxiety and depression include: normal mood states that shade into more intense or prolonged pathological mood states (e.g., panic or anhedonia), syndromes that involve covarying nonmood symptoms (e.g., autonomic arousal or vegetative signs), and specific diagnostic entities (e.g., panic disorder or melancholia; Klerman, 1980). Despite widespread awareness of these distinctions, multiple levels of meaning are often intermixed within a single report. Such terminological imprecision is problematic, because the conclusions that can be drawn about the relation between anxiety and depression are not necessarily the same across all of these levels of meaning. Our review focuses on the assessment of anxiety and depression on two levelsfirst, mood and, second, symptom cluster or syndrome-although we also examine the implications of these results for specific diagnostic entities.
在廣泛的焦慮和抑鬱文獻中，造成混淆的一個因素是這些術語的多種用法。焦慮和抑鬱的幾個可區分的含義包括：正常的情緒狀態，逐漸轉變為更強烈或持久的病理情緒狀態（例如，驚恐或無快感），涉及共變的非情緒症狀的綜合症（例如，自主神經興奮或植物性徵兆），以及特定的診斷實體（例如，驚恐障礙或憂鬱症；Klerman, 1980）。儘管對這些區別的廣泛認識，單一報告中常常混合多個層次的含義。這種術語的不精確性是有問題的，因為關於焦慮和抑鬱之間關係的結論在這些不同的含義層次上不一定相同。我們的回顧重點在於從兩個層面評估焦慮和抑鬱：首先是情緒，其次是症狀群或綜合症——儘管我們也檢視這些結果對特定診斷實體的影響。

Anxious and depressed moods represent the defining cores of their corresponding disorders, and a number of measures focus on these affects per se. Most common, however, are broader measures that also assess symptoms associated with anxious and depressed mood. These measures have diverse origins and intents that range from rationally derived scales that are intended to assess defined clinical syndromes to psychometrically developed scales designed to assess empirically derived clusters of symptoms. Regardless of origin, however-and despite the fact that they assess a diverse range of symptomsthese measures all tend to be homogeneous (internal consistency reliability coefficients are typically .80 or higher), and their scores are continuously distributed. Because of these properties, these scales are typically scored dimensionally. Nevertheless, cutoff scores on these measures are sometimes used to delineate the mere presence or absence of anxiety or depressive syndromes. Fortunately, this questionable practice has waned since the advent of the DSM- $III$$III$IIII I I (American Psychiatric Association, 1980) with its specific diagnostic criteria and because various writers have pointed out the pitfalls of such usage (e.g., Beck, Steer, & Garbin, 1988; Kendall, Hollon, Beck, Hammen, & Ingram, 1987). Therefore, our review will be limited to reports that have used dimensional scoring.
焦慮和抑鬱情緒代表了其相應疾病的定義核心，許多測量工具專注於這些情緒本身。然而，最常見的是更廣泛的測量工具，這些工具還評估與焦慮和抑鬱情緒相關的症狀。這些測量工具的來源和目的各異，範圍從理性推導的量表，旨在評估明確的臨床綜合症，到心理測量學上發展的量表，旨在評估經驗上推導的症狀群。無論來源如何——儘管它們評估的症狀範圍多樣——這些測量工具通常都趨於同質性（內部一致性信度係數通常為 0.80 或更高），其分數呈連續分佈。由於這些特性，這些量表通常以維度方式進行評分。然而，這些測量工具的截止分數有時用於劃分焦慮或抑鬱綜合症的存在或缺失。 幸運的是，自從 DSM- $III$$III$IIII I I （美國精神醫學會，1980 年）的出現，這種可疑的做法已經減少，因為各種作者指出了這種用法的陷阱（例如，Beck、Steer 和 Garbin，1988 年；Kendall、Hollon、Beck、Hammen 和 Ingram，1987 年）。因此，我們的回顧將限於使用維度評分的報告。

In determining whether anxiety and depression represent different aspects of one continuum or instead exist as discrete phenomena, evaluation of their discriminant validity is obviously indispensable. However, discriminant (i.e., between-affects) comparisons cannot be interpreted meaningfully outside the context of the convergent (i.e., within-affects) validity of measures of each affect or syndrome separately, which, in the case of clinical ratings, must also include evidence of interrater reliability Therefore, our review encompasses all of these basic properties. We examine self-report and clinical ratings both separately and in relation to each other; similarly, mood and syndrome measures are examined both separately and in relation to one another. We focus on the most widely used measures, but other measures are referenced when appropriate. Finally, results are reported separately for patient and nonpatient samples whenever possible. ${}^{\text{?}}$${}^{\text{? }}$^("? "){ }^{\text {? }}
在判斷焦慮和抑鬱是否代表同一連續體的不同方面，或是作為離散現象存在時，評估它們的區別效度顯然是不可或缺的。然而，區別（即情感之間）比較在沒有各自情感或綜合症的收測量的收斂（即情感內部）效度的背景下無法有意義地解釋，而在臨床評分的情況下，這也必須包括評估者之間可靠性的證據。因此，我們的回顧涵蓋了所有這些基本特性。我們分別檢視自我報告和臨床評分，並將它們相互關聯；同樣，情緒和綜合症的測量也分別檢視並相互關聯。我們專注於最廣泛使用的測量工具，但在適當的情況下也會提及其他測量工具。最後，結果在可能的情況下分別報告患者和非患者樣本。

The most commonly used measures of anxious and depressed mood are scales from the Profile of Mood States (POMS; McNair, Lorr, & Droppleman, 1971) and the Multiple Affect Adjective Check List (MAACL; Zuckerman & Lubin, 1965) or its recent revision (Zuckerman & Lubin, 1985; we will use MAACL to refer to both the original and revised forms). On the basis of extensive factor analyses, we recently developed the Positive and Negative Affect Schedule-Expanded Form (PANAS-X; Watson & Clark, 1990), which contains specific affect scales for fear (anxiety) and sadness (depression), and we report data on these scales also.
最常用的焦慮和抑鬱情緒測量工具是情緒狀態概況量表（POMS；McNair, Lorr, & Droppleman, 1971）和多重情感形容詞檢查表（MAACL；Zuckerman & Lubin, 1965）或其最近的修訂版（Zuckerman & Lubin, 1985；我們將使用 MAACL 來指代原始和修訂形式）。基於廣泛的因素分析，我們最近開發了擴展版的正負情感量表（PANAS-X；Watson & Clark, 1990），該量表包含針對恐懼（焦慮）和悲傷（抑鬱）的特定情感量表，我們也報告了這些量表的數據。

Validity. The convergent and discriminant validity correlations between the respective POMS and MAACL scales are shown in Tablel. Although the convergence in three nonpatient samples was moderately high, it was unacceptably low in the one patient sample available (Zuckerman & Lubin, 1985). Moreover, for both types of subjects, the discriminant coefficients within each measure were higher than the convergent coefficients across the measures (significantly so in the patient sample). These results obviously do not form acceptable convergent and discriminant validity patterns; the data for anxiety are especially problematic. Thus, these data demonstrate that the MAACL and POMS are not measuring two distinct affects in the same way.
有效性。各自的 POMS 和 MAACL 量表之間的收斂和區別效度相關性如表 1 所示。儘管在三個非病人樣本中的收斂性中等偏高，但在唯一的病人樣本中則低得不可接受（Zuckerman & Lubin, 1985）。此外，對於兩種類型的受試者，每個量表內的區別係數均高於各量表之間的收斂係數（在病人樣本中顯著如此）。這些結果顯然不形成可接受的收斂和區別效度模式；焦慮的數據尤其成問題。因此，這些數據顯示 MAACL 和 POMS 並未以相同方式測量兩種不同的情感。

This discouraging pattern is at least partially due to important differences in their rating formats (checklist vs. 5 -point rating scale). This hypothesis is supported by the fact that using the same rating format and time frame, we have found a more acceptable convergent and discriminant validity pattern ( 85 vs . 66) between the POMS and PANAS-X scales in a sample of 563 college students (Watson & Clark, 1990), although it must also be noted that these two instruments have some terms in common. Given that these two types of measures do not converge well, which provides the more trustworthy data?
這種令人沮喪的模式至少部分是由於它們的評分格式（檢查表與五點評分量表）之間的重要差異。這一假設得到了支持，因為在使用相同的評分格式和時間範圍時，我們在 563 名大學生的樣本中發現了 POMS 和 PANAS-X 量表之間更可接受的收斂和區別效度模式（85 對 66）（Watson & Clark，1990），儘管也必須指出這兩種工具有一些共同的術語。考慮到這兩種類型的測量不太收斂，哪一種提供了更可靠的數據？

Several lines of evidence suggest that the rating scales yield somewhat more valid results than the MAACL. First, there are
幾條證據表明，評分量表產生的結果比 MAACL 更有效。首先，有

Table 1 表格 1
Convergent and Discriminant Validity Correlations for Two Measures of Self-Rated Depressed and Anxious Mood in Patient and Nonpatient Samples
收斂效度和區別效度在患者和非患者樣本中自評抑鬱和焦慮情緒的兩個測量之間的相關性

Note. MAACL $=$$=$== Multiple Affect Adjective Check List; and POMS $=$$=$== Profile of Mood States. Correlations in nonpatient samples are shown in the lower half of the correlation matrix, and those in patient samples, in the upper half. Sample sizes for convergent (across-instruments) correlations, shown in boldface, are 270 and 123 in nonpatient and patient samples, respectively. Sample sizes for discriminant (within-instruments) correlations, shown in italics, range from 90 to $2,524(Mdn=933)$$2,524(Mdn=933)$2,524(Mdn=933)2,524(M d n=933).
注意。MAACL $=$$=$== 多重情感形容詞檢查表；以及 POMS $=$$=$== 情緒狀態概況。非病人樣本的相關性顯示在相關矩陣的下半部分，病人樣本的相關性顯示在上半部分。收斂（跨工具）相關性的樣本大小，以粗體顯示，分別為非病人樣本的 270 和病人樣本的 123。區別（內部工具）相關性的樣本大小，以斜體顯示，範圍從 90 到 $2,524(Mdn=933)$$2,524(Mdn=933)$2,524(Mdn=933)2,524(M d n=933) 。
serious psychometric problems associated with the use of a checklist format (Fogel, Curtis, Kordasz, & Smith, 1966; Herron, 1969). Second, in a combined patient sample (Fogel et al., 1966; Zuckerman & Lubin, 1985), the convergent and discriminant validity pattern of the MAACL scales with singleitem self-ratings of anxious and depressed mood was also relatively poor ( 51 vs. .45 for the average convergent and discriminant validity correlations, respectively). Third, the convergent and discriminant validity patterns with syndromal measures of anxiety and depression (shown in Table 2) are somewhat better for the rating scales than for the MAACL in both nonpatient and patient samples. Specifically, the average convergent correlation for the MAACL ( $r=.55$$r=.55$r=.55r=.55 ) is significantly lower than that for the POMS $(r=.77)$$(r=.77)$(r=.77)(r=.77) and the PANAS-X $(r=.67)$$(r=.67)$(r=.67)(r=.67). Moreover, whereas the POMS also has a significantly higher average discriminant correlation than the MAACL ( $r\mathrm{s}=.68\mathrm{vs}.49$$r\mathrm{s}=.68\mathrm{vs}.49$rs=.68vs.49r \mathrm{~s}=.68 \mathrm{vs} .49, respectively), the PANAS-X does not ( $r=.52$$r=.52$r=.52r=.52 ). Thus, in terms of the
與使用檢查表格式相關的嚴重心理測量問題（Fogel, Curtis, Kordasz, & Smith, 1966; Herron, 1969）。其次，在一個合併的病人樣本中（Fogel et al., 1966; Zuckerman & Lubin, 1985），MAACL 量表與單項自評焦慮和抑鬱情緒的收斂和區別效度模式也相對較差（平均收斂和區別效度相關性分別為 51 和 0.45）。第三，與焦慮和抑鬱的綜合測量（見表 2）相比，對於評分量表的收斂和區別效度模式在非病人和病人樣本中均優於 MAACL。具體而言，MAACL 的平均收斂相關性（ $r=.55$$r=.55$r=.55r=.55 ）顯著低於 POMS（ $(r=.77)$$(r=.77)$(r=.77)(r=.77) ）和 PANAS-X（ $(r=.67)$$(r=.67)$(r=.67)(r=.67) ）。此外，雖然 POMS 的平均區別相關性也顯著高於 MAACL（ $r\mathrm{s}=.68\mathrm{vs}.49$$r\mathrm{s}=.68\mathrm{vs}.49$rs=.68vs.49r \mathrm{~s}=.68 \mathrm{vs} .49 ），但 PANAS-X 則沒有（ $r=.52$$r=.52$r=.52r=.52 ）。因此，在這方面的
squared multiple correlation difference between the average convergent and discriminant correlations, the PANAS-X showed the greatest difference (.18), followed by the POMS (. 15 in patient and .06 in nonpatient samples) and then by the MAACL ( 12 in patient and .04 in nonpatient samples). It must be noted, however, that although the PANAS-X scales appear to have promise as measures of anxious and depressed mood, they have not yet been tested in patient samples.
平方多重相關性在平均收斂和區辨相關性之間的差異中，PANAS-X 顯示出最大的差異（0.18），其次是 POMS（在患者樣本中為 0.15，在非患者樣本中為 0.06），然後是 MAACL（在患者樣本中為 0.12，在非患者樣本中為 0.04）。然而，必須指出的是，儘管 PANAS-X 量表似乎在測量焦慮和抑鬱情緒方面具有潛力，但尚未在患者樣本中進行測試。

Summary and conclusions. We draw a number of conclusions from these data. First, the MAACL scales do not yield discriminable measures of depressed and anxious mood and are probably not the best available measures for assessing these specific affects. In contrast, scales with a Likert rating format (POMS, PANAS-X) have acceptable convergent validity, both with each other and with syndromal measures of depression and anxiety. Although the level of convergence is not so high as to suggest that these scales could substitute for syndromal measures, they likely yield valid assessments of their core mood states. However, even with valid factor-analytically derived scales, the overlap between anxious and depressed mood is substantial, which indicates that these basic affects are at best only partially differentiable. This overlap is somewhat stronger in patient than nonpatient samples, probably, in part, because of the infrequent occurrence of intense negative moods in normal subjects.
摘要與結論。我們從這些數據中得出幾個結論。首先，MAACL 量表無法提供可區分的抑鬱和焦慮情緒測量，可能不是評估這些特定情感的最佳可用測量工具。相對而言，使用李克特評分格式的量表（POMS、PANAS-X）在彼此之間以及與抑鬱和焦慮的綜合測量之間具有可接受的收斂效度。儘管收斂程度並不高到可以替代綜合測量，但它們可能提供對其核心情緒狀態的有效評估。然而，即使是有效的因素分析衍生量表，焦慮和抑鬱情緒之間的重疊也相當顯著，這表明這些基本情感至多只能部分區分。在患者樣本中，這種重疊比非患者樣本更強，這可能部分是因為正常受試者中強烈負面情緒的發生頻率較低。

In seeking to explain this overlap, some may argue that it reflects a simple, probabilistic co-occurrence between etiologically independent moods. However, accumulating data suggest that it instead represents a shared general negative affect (NA) component that is an inherent and important aspect of each mood state (Watson & Clark, 1991). In this regard, it is important to emphasize that this shared variance remains strong even in ratings of current, momentary (i.e., state) mood (Mayer & Gaschke, 1988; Watson, 1988b; Watson & Tellegen, 1985). Moreover, this nonspecific NA encompasses not only anxiety and depression, but other negative mood states as well, so that simi-
在尋求解釋這種重疊時，有些人可能會認為這反映了病因獨立的情緒之間的簡單概率共現。然而，累積的數據表明，這實際上代表了一種共享的普遍負面情感（NA）成分，這是每種情緒狀態固有且重要的方面（Watson & Clark, 1991）。在這方面，重要的是要強調，即使在對當前瞬時（即狀態）情緒的評價中，這種共享變異仍然很強（Mayer & Gaschke, 1988；Watson, 1988b；Watson & Tellegen, 1985）。此外，這種非特定的 NA 不僅包括焦慮和抑鬱，還包括其他負面情緒狀態，因此類似的情況也存在。

Table 2 表 2
Convergent and Discriminant Validity Correlations Between Self-Rated Depressed and Anxious Mood and Syndromes in Patient and Nonpatient Samples
自評抑鬱和焦慮情緒及症狀在患者和非患者樣本中的聚合效度和區別效度相關性

Note. Convergent correlations are shown in boldface. The Sadness and Fear scales of the Positive and Negative Affect Schedule-Expanded form measure depression and anxiety, respectively.
注意。收斂相關以粗體顯示。正負情感量表擴展版的悲傷和恐懼量表分別測量抑鬱和焦慮。
lar data can be compiled for other negative emotions, such as anger and guilt (Gotlib, 1984; Watson & Clark, 1984, 1991; Watson & Tellegen, 1985). To be sure, affect-specific variance can also be identified (Watson & Clark, 1990, 1991, in press-a), but the general component in negative mood scales is invariably substantial.
大型數據可以用於編制其他負面情緒的資料，例如憤怒和內疚（Gotlib, 1984；Watson & Clark, 1984, 1991；Watson & Tellegen, 1985）。確實，情感特定的變異也可以被識別（Watson & Clark, 1990, 1991，待出版），但負面情緒量表中的一般成分始終是相當可觀的。

We believe that explicit recognition of the fact that negative moods are only partially discriminable will increase the validity of anxiety and depressive diagnostic criteria. In later sections we discuss aspects of anxiety and depressive syndromesand alternative strategies for mood and symptom assessmentthat enhance their differentiation.
我們相信，明確承認負面情緒僅部分可辨識的事實將提高焦慮和抑鬱診斷標準的有效性。在後面的部分中，我們將討論焦慮和抑鬱綜合症的各個方面以及增強其區分的情緒和症狀評估的替代策略。

The most widely used self-report measures of anxious and depressive symptomatology include: the Beck Depression Inventory (BDI; Beck, Ward, Mendelson, Mock, & Erbaugh, 1961) and the Beck Anxiety Inventory (BAI; Beck, Epstein, Brown, & Steer, 1988); the Symptom Checklist-90 (SCL-90; Derogatis, Lipman, & Covi, 1973) Depression and Anxiety scales; scales scored from the item pool of the Minnesota Multiphasic Personality Inventory (MMPI; Hathaway & McKinley, 1943), such as the Taylor Manifest Anxiety Scale (Taylor, 1953) for anxiety and the MMPI Scale 2 for depression; the Self-Rating Depression Scale (SDS) and the Self-Rating Anxiety Scale (SAS; Zung, 1965, 1971); Costello-Comrey Anxiety Scale and Costello-Comrey Depression Scale (CC-A and CC-D; Costello & Comrey, 1967); State-Trait Anxiety Inventory (STAI; Spielberger, Gorsuch, & Lushene, 1970); Institute for Personality and Ability Testing Anxiety Scale Questionnaire (Krug, Scheier, & Cattell, 1976); and the Center for Epidemiological Studies Depression Scale (Radloff, 1977). The Inventory to Diagnose Depression (Zimmerman & Coryell, 1987) and Inventory for Depressive Symptomatology, which is available in both self- and clinician-rated formats (Rush et al., 1986), have appeared more recently.
最廣泛使用的焦慮和抑鬱症狀自我報告量表包括：貝克抑鬱量表（BDI；貝克、沃德、門德爾森、莫克和厄巴赫，1961 年）和貝克焦慮量表（BAI；貝克、艾普斯坦、布朗和斯蒂爾，1988 年）；症狀檢查表-90（SCL-90；德羅加提斯、利普曼和科維，1973 年）中的抑鬱和焦慮量表；來自明尼蘇達多相人格測驗（MMPI；哈索威和麥金利，1943 年）題庫的量表，如泰勒顯性焦慮量表（Taylor，1953 年）用於焦慮和 MMPI 第 2 量表用於抑鬱；自評抑鬱量表（SDS）和自評焦慮量表（SAS；宗，1965 年，1971 年）；科斯特羅-科姆雷焦慮量表和科斯特羅-科姆雷抑鬱量表（CC-A 和 CC-D；科斯特羅和科姆雷，1967 年）；狀態-特質焦慮量表（STAI；斯皮爾伯格、戈爾薩赫和盧申，1970 年）；人格與能力測試研究所焦慮量表問卷（克魯格、謝爾和卡特爾，1976 年）；以及流行病學研究中心抑鬱量表（拉德洛夫，1977 年）。 抑鬱症診斷清單（Zimmerman & Coryell, 1987）和抑鬱症狀清單，均有自評和臨床評估格式（Rush et al., 1986），最近出現。

It is important to note that these scales (and their clinicianrated counterparts) typically assess what may be called modal anxiety and depressive symptomatology as they focus on the core aspects of each syndrome type rather than on all possible variants. Depression scales primarily target symptoms of nonpsychotic, nonmelancholic depression; melancholic symptoms (e.g., diurnal variation) are sometimes included, but more severe psychotic symptoms (e.g., delusions of guilt) are rarely assessed. Similarly, anxiety symptom scales typically target generalized anxiety and panic attacks and rarely include more than a few items to target obsessive-compulsive disorder, social or simple phobias, or postraumatic stress disorder (although specific scales have been developed to assess some of these disorders). We use the terms syndromal anxiety and syndromal depression in this article to describe this modal scale content, but it must be recognized that item content varies even among the most widely used scales; this content heterogeneity is an issue we discuss later.
重要的是要注意，這些量表（及其臨床評估的對應量表）通常評估的可以稱為模式焦慮和抑鬱症狀，因為它們專注於每種綜合症類型的核心方面，而不是所有可能的變體。抑鬱量表主要針對非精神病性、非憂鬱性抑鬱的症狀；憂鬱症狀（例如，日間變化）有時會被納入，但更嚴重的精神病症狀（例如，罪惡妄想）則很少被評估。同樣，焦慮症狀量表通常針對廣泛性焦慮和驚恐發作，並且很少包括超過幾個項目來針對強迫症、社交或簡單恐懼症或創傷後壓力症候群（儘管已經開發了特定的量表來評估其中一些疾病）。在本文中，我們使用綜合症焦慮和綜合症抑鬱這些術語來描述這種模式量表內容，但必須認識到，即使在最廣泛使用的量表中，項目內容也存在變異；這種內容異質性是我們稍後討論的問題。

Validity: A large number of studies have examined the convergent and discriminant validity patterns of self-report measures of syndromal anxiety and depression. These data are summarized in the first two columns in Table 3. The average con-
有效性：大量研究已檢視自我報告的症狀性焦慮和抑鬱量表的聚合效度和區別效度模式。這些數據在表 3 的前兩列中進行了總結。平均共-

Table 3 表 3
Convergent and Discriminant Validities for Syndromal Measures of Depression and Anxiety by Self- and Clinical Raters in Patient and Nonpatient Samples
抑鬱和焦慮的綜合效度與區別效度：由自評者和臨床評估者在患者和非患者樣本中的綜合量表

${}^2$${}^2$^(2){ }^{2} From Watson and Clark (1984). This is median of 9 anxiety-negative affect measures, which were not subdivided by sample type. ${}^{\text{b}}$${}^{\text{b }}$^("b "){ }^{\text {b }} This does not include Minnesota Multiphasic Personality Inventory data on 50,000 medical patients ( $r=.61;$$r=.61;$r=.61;r=.61 ; Swenson, Pearson, & Osborne, 1973); see text for further information. ${}^{\mathrm{c}}$${}^{\mathrm{c}}$^(c){ }^{\mathrm{c}} From Eaton and Ritter (1988; $n=$$n=$n=n= 2,768 community adults).
${}^2$${}^2$^(2){ }^{2} 來自 Watson 和 Clark (1984)。這是 9 個焦慮-負面情感測量的中位數，未按樣本類型細分。 ${}^{\text{b}}$${}^{\text{b }}$^("b "){ }^{\text {b }} 這不包括對 50,000 名醫療病人的明尼蘇達多相人格測驗數據 ( $r=.61;$$r=.61;$r=.61;r=.61 ; Swenson, Pearson, & Osborne, 1973)；詳情請參見文本。 ${}^{\mathrm{c}}$${}^{\mathrm{c}}$^(c){ }^{\mathrm{c}} 來自 Eaton 和 Ritter (1988; $n=$$n=$n=n= 2,768 名社區成年人)。
vergent correlation among five measures of depressive symptomatology (BDI, MMPI Scale 2, SCL-90 Depression scale, CC-D, and SDS) is in the low. 70 s , with no difference due to sample type. Three figures are given for measures of anxious symptomatology, the median convergent coefficient from nine scales examined in Watson and Clark’s (1984) review (which did not distinguish between sample types), and two average values -calculated separately for nonpatient and patient samplesfrom subsequently published studies that covered five measures of anxiety (SAS, CC-A, Taylor Manifest Anxiety Scale, STAITrait, and the Institute for Personality and Ability Testing Anxiety Scale Questionnaire). On the whole it appears that self-report measures of anxiety may show somewhat greater convergence than those for depression, especially in patient samples, but clearly the convergent validity of both syndromes is well-established. ${}^2$${}^2$^(2){ }^{2} This high degree of convergence indicates, in part, that the various scales are targeting the same construct; indeed, scales for each syndrome contain many common items (e.g,
五種抑鬱症狀量表（BDI、MMPI 第 2 量表、SCL-90 抑鬱量表、CC-D 和 SDS）之間的趨同相關性在低 70s，且樣本類型之間沒有差異。對於焦慮症狀量表，提供了三個數據，來自 Watson 和 Clark（1984）回顧中檢查的九個量表的中位數趨同係數（未區分樣本類型），以及兩個平均值——分別針對非病人和病人樣本計算，這些樣本來自隨後發表的研究，涵蓋了五種焦慮量表（SAS、CC-A、Taylor 顯性焦慮量表、STAI 特質和人格與能力測試研究所焦慮量表問卷）。總的來看，自我報告的焦慮量表似乎顯示出比抑鬱量表更大的趨同，特別是在病人樣本中，但顯然這兩種綜合症的趨同效度都是確立的。這種高度的趨同程度部分表明，各種量表針對的是相同的構念；事實上，每種綜合症的量表都包含許多共同項目（例如，

Gotlib & Cane, 1989; Kavan, Pace, Ponterotto, & Barone, 1990).
Gotlib 和 Cane, 1989; Kavan, Pace, Ponterotto, 和 Barone, 1990).

Turning to the issue of discriminant validity, however, one again finds disturbingly high correlations. When paired anxiety and depression scales (i.e., two scales from a single instrument, such as the SCL-90, the Beck, Zung, or Costello-Comrey scales, or the MMPI) are compared, the overall correlations between them are .66 and .70 in patient and nonpatient samples, respectively. When scales from different instruments are compared, the values are only slightly lower ( $r=.64$$r=.64$r=.64r=.64 and .62 , respectively). This pattern is quite similar to that observed with the pure mood scales: Whereas diverse self-report measures of anxiety and depression yield strongly convergent assessments of their respective syndromes, there is little specificity in their measurement, especially in nonpatient samples. Rather, the data suggest the presence of a large nonspecific component that is shared by both syndromes.
然而，轉向判別效度的問題時，再次發現令人擔憂的高相關性。當配對的焦慮和抑鬱量表（即來自單一工具的兩個量表，如 SCL-90、Beck、Zung 或 Costello-Comrey 量表，或 MMPI）進行比較時，患者和非患者樣本之間的整體相關性分別為 .66 和 .70。當比較來自不同工具的量表時，這些值僅略低（分別為 $r=.64$$r=.64$r=.64r=.64 和 .62）。這一模式與純情緒量表的觀察結果非常相似：雖然多樣的自我報告焦慮和抑鬱測量提供了對各自綜合症的強烈收斂評估，但在其測量中幾乎沒有特異性，特別是在非患者樣本中。相反，數據表明這兩種綜合症之間存在一個大型的非特異性成分。

Scale-level analyses. One concern with summary correlations is that they may mask significant differences among measures. That is, some measures may show strong convergent and discriminant validity patterns that are overwhelmed by data from less well-constructed scales. Therefore, we examined the correlational patterns for each of the well-established measures separately. The results are shown in Table 4, and several aspects of the table deserve comment. First, most of the correlations are based on only one or two studies and, therefore, are not definitive. Second, broadly speaking, measures with higher convergent validity typically have higher discriminant coefficients as well. This covariation suggests that some measures are
量表層級分析。摘要相關性的一個擔憂是，它們可能掩蓋了測量之間的顯著差異。也就是說，一些測量可能顯示出強烈的收斂效度和區別效度模式，但卻被來自較差構建量表的數據所淹沒。因此，我們分別檢查了每個已建立的測量的相關模式。結果顯示在表 4 中，該表的幾個方面值得評論。首先，大多數相關性僅基於一到兩項研究，因此並不具決定性。其次，廣義而言，具有較高收斂效度的測量通常也具有較高的區別係數。這種共變異表明某些測量是
more highly loaded with the nonspecific distress factor than others. Such measures provide a highly valid assessment of generalized distress but are not particularly useful for discriminating anxious from depressive syndromes.
比其他人更高度負載非特定痛苦因素。這些措施提供了對一般性痛苦的高度有效評估，但對於區分焦慮症和抑鬱症綜合徵並不是特別有用。

Third, the Beck inventories and the Costello-Comrey scales -both of which used factor-analytic techniques in the development of one or both scales–appear to offer the best convergent and discriminant validity patterns, although cautionary notes are warranted in both cases. The BAI is new and has not yet been studied much by researchers other than its creators. Similarly, data for the CC-A and CC-D are sparse. Finally, although the convergent and discriminant validity patterns are the best for these scales, it still must be acknowledged that their discriminant correlations average approximately .56. As was discussed earlier with regard to mood, it has been argued that this overlap simply reflects the co-occurrence of etiologically distinct syndromes. Again however, a more compelling explanation is that a nonspecific distress factor forms an inherent core component of both syndromes. This nonspecific distress factor has been identified repeatedly by many researchers and has been given many labels (e.g., neuroticism, general maladjustment, or negative emotionality). We have chosen to call it negative affectivity or trait NA (Watson & Clark, 1984) because of its close association with the general, higher order mood dimension. At this point, a brief digression into the nature of PA and NA is necessary.
第三，貝克量表和科斯特羅-科梅瑞量表——這兩者在開發過程中都使用了因素分析技術——似乎提供了最佳的收斂效度和區別效度模式，儘管在這兩種情況下都需要謹慎對待。BAI 是新的，除了其創建者之外，尚未受到研究者的廣泛研究。同樣，CC-A 和 CC-D 的數據也很稀少。最後，儘管這些量表的收斂和區別效度模式是最佳的，但仍必須承認它們的區別相關性平均約為 0.56。正如之前討論的情緒問題，有人認為這種重疊僅僅反映了病因上不同的綜合症的共現。然而，更有說服力的解釋是，一種非特異性痛苦因素構成了這兩種綜合症的內在核心組成部分。這種非特異性痛苦因素已被許多研究者反復識別，並被賦予了許多標籤（例如，神經質、一般適應不良或負面情緒性）。 我們選擇將其稱為負向情感性或特質 NA（Watson & Clark, 1984），因為它與一般的高階情緒維度有著密切的關聯。在這一點上，有必要簡要探討 PA 和 NA 的本質。

Positive and negative affect. Recent research has produced strong evidence that PA and NA are the dominant dimensions in self-reported mood, both in the United States and in other cultures (Diener, Larsen, Levine, & Emmons, 1985; Stone,
正向和負向情感。最近的研究提供了強有力的證據，表明正向情感（PA）和負向情感（NA）是自我報告情緒的主要維度，無論是在美國還是在其他文化中（Diener, Larsen, Levine, & Emmons, 1985; Stone,

Table 4 表 4
Convergent and Discriminant Validities for Self-Rated Anxiety and Depression Measures in Patient and Nonpatient Samples
自評焦慮和抑鬱量表在患者和非患者樣本中的收斂效度和區別效度

Note. The numbers of studies and subjects ( $N$$N$NN ) may not apply to every figure in a row; sample sizes for convergent validity are typically higher than for discriminant validity. Beck = Beck Depression Inventory and Anxiety Inventory; Costello-Comrey = Costello-Comrey Depression Scale and Anxiety Scale; MMPI = Minnesota Multiphasic Personality Inventory; SDS-SAS = Self-Rating Depression Scale and Self-Rating Anxiety Scale; SCL-90 = Symptom Check List-90 Depression and Anxiety scales; STAI = State-Trait Anxiety Inventory; TMAS = Taylor Manifest Anxiety Scale (an MMPI-based scale).
注意。研究和受試者的數量（ $N$$N$NN ）可能不適用於每一行的每個數字；收斂效度的樣本量通常高於區別效度。Beck = 貝克抑鬱量表和焦慮量表；Costello-Comrey = Costello-Comrey 抑鬱量表和焦慮量表；MMPI = 明尼蘇達多相人格測驗；SDS-SAS = 自評抑鬱量表和自評焦慮量表；SCL-90 = 症狀檢查表-90 抑鬱和焦慮量表；STAI = 狀態-特質焦慮量表；TMAS = 泰勒顯性焦慮量表（基於 MMPI 的量表）。

1981; Tellegen, 1985; Watson, Clark, & Tellegen, 1984; Watson & Tellegen, 1985; Zevon & Tellegen, 1982). Briefly, NA represents the extent to which a person is feeling upset or unpleasantly engaged rather than peaceful and encompasses various aversive states including upset, angry, guilty; afraid, sad, scornful, disgusted, and worried; such states as calm and relaxed best represent the lack of NA. In contrast, PA reflects the extent to which a person feels a zest for life and is most clearly defined by such expressions of energy and pleasurable engagement as $ac$$ac$aca c tive, delighted, interested, enthusiastic, and proud; the absence of PA is best captured by terms that reflect fatigue and languor (e.g., tired or sluggish).
1981；Tellegen，1985；Watson、Clark 和 Tellegen，1984；Watson 和 Tellegen，1985；Zevon 和 Tellegen，1982）。簡而言之，負向情緒（NA）代表一個人感到不安或不愉快的程度，而不是平靜，並涵蓋各種厭惡的狀態，包括不安、生氣、內疚；害怕、悲傷、輕蔑、厭惡和擔心；如冷靜和放鬆等狀態最能代表缺乏負向情緒。相對而言，正向情緒（PA）反映一個人對生活的熱情程度，最明確的定義是能量和愉快參與的表達，如積極、愉快、感興趣、熱情和自豪；缺乏正向情緒最能用反映疲勞和無精打采的術語來捕捉（例如，疲倦或遲鈍）。

Despite their opposite-sounding labels, these two mood dimensions are largely independent of one another, and they have distinctive correlational patterns with other variables. Briefly, only PA is related to diverse measures of social activity, exercise, and reports of pleasant events, whereas NA alone is correlated with health complaints, perceived stress, and unpleasant events (L. A. Clark & Watson, 1988, 1989; Watson, 1988a; Watson & Pennebaker, 1989). Furthermore, PA (and depressive phenom-ena)-but not NA (or anxiety)-has been linked to the body’s circadian cycle (L. A. Clark, Watson, & Leeka, 1989; Healy & Williams, 1988; Thayer, 1987) and to seasonal variations (Kasper & Rosenthal, 1989; Smith, 1979). Finally, the two mood dimensions are differentially related to two major personality traits: As mentioned earlier, state NA is associated with measures of trait NA or neuroticism (Costa & McCrae, 1980; Eysenck & Eysenck, 1968, 1975; Tellegen, 1985; Watson & Clark, 1984), whereas state PA is correlated with measures of positive affectivity (trait PA; Tellegen, 1985) or extraversion (Costa & McCrae, 1980; Eysenck & Eysenck, 1968,1975). Persons high in trait PA are cheerful, enthusiastic, and vigorous; but in addition to this core mood component, they also tend to be socially masterful, to be forceful leaders who enjoy being the center of attention, and to be achievement oriented (Tellegen, 1985; Watson & Clark, in press-b).
儘管這兩個情緒維度的標籤聽起來相反，但它們在很大程度上是相互獨立的，並且與其他變數有著獨特的相關模式。簡而言之，只有正向情緒（PA）與社交活動、運動和愉快事件的多種測量相關，而負向情緒（NA）則僅與健康問題、感知壓力和不愉快事件相關（L. A. Clark & Watson, 1988, 1989; Watson, 1988a; Watson & Pennebaker, 1989）。此外，正向情緒（及抑鬱現象）與身體的晝夜節律有關（L. A. Clark, Watson, & Leeka, 1989; Healy & Williams, 1988; Thayer, 1987），而負向情緒（或焦慮）則沒有這種關聯，並且與季節變化有關（Kasper & Rosenthal, 1989; Smith, 1979）。最後，這兩個情緒維度與兩個主要的人格特質的關係不同：如前所述，狀態負向情緒與特質負向情緒或神經質的測量相關（Costa & McCrae, 1980; Eysenck & Eysenck, 1968, 1975; Tellegen, 1985; Watson & Clark, 1984），而狀態正向情緒則與正向情感性（特質正向情緒；Tellegen, 1985）或外向性（Costa & McCrae, 1980; Eysenck & Eysenck, 1968, 1975）相關。 特質 PA 高的人通常是快樂、熱情和充滿活力的；除了這一核心情緒成分外，他們還傾向於在社交上非常擅長，成為強勢的領導者，喜歡成為注意的焦點，並且以成就為導向（Tellegen, 1985；Watson & Clark, 即將出版-b）。

Although investigation of the shared factor in anxiety and depression-that is, general NA-will increase our understanding of important aspects of these syndromes, their differentiation will depend on the identification of distinctive (i.e., specific) factors that they do not have in common. Several converging lines of evidence suggest that an important specific factor that marks depression is the absence of PA. For example, Tellegen (1985) factor analyzed self-report measures of NA, PA, anxiety, and depression. The resulting two-factor solution indicated that anxiety was more highly associated with the NA factor, whereas depression was a better marker of low PA.
儘管對焦慮和抑鬱之間的共同因素——即一般負情緒（NA）的研究將增進我們對這些綜合症重要方面的理解，但它們的區分將依賴於識別它們不共有的獨特（即特定）因素。幾條相互印證的證據表明，抑鬱的一個重要特定因素是缺乏正情緒（PA）。例如，Tellegen（1985）對負情緒、正情緒、焦慮和抑鬱的自我報告量表進行了因素分析。結果的兩因素解決方案顯示，焦慮與負情緒因素的關聯性更高，而抑鬱則是低正情緒的更好標誌。

Similarly, and apparently without knowledge of Tellegen’s (1985) work, a team of British researchers developed the Hospital Anxiety and Depression Scale (HADS; Zigmond & Snaith, 1983), in which the depressive items primarily assess positive affectivity (e.g., " 1 look forward with enjoyment to things"), whereas the anxiety items are typical of self-reported anxiety symptom scales (e.g., “I feel tense or wound up”). As shown in Table 3 (last correlation, second column), the average correlation between the HADS scales across two patient samples (Aylard, Gooding, McKenna, & Snaith, 1987; Bramley, Easton, Morley, & Snaith, 1988) was .11. This clearly represents better discriminant validity than is typically seen and lends support
同樣地，顯然在不知 Tellegen（1985）工作的情況下，一組英國研究人員開發了醫院焦慮與抑鬱量表（HADS；Zigmond & Snaith，1983），其中抑鬱項目主要評估正向情感（例如，“我期待著愉快的事情”），而焦慮項目則是典型的自我報告焦慮症狀量表（例如，“我感到緊張或焦慮”）。如表 3 所示（最後的相關性，第二列），在兩個患者樣本之間（Aylard, Gooding, McKenna, & Snaith, 1987；Bramley, Easton, Morley, & Snaith, 1988），HADS 量表的平均相關性為 0.11。這顯然代表了比通常所見的更好的區別效度，並提供了支持。
to the notion that low PA plays an important role in distinguishing depression from anxiety. Later we discuss other evidence in regard to the specific role of low PA in depression and also identify a specific anxiety factor. Before doing so, however, we summarize the self-report findings and examine whether the patterns observed with self-report measures can also be seen in clinical ratings.
關於低正情緒在區分抑鬱與焦慮中的重要角色的觀念。稍後我們將討論其他證據，涉及低正情緒在抑鬱中的具體角色，並確定一個特定的焦慮因素。然而，在此之前，我們將總結自我報告的發現，並檢視自我報告量表中觀察到的模式是否也能在臨床評分中看到。

Summary and conclusions. Self-report symptom measures of anxious and depressive symptomatology show substantial convergent validity. Depression measures show little, if any, difference in the level of convergence between patient and nonpatient samples. Anxiety measures may display somewhat greater convergence in patient samples, but further data are needed to establish this effect firmly.
摘要與結論。自我報告的焦慮和抑鬱症狀量表顯示出相當的收斂效度。抑鬱量表在患者和非患者樣本之間的收斂程度幾乎沒有差異。焦慮量表在患者樣本中可能顯示出稍微更高的收斂性，但需要進一步的數據來確定這一效果。

For discriminant validity, however, the data are less encouraging, with average discriminant correlations in the range from .62 to 70 (see Table 3). Nevertheless, the squared multiple correlation difference between convergent and discriminant coefficients averaged approximately .13 and .17 in nonpatient and patient samples, respectively (cf. the $.15-18$$.15-18$.15-18.15-18 squared multiple correlation difference reported for mood rating scales). Moreover, two scale pairs-the Beck inventories and the CostelloComrey scales-showed better convergent and discriminant validity patterns than did other sets of measures. However, the discriminant correlations are still substantial, and each can benefit from more research.
然而，就區別效度而言，數據的表現不太令人鼓舞，平均區別相關性在 0.62 到 0.70 之間（見表 3）。儘管如此，非病人和病人樣本中，收斂和區別係數之間的平方多重相關差異分別平均約為 0.13 和 0.17（參見 $.15-18$$.15-18$.15-18.15-18 所報告的情緒評分量表的平方多重相關差異）。此外，兩組量表——貝克量表和科斯特羅-科梅里量表——顯示出比其他測量組更好的收斂和區別效度模式。然而，區別相關性仍然相當可觀，每個量表都可以從更多的研究中受益。

As with the mood data, these results suggest that a strong nonspecific distress factor-which we interpret as state NA in mood ratings and as trait NA syndromally-dominates self-ratings of anxious and depressive symptomatology and may account for most of their overlap. Again, we believe that this substantial nonspecific component is an important and inseparable part of these syndromes and ought to be explicitly acknowledged in the official diagnostic system. Finally, theoretical and empirical advances in mood and personality suggest the importance of a second major factor, namely, PA, in differentiating depression and anxiety. Specifically, depression-but not anxiety-is associated with low PA, and inclusion of PA-related items in depression scales may enhance their discriminant validity.
與情緒數據相似，這些結果表明一個強烈的非特異性痛苦因素——我們將其解釋為情緒評分中的狀態負面情緒（state NA）以及作為特質負面情緒（trait NA）綜合症——主導了自我評估的焦慮和抑鬱症狀，並可能解釋了它們大部分的重疊。我們再次認為，這一實質性的非特異性成分是這些綜合症的重要且不可分割的部分，應在官方診斷系統中明確承認。最後，情緒和人格的理論及實證進展表明，第二個主要因素，即正面情緒（PA），在區分抑鬱和焦慮方面的重要性。具體而言，抑鬱——而非焦慮——與低正面情緒相關，將與正面情緒相關的項目納入抑鬱量表可能增強其區分效度。

Interrater reliability: Clinical ratings of mood are typically based on 1-3 items embedded in broader syndrome rating scales (rather than existing as independent measures) and are therefore rarely reported separately. However, we located six studies that reported the interrater reliability of clinically rated mood, and their results suggest that the conditions under which mood is assessed are critically important. Specifically, five studies used either joint interviews or separate structured interviews with heterogeneous patient samples, and their average interrater reliabilities were .67 for both depressed and anxious mood. In contrast, poor reliability was obtained for both depressed $(r=.37)$$(r=.37)$(r=.37)(r=.37) and anxious $(r=.19)$$(r=.19)$(r=.19)(r=.19) mood in one study that used independent, unstructured clinical interviews with a homogeneous sample (Cicchetti & Prusoff, 1983). That these poor
評分者間信度：臨床情緒評分通常基於嵌入在更廣泛綜合症評分量表中的 1-3 項目（而不是作為獨立的測量工具），因此很少單獨報告。然而，我們找到六項研究報告了臨床評定情緒的評分者間信度，結果顯示情緒評估的條件至關重要。具體而言，五項研究使用了聯合訪談或對異質患者樣本的分開結構化訪談，其抑鬱和焦慮情緒的平均評分者間信度均為 0.67。相比之下，在一項使用獨立、非結構化臨床訪談的同質樣本的研究中，抑鬱 $(r=.37)$$(r=.37)$(r=.37)(r=.37) 和焦慮 $(r=.19)$$(r=.19)$(r=.19)(r=.19) 情緒的信度較差（Cicchetti & Prusoff, 1983）。這些差的信度
reliabilities were not simply a function of inadequate measures or ill-trained raters is supported by the fact that the reliability coefficients rose to .72 and .40 , respectively, when the study population evidenced a greater range of moods after 16 weeks of treatment.
可靠性並不僅僅是由於測量不足或評分者訓練不當所造成的，這一點得到了支持，因為當研究人群在接受 16 週治療後顯示出更大的情緒範圍時，可靠性係數分別上升至 0.72 和 0.40。

We know of only one study in which the interrater reliability of others’ ratings of mood in normal subjects has been investigated (Watson & Clark, 1991). Ratings were made on scales from the PANAS-X (Watson & Clark, 1990) by nonprofessional peers solely on the basis of acquaintance, without benefit of interview or training. As in Cicchetti and Prusoff’s (1983) study, the pairwise correlation between any two judges was fairly low ( $r=.19$$r=.19$r=.19r=.19 to .37 ). However, when the data of 4 raters were aggregated, moderate ( 49 and .58 for the Sadness, or depression, and Fear, or anxiety, scales, respectively) to high (.70 for PA) reliabilities were obtained. ${}^3$${}^3$^(3){ }^{3}
我們只知道一項研究調查了正常受試者情緒評分的評分者間信度（Watson & Clark, 1991）。評分是由非專業的同儕根據相識程度，僅使用 PANAS-X（Watson & Clark, 1990）上的量表進行的，沒有進行面試或培訓。與 Cicchetti 和 Prusoff（1983）的研究類似，任何兩位評分者之間的配對相關性相當低（ $r=.19$$r=.19$r=.19r=.19 到.37）。然而，當 4 位評分者的數據被聚合時，獲得了中等（悲傷或抑鬱量表為 49 和.58）到高（正情緒為.70）的信度。 ${}^3$${}^3$^(3){ }^{3}

Together, these data suggest that mood can be reliably rated under appropriate conditions (i.e., adequate range of subject moods, use of joint or standardized interviews, or use of aggregated multiple ratings). However, we found no data to indicate whether the relatively small mood variations seen among highly distressed patients (e.g., at intake) can be rated reliably.
這些數據共同表明，在適當的條件下（即，受試者情緒的足夠範圍、使用聯合或標準化的訪談，或使用聚合的多重評分），情緒可以被可靠地評估。然而，我們未找到數據顯示在高度痛苦的患者中（例如，在入院時）所見的相對較小的情緒變化是否可以可靠地評估。

Validity. We found no studies that used more than one clinical measure to assess patients’ moods, but several have reported convergent correlations between mood and global ratings or syndrome measures. First, Maier, Buller, Philipp, and Heuser (1988) found a convergent correlation of 65 between ratings of anxious mood and global ratings of anxious symptomatology (on the Covi Anxiety scale; Lipman, 1982) in two patient samples. Unfortunately, discriminant validity was not examined. Correlations between single-item ratings of depressed and anxious mood with total scale score on the Hamilton Rating Scale for Depression (HRSD; Hamilton, 1960) have also been reported in heterogeneous patient populations (Hamilton, 1967; Mowbray, 1972). The (part-whole) convergent correlations (i.e., depressed mood with total HRSD score) were. 59 and .78 , respectively, whereas the discriminant correlations (anxious mood with total HRSD score) were .25 and .60 . Although there is a clear level difference in both types of coefficients across the two studies, it is interesting to note that the squared multiple correlation difference between the convergent and discriminant correlations was virtually the same ( $29\mathrm{vs}..25$$29\mathrm{vs}..25$29vs..2529 \mathrm{vs} . .25 ). This similarity suggests that Mowbray’s (1972) ratings contained a larger (and more typical) nonspecific component than did Hamilton’s (1967) ratings but that the pattern of correlations was otherwise comparable.
有效性。我們未找到使用多於一項臨床指標來評估患者情緒的研究，但幾項研究報告了情緒與整體評分或綜合症狀指標之間的趨同相關性。首先，Maier、Buller、Philipp 和 Heuser（1988）在兩個患者樣本中發現焦慮情緒評分與焦慮症狀的整體評分（在 Covi 焦慮量表上；Lipman，1982）之間的趨同相關性為 65。不幸的是，未檢查區別效度。還有報告在異質患者群體中，抑鬱和焦慮情緒的單項評分與漢密爾頓抑鬱評分量表（HRSD；Hamilton，1960）的總分之間的相關性（Hamilton，1967；Mowbray，1972）。（部分-整體）趨同相關性（即抑鬱情緒與 HRSD 總分）分別為.59 和.78，而區別相關性（焦慮情緒與 HRSD 總分）則為.25 和.60。 儘管在這兩項研究中兩種類型的係數存在明顯的水平差異，但有趣的是，收斂和區辨相關之間的平方多重相關差異幾乎是相同的（ $29\mathrm{vs}..25$$29\mathrm{vs}..25$29vs..2529 \mathrm{vs} . .25 ）。這種相似性表明，Mowbray（1972）的評分包含了比 Hamilton（1967）的評分更大（且更典型）的非特定成分，但相關的模式在其他方面是可比的。

Corroborating this hypothesis, the discriminant coefficients for depressed and anxious mood per se were strikingly different in the two studies: Hamilton (1967) found no relation between depressed and anxious mood ( $r=.01;N=272$$r=.01;N=272$r=.01;N=272r=.01 ; N=272 ), whereas Mowbray (1972) reported a more typical correlation of $43(N=347)$$43(N=347)$43(N=347)43(N=347). We cannot explain this discrepancy except to say that it is our impression that scales’ creators typically find better discrimination than do others. However, it is not clear if this enhanced discrimination results because the authors are capable of using their scales more sensitively than others or if it is somehow artifactual. As the Hamilton scales are widely used, it will be possible to investigate the discriminant validity of clinical mood ratings in larger samples.
支持這一假設的是，兩項研究中抑鬱和焦慮情緒的判別係數截然不同：漢密爾頓（1967）發現抑鬱和焦慮情緒之間沒有關係（ $r=.01;N=272$$r=.01;N=272$r=.01;N=272r=.01 ; N=272 ），而莫布雷（1972）報告了一個更典型的相關性（ $43(N=347)$$43(N=347)$43(N=347)43(N=347) ）。我們無法解釋這一差異，除了說我們的印象是，量表的創建者通常能夠找到比其他人更好的區分。然而，目前尚不清楚這種增強的區分是否是因為作者能夠比其他人更敏感地使用他們的量表，還是某種人為因素造成的。由於漢密爾頓量表被廣泛使用，因此可以在更大樣本中調查臨床情緒評分的判別效度。

Correlations among peer ratings of mood in normal subjects also suggest the presence of a strong general factor (Watson & Clark, 1991). Whereas the discriminant correlations between PA and the negative moods of Sadness (depression) and Fear (anxiety) were appropriately low ( $r\mathrm{s}=-.33$$r\mathrm{s}=-.33$rs=-.33r \mathrm{~s}=-.33 and -.22 , respectively), ratings of these two negative moods were strongly correlated $(r=.65)$$(r=.65)$(r=.65)(r=.65). Taken as a whole, the data suggest that clinical raters generally agree with regard to the presence or absence of anxious and depressed moods. As with self-ratings, however, these data also show evidence of a strong general distress factor.
正常受試者之間的情緒同儕評分相關性也暗示了存在一個強大的一般因素（Watson & Clark, 1991）。而正向情緒（PA）與悲傷（抑鬱）和恐懼（焦慮）這兩種負面情緒之間的區別相關性適當地較低（ $r\mathrm{s}=-.33$$r\mathrm{s}=-.33$rs=-.33r \mathrm{~s}=-.33 和 -.22，分別），但這兩種負面情緒的評分卻高度相關 $(r=.65)$$(r=.65)$(r=.65)(r=.65) 。整體而言，數據顯示臨床評估者在焦慮和抑鬱情緒的存在或缺失上通常達成一致。然而，與自我評估一樣，這些數據也顯示出強烈的一般困擾因素的證據。

The most widely used clinician-based symptom or syndrome measures of anxiety and depression include: the aforementioned HRSD and its counterpart, the Hamilton Rating Scale for Anxiety (HRSA; Hamilton, 1959), for which alternative scoring methods have recently been developed (Riskind, Beck, Brown, & Steer, 1987); the anxiety and depression subscales of the Schedule for Affective Disorder and Schizophrenia (Endicott & Spitzer, 1978); and the Covi Anxiety and Raskin Depression scales (Lipman, 1982). In addition, the Clinical Anxiety Scale (a modification of the HRSA; Snaith, Baugh, Clayden, Husain, & Sipple, 1982) and the Montgomery-Asberg Depression Rating Scale (Montgomery & Asberg, 1979) have been used in a number of British studies. Finally, as mentioned earlier, the clinician-rated Inventory for Depressive Symptomatology (Rush et al., 1986) was recently developed.
最廣泛使用的臨床醫生基於症狀或綜合徵的焦慮和抑鬱測量工具包括：前述的哈密爾頓抑鬱量表（HRSD）及其對應的哈密爾頓焦慮評分量表（HRSA；哈密爾頓，1959），最近已開發出替代計分方法（Riskind, Beck, Brown, & Steer, 1987）；情感障礙和精神分裂症的日程表中的焦慮和抑鬱子量表（Endicott & Spitzer, 1978）；以及科維焦慮量表和拉斯金抑鬱量表（Lipman, 1982）。此外，臨床焦慮量表（HRSA 的修改版；Snaith, Baugh, Clayden, Husain, & Sipple, 1982）和蒙哥馬利-阿斯伯格抑鬱評分量表（Montgomery & Asberg, 1979）在多項英國研究中被使用。最後，如前所述，臨床醫生評定的抑鬱症狀清單（Rush et al., 1986）最近也被開發出來。

Interrater reliability. As with clinical mood ratings, the interrater reliability of clinical symptom ratings appears to be strongly influenced by the conditions of data collection. Higher reliabilities have been found
評估者間信度。與臨床情緒評分類似，臨床症狀評分的評估者間信度似乎受到數據收集條件的強烈影響。已發現更高的信度。
when ratings are made on heterogeneous populations by highly trained interviewers with similar backgrounds, and are based on exactly the same information (joint interviews, live observation, videotapes, and audiotapes . . . ). If any of these conditions are altered, reliabilities suffer predictably. (L. A. Clark, 1989, p. 90)
當由背景相似的高訓練訪談者對異質人群進行評分，並且基於完全相同的信息（聯合訪談、現場觀察、錄影帶和錄音帶……）時。如果這些條件中的任何一個被改變，可靠性會可預測地下降。（L. A. Clark, 1989, p. 90）

Reliabilities in one study in which none of these conditions were met (Cicchetti & Prusoff, 1983) ranged down to 46 . Interestingly, sample type appears to be less important than the range of symptomatology in the sample (i.e., higher reliability is obtained with greater range). Furthermore, specific depression symptom measures (e.g., HRSD) are slightly more reliable than global measures of depression (interrater $rs=.85\mathrm{vs}..78$$rs=.85\mathrm{vs}..78$rs=.85vs..78r s=.85 \mathrm{vs} . .78 ), but both are affected by the same parameters. Unfortunately, suffcient data do not exist to examine this issue for anxiety ratings nor to determine whether other structured depression ratings also show consistently higher reliabilities than do global ratings.
在一項未滿足這些條件的研究中（Cicchetti & Prusoff, 1983），可靠性低至 46。有趣的是，樣本類型似乎不如樣本中症狀範圍重要（即，範圍越大，可靠性越高）。此外，特定的抑鬱症狀測量（例如，HRSD）比抑鬱的整體測量（評分者間 $rs=.85\mathrm{vs}..78$$rs=.85\mathrm{vs}..78$rs=.85vs..78r s=.85 \mathrm{vs} . .78 ）稍微更可靠，但兩者都受到相同參數的影響。不幸的是，沒有足夠的數據來檢查焦慮評分的這一問題，也無法確定其他結構化的抑鬱評分是否也顯示出比整體評分更一致的高可靠性。
L. A. Clark’s (1989) review also revealed greater variability in the reliability of anxiety symptom ratings. In eight studies ( $N=$$N=$N=N= 538) that examined the reliability of clinical ratings of anxious symptomatology, coefficients ranged from .26 to .95 , with a mean of 76 . On closer inspection, however, it appears that this
L. A. Clark（1989）的評論也顯示出焦慮症狀評分的可靠性變異性更大。在八項研究（ $N=$$N=$N=N= 538）中，檢視了焦慮症狀的臨床評分可靠性，係數範圍從 0.26 到 0.95，平均為 0.76。然而，仔細檢查後，似乎這個

variability reflects the fact that studies of anxious symptomatology have been performed under widely varying conditions. Specifically, the average reliability of five studies that used joint interviews and well-defined criteria was 84 , whereas that of four that used either separate interviews, no specific criteria, or both was $.47.{}^4$$.47.{}^4$.47.^(4).47 .{ }^{4} Thus, under optimal conditions, ratings of anxious and depressive symptomatology show similarly high reliabilities, whereas under less favorable circumstances, the same low reliabilities are seen.
變異性反映了焦慮症狀學研究是在廣泛變化的條件下進行的事實。具體而言，五項使用聯合訪談和明確標準的研究的平均可靠性為 84，而四項使用分開訪談、沒有具體標準或兩者皆有的研究則為 $.47.{}^4$$.47.{}^4$.47.^(4).47 .{ }^{4} 。因此，在最佳條件下，焦慮和抑鬱症狀的評分顯示出同樣高的可靠性，而在不太有利的情況下，則顯示出相同的低可靠性。

Convergent validity. In contrast to the large number of studies that have reported correlations among self-reported symptom measures, relatively few studies have examined the convergent validity of clinical rating scales, and none have used nonpatient samples. Available data are summarized in the last column of Table 3. Convergence among clinical ratings of depressive symptomatology was uniformly high (average correlation of .83). Most of the studies compared the HRSD to other measures, but good convergence was also found among other scales in one study (Deluty, Deluty, & Carver, 1986). Further research needs to be undertaken to confirm this finding with other scales and also with the new scoring system for the HRSD developed by Riskind et al. (1987).
收斂效度。與大量報告自我報告症狀測量之間相關性的研究相比，檢驗臨床評分量表的收斂效度的研究相對較少，且沒有使用非病人樣本。可用數據在表 3 的最後一欄中進行了總結。抑鬱症狀的臨床評分之間的收斂性普遍較高（平均相關係數為 0.83）。大多數研究將 HRSD 與其他測量進行比較，但在一項研究中（Deluty, Deluty, & Carver, 1986）也發現其他量表之間有良好的收斂性。需要進一步研究以確認這一發現，並與 Riskind 等人（1987）開發的 HRSD 新評分系統進行比較。

Only a few studies have investigated the convergent validity of anxiety symptom rating scales, but they have yielded an average convergence correlation of 74 . Although this is an acceptably high level of convergence, it is nevertheless significantly lower than that obtained for depressive symptoms. A brief content analysis of commonly used anxiety symptom scales indicates that this lower convergence may occur because the various measures have somewhat different foci. That is, as mentioned earlier, the relative assessment weight assigned to the various facets of anxiety (e.g., general anxious mood, cognitive worry, physical tension, symptoms of autonomic hyperarousal, other somatic symptoms, and even specific fears) varies considerably across scales. Therefore, more precise information about anxiety symptom ratings might be obtained if specific scales were developed for each of these facets. It is noteworthy, however, that patients’ self-ratings of anxious symptomatologywhich are similarly varied in content-are significantly more convergent than the clinical ratings. Thus, clinicians may be more sensitive to the heterogeneous nature of anxiety symptoms than are patients and, accordingly, make more differentiated ratings than do patients, who may instead generalize their overall level of subjective distress across a wide range of specific symptoms. We make a similar point with regard to depressive ratings later.
只有少數研究探討了焦慮症狀評分量表的收斂效度，但它們的平均收斂相關性為 74。雖然這是一個可接受的高收斂水平，但仍然顯著低於抑鬱症狀的收斂水平。對常用焦慮症狀量表的簡要內容分析表明，這種較低的收斂可能是因為各種量表的重點有所不同。也就是說，如前所述，對焦慮的各個方面（例如，普遍焦慮情緒、認知擔憂、身體緊張、自主神經過度興奮的症狀、其他躯體症狀，甚至特定恐懼）所分配的相對評估權重在各量表之間差異顯著。因此，如果為這些方面中的每一個開發特定的量表，可能會獲得更精確的焦慮症狀評分信息。然而，值得注意的是，患者對焦慮症狀的自我評分——其內容同樣多樣——的收斂性顯著高於臨床評分。 因此，臨床醫生可能對焦慮症狀的異質性比患者更敏感，因此，他們的評分可能比患者更具區別性，而患者則可能將其整體主觀痛苦的程度概括到一系列具體症狀中。我們稍後在抑鬱評分方面也提出了類似的觀點。

Discriminant validity. Several studies have examined the discriminant validity of depressive and anxious symptom rating scales (again, see Table 3, last column). In three studies that used the original Hamilton scales ( $N=191)$$N=191)$N=191)N=191), the average discriminant correlation was quite high $(r=.77)$$(r=.77)$(r=.77)(r=.77), in part, because of item overlap. In contrast, an average discriminant correlation of .39 was obtained either with the revised Hamilton scales or with other clinician-rated anxiety and depressive symptom measures. This figure is quite close to that $(r=.43)$$(r=.43)$(r=.43)(r=.43) found between anxiety and depression rating scales developed from the Diagnostic Interview Schedule (Robins, Helzer, Croughan, & Ratcliff, 1981) for use in a large ( $N=2,768$$N=2,768$N=2,768N=2,768 ) community sample (Eaton & Ritter, 1988). Thus, a discriminant coefficient of ap-
區別效度。幾項研究檢視了抑鬱和焦慮症狀評分量表的區別效度（再次參見表 3，最後一欄）。在三項使用原始漢密爾頓量表的研究中，平均區別相關性相當高，部分原因是項目重疊。相對而言，使用修訂版漢密爾頓量表或其他臨床評估的焦慮和抑鬱症狀測量時，獲得的平均區別相關性為 0.39。這一數字與從診斷訪談日程（Robins, Helzer, Croughan, & Ratcliff, 1981）開發的焦慮和抑鬱評分量表之間的相關性相當接近，該量表用於一個大型社區樣本（Eaton & Ritter, 1988）。因此，區別係數為 ap-
proximately $.40-.45$$.40-.45$.40-.45.40-.45 appears to represent a reasonable estimate of the correlation between clinical ratings of anxious and depressive symptomatology in both patient and nonpatient samples. Although this still represents substantial overlap, it is clearly a more acceptable level of discriminant validity than has been obtained with self-ratings.
大約 $.40-.45$$.40-.45$.40-.45.40-.45 似乎代表了患者和非患者樣本中焦慮和抑鬱症狀的臨床評分之間相關性的合理估計。儘管這仍然代表了相當大的重疊，但顯然這比自我評分所獲得的區別效度水平更為可接受。

We noted before, however, that when self-report depression measures contained items that reflected low PA, the discrimination between anxiety and depression was even sharper, and it is noteworthy that this phenomenon is replicated in clinical ratings (mean $r=.11$$r=.11$r=.11r=.11; see Table 3, last row of correlations). In two studies (Aylard et al., 1987; Bramley et al., 1988), the Clinical Anxiety Scale and the Montgomery-Asberg Depression Rating Scale have been used. In a third study Vye (1986) used global measures of depressive and anxious symptoms, but it is clear from Vye’s description that low PA (especially the lack of interest or pleasure) played a major role in the conceptualization of depression. It is important to note that with the exception of the Montgomery-Asberg Depression Rating Scale, the clinical rating scales used in these studies were not themselves atypical. Recall also that the British research apparently was conceived independently of Tellegen’s (1985) model. Thus, the lower discriminant correlations resulted not so much from using unusual scales as from the distinctive way in which these clinicians interpreted the scale items. We examine the validity of this alternative conceptualization later, but first we summarize the findings for clinical ratings of anxious and depressive symptomatology.
我們之前提到，當自我報告的抑鬱量表包含反映低正情緒（PA）的項目時，焦慮與抑鬱之間的區分更加明顯，值得注意的是，這一現象在臨床評分中也得到了重複（平均 $r=.11$$r=.11$r=.11r=.11 ；見表 3，相關性最後一行）。在兩項研究中（Aylard 等，1987；Bramley 等，1988），使用了臨床焦慮量表和 Montgomery-Asberg 抑鬱評定量表。在第三項研究中，Vye（1986）使用了抑鬱和焦慮症狀的全球測量，但從 Vye 的描述中可以清楚看出，低正情緒（特別是缺乏興趣或快樂）在抑鬱的概念化中扮演了重要角色。值得注意的是，除了 Montgomery-Asberg 抑鬱評定量表外，這些研究中使用的臨床評分量表本身並不異常。還要回想一下，英國的研究顯然是獨立於 Tellegen（1985）的模型構思的。因此，較低的區別相關性並不是因為使用了不尋常的量表，而是因為這些臨床醫生對量表項目的解釋方式具有獨特性。 我們稍後將檢視這種替代概念化的有效性，但首先我們總結有關焦慮和抑鬱症狀的臨床評分結果。

Summary and conclusions. Clinical ratings of syndromal anxiety and depression have good interrater reliability and are highly convergent within affect when (a) the raters are similarly and adequately trained, (b) the rating criteria are clearly specified, © the ratings are based on the same information, and (d) there is adequate within-sample variability. Clinical ratings of mood are affected by similar considerations; they are somewhat less reliable than syndromal ratings because mood is typically measured with only 1-3 items. Sample type per se does not appear to affect the reliability of ratings, but data that pertains to possible effects on convergent validity are lacking.
摘要與結論。臨床對症焦慮和抑鬱的評分具有良好的評分者間可靠性，並且在情感上高度一致，當（a）評分者接受了相似且充分的訓練，（b）評分標準明確規定，（c）評分基於相同的信息，以及（d）樣本內變異性足夠時。情緒的臨床評分受到類似考量的影響；由於情緒通常僅用 1-3 個項目來測量，因此其可靠性略低於對症評分。樣本類型本身似乎不影響評分的可靠性，但缺乏與可能影響一致性效度相關的數據。

The reliability of anxiety and depressive symptom ratings are similar, and the convergent validity coefficients for both are acceptable, but clinical ratings of anxiety symptoms are somewhat less convergent than those for depression ( $s=.83\mathrm{vs}.74$$s=.83\mathrm{vs}.74$s=.83vs.74s=.83 \mathrm{vs} .74, respectively). Although further studies, especially ones that examine the various facets of anxiety, are needed to establish the validity of anxiety syndrome scales definitively, the data so far obtained suggest that good convergence can be expected.
焦慮和抑鬱症狀評分的可靠性相似，兩者的聚合效度係數均可接受，但焦慮症狀的臨床評分與抑鬱症狀的聚合性稍微較低（ $s=.83\mathrm{vs}.74$$s=.83\mathrm{vs}.74$s=.83vs.74s=.83 \mathrm{vs} .74 ，分別）。儘管仍需進一步研究，特別是檢視焦慮的各個面向，以確定焦慮綜合症量表的效度，但迄今為止獲得的數據顯示，良好的聚合性是可以期待的。

Whereas clinical ratings of the two moods or syndromes overlap substantially, a greater level of discrimination, in comparison with self-ratings, is clear nevertheless. This enhanced discrimination suggests that when clinicians make ratings, they give more weight to specific factors that distinguish anxiety from depression than do patients. However, although clinical ratings are typically used as a standard against which to judge self-reports, the relative validity (e.g., the clinical utility) of the
儘管兩種情緒或綜合症的臨床評分有很大重疊，但與自我評分相比，區分的程度仍然明顯更高。這種增強的區分表明，當臨床醫生進行評分時，他們對區分焦慮和抑鬱的特定因素給予了比患者更多的重視。然而，儘管臨床評分通常被用作評判自我報告的標準，但其相對有效性（例如，臨床實用性）仍然存在。

two types of judgments has not been systematically compared. Thus, it must not be assumed a priori that increased differentiation is necessarily valid or desirable. Greater clinical differentiation may stem from the fact that clinicians are prepared to see (if not force) differences, perhaps by virtue of their training. If a revised diagnostic system were to recognize the existence of mixed anxiety-depression, it would not be surprising if clinicians subsequently viewed these symptoms more similarly to the way patients now report them. ${}^5$${}^5$^(5){ }^{5}
兩種類型的判斷尚未進行系統性的比較。因此，不能先入為主地假設增加的區分必然是有效或可取的。更大的臨床區分可能源於臨床醫生準備去看（如果不是強迫）差異，也許是因為他們的訓練。如果修訂的診斷系統能夠認可混合焦慮-抑鬱的存在，那麼臨床醫生隨後以更相似的方式看待這些症狀，與患者目前的報告方式並不令人驚訝。

As in self-reports, positive and negative mood states are relatively independent in peer ratings. In addition, it appears that if clinicians conceptualize depression as having a substantial component of low PA (even if they do not explicitly use this terminology), they rate it as clearly distinctive from anxiety. Again, however, the validity of this approach requires further examination. We discuss both of these validity issues, but first we examine the convergent and discriminant validity of self-reports versus clinical ratings.
在自我報告中，正面和負面情緒狀態在同儕評價中相對獨立。此外，似乎如果臨床醫生將抑鬱症概念化為具有相當程度的低正情緒（即使他們不明確使用這一術語），他們會將其評價為與焦慮明顯不同。然而，這種方法的有效性仍需進一步檢驗。我們討論這兩個有效性問題，但首先我們檢視自我報告與臨床評價的收斂效度和區別效度。

Validity. Data relevant to the convergence between self- and clinically rated anxious and depressed mood exist widely; unfortunately, however, they are usually embedded in broader syndromal measures and are, therefore, seldom reported. Available data are summarized in the top half of Table 5 . With one exception (Fogel et al., 1966, who used the MAACL, the validity of which, as we have shown, is questionable), the convergent and discriminant correlations covaried, both across studies and during retesting within the same study. That is, replicating the pattern observed within self-report measures, higher convergent correlations were generally accompanied by higher discriminant correlations (cf. Table 4). As we have also seen before, the distribution of these correlations was bimodal: In three studies with patient samples, moderately high convergence was accompanied by correspondingly high discriminant coefficients (these are labeled good convergence in Table 5), whereas three others reported both poor convergence and low discriminant correlations (poor convergence in Table 5). Remarkably, however, the squared multiple correlation difference between the convergent and discriminant coefficients was virtually identical in both instances.
有效性。與自我評估和臨床評估的焦慮和抑鬱情緒之間的收斂相關的數據廣泛存在；然而，不幸的是，它們通常嵌入在更廣泛的綜合測量中，因此很少被報告。可用數據在表 5 的上半部分進行了總結。除了有一個例外（Fogel 等，1966 年，他們使用的 MAACL，其有效性如我們所示，存在疑問），收斂和區別相關性在研究之間以及在同一研究的重測中都呈現共變。也就是說，重複自我報告測量中觀察到的模式，較高的收斂相關性通常伴隨著較高的區別相關性（參見表 4）。正如我們之前所見，這些相關性的分佈是雙峰的：在三項以患者樣本為基礎的研究中，中等高的收斂伴隨著相應高的區別係數（在表 5 中標記為良好收斂），而另外三項則報告了低收斂和低區別相關性（在表 5 中標記為差收斂）。 然而，值得注意的是，收斂和區辨係數之間的平方多重相關差異在兩種情況下幾乎是相同的。

As before, the studies that obtained poor convergence used homogeneous samples, questionable measures, or both. One of the studies that obtained poor convergence used items from the SDS, in which the frequency of symptoms rather than their severity is rated, as the source of the mood self-ratings and used the Hamilton scales for the clinical ratings (Carroll, Fielding, & Blashki, 1973). This format difference may have led to poor convergence. Unfortunately, discriminant validity was not reported in this study. In contrast, the studies with good convergence used reliable measures, standardized rating systems, and heterogeneous patient samples.
如前所述，獲得較差收斂的研究使用了同質樣本、可疑的測量工具，或兩者兼而有之。其中一項獲得較差收斂的研究使用了 SDS 中的項目，該項目評估的是症狀的頻率而非其嚴重程度，作為情緒自評的來源，並使用漢密爾頓量表進行臨床評分（Carroll, Fielding, & Blashki, 1973）。這種格式差異可能導致了較差的收斂。不幸的是，這項研究並未報告區別效度。相對而言，收斂良好的研究使用了可靠的測量工具、標準化的評分系統和異質的患者樣本。

Correlations between self- and others’ ratings of mood have also been reported in nonpatient samples, for which peers or spouses rather than clinicians served as judges (Costa & McCrae, 1988; Watson & Clark, 1991; Zuckerman & Lubin,
自我與他人情緒評價之間的相關性在非病人樣本中也有報導，其中同儕或配偶而非臨床醫生擔任評判者（Costa & McCrae, 1988; Watson & Clark, 1991; Zuckerman & Lubin）
1985). The average convergent correlation for anxiety in these studies (Table 5 , line 3 ) was virtually the same as that obtained in the good convergence patient samples (Table 5, line 1). In contrast, the convergent coefficient for depression and the discriminant correlation were both somewhat lower. This pattern probably results, in part, from the relatively low mean levels of these affects (especially depression) in nonpatient samples. Nonetheless, the squared multiple correlation difference between the convergent and discriminant correlations (.10) was twice as great as that found in the patient samples.
1985 年）。這些研究中焦慮的平均收斂相關性（表 5，第 3 行）幾乎與良好收斂患者樣本中獲得的相同（表 5，第 1 行）。相比之下，抑鬱的收斂係數和區別相關性均稍低。這種模式可能部分是由於非患者樣本中這些情感（特別是抑鬱）的平均水平相對較低。然而，收斂和區別相關性之間的平方多重相關性差異（0.10）是患者樣本中發現的兩倍。

Finally, two studies-one with patient (Vye, 1986) and one with nonpatient subjects (Watson & Clark, 1991)-examined the convergent and discriminant validity patterns of ratings of PA and NA. The nonpatients were rated by three or more untrained peers with whom they were well acquainted, whereas the patients were rated by a single clinician. The results are shown in the bottom half of Table 5 and demonstrate clear convergent and discriminant validity patterns in both cases. Compared with ratings of anxiety and depression, the squared multiple correlation differences were notably larger in both studies, although it is impossible to compare the studies to each other because of their many methodological differences. These data again suggest that the discrimination between anxiety and depression will be greatly enhanced if the link between low PA and depression can be firmly established.
最後，兩項研究——一項針對患者（Vye, 1986）和一項針對非患者受試者（Watson & Clark, 1991）——檢視了正向情緒（PA）和負向情緒（NA）評分的聚合效度和區別效度模式。非患者由三位或更多與他們相熟的未受訓的同儕進行評分，而患者則由一位臨床醫生進行評分。結果顯示在表 5 的下半部分，並在兩種情況下都顯示出明確的聚合效度和區別效度模式。與焦慮和抑鬱的評分相比，兩項研究中的平方多重相關差異顯著較大，儘管由於方法上的許多差異，無法將這些研究相互比較。這些數據再次表明，如果能夠確立低正向情緒與抑鬱之間的聯繫，則焦慮和抑鬱之間的區別將大大增強。

Summary and conclusions. These data have many parallels to those discussed earlier: (a) Moderate to good convergent validity is obtained when adequate and comparable scales are used in heterogeneous samples, and (b) the convergent and discriminant correlations covary, which suggests that a general distress factor underlies both types of mood ratings to a considerable extent. Convergent and discriminant correlations will both be higher when this nonspecific factor is rated reliably (e.g., through the use of multiple raters and well-constructed scales) than when it is not.
摘要與結論。這些數據與之前討論的數據有許多相似之處：（a）當在異質樣本中使用適當且可比較的量表時，獲得中等到良好的收斂效度，以及（b）收斂和區別相關性共變，這表明一個一般的困擾因素在相當程度上是兩種類型的情緒評分的基礎。當這個非特定因素被可靠地評估（例如，通過使用多位評估者和精心設計的量表）時，收斂和區別相關性都會更高，而不是在未可靠評估的情況下。

Furthermore, NA and PA show clear convergent and discriminant validity patterns across self- and clinical ratings. Given that anxiety and depression both involve NA, whereas only (low) PA is related to depression, strengthening the PA component of depression measures will improve the discrimination between these syndromes.
此外，負向情緒（NA）和正向情緒（PA）在自我評估和臨床評估中顯示出明顯的收斂和區別效度模式。考慮到焦慮和抑鬱都涉及負向情緒，而只有（低）正向情緒與抑鬱相關，加強抑鬱測量中的正向情緒成分將改善這些綜合症之間的區別。

Validity. A remarkable number of studies have examined the convergence of self- and clinically rated depression. Notably fewer have examined comparable data for anxiety, and we found only three studies in which the discriminant validity of these ratings was investigated. These data are summarized in Table 6. Convergence between self- and clinical raters is highest ( $r=.71$$r=.71$r=.71r=.71 ) for specific, multi-item measures of depression (e.g., the HRSD). Indeed, it seems that the level of convergence is as high as the reliabilities of these scales permit. Moreover, there is no indication that the results differ systematically between patient and nonpatient samples (Beck et al., 1988), so they have been combined for presentation in Table 6. Convergence between
有效性。相當多的研究已經檢視了自我評估與臨床評估的抑鬱症之間的收斂性。值得注意的是，對於焦慮的可比數據進行研究的數量明顯較少，我們僅找到三項研究調查了這些評分的區別效度。這些數據在表 6 中進行了總結。自我評估與臨床評估之間的收斂性在特定的多項目抑鬱測量（例如 HRSD）中最高（ $r=.71$$r=.71$r=.71r=.71 ）。事實上，收斂程度似乎與這些量表的信度相當。此外，沒有跡象表明患者和非患者樣本之間的結果系統性地存在差異（Beck et al., 1988），因此它們已被合併以便在表 6 中呈現。自我評估與臨床評估之間的收斂性

Table 5 表 5
Convergent and Discriminant Validities for Mood Measures: Self-Versus Clinical Raters in Patient and Nonpatient Samples
情緒測量的收斂效度與區別效度：患者與非患者樣本中的自我評估與臨床評估者比較

Note. $\mathrm{NA}=$$\mathrm{NA}=$quadNA=\quad \mathrm{NA}= negative affect; and $\mathrm{PA}=$$\mathrm{PA}=$PA=\mathrm{PA}= positive affect.
注意。 $\mathrm{NA}=$$\mathrm{NA}=$quadNA=\quad \mathrm{NA}= 負面情感；以及 $\mathrm{PA}=$$\mathrm{PA}=$PA=\mathrm{PA}= 正面情感。
${}^{\cdot }$${}^{\cdot }$^(*){ }^{\cdot} For anxiety, $n=244.{}^{\mathrm{b}}$$n=244.{}^{\mathrm{b}}$n=244.^(b)n=244 .{ }^{\mathrm{b}} For discriminant validity, no. of studies $=2$$=2$=2=2 and $n=220$$n=220$n=220n=220.
${}^{\cdot }$${}^{\cdot }$^(*){ }^{\cdot} 對於焦慮， $n=244.{}^{\mathrm{b}}$$n=244.{}^{\mathrm{b}}$n=244.^(b)n=244 .{ }^{\mathrm{b}} 對於區別效度，研究數量 $=2$$=2$=2=2 和 $n=220$$n=220$n=220n=220 。

Correlations between self- and clinical ratings of anxiety are more variable, and it is clearly important to distinguish between studies that have used reliable versus unreliable measures (or rating conditions). However, even with reliable measures or conditions, convergence is slightly lower than for specific depression measures ( $r=.64\mathrm{vs}..71$$r=.64\mathrm{vs}..71$r=.64vs..71r=.64 \mathrm{vs} . .71 ), perhaps because of the greater sensitivity of clinicians to the heterogeneity of anxiety symptoms. When the clinical ratings are of poor or unknown reliability, correlations with self-reported anxious symptomatology are unacceptably low ( $r=.37$$r=.37$r=.37r=.37 ). All studies in which the convergent validity of anxiety ratings has been examined have used patient samples, so the level of convergence in nonpatient samples is unknown.
自我評估與臨床評估之間的焦慮相關性變化較大，明顯需要區分使用可靠與不可靠測量（或評估條件）的研究。然而，即使在可靠的測量或條件下，焦慮評估的收斂性仍略低於特定的抑鬱測量（ $r=.64\mathrm{vs}..71$$r=.64\mathrm{vs}..71$r=.64vs..71r=.64 \mathrm{vs} . .71 ），這或許是因為臨床醫生對焦慮症狀的異質性更為敏感。當臨床評估的可靠性差或未知時，自我報告的焦慮症狀與之之間的相關性則低得不可接受（ $r=.37$$r=.37$r=.37r=.37 ）。所有檢驗焦慮評估收斂效度的研究均使用了患者樣本，因此非患者樣本中的收斂程度尚不清楚。

Three studies presented a multitrait (anxiety vs. depression), multimethod (self- vs. clinician rating) matrix for syndromal measures (Bramley et al., 1988; D. A. Clark, Beck, & Brown, 1989; Vye, 1986). All of them used different measures, but the results were remarkably similar nonetheless and yielded an overall heterotrait-heteromethod correlation of .34. In each study there was evidence that these correlations were not symmetrical in both directions (i.e., the correlation of clinicianrated depression with self-rated anxiety differed from that of clinician-rated anxiety with self-rated depression), but the differences were not systematic across studies and are probably measure specific. It is noteworthy that two of the three studies were cited earlier because they emphasized the low PA aspect of depression, yet the convergent and discriminant validity pattern was the same in third study, which used the revised Hamilton scales and the Beck inventories.
三項研究呈現了一個多特質（焦慮與抑鬱）、多方法（自評與臨床評估）矩陣，用於綜合性測量（Bramley et al., 1988; D. A. Clark, Beck, & Brown, 1989; Vye, 1986）。它們使用了不同的測量工具，但結果卻異常相似，並產生了整體的異特質-異方法相關係數為 0.34。在每項研究中都有證據顯示這些相關性在兩個方向上並不對稱（即，臨床評估的抑鬱與自評的焦慮之間的相關性與臨床評估的焦慮與自評的抑鬱之間的相關性不同），但這些差異在各研究之間並不系統，可能是特定於測量工具的。值得注意的是，三項研究中有兩項之前被引用，因為它們強調了抑鬱的低正情緒（PA）方面，然而第三項研究的收斂與區別效度模式卻是相同的，該研究使用了修訂版的漢密爾頓量表和貝克量表。

Summary and conclusions. The convergent validity between well-established self-report and clinical measures of depression is high-nearly as high as the reliabilities and convergent validity estimates within each type of rating (see Tables 3-5). The convergent validity between reliable self- and clinical measures
摘要與結論。已建立的自我報告與臨床抑鬱測量之間的收斂效度很高，幾乎與每種類型評分內的信度和收斂效度估計一樣高（見表 3-5）。可靠的自我報告與臨床測量之間的收斂效度。
of anxiety is slightly lower, although it is certainly still acceptable. Greater sensitivity of clinicians to the heterogeneous content of anxiety measures (wherein patients respond more on the basis of their general affective distress level) may contribute to this lower convergence. Because both affect and rater type are varied in these analyses, the discriminant correlations are (not surprisingly) somewhat lower than the within-methods discriminant validities, which were in the ${.60}^{\prime }$${.60}^{\prime }$.60^(').60^{\prime} s for self-report and approximately $.40-.45$$.40-.45$.40-.45.40-.45 for clinical ratings. There was good agreement across studies, however, so the overall figure of 34 likely represents an accurate lower bound estimate of the true correlation between syndromal measures of anxiety and depression.
焦慮的程度略低，儘管這仍然是可以接受的。臨床醫生對焦慮測量的異質內容的敏感性較高（患者的反應更多是基於他們的一般情感困擾水平）可能導致這種較低的一致性。由於這些分析中情感和評估者類型的變化，區別性相關性（並不令人驚訝）略低於方法內的區別效度，自我報告的區別效度在 ${.60}^{\prime }$${.60}^{\prime }$.60^(').60^{\prime} 範圍內，而臨床評分約為 $.40-.45$$.40-.45$.40-.45.40-.45 。然而，各研究之間的協議良好，因此整體數字 34 可能代表了焦慮和抑鬱的綜合測量之間真實相關性的準確下限估計。

We have presented a great deal of evidence that examines the convergent and discriminant validity patterns of measures of anxiety and depression. First, a large number of studies provide consistent evidence that under optimal conditions the convergence among reliable syndromal depression ratings averages in the low .80 s for clinical ratings and approximately .70 both within self-report and for self-versus clinical ratings. Optimal conditions include trained raters, access to the same information, well-defined criteria, and an adequate range of symptomatology. There are no systematic differences between patient and nonpatient samples per se for either type of rating.
我們提供了大量證據，檢視焦慮和抑鬱測量的收斂和區別效度模式。首先，大量研究提供了一致的證據，表明在最佳條件下，可靠的綜合性抑鬱評分之間的收斂平均值在臨床評分中約為低於 0.80，而在自我報告和自我與臨床評分之間約為 0.70。最佳條件包括受過訓練的評估者、獲取相同的信息、明確的標準以及足夠的症狀範圍。對於任何類型的評分，患者和非患者樣本之間並不存在系統性的差異。

The data for syndromal anxiety are fewer and suggest somewhat less consistency in clinical ratings as compared with those for depression. The convergence among self-report measures of syndromal anxiety is comparable to that of depression, with some evidence that it may be higher in patient samples. Convergence between self- and clinical ratings of syndromal anxiety is slightly lower than between those for depression, presumably because of the greater variability across the clinical ratings,
有關症狀性焦慮的數據較少，並且與抑鬱症的臨床評分相比，顯示出一致性稍低。症狀性焦慮的自我報告量表之間的收斂性與抑鬱症相當，並有一些證據表明在患者樣本中可能更高。症狀性焦慮的自我評分與臨床評分之間的收斂性略低於抑鬱症，這可能是由於臨床評分之間的變異性更大。

Table 6 表 6
Convergent and Discriminant Validities for Syndrome Measures: Self-Versus Clinical Raters in Patient and Nonpatient Samples
綜合效度與區別效度在症狀測量中的應用：自評與臨床評估者在患者與非患者樣本中的比較

which itself may stem from clinical sensitivity to the greater heterogeneity of the anxiety disorders as compared with the depressive disorders.
這本身可能源於對焦慮障礙相較於抑鬱障礙的更大異質性的臨床敏感性。

Levels of discriminant validity are affected by several parameters. First, there is only modest discriminant validity between self-report measures of anxious and depressive symptomatology in nonpatient samples, for which a large general NA factor accounts for most of the reliable score variation. In contrast, a moderate degree of differentiation can be found in patient samples. This increased differentiation in patient samples is consistent with the results of Hiller, Zaudig, and von Bose (1989) who found that the overlap between depressive and anxious symptoms decreased as severity of psychopathology increased. Nevertheless, it is important to recognize that even in patient samples, self-ratings of anxiety and depression typically provide more information about the overall level of subjective distress than about the relative salience of depressive versus anxious symptomatology. Second, instruments do vary in their convergent and discriminant validity patterns. Two sets of paired measures-the Beck inventories and the Costello-Comrey scales-apparently provide a more differentiated correlational pattern in both patient and nonpatient samples, but more data are needed on each set of scales to establish this finding conclusively.
判別效度的水平受到幾個參數的影響。首先，在非病人樣本中，自我報告的焦慮和抑鬱症狀之間的判別效度僅為適中，這主要是由一個大型的普遍負情緒因素解釋了大部分可靠的分數變異。相對而言，在病人樣本中可以發現中等程度的區分。病人樣本中這種增加的區分與 Hiller、Zaudig 和 von Bose（1989）的研究結果一致，他們發現隨著精神病理學嚴重程度的增加，抑鬱和焦慮症狀之間的重疊減少。然而，重要的是要認識到，即使在病人樣本中，自我評估的焦慮和抑鬱通常提供的更多是關於主觀痛苦的整體水平的信息，而不是抑鬱與焦慮症狀之間的相對顯著性。其次，各種工具在其收斂和判別效度模式上確實存在差異。 兩組配對測量——貝克量表和科斯特羅-康瑞量表——顯然在患者和非患者樣本中提供了更具區分性的相關模式，但需要更多數據來確定每組量表的這一發現。

The convergent and discriminant validity of clinical ratings ranges from very poor (with the original Hamilton scales) to very good (in several studies in which the researchers conceptualized depression largely in terms of a lack of pleasure or interest, i.e., low PA). It is noteworthy that in the latter studies, the convergent and discriminant validity pattern was improved by lowering the discriminant coefficients without simultaneously lowering the convergent correlations also (which is the more typical pattern). Some data (to be discussed subsequently) support the validity of this conceptualization, but it has yet to be tested widely.
臨床評分的收斂效度和區別效度範圍從非常差（使用原始的漢密爾頓量表）到非常好（在幾項研究中，研究者主要將抑鬱概念化為缺乏快樂或興趣，即低正情緒）。值得注意的是，在後者的研究中，通過降低區別係數而不降低收斂相關性，改善了收斂和區別效度的模式（這是更典型的模式）。一些數據（稍後將討論）支持這一概念化的有效性，但尚未得到廣泛測試。

In the majority of studies, discriminant coefficients are in the low 40 s , a level of correlation that suggests both significant overlap and substantial differentiability between the two syndromes. These data indicate that anxiety and depressive syndromes share a significant nonspecific component of general-
在大多數研究中，判別係數在低 40 的範圍內，這一相關水平表明兩種綜合症之間存在顯著的重疊和相當的可區分性。這些數據表明，焦慮和抑鬱綜合症共享一個顯著的非特異性成分。
ized affective distress but that they can also be meaningfully differentiated on the basis of one or more distinctive factors. Thus, two or more constructs are needed to explain the correlational data for anxiety and depressive phenomena, both at the mood and syndromal level: a general nonspecific (NA) factor that is common to the moods or syndromes and one or more specific factors that distinguish them. We turn now to an examination of these specific factors.
雖然情感困擾是相似的，但它們也可以根據一個或多個獨特因素進行有意義的區分。因此，需要兩個或更多的構念來解釋焦慮和抑鬱現象的相關數據，無論是在情緒還是症狀層面：一個普遍的非特異性（NA）因素，這是情緒或症狀的共同特徵，以及一個或多個區分它們的特定因素。我們現在將轉向對這些特定因素的檢視。

Several important points have not yet been established by these data. First, we need to go beyond these correlational results to determine the number and nature of the specific factors involved in the differentiation of anxiety and depressive syndromes. The second issue concerns the distress level at which self-report measures begin to provide a notable degree of discrimination. It has been observed that general medical samples typically score higher than nonpatient samples, but lower than psychiatric patients, on various measures of both depression and anxiety (Klerman, 1989) and that mixed affective symptoms are especially common in this population (Katon & RoyByrne, 1989, 1991). Do these patients more closely resemble psychiatric patients or nonpatients in the degree of specificity obtainable from their self-reports? Third, we cannot determine from these data the patterning of the general and specific factors within individual persons. For instance, it may be that every anxious or depressed patient shows an elevated level of the general factor and an additional elevation on one and only one specific factor. If so, then anxiety and depression are best viewed as distinct disorders that share some symptoms.
幾個重要的點尚未由這些數據確立。首先，我們需要超越這些相關結果，以確定在焦慮和抑鬱綜合症的區分中涉及的具體因素的數量和性質。第二個問題涉及自我報告測量開始提供顯著區分程度的困擾水平。已觀察到，一般醫療樣本在各種抑鬱和焦慮測量上通常得分高於非病人樣本，但低於精神病患者（Klerman, 1989），而且在這一人群中混合情感症狀尤其常見（Katon & RoyByrne, 1989, 1991）。這些患者在自我報告中所能獲得的特異性程度上更接近精神病患者還是非病人？第三，我們無法從這些數據中確定個體內一般因素和特定因素的模式。例如，可能每位焦慮或抑鬱患者都顯示出一般因素的升高水平，並且在一個且僅一個特定因素上也有額外的升高。 如果是這樣，那麼焦慮和抑鬱最好被視為不同的疾病，雖然它們有一些共同的症狀。

Alternatively, some patients may have only a general factor elevation, whereas others may show significant elevations on all of the factors. This situation would indicate a more complicated relation between anxious and depressive phenomena and would necessitate, in turn, a more complex diagnostic system. Patients with only general factor elevations might receive a diagnosis of generalized affective disorder, a designation that might find particular use in general medical populations (Katon & Roy-Byrne, 1989). On the other hand, those with elevations on
另一方面，一些患者可能只有一般因素的升高，而其他患者則可能在所有因素上顯示出顯著的升高。這種情況表明焦慮和抑鬱現象之間的關係更為複雜，並因此需要一個更複雜的診斷系統。只有一般因素升高的患者可能會被診斷為廣泛性情感障礙，這一稱謂在一般醫療人群中可能特別有用（Katon & Roy-Byrne, 1989）。另一方面，那些在所有因素上都有升高的患者則可能會...
all factors would receive either two diagnoses (e.g., major depression and an anxiety disorder such as generalized anxiety disorder [GAD] or panic, depending on other features of the symptom picture) or a specific diagnosis of mixed affective disorder, which would be warranted if such patients were shown to be clinically distinct from those with only one type of disorder (e.g., see Akiskal, 1990; Tyrer, 1984; Van Valkenburg, Akiskal, Puzantian, & Rosenthal, 1984).
所有因素將會獲得兩個診斷（例如，重度憂鬱症和焦慮症，如廣泛性焦慮症[GAD]或驚恐症，具體取決於症狀特徵的其他方面）或混合情感障礙的特定診斷，如果這些患者被證明在臨床上與僅有一種障礙的患者有明顯區別，則這是合理的（例如，參見 Akiskal, 1990; Tyrer, 1984; Van Valkenburg, Akiskal, Puzantian, & Rosenthal, 1984）。
There are three types of evidence for specific factors that differentiate anxiety and depression: (1) the effect of content and context in assessing depression and anxiety; (2) factor-analytic data that indicate the presence of specific factors; and (3) recent work in the structure of mood that suggests that PA is an important, specific component of depression.
有三種類型的證據顯示區分焦慮和抑鬱的特定因素：(1) 在評估抑鬱和焦慮時內容和背景的影響；(2) 因素分析數據顯示特定因素的存在；以及 (3) 最近在情緒結構方面的研究表明，正情緒（PA）是抑鬱的一個重要特定組成部分。

Content analyses. We noted earlier that the Beck inventories and Costello-Comrey scales exhibited somewhat better convergent and discriminant validity patterns than other measures. A content analysis of these scales, to contrast them with those that showed poorer discriminant validity, may suggest reasons for their preferred pattern and offer insight into factors that differentiate the syndromes. The BDI taps a broad range of items generally considered diagnostic of major depressive disorder (e.g., sadness or unhappiness, loss of interest, guilt, suicidal tendencies, and appetite disturbance). It also includes items that indicate general life dissatisfaction, hopelessness, or low self-esteem, factors that one may also see in such other $DSM-III-R$$DSM-III-R$DSM-III-RD S M-I I I-R diagnoses as dysthymia, adjustment reactions, overanxious disorder, personality disorder, and so on. Finally, it assesses symptoms that are common to several depressive and anxiety disorders (e.g., irritability, poor concentration or indecisiveness, insomnia, and fatigue). In contrast, the BAI focuses specifically on the physiological aspects of anxiety. Three of the 20 items are anxious mood terms, and 3 others assess specific fears, whereas the remaining 14 items all assess the symptoms of autonomic hyperactivity and motor tension associated with GAD (and panic disorder as well, if the appropriate temporal features are present). This analysis suggests that the discriminant validity of the Beck inventories stems largely from the content specificity of the BAI. Indeed, in patient samples, the BDI is more highly correlated with other anxiety inventories than it is with the BAI ( $rs=.61$$rs=.61$rs=.61r s=.61 and .49 , respectively; see Table 4); however, no difference is found in nonpatient samples ( $r=.61$$r=.61$r=.61r=.61 in both cases), perhaps because of the infrequent occurrence of significant physiological symptomatology in these subjects.
內容分析。我們之前提到，貝克量表和科斯特羅-科梅瑞量表的聚合效度和區別效度模式比其他測量工具稍好。對這些量表進行內容分析，以便與那些顯示較差區別效度的量表進行對比，可能會揭示其優越模式的原因，並提供有關區分症狀的因素的見解。貝克抑鬱量表涵蓋了一系列通常被認為是重度抑鬱症診斷的項目（例如，悲傷或不快、失去興趣、內疚、自殺傾向和食慾障礙）。它還包括表明一般生活不滿、絕望或低自尊的項目，這些因素在其他診斷如持續性抑鬱症、適應反應、過度焦慮症、人格障礙等中也可能出現。最後，它評估了幾種抑鬱和焦慮障礙中常見的症狀（例如，易怒、注意力不集中或優柔寡斷、失眠和疲勞）。相比之下，貝克焦慮量表專注於焦慮的生理方面。 在 20 個項目中，有 3 個是焦慮情緒術語，另外 3 個評估特定的恐懼，而剩下的 14 個項目則評估與廣泛性焦慮症（如果存在適當的時間特徵，還包括驚恐障礙）相關的自律神經亢進和運動緊張的症狀。這項分析表明，貝克量表的區別效度在很大程度上源於 BAI 的內容特異性。事實上，在患者樣本中，BDI 與其他焦慮量表的相關性高於與 BAI 的相關性（分別為 $rs=.61$$rs=.61$rs=.61r s=.61 和.49；見表 4）；然而，在非患者樣本中則未發現差異（在兩種情況下均為 $r=.61$$r=.61$r=.61r=.61 ），這可能是因為這些受試者中顯著的生理症狀不常出現。

The CC-D scale focuses more narrowly on the symptoms of depressed mood, loss of interest or pleasure, and worthlessness. It does not assess physiological or vegetative changes, fatigue, or suicidal ideation. However, its strength may lie in the fact that it assesses the loss of interest or pleasure particularly well by including a number of (reverse-keyed) positively worded items (e.g., “Living is a wonderful adventure for me”). Similarly to the BAI, the nine items of the CC-A focus on the anxious mood, motor tension, and vigilance components of GAD. Thus, the
CC-D 量表更狹義地聚焦於抑鬱情緒、失去興趣或快樂以及無價值感的症狀。它不評估生理或植物性變化、疲勞或自殺意念。然而，它的優勢可能在於它特別好地評估了失去興趣或快樂，通過包含多個（反向鍵入）正面措辭的項目（例如，“生活對我來說是一場美妙的冒險”）。與 BAI 類似，CC-A 的九個項目聚焦於廣泛性焦慮症的焦慮情緒、運動緊張和警覺性成分。因此，
increased discriminative power of the Costello-Comrey scales appears to reflect the fact that in addition to nonspecific distress, they assess symptomatology specific to both depression (i.e., lack of PA) and anxiety (i.e., motor tension and vigilance).
Costello-Comrey 量表的辨別力增強似乎反映了這樣一個事實：除了非特異性痛苦外，它們還評估了與抑鬱（即缺乏正向情緒）和焦慮（即運動緊張和警覺性）特定相關的症狀。

In contrast to this clear content differentiation, the MMPI scales, STAI, SDS, and SAS each contain symptoms that are common to the two syndromes (e.g., restlessness and fatigue). Moreover, each assesses symptoms more characteristic of the other syndrome. For example, the SDS scale includes tachycardia, whereas the STAI measures blue mood, crying, and unhappiness. The MMPI scales are quite heterogeneous in content; both the depression and anxiety scales contain many items similar to those in the BDI, plus others that are more generally related to anxiety (e.g., worry, obsessiveness and brooding, and hypersensitivity).
與這種明確的內容區分相對，MMPI 量表、STAI、SDS 和 SAS 各自包含兩種綜合症的共同症狀（例如，坐立不安和疲勞）。此外，每個量表還評估了更具特徵性的其他綜合症症狀。例如，SDS 量表包括心跳過速，而 STAI 則測量憂鬱情緒、哭泣和不快。MMPI 量表在內容上相當異質；抑鬱和焦慮量表都包含許多與 BDI 相似的項目，以及其他更普遍與焦慮相關的項目（例如，擔憂、強迫和沉思，以及過度敏感）。

However, content considerations alone are not sufficient to explain the better convergent and discriminant validity patterns of the Beck and Costello-Comrey scales. The content assessed in the SCL-90 scales is quite similar to that of the Beck inventories, and yet their discriminative power is substantially less. One possible explanation for this difference in discriminant validity is that the SCL-90 items are intermixed in a single inventory with a single response format, whereas the Beck scales have different formats and are administered as separate measures. If this explanation is correct, then the enhanced discriminative power of the Beck versus the SCL 90 may be partially due to method variance. In this regard, it is noteworthy that the Costello-Comrey scales are combined in a single inventory with no difference in format between the two scales. Because the discriminative power of the Costello-Comrey scales cannot be attributed to method variance, their content deserves especially close attention.
然而，僅僅考慮內容並不足以解釋貝克（Beck）和科斯特羅-科梅瑞（Costello-Comrey）量表在收斂效度和區別效度上的較好模式。SCL-90 量表所評估的內容與貝克量表相當相似，但其區別能力卻明顯較低。這種區別效度差異的一個可能解釋是，SCL-90 的項目混合在一個單一的量表中，並使用單一的反應格式，而貝克量表則有不同的格式並作為獨立的測量工具進行施測。如果這一解釋是正確的，那麼貝克量表相對於 SCL-90 的增強區別能力可能部分是由於方法變異所致。在這方面，值得注意的是，科斯特羅-科梅瑞量表是合併在一個單一的量表中，兩個量表之間沒有格式上的差異。由於科斯特羅-科梅瑞量表的區別能力無法歸因於方法變異，因此其內容特別值得關注。

Factors that influence reported symptom levels. In general, patients tend to rate their symptoms as more severe than do clinicians (Katon & Roy-Byrne, 1989). Furthermore, for depressive symptomatology, level of severity (whether self- or clinicianrated) is negatively correlated with self-clinician discrepancy scores (Rush, Hiser, & Giles, 1987; Tondo, Burrai, Scamonatti, Weissenburger, & Rush, 1988; Zimmerman, Coryell, Wilson, & Corenthal, 1986). That is, whereas self- and clinical ratings are highly similar for severely depressed patients, these ratings show substantial discrepancy at lower levels of disturbance. This finding needs replication with anxious patients.
影響報告症狀水平的因素。一般而言，患者傾向於將其症狀評價為比臨床醫生更嚴重（Katon & Roy-Byrne, 1989）。此外，對於抑鬱症狀，嚴重程度（無論是自我評估還是臨床評估）與自我與臨床評估之間的差異分數呈負相關（Rush, Hiser, & Giles, 1987; Tondo, Burrai, Scamonatti, Weissenburger, & Rush, 1988; Zimmerman, Coryell, Wilson, & Corenthal, 1986）。也就是說，對於重度抑鬱的患者，自我評估和臨床評估非常相似，而在較低的困擾水平上，這些評估顯示出顯著的差異。這一發現需要在焦慮患者中進行重複驗證。

Taken together, these data suggest that, as compared with patients’ self-ratings, clinicians’ judgments more strongly reflect specific, clinically prominent symptoms (e.g., marked anhedonia and psychomotor retardation) and are less influenced by the patients’ general level of affective distress. Moreover, they further suggest that the responses of less disturbed patients primarily reflect their standing on the general distress factor, whereas severely disturbed patients focus additionally on their specific symptoms. In this regard, it is noteworthy that clinically diagnosed depressed patients tend to rate themselves as having more severe symptoms of both types than do anxiety patients of equal (clinician-rated) severity (L. A. Clark, 1989), which indicates that depressed patients may experience higher levels of general distress than do anxious patients. In this context, it is also useful to recall a suggestion made earlier with regard to anxious symptomatology, namely, that patients’ self-
綜合來看，這些數據表明，與患者的自我評價相比，臨床醫生的判斷更強烈地反映出特定的、臨床上突出的症狀（例如，明顯的快感缺失和精神運動遲緩），並且不太受患者一般情緒困擾程度的影響。此外，這些數據進一步表明，較少困擾的患者的反應主要反映他們在一般困擾因素上的狀態，而嚴重困擾的患者則額外關注他們的特定症狀。在這方面，值得注意的是，臨床診斷的抑鬱症患者往往自評其症狀的嚴重程度高於同等（臨床評估）嚴重程度的焦慮患者（L. A. Clark, 1989），這表明抑鬱症患者可能經歷的普遍困擾程度高於焦慮患者。在這個背景下，回顧之前對焦慮症狀的建議也是有用的，即患者的自我評價——
ratings may be more convergent than clinical ratings because clinicians focus more on specific scale content, whereas patients emphasize their general distress level.
評分可能比臨床評分更具一致性，因為臨床醫生更專注於特定的量表內容，而患者則更強調他們的整體痛苦程度。

Analyses of the Hamilton scales. Ten studies have compared HRSD and HRSA scores in different diagnostic groups or in relation to treatment. Because the revised Hamilton scales show much clearer convergent and discriminant validity patterns, studies with the original scoring will be reviewed briefly and then compared to the results of two studies with the revised scales. Finally, a content comparison of the original and revised scoring systems suggests why the revised system yields more discriminating results.
漢密爾頓量表的分析。十項研究比較了不同診斷組別或與治療相關的 HRSD 和 HRSA 分數。由於修訂版漢密爾頓量表顯示出更明確的收斂和區別效度模式，因此將簡要回顧使用原始計分的研究，然後與兩項使用修訂量表的研究結果進行比較。最後，對原始和修訂計分系統的內容比較顯示了為什麼修訂系統能夠產生更具區別性的結果。

Four studies compared (original) Hamilton scale levels in panic disorder patients, without or without an additional depressive disorder (Breier et al., 1984; Buller, Maier, & Benkert, 1986; Ganellen & Zola, 1989; Lesser et al., 1988). Without exception, patients with secondary depression scored significantly higher on both Hamilton scales as compared with those without. DiNardo and Barlow (1990) found similar patterns on both scales in eight diagnostic groups. Specifically, dysthymics scored highest and phobics, lowest on both Hamilton scales. Patients with other anxiety disorders (panic, GAD, and agoraphobia) and major depression had intermediate scores on both scales.
四項研究比較了恐慌症患者的（原始）漢密爾頓量表水平，無論是否伴隨有額外的抑鬱症（Breier et al., 1984; Buller, Maier, & Benkert, 1986; Ganellen & Zola, 1989; Lesser et al., 1988）。毫無例外，伴隨抑鬱的患者在兩個漢密爾頓量表上的得分均顯著高於那些沒有抑鬱的患者。DiNardo 和 Barlow（1990）在八個診斷組中發現了兩個量表上類似的模式。具體而言，持續性心情障礙患者在兩個漢密爾頓量表上的得分最高，而恐懼症患者的得分最低。其他焦慮症患者（恐慌症、廣泛性焦慮症和廣場恐懼症）及重度抑鬱症患者在兩個量表上的得分則為中等。

Similarly, four treatment studies with depressed, anxious, or mixed patient groups with diverse interventions all found that scores on both Hamilton scales were reduced after treatment (Borkovec & Mathews, 1988; Borkovec et al., 1987; Lesser et al., 1988; Widlocher, Lecrubier, & Le Goc, 1983). A fifth study (Grunhaus, Rabin, & Greden, 1986) found that pure depressed patients had lower HRSD scores after treatment than did patients with an additional panic disorder, who scored higher than the depressed patients on both anxious and depressed mood. Taken together, these data are congruent with the earlier correlational findings in demonstrating the influence of a strong nonspecific factor in the original Hamilton scales.
類似地，四項針對抑鬱、焦慮或混合患者群體的治療研究，無論介入方式多樣，都發現治療後哈密爾頓量表的得分均有所降低（Borkovec & Mathews, 1988; Borkovec et al., 1987; Lesser et al., 1988; Widlocher, Lecrubier, & Le Goc, 1983）。第五項研究（Grunhaus, Rabin, & Greden, 1986）發現，純抑鬱患者的 HRSD 得分在治療後低於有額外驚恐障礙的患者，而後者在焦慮和抑鬱情緒上得分均高於抑鬱患者。綜合來看，這些數據與早期的相關性研究結果一致，顯示出原始哈密爾頓量表中強烈的非特異性因素的影響。

DiNardo and Barlow (1990) compared the same eight diagnostic groups with the revised Hamilton scoring system and obtained notably different results. On the revised HRSA, patients with agoraphobia, obsessive-compulsive disorder, panic, and mixed anxiety-depression diagnoses all scored higher than did those with GAD, dysthymia, major depression, or phobias. In contrast, patients with major depression and dysthymia scored higher on the revised HRSD than did those with obses-sive-compulsive disorder, agoraphobia, or mixed diagnoses, who in turn scored higher than did those with GAD, panic disorder, or phobias. Thus, on the revised Hamilton scales, the ordering of the diagnostic groups conformed much more closely to the theoretically expected pattern.
DiNardo 和 Barlow (1990) 將相同的八個診斷組別與修訂版的漢密爾頓評分系統進行比較，並獲得了顯著不同的結果。在修訂版的 HRSA 中，患有廣場恐懼症、強迫症、驚恐症和混合焦慮-抑鬱診斷的患者得分均高於患有廣泛性焦慮症、持續性心情低落、重度抑鬱或恐懼症的患者。相反，重度抑鬱和持續性心情低落的患者在修訂版 HRSD 上的得分高於患有強迫症、廣場恐懼症或混合診斷的患者，而後者的得分又高於患有廣泛性焦慮症、驚恐症或恐懼症的患者。因此，在修訂版的漢密爾頓量表上，診斷組別的排序與理論上預期的模式更為一致。

How were the Hamilton scales rescored to yield these improved results? The most systematic change involved physiological items: Specifically, two physiological items were dropped from the depression scale entirely, and four were reassigned from depression to anxiety. Benshoof, Moras, DiNardo, and Barlow (1989) provide item data that support the validity of this revised scoring scheme. They compared depressed patients
漢密爾頓量表是如何重新評分以產生這些改進結果的？最系統的變化涉及生理項目：具體而言，兩個生理項目完全從抑鬱量表中刪除，四個則從抑鬱重新分配到焦慮。Benshoof、Moras、DiNardo 和 Barlow（1989）提供了支持這一修訂評分方案有效性的項目數據。他們比較了抑鬱患者。
with three anxiety groups on each of the Hamilton items. Although none of the 15 revised HRSA items differentiated the depressed patients from all anxiety groups, this was largely because the GAD group tended to overlap with the depressed patients. Importantly, the items that showed the clearest differentiation between the depressive patients and the panic and agoraphobic patients were physiological in nature, that is, cardiovascular, autonomic, and respiratory symptoms. Thus, these data again suggest the importance of physiological signs for differentiating anxiety from depression.
在每個漢密爾頓項目上有三個焦慮組。儘管 15 個修訂的 HRSA 項目並未將抑鬱患者與所有焦慮組區分開來，但這主要是因為廣泛性焦慮症組與抑鬱患者之間存在重疊。重要的是，顯示抑鬱患者與驚恐症和廣場恐懼症患者之間最明顯區分的項目是生理性質的，即心血管、自主神經和呼吸症狀。因此，這些數據再次表明生理徵兆在區分焦慮與抑鬱中的重要性。

Symptom rating studies. L. A. Clark (1989) found that only a small subset of anxiety-related symptoms, panic attacks (including the associated autonomic symptoms) and agoraphobic avoidance, reliably differentiated anxious from depressed patients. Similarly, the most differentiating depression symptoms were those generally associated with the melancholic subsyndrome (e.g., profound loss of pleasure and early morning awakening). However, most symptoms (e.g., irritability, anxious mood, and disturbances of sleep and appetite) failed to discriminate the two types of patients, primarily because they were highly prevalent in both groups. These findings are consistent with demonstrations that self-reports of NA, and of anxiety in particular, are correlated with other types of complaints, especially those of physical health (indigestion, sore throat, itchiness, joint pain, etc.; Watson & Clark, in press-a; Watson & Pennebaker, 1989). In fact, Watson and Pennebaker (1989) proposed that the concept of NA be expanded further into an extremely broad dimension of somatopsychic distress; such a dimension would encompass the nonspecific component common to both depressive and anxious disorders. In contrast, those symptoms that differ in frequency between the two types of patients reflect the unique aspects of these syndromes.
症狀評估研究。L. A. Clark（1989）發現，只有一小部分與焦慮相關的症狀，如驚恐發作（包括相關的自律神經症狀）和廣場恐懼的迴避行為，能可靠地區分焦慮患者和抑鬱患者。同樣，最能區分抑鬱症狀的是那些通常與憂鬱亞型相關的症狀（例如，深刻的快樂喪失和清晨早醒）。然而，大多數症狀（例如，易怒、焦慮情緒以及睡眠和食慾的干擾）未能區分這兩類患者，主要是因為它們在兩組中都高度普遍。這些發現與自我報告的負面情緒（NA）和特別是焦慮與其他類型的抱怨（尤其是身體健康方面的抱怨，如消化不良、喉嚨痛、癢、關節痛等）相關的證據一致（Watson & Clark，待發表；Watson & Pennebaker，1989）。事實上，Watson 和 Pennebaker（1989）提出，NA 的概念應進一步擴展為一個極為廣泛的身心痛苦維度；這樣的維度將涵蓋抑鬱症和焦慮症共同的非特異性成分。 相對而言，兩類患者之間在頻率上有所不同的症狀反映了這些綜合症的獨特特徵。
L. A. Clark (1989) presented evidence to indicate that rating context also influences clinical judgments of anxiety and depressive symptoms. Specifically, when ratings were made as part of the clinical diagnostic process, greater differentiation was found between anxiety and depression symptom ratings than when the ratings were made independently of diagnosis. Thus, the correlational data showing that clinicians (more than patients) focus on the distinctive features of these disorders may stem, at least in part, from the necessity of assigning diagnoses. If this explanation is correct, it further suggests that if a nonspecific affective disorder diagnosis were available to clinicians, clinical ratings might subsequently show less differentiation than they do currently, because of a decreased need to distinguish between the two types of syndromes.
L. A. Clark (1989) 提出了證據表明評分背景也會影響對焦慮和抑鬱症狀的臨床判斷。具體而言，當評分作為臨床診斷過程的一部分時，焦慮和抑鬱症狀評分之間的區分程度比在獨立於診斷的情況下進行評分時更大。因此，顯示臨床醫生（比患者更）專注於這些疾病的獨特特徵的相關數據，可能部分源於分配診斷的必要性。如果這一解釋是正確的，那麼它進一步暗示，如果臨床醫生能夠使用非特定情感障礙的診斷，臨床評分可能會顯示出比目前更少的區分，因為對於區分這兩種類型的綜合症的需求減少。

Clinical ratings may also be influenced by the setting in which the data are collected. For example, studies of anxiety clinic patients typically report a lower frequency of depressive diagnoses than do studies with samples from other sites. Across 13 studies ( 11 were reviewed by L. A. Clark, 1989; the others were carried out by Buller et al., 1986, and Maier et al., 1988) that examined the prevalence of depression in patients with agoraphobia, panic disorder, or both ( $N=682$$N=682$N=682N=682 ), 64% (range, $41\mathrm{\%}-92\mathrm{\%}$$41\mathrm{\%}-92\mathrm{\%}$41%-92%41 \%-92 \% ) were found to have a depressive disorder. In contrast, five studies with corresponding patients from anxiety clinics found an average depression prevalence of only $21.5\mathrm{\%}$$21.5\mathrm{\%}$21.5%21.5 \% (range, $10\mathrm{\%}-39\mathrm{\%}$$10\mathrm{\%}-39\mathrm{\%}$10%-39%10 \%-39 \%; Barlow, DiNardo, Vermilyea, Vermilyea, & Blanchard, 1986; Benshoof et al., 1989; de Ruiter, Rijken, Barssen,
臨床評分也可能受到數據收集環境的影響。例如，焦慮診所患者的研究通常報告的抑鬱診斷頻率低於來自其他地點樣本的研究。在 13 項研究中（11 項由 L. A. Clark 於 1989 年回顧；其他由 Buller 等人於 1986 年和 Maier 等人於 1988 年進行），檢查了患有廣場恐懼症、驚恐障礙或兩者兼有的患者中抑鬱症的流行率，發現 64%（範圍， $41\mathrm{\%}-92\mathrm{\%}$$41\mathrm{\%}-92\mathrm{\%}$41%-92%41 \%-92 \% ）的患者被診斷為抑鬱症。相比之下，五項來自焦慮診所的相應患者的研究發現，抑鬱症的平均流行率僅為 $21.5\mathrm{\%}$$21.5\mathrm{\%}$21.5%21.5 \% （範圍， $10\mathrm{\%}-39\mathrm{\%}$$10\mathrm{\%}-39\mathrm{\%}$10%-39%10 \%-39 \% ；Barlow, DiNardo, Vermilyea, Vermilyea, & Blanchard, 1986；Benshoof 等人，1989；de Ruiter, Rijken, Barssen，
van Schaik, & Kraaimat, 1989; DiNardo & Barlow, 1990; Sanderson, DiNardo, Rapee, & Barlow, 1990). It is not clear from these data whether the observed prevalence differences (a) are veridical and reflect systematic variations in health care seeking, (b) stem from a diagnostic bias against finding cross-affect disorders in specific-affect clinics, or © result from self-perception differences in patients that affect their symptom reporting during interviews. Of course, all of these factors could be operating simultaneously. Furthermore, it must be noted that four of the five anxiety clinic studies were done at a single site, and all five used the Anxiety Disorder Interview Schedule (DiNardo, O’Brien, Barlow, Waddell, & Blanchard, 1983) for diagnostic purposes. Thus, these findings clearly need to be replicated in other clinics with other diagnostic procedures.
van Schaik 和 Kraaimat, 1989; DiNardo 和 Barlow, 1990; Sanderson, DiNardo, Rapee 和 Barlow, 1990)。從這些數據中尚不清楚觀察到的流行率差異是否 (a) 是真實的，並反映了尋求醫療的系統性變化，(b) 源於對特定情感診所中交叉情感障礙的診斷偏見，或 (c) 由於患者自我感知的差異影響了他們在面談中的症狀報告。當然，所有這些因素可能同時發生。此外，必須指出的是，五項焦慮診所研究中有四項是在單一地點進行的，且所有五項都使用了焦慮障礙面談時間表（DiNardo, O’Brien, Barlow, Waddell 和 Blanchard, 1983）作為診斷目的。因此，這些發現顯然需要在其他診所和其他診斷程序中進行複製。

In addition to properties of the assessment instruments themselves, a number of factors appear to affect ratings of syndromal anxiety and depression. Patient self-ratings seem to be influenced more strongly by general distress levels than are clinical ratings. Moreover, in depression the importance of general distress-in relation to specific depressive symptomsmay be greater in milder levels of the syndrome. This issue has not been examined for anxiety syndromes, however. The context in which clinical ratings are made is also an important factor: Ratings of syndromal anxiety and depression are more distinctive when they are made as part of the diagnostic process. Thus, the existence of a nonspecific affective diagnosis may increase the observed overlap in clinical ratings because of a decreased need for diagnostic differentiation. Furthermore, various treatment studies have shown significant nonspecific changes in both anxious and depressed patients and thereby suggested that decreased differentiation will not have deleterious treatment effects. Finally, the clinical setting itself may also affect ratings. Secondary depression is reported less frequently in anxiety disorder clinics than in other sites, but additional research is needed to determine the replicability of this finding in additional settings and with other assessment instruments, and if replicable, the extent to which this finding represents true prevalence differences or, rather, reflects perceptual differences on the part of clinicians, patients, or both.
除了評估工具本身的特性外，還有多種因素似乎影響對症狀性焦慮和抑鬱的評分。患者的自我評分似乎受到一般困擾程度的影響，這種影響比臨床評分更強。此外，在抑鬱中，一般困擾的重要性相對於特定的抑鬱症狀在症狀較輕的情況下可能更大。然而，這一問題尚未對焦慮症狀進行研究。臨床評分所處的背景也是一個重要因素：當症狀性焦慮和抑鬱的評分作為診斷過程的一部分時，這些評分更具區別性。因此，非特異性情感診斷的存在可能會增加臨床評分的重疊，因為對診斷區分的需求減少。此外，各種治療研究顯示焦慮和抑鬱患者均有顯著的非特異性變化，從而表明減少區分不會對治療效果產生不利影響。最後，臨床環境本身也可能影響評分。 次級憂鬱在焦慮症診所的報告頻率低於其他場所，但需要進一步研究以確定這一發現是否在其他環境和使用其他評估工具時可重複，如果可重複，則需要探討這一發現在多大程度上代表了真實的流行率差異，或是反映了臨床醫生、患者或兩者的感知差異。

The discriminative power of syndromal scales, whether based on self- or clinical ratings, depends on having clearly defined, nonoverlapping content. For self-report scales, moreover, rating format may also be a factor. The greatest discriminative power for syndromal ratings of anxiety is obtained when physiological symptoms (i.e., autonomic hyperactivity) are emphasized along with tension, fear, and anxious mood. Similarly, measures of depressive symptomatology that emphasize the loss of pleasure (i.e., an absence of PA) and other symptoms of melancholia appear to be more distinctive than those that do not. Furthermore, clinicians who conceptualize depression in terms of loss of pleasure and low PA produce more differentiated ratings even if they use a standard depression rating scale.
綜合症量表的區分能力，無論是基於自我評估還是臨床評估，都取決於內容的明確定義和不重疊。此外，對於自我報告量表，評分格式也可能是一個因素。當強調生理症狀（即，自主神經系統過度活躍）以及緊張、恐懼和焦慮情緒時，焦慮的綜合症評分具有最大的區分能力。同樣，強調快樂喪失（即，缺乏正情緒）和其他憂鬱症狀的憂鬱症量表似乎比那些不強調的更具區別性。此外，將憂鬱症概念化為快樂喪失和低正情緒的臨床醫生，即使使用標準的憂鬱症評分量表，也會產生更具區分性的評分。

In a previous review (L. A. Clark & Watson, 199 1b), we factor analyzed the 10 most commonly used anxiety and depression
在之前的評論中（L. A. Clark & Watson, 1991b），我們對 10 種最常用的焦慮和抑鬱進行了因素分析。
scales (both mood and syndromal). The first factor-most clearly marked by the BDI and MMPI anxiety scales-was very broad and general; it was easily identifiable as general NA, demoralization, or somatopsychic distress. The emergence of this factor reinforces our earlier conclusion that nonspecific distress is inherent in the syndromes of depression and anxiety and is largely responsible for their co-occurrence. In contrast, the second factor was primarily represented by the CC-A scale, which emphasizes fearful mood, anxious vigilance, and motor tension, content that is also found in the BAI (which because of its recent development, lacked sufficient data to be included in the analysis). Thus, these data parallel the content analyses described earlier. However, the absence of a specifically depressive factor raises the question of whether such a factor will emerge if additional items with content peculiar to depression are included in these types of analyses.
量表（包括情緒和綜合症）。第一個因素——最明顯地由 BDI 和 MMPI 焦慮量表標示——非常廣泛且一般；它很容易被識別為一般的負向情緒、沮喪或身心痛苦。這一因素的出現強化了我們之前的結論，即非特異性痛苦是抑鬱和焦慮綜合症固有的，並在很大程度上負責它們的共現。相對而言，第二個因素主要由 CC-A 量表代表，該量表強調恐懼情緒、焦慮警覺和運動緊張，這些內容在 BAI 中也有出現（由於其最近的發展，缺乏足夠的數據以納入分析）。因此，這些數據與之前描述的內容分析相呼應。然而，缺乏特定的抑鬱因素引發了這樣的問題：如果在這類分析中納入具有抑鬱特有內容的額外項目，是否會出現這樣的因素。

Symptom-level analyses. A number of studies (reviewed by L. A. Clark & Watson, 1991b) have directly factor analyzed selfor clinical ratings of general neurotic symptoms and identified separate depression and anxiety factors, or more rarely, a single bipolar factor. Two general patterns can be discerned. First, in many studies the so-called depression factor was quite broad, encompassing many nonspecific symptoms of distress in addition to more distinctively depressive phenomena, whereas the so-called anxiety factor was more narrowly focused on physiological signs of anxiety and shakiness or tension. These data replicate the pattern observed in our content analysis of the Beck inventories and of the scale-level factor analysis. The second pattern-seen particularly in studies with variants of the SCL-90-was a tripartite division of depression and anxiety items into: (a) a general neurotic factor, which includes feelings of inferiority and rejection, oversensitivity to criticism, selfconsciousness, social distress, and sometimes also depressed and anxious mood; (b) a specific anxiety factor, which is focused on feelings of tension, nervousness, shakiness, and panic (wherein explicitly somatic items often form yet another factor); and © a specific depression factor that includes loss of interest or pleasure, anorexia, and crying spells, and sometimes hopelessness, loneliness, suicidal ideation, and depressed mood as well. This factor seems clearly related to PA and also seems to reflect the lack of energy and zest that characterizes the low end of this dimension.
症狀層級分析。許多研究（由 L. A. Clark 和 Watson 於 1991b 年回顧）直接對一般神經症症狀的自我或臨床評分進行了因素分析，並識別出獨立的抑鬱和焦慮因素，或更少見的單一雙極因素。可以辨別出兩種一般模式。首先，在許多研究中，所謂的抑鬱因素相當廣泛，除了更明顯的抑鬱現象外，還包括許多非特異性的痛苦症狀，而所謂的焦慮因素則更狹窄地集中於焦慮的生理徵兆以及顫抖或緊張。這些數據重複了我們對貝克量表的內容分析和量表層級因素分析中觀察到的模式。 第二種模式——特別是在使用 SCL-90 變體的研究中可見——是將抑鬱和焦慮項目劃分為三個部分：(a) 一個一般神經質因素，包括自卑感和被拒絕感、對批評的過度敏感、自我意識、社交困擾，有時還包括抑鬱和焦慮的情緒；(b) 一個特定的焦慮因素，專注於緊張、緊張不安、顫抖和恐慌的感覺（其中明確的身體症狀項目通常形成另一個因素）；以及(c) 一個特定的抑鬱因素，包括失去興趣或快樂、厭食症和哭泣發作，有時還包括絕望、孤獨、自殺意念和抑鬱情緒。這個因素似乎明顯與 PA 有關，並且似乎反映了這一維度低端特徵的缺乏能量和熱情。

Beck (1972) obtained similar results in his review of 13 factoranalytic studies of depressive symptoms. He noted three factors that appeared in all studies: One factor, marked by self-deprecation, low self-esteem, sad affect, self-blame, and so on, corresponds to the general distress dimension we have noted repeatedly. A second factor, marked by apathy, emotional withdrawal, fatigue, loss of sexual interest, and lack of social participation, was more specifically depressive and clearly reflects the lack of pleasure and social-interpersonal engagement that is characteristic of low PA. General somatic complaints and difficulties constituted the third invariant factor. Most studies also found a specific anxiety factor, defined by such items as tension and agitation.
貝克（1972）在對 13 項抑鬱症狀的因素分析研究進行回顧時獲得了類似的結果。他注意到所有研究中出現的三個因素：第一個因素以自我貶低、低自尊、悲傷情感、自責等為特徵，對應於我們反覆提到的一般痛苦維度。第二個因素以冷漠、情感撤退、疲勞、性興趣喪失和缺乏社交參與為特徵，更具體地反映了抑鬱，明顯顯示出低正情緒（PA）所特有的缺乏快樂和社交人際互動。一般的身體不適和困難構成了第三個不變因素。大多數研究還發現了一個特定的焦慮因素，以緊張和激動等項目來定義。

A general concern with item-level analyses is that the results are influenced by base rate differences, that is, systematic differences in the frequency of anxiety and depressive symptoms may lead to the identification of spurious specific factors. Fortu-
對於項目層級分析的一個普遍關注是，結果受到基準率差異的影響，即焦慮和抑鬱症狀的頻率系統性差異可能導致虛假的特定因素的識別。
nately, however, base rate differences do not appear to be a major influence in this area. For example, in a sample of 364 outpatients, Prusoff and Klerman (1974) reported mean selfreported symptom levels that ranged from 1.5 to 3.1 on a 1-4 scale, with no systematic difference in the base rates of specific anxiety and depressive items. Using physicians as raters for the same set of symptoms, Lipman, Rickels, Covi, Derogatis, and Uhlenhuth (1969) found a range of 1.7 to 3.0 in a sample of 837 outpatients. Although there was an overall difference in the mean frequency of depressive (2.2) versus anxious (2.7) symptoms, neither set of symptoms showed a truncation of range sufficient to restrict the interitem correlations greatly. Nevertheless, future researchers in this area must be alert to potential artifacts due to differential endorsement rates. ${}^7$${}^7$^(7){ }^{7}
然而，基準率的差異似乎並不是這個領域的主要影響因素。例如，在一個包含 364 名門診病人的樣本中，Prusoff 和 Klerman（1974）報告的自我報告症狀水平在 1 到 4 的量表上範圍從 1.5 到 3.1，且特定焦慮和抑鬱項目的基準率沒有系統性的差異。使用醫生作為同一組症狀的評估者，Lipman、Rickels、Covi、Derogatis 和 Uhlenhuth（1969）在 837 名門診病人的樣本中發現範圍為 1.7 到 3.0。儘管抑鬱（2.2）與焦慮（2.7）症狀的平均頻率存在整體差異，但這兩組症狀的範圍都沒有足夠的截斷來大幅限制項目間的相關性。然而，未來在這個領域的研究者必須警惕由於不同的認可率而產生的潛在假象。

Summary and conclusions. Consistent with the earlier content analyses, factor analytic studies of symptoms demonstrate that a rather distinctive anxiety factor that focuses on nervous tension and autonomic symptomatology can be found and that a highly general distress factor that encompasses but is not limited to depressive symptoms also frequently appears. Furthermore, these data extend the scale content analyses by demonstrating that a specific depression factor, which represents a more severe depressive syndrome, is also identifiable. Symptoms related to the absence of PA (e.g., loss of interest or pleasure, apathy, hopelessness, extreme fatigue, lethargy, and psychomotor retardation) are common markers of this cluster. This factor may also contain some NA-related items (e.g., depressed mood) but does not include such nonspecific symptoms as low self-esteem, which appear instead on the general factor.
摘要與結論。與早期的內容分析一致，症狀的因素分析研究顯示出一個相當獨特的焦慮因素，專注於神經緊張和自主神經症狀，並且經常出現一個高度一般性的痛苦因素，涵蓋但不限於抑鬱症狀。此外，這些數據擴展了量表內容分析，顯示出一個特定的抑鬱因素，代表著更嚴重的抑鬱綜合症。與缺乏正情緒（例如，失去興趣或快樂、冷漠、絕望、極度疲勞、嗜睡和精神運動遲緩）相關的症狀是這一群體的常見標誌。這個因素可能還包含一些與負情緒相關的項目（例如，抑鬱情緒），但不包括如低自尊等非特異性症狀，這些症狀則出現在一般因素中。

These data thus support and extend the conclusion drawn from the correlational studies that syndromal anxiety and depression share a significant nonspecific component of generalized affective distress but, nevertheless, can be differentiated on the basis of additional distinctive factors. Thus, the marked physiological hyperarousal associated with GAD (and panic attacks, if onset is sudden) appears to be relatively specific to anxiety, whereas the various manifestations of low PA (apathy, behavioral withdrawal, or retardation) are distinctly characteristic of depression, especially its melancholic subsyndrome.
這些數據因此支持並擴展了從相關研究中得出的結論，即綜合症焦慮和抑鬱共享一個顯著的非特異性成分，即廣泛的情感困擾，但仍然可以根據額外的獨特因素進行區分。因此，與廣泛性焦慮症（GAD）相關的明顯生理過度警覺（如果突然發作則伴隨恐慌發作）似乎相對特定於焦慮，而低正情緒（PA）的各種表現（冷漠、行為撤退或遲緩）則明顯特徵化抑鬱，特別是其憂鬱亞綜合症。

Extensive theoretical and empirical work is converging on the conclusion that the relative absence of positive mood and pleasurable experiences are critical in distinguishing depression from anxiety. We have introduced some of these data in the course of our review. We briefly summarize other aspects of this research now. For further discussions, see L. A. Clark and Watson (1991b), Depue, Krauss, and Spoont (1987), Kendall and Watson (1989), Tellegen (1985), Watson, Clark, and Carey (1988), and Watson and Clark (in press-b).
廣泛的理論和實證研究正在趨向於得出結論：積極情緒和愉快經驗的相對缺乏在區分抑鬱症和焦慮症方面至關重要。我們在回顧中介紹了一些這些數據。現在我們簡要總結這項研究的其他方面。欲進一步討論，請參見 L. A. Clark 和 Watson (1991b)、Depue、Krauss 和 Spoont (1987)、Kendall 和 Watson (1989)、Tellegen (1985)、Watson、Clark 和 Carey (1988) 以及 Watson 和 Clark (即將出版-b)。

Most affective states are rather pure markers of either PA or NA. A few, however, are combinations of the two dimensions. Most notably, terms that reflect depressed mood (e.g., sad or blue) or interpersonal disengagement (e.g., lonely or alone) represent a mixture of relatively high NA and moderately low PA (Watson & Clark, 1984; Watson & Tellegen, 1985). These mood data suggest that whereas anxious mood is essentially a state of high NA, depressed mood is a more complex affect that in-
大多數情感狀態都是純粹的正向情感（PA）或負向情感（NA）標記。然而，有一些則是這兩個維度的組合。最明顯的是，反映抑鬱情緒的術語（例如，悲傷或憂鬱）或人際脫離（例如，孤獨或獨自）代表了相對較高的負向情感和適度較低的正向情感的混合（Watson & Clark, 1984; Watson & Tellegen, 1985）。這些情緒數據表明，焦慮情緒本質上是一種高負向情感的狀態，而抑鬱情緒則是一種更複雜的情感。
cludes a significant secondary component of low PA. Consistent with this idea, many existing measures of general anxious symptomatology are predominantly measures of trait NA, whereas corresponding depression scales, although they are strongly related to trait NA, also have a significant low PA component (Watson & Clark, 1984; Watson & Kendall, 1989). ${}^8$${}^8$^(8){ }^{8} This pattern is consistent with the idea that a core set of symptoms specific to depression and quasi-independent of both general NA and a specific anxiety cluster may be identified.
包括一個顯著的低正情緒（PA）次要成分。與此觀點一致的是，許多現有的普遍焦慮症狀量表主要是測量特質負情緒（NA），而相應的抑鬱量表雖然與特質 NA 有很強的關聯，但也具有顯著的低 PA 成分（Watson & Clark, 1984；Watson & Kendall, 1989）。這一模式與這樣的觀點一致，即可以識別出一組特定於抑鬱的核心症狀，並且幾乎獨立於一般的 NA 和特定的焦慮群體。

Data that relates trait NA and PA to symptoms of depression and anxiety support the utility of the PA dimension in differential diagnosis. Watson et al. (1988) found that trait NA was significantly associated with the majority of anxiety symptoms and with 19 of 20 depressive symptoms, whereas trait PA was much more strongly and consistently related to the depressive than to the anxious symptoms. Similarly, trait NA was correlated with the presence of both depressive and anxiety diagnoses, whereas trait PA was consistently related only to the depressive disorders. Thus, NA was nonspecific and reflected the general presence of anxious and depressive symptoms or disorder, whereas PA was specific to depression.
與特質負向情緒（NA）和正向情緒（PA）相關的數據支持了 PA 維度在鑑別診斷中的實用性。Watson 等人（1988）發現，特質 NA 與大多數焦慮症狀以及 20 種抑鬱症狀中的 19 種顯著相關，而特質 PA 則與抑鬱症狀的關聯性遠比與焦慮症狀的關聯性強且一致。同樣，特質 NA 與抑鬱和焦慮診斷的存在相關，而特質 PA 則僅與抑鬱症狀相關。因此，NA 是非特異性的，反映了焦慮和抑鬱症狀或疾病的一般存在，而 PA 則特定於抑鬱。

Studies of dysfunctional cognitions have revealed a similar pattern with measures of positive and negative thinking. For example, the Automatic Thoughts Questionnaire (Hollon & Kendall, 1980) was designed to assess the frequency of negative self-referent thoughts in depression. Whereas depressed patients do score higher on the Automatic Thoughts Questionnaire than various psychiatric groups (Hollon, Kendall, & Lumry, 1986), generally anxious subjects obtain similar scores to depressed subjects (Kendall & Ingram, 1989). Recently, however, measures of positive thinking have been developed that are relatively independent of negative cognitions (e.g., Ingram & Wisnicki, 1988) and show evidence of being more specifically related to depression. For example, the addition of 10 positive statements to the Automatic Thoughts Questionnaire significantly increased its ability to differentiate a group of depressed subjects from a heterogeneous group of psychiatric patients, which included some with panic disorder (Kendall, Howard, & Hays, 1989).
對於功能失調的認知研究顯示出與正向和負向思維測量相似的模式。例如，自動思維問卷（Hollon & Kendall, 1980）旨在評估抑鬱症中負向自我參照思維的頻率。雖然抑鬱患者在自動思維問卷上的得分高於各種精神病群體（Hollon, Kendall, & Lumry, 1986），但一般焦慮的受試者與抑鬱受試者的得分相似（Kendall & Ingram, 1989）。然而，最近已開發出相對獨立於負向認知的正向思維測量（例如，Ingram & Wisnicki, 1988），並顯示出與抑鬱症更具特定相關性的證據。例如，向自動思維問卷中添加 10 個正向陳述顯著提高了其區分一組抑鬱受試者與一組異質精神病患者（其中包括一些驚恐障礙患者）的能力（Kendall, Howard, & Hays, 1989）。

Watson and Kendall (1989) summarized extensive data in regard to those factors that anxiety and depression share in contrast to those that differentiate these syndromes. Several
沃森和肯達爾（1989）總結了大量數據，關於焦慮和抑鬱之間的共同因素，以及區分這些綜合症的因素。幾個

specific factors they identified can be conceptualized in terms of low PA. For example, the loss or absence of pleasurable life experiences seen in depressive syndromes is clearly associated with low levels of positive mood. It is noteworthy that this phenomenon is not tied to a particular causal model: Behavioral researchers focus on the insufficiency of environmental reinforcers (e.g., Foa, Rothbaum, & Kozak, 1989; Rehm, 1989), cognitive theorists have suggested that depressive mood states bias against processing of positive self-relevant information, and still others emphasize the inability of depressed persons to enjoy pleasant events for reasons that are either psychological (Costello, 1972) or biological (Klein, 1974; Meehl, 1975) in origin. Similarly, the behavioral deficits seen in depressive syndromes can be interpreted as manifestations of low levels of positive emotional arousal (Safran & Greenberg, 1989). Moreover, PA-but not NA-shows seasonal and diurnal variations, which have also been documented for depression but not for anxiety (e.g., Depue et al., 1987; Kasper & Rosenthal, 1989; Healy & Williams, 1988). Thus, several lines of research suggest that the lack of positive affective experience is specifically associated with depressive symptomatology and differentiates it from anxiety-related phenomena.
他們所識別的具體因素可以從低正向情緒的角度來概念化。例如，在抑鬱綜合症中，愉快生活經歷的喪失或缺失明顯與低水平的正向情緒相關。值得注意的是，這一現象並不與特定的因果模型相關：行為研究者專注於環境增強物的不足（例如，Foa、Rothbaum 和 Kozak，1989；Rehm，1989），認知理論家則建議抑鬱情緒狀態對正向自我相關信息的處理存在偏見，還有其他人強調抑鬱者因心理（Costello，1972）或生物（Klein，1974；Meehl，1975）原因而無法享受愉快事件的能力。同樣，抑鬱綜合症中所見的行為缺陷可以解釋為低水平正向情緒喚起的表現（Safran & Greenberg，1989）。此外，正向情緒—而非負向情緒—顯示出季節性和日間變化，這一點在抑鬱症中也有記錄，但在焦慮症中則沒有（例如，Depue 等，1987；Kasper & Rosenthal，1989；Healy & Williams，1988）。 因此，幾條研究線索表明，缺乏正向情感體驗與抑鬱症狀特別相關，並將其與焦慮相關現象區分開來。

The conclusions that emerge from the psychometric data are clear and can be stated succinctly: Anxious and depressed syndromes share a significant nonspecific component that encompasses general affective distress and other common symptoms, whereas these syndromes are distinguished by physiological hyperarousal (specific to anxiety) versus the absence of PA (specific to depression). This tripartite view implies that a complete description of the affective domain requires assessing both the common and unique elements of the syndromes: general distress, the physiological tension and hyperarousal of anxiety, and the pervasive anhedonia of depression. Neither general distress nor the syndrome-specific symptom clusters alone can completely describe these syndromes; rather, they jointly define the domain. These psychometric results thus provide a theoreticalempirical framework for interpreting a great deal of clinical and epidemiologic data and for developing a more satisfactory nosology in this area. That is, we believe the problems of diagnostic comorbidity and optimal classification of anxious and depressive disorders can be understood best in terms of this recurring, tripartite division of symptoms.
從心理測量數據中得出的結論是明確的，可以簡潔地表述：焦慮和抑鬱綜合症共享一個重要的非特異性成分，涵蓋一般情感困擾和其他常見症狀，而這些綜合症則以生理過度警覺（特有於焦慮）與缺乏正情緒（特有於抑鬱）來區分。這種三元觀點意味著，對情感領域的完整描述需要評估綜合症的共同和獨特元素：一般困擾、焦慮的生理緊張和過度警覺，以及抑鬱的普遍快感缺失。僅僅依賴一般困擾或特定綜合症的症狀群無法完全描述這些綜合症；相反，它們共同定義了這一領域。因此，這些心理測量結果為解釋大量臨床和流行病學數據提供了一個理論實證框架，並為在這一領域發展更滿意的分類系統奠定了基礎。 也就是說，我們認為診斷共病問題和焦慮及抑鬱障礙的最佳分類，最能通過這種反覆出現的三分法來理解。

However, the question of how these factors are combined in persons remains unanswered. Are anxiety and depression entities that are strongly correlated because of their many common symptoms, yet whose specific components differentiate them sufficiently to define them as distinct syndromes? Or have attempts to differentiate anxiety and depression failed in part because there are sizable groups of patients who cannot be meaningfully categorized simply as either anxious or depressed because they either exhibit a wide variety of both types of specific symptoms or else show primarily nonspecific symptoms?
然而，這些因素在個體中如何結合的問題仍然沒有答案。焦慮和抑鬱是否因其許多共同症狀而強烈相關，但其具體成分又足以使它們被定義為不同的綜合症？還是說，區分焦慮和抑鬱的嘗試部分失敗，因為有相當大的一群患者無法僅僅被有意義地分類為焦慮或抑鬱，因為他們要麼表現出各種各樣的特定症狀，要麼主要顯示非特定症狀？

The data suggest that both of these views may be true and highlight the importance of distinguishing between symptom and diagnostic levels. That is, the correlation between ratings of anxious and depressive symptomatology may simply reflect the
數據表明這兩種觀點可能都是正確的，並強調區分症狀和診斷層級的重要性。也就是說，焦慮和抑鬱症狀評分之間的相關性可能僅僅反映了
fact that anxiety and depression share many distress symptoms rather than indicate diagnostic overlap. Certainly it is possible to identify many patients who have a diagnosis of anxiety but not depression or vice versa. For instance, to turn the comorbidity data reported by L. A. Clark (1989) around, one third of patients whose primary diagnosis is panic or agoraphobia do not have a depression diagnosis, whereas over two thirds of those with simple or social phobias do not. Similarly, roughly half of all patients diagnosed with primary depression have no anxiety diagnosis. Thus, patients with depressive disorders may have substantial anxious symptomatology, or vice versa, because of shared symptoms, without showing the full disorder in the other domain.
焦慮和抑鬱共享許多痛苦症狀的事實，而不是表明診斷上的重疊。當然，可以確定許多患者的診斷是焦慮而不是抑鬱，反之亦然。例如，根據 L. A. Clark（1989）報告的共病數據，三分之一的主要診斷為驚恐症或廣場恐懼症的患者並沒有抑鬱診斷，而超過三分之二的簡單或社交恐懼症患者則沒有。同樣，大約一半的主要抑鬱症患者沒有焦慮診斷。因此，抑鬱症患者可能因為共享症狀而表現出相當程度的焦慮症狀，反之亦然，而不會在另一個領域顯示出完整的疾病。

On the other hand, researchers have amply documented that many affective disorder patients show a mixed anxious-depressed symptom picture that cannot easily be characterized as one type of disorder or the other (Downing & Rickels, 1974; Gersh & Fowles, 1982; Hollister et al., 1967; Paykel, 1971, 1972). Such patients, who may or may not meet criteria for current $DSM-III-R$$DSM-III-R$DSM-III-RD S M-I I I-R diagnoses, are frequently identified in general medical samples (Katon & Roy-Byrne, 1991 ; Klerman, 1989).
另一方面，研究人員充分記錄了許多情感障礙患者表現出混合的焦慮-抑鬱症狀，這種症狀圖景不容易被歸類為某一類型的障礙（Downing & Rickels, 1974；Gersh & Fowles, 1982；Hollister et al., 1967；Paykel, 1971, 1972）。這些患者可能符合或不符合當前診斷的標準，經常在一般醫療樣本中被識別出來（Katon & Roy-Byrne, 1991；Klerman, 1989）。

Moreover, research on comorbid diagnostic patterns has demonstrated that patients who meet criteria for a diagnosis of both anxiety and depression represent a distinctive group, with significantly poorer treatment response and outcome, more severe clinical presentation of both syndromes, and greater chronicity (Breier et al., 1984; Clancy, Noyes, Hoenk, & Slymen, 1978; Grunhaus, 1987, 1988; Grunhaus et al., 1986; Maser & Cloninger, 1990; Stavrakaki & Vargo, 1986; Van Valkenburg et al., 1984). They also scored significantly higher on a factor-analytic measure of general distress (NA) than did patients with a diagnosis of only anxiety or only depression (L. A. Clark & Watson, 1991b). The synergistic aspect of this comorbidity, which Grunhaus (1988) suggested reflects a “dual diathesis” (p. 1214), is inconsistent with the notion of simple co-occurrence of distinctive syndromes due to shared symptoms.
此外，對共病診斷模式的研究顯示，符合焦慮和抑鬱診斷標準的患者代表一個獨特的群體，其治療反應和結果顯著較差，兩種綜合症的臨床表現更為嚴重，且慢性程度更高（Breier et al., 1984; Clancy, Noyes, Hoenk, & Slymen, 1978; Grunhaus, 1987, 1988; Grunhaus et al., 1986; Maser & Cloninger, 1990; Stavrakaki & Vargo, 1986; Van Valkenburg et al., 1984）。他們在一般痛苦的因素分析測量（NA）上得分顯著高於僅有焦慮或僅有抑鬱診斷的患者（L. A. Clark & Watson, 1991b）。Grunhaus（1988）所建議的這種共病的協同作用，反映了一種“雙重易感性”（第 1214 頁），與由於共享症狀而導致的獨特綜合症的簡單共存概念不一致。

Although a complete exploration of the implications of this tripartite model for the classification of affective disorders is beyond our scope, a few observations are in order. In general, the data indicate that elevated levels of the nonspecific component will nearly always be evident in anxious or depressed patients; indeed, dysfunctionally high NA essentially signals the presence of these disorders (although lower levels of trait NA may be seen in subjects with highly circumscribed disorders, such as simple phobias). Thus, elevated NA suggests the general relevance of anxiety-depression diagnoses (and perhaps other diagnoses as well) but in and of itself offers little basis for finer discrimination; rather, further differentiation is provided by the two specific factors. Relatively low or high levels on both of these factors together suggests a mixed mood disorder, and we submit that the data support the addition of a diagnosis of mixed anxiety-depression to the current classification system.
儘管對這一三元模型在情感障礙分類中的含義進行全面探討超出了我們的範疇，但有幾點觀察是必要的。一般來說，數據顯示，非特異性成分的升高幾乎總是會在焦慮或抑鬱患者中明顯存在；事實上，功能性過高的負面情緒基本上標誌著這些障礙的存在（儘管在具有高度局限性障礙的受試者中，如簡單恐懼症，可能會看到較低的特質負面情緒水平）。因此，升高的負面情緒暗示了焦慮-抑鬱診斷的一般相關性（也許還有其他診斷），但就其本身而言，對於更細緻的區分提供的依據有限；相反，進一步的區分由這兩個特定因素提供。這兩個因素的相對低或高水平共同暗示了一種混合情緒障礙，我們認為數據支持在當前分類系統中增加混合焦慮-抑鬱診斷。

It will be important to draw up the criteria for this disorder in such as way as to discourage its use as a “diagnostic landfill” For example, we believe the field has sufficient psychometric sophistication to permit quantification of level of affective dis-
制定此疾病的標準將非常重要，以避免其被用作“診斷垃圾場”。例如，我們認為該領域擁有足夠的心理測量學複雜性，可以量化情感障礙的程度
tress, similar to the use of specific IQ levels in the diagnosis of mental retardation. Furthermore, specific numbers of clinically significant symptoms or other criteria can be used to designate mild, moderate, and severe variants of the diagnosis. If developed in this way, the diagnosis will not represent a retreat from the goal of diagnostic specificity.
髮絲，類似於在智力障礙診斷中使用特定智商水平。此外，可以使用臨床上顯著症狀的具體數量或其他標準來指定診斷的輕度、中度和重度變異。如果以這種方式發展，診斷將不會代表對診斷特異性目標的退縮。

Patients whose predominant symptoms are nonspecific (distress, demoralization, irritability, mild disturbances of sleep and appetite, distractibility, and vague somatic complaints) and who show low (or moderate) levels of both specific factors-that is, show neither marked psychophysiological symptoms nor an-hedonia-will receive a diagnosis of mixed anxiety-depression, mild (or moderate). This category is essentially the diagnosis that is already recognized in the 10 th edition of the International Classification of Diseases and Related Health Problems (World Health Organization, 1990) and that is slated for the DSM-IV field trials. Patients with this diagnosis are probably most prevalent in general medical populations but are certainly not uncommon in psychiatric settings.
主要症狀為非特異性（痛苦、沮喪、易怒、輕微的睡眠和食慾障礙、注意力不集中以及模糊的身體不適）的患者，若在特定因素上顯示低（或中等）水平，即不顯示明顯的心理生理症狀或無快感，將被診斷為輕度（或中度）混合焦慮抑鬱症。這一類別本質上是國際疾病分類第十版（世界衛生組織，1990 年）中已被認可的診斷，並計劃納入 DSM-IV 的實地試驗。這一診斷的患者在一般醫療人群中可能最為普遍，但在精神科環境中也絕對不罕見。

On the other hand, patients who report not only very high levels of general distress but also both anhedonia and psychophysiological hyperarousal will be diagnosed as mixed anxietydepression, severe. This diagnosis may potentially be reserved for patients who fully meet criteria for both an anxiety and a depressive disorder, either simultaneously or longitudinally. Although the two component diagnoses can, of course, be assigned independently, use of the diagnosis mixed anxiety-depression, severe recognizes the synergistic quality of the dual diagnosis. Such a diagnosis will represent a lifetime diagnosis, much as bipolar disorder does currently, because episodes of marked anxiety and anhedonic depressive episodes do not necessarily occur simultaneously in these patients (Breier et al., 1984). To follow the bipolar disorder analogy, alternate forms such as mixed anxiety-depression-depressed may be used to designate the current episode.
另一方面，報告不僅有非常高的普遍痛苦水平，還有無快感和心理生理過度興奮的患者將被診斷為混合焦慮抑鬱症，重度。這一診斷可能專門保留給完全符合焦慮症和抑鬱症標準的患者，無論是同時出現還是隨時間推移出現。雖然這兩個組成診斷當然可以獨立分配，但使用混合焦慮抑鬱症，重度的診斷承認了雙重診斷的協同特性。這樣的診斷將代表終身診斷，正如目前的雙相情感障礙一樣，因為這些患者的顯著焦慮和無快感的抑鬱發作不一定同時發生（Breier et al., 1984）。為了跟隨雙相情感障礙的類比，可以使用混合焦慮抑鬱-抑鬱等替代形式來指定當前的發作。

Finally, the consideration of lifetime diagnoses leads us to the issue of the role of chronicity in defining psychiatric syndromes. The shared general distress factor is manifested both as a transient state and as a more stable trait. The relative stability of trait NA is well documented, with 12-year retest correlations of .70 and higher (L. A. Clark & Watson, 1991a). Moreover, genetic studies with diverse methodologies have consistently shown a significant heritability for trait NA (e.g., Carey & Gottesman, 1981; Jardine, Martin, & Henderson, 1984; Kendler, Heath, Martin, & Eaves, 1987; Loehlin, Willerman, & Horn, 1987; Pedersen, Plomin, McClearn, & Friberg, 1988; Rose, 1988; Tellegen et al., 1988; see L. A. Clark & Watson, 1991 (a, for a review).
最後，對於終生診斷的考量使我們面臨慢性在定義精神病症候群中的角色問題。共同的普遍痛苦因素既表現為短暫的狀態，也表現為更穩定的特徵。特徵性負情緒的相對穩定性有充分的文獻記載，12 年的重測相關性達到 0.70 或更高（L. A. Clark & Watson, 1991a）。此外，使用多種方法的遺傳研究一致顯示特徵性負情緒具有顯著的遺傳性（例如，Carey & Gottesman, 1981；Jardine, Martin, & Henderson, 1984；Kendler, Heath, Martin, & Eaves, 1987；Loehlin, Willerman, & Horn, 1987；Pedersen, Plomin, McClearn, & Friberg, 1988；Rose, 1988；Tellegen et al., 1988；參見 L. A. Clark & Watson, 1991（a，回顧）。

These data indicate that the high levels of general distress and nonspecific symptoms reported by many patients are likely to be a manifestation of trait NA, which is rather chronic in nature. Indeed, in Hays’s (1964) investigation of modes of illness onset, he described an anxious-depressed group with long-standing neurotic symptoms who later developed depression (see also Gersh & Fowles, 1982). Breslau and Davis (1985) also noted that when the duration requirement for GAD was increased from 1 to 6 months, the lifetime rate of major depressive disorder increased from $23\mathrm{\%}$$23\mathrm{\%}$23%23 \% to $67\mathrm{\%}$$67\mathrm{\%}$67%67 \%. Some of this increase may represent state effects (i.e., some subjects may develop a
這些數據表明，許多患者報告的高水平一般困擾和非特異性症狀可能是特質神經質（NA）的表現，這種特質本質上相當慢性。事實上，在 Hays（1964）對疾病發作模式的研究中，他描述了一組焦慮抑鬱的患者，這些患者有長期的神經症狀，後來發展為抑鬱症（另見 Gersh & Fowles，1982）。Breslau 和 Davis（1985）也注意到，當廣泛性焦慮症（GAD）的持續時間要求從 1 個月增加到 6 個月時，重度抑鬱症的終生發病率從 $23\mathrm{\%}$$23\mathrm{\%}$23%23 \% 增加到 $67\mathrm{\%}$$67\mathrm{\%}$67%67 \% 。這一增長中的部分可能代表狀態效應（即，一些受試者可能會發展出一種）。
depressive syndrome in response to persistent anxiety). ${}^9$${}^9$^(9){ }^{9} However, because these were lifetime depression rates, it is also likely that chronicity itself is an important criterion in the diagnosis of mixed anxiety-depression. That is, we have already noted that patients who meet criteria for both an anxious and a depressive disorder are higher on NA and typically show a more chronic course than those with only one type of disorder. Breslau and Davis’s (1985) data suggest that the reverse is also true: As subjects with more chronic (i.e., trait) NA are identified, the prevalence of a mixed syndrome also increases. Of course, if mixed anxiety-depression is marked by chronicity, the potential overlap with the Axis II personality disorders must also be considered, a topic that is beyond the scope of this article (see Widiger & Shea, 1991).
持續焦慮所引起的抑鬱綜合症）。 ${}^9$${}^9$^(9){ }^{9} 然而，由於這些是終生抑鬱率，因此慢性本身也可能是混合焦慮-抑鬱診斷中的一個重要標準。也就是說，我們已經注意到，符合焦慮和抑鬱障礙標準的患者在負面情緒（NA）上較高，並且通常顯示出比僅有一種障礙的患者更慢性的病程。Breslau 和 Davis（1985）的數據表明，反之亦然：隨著識別出更多慢性（即，特質）負面情緒的受試者，混合綜合症的流行率也會增加。當然，如果混合焦慮-抑鬱以慢性為特徵，則必須考慮與第二軸人格障礙的潛在重疊，這是一個超出本文範疇的主題（見 Widiger & Shea, 1991）。

In conclusion, the data we have reviewed provide a framework for understanding affective syndromes in terms of their specific and nonspecific components. In particular, we feel they argue strongly for the development of a new diagnostic category that formally recognizes the importance of the pervasive and highly general trait of neuroticism and negative affectivity. This factor emerges as a ubiquitous and inescapable force in psychometric data. Currently, its strong-yet often un-recognized-presence seriously hampers attempts to forge a satisfactory diagnostic taxonomy in this area. By formally recognizing the existence of this important dimension, psychiatric classification will be operating from a position of much greater strength and will have advanced significantly toward its ultimate nosological goal.
總結來說，我們所審查的數據提供了一個框架，以理解情感綜合症的特定和非特定組成部分。特別是，我們認為這些數據強烈主張發展一個新的診斷類別，正式承認神經質和負面情感這一普遍且高度一般化特徵的重要性。這一因素在心理測量數據中顯現為一種無處不在且不可避免的力量。目前，它強而有力但常常未被認識的存在，嚴重妨礙了在這一領域建立令人滿意的診斷分類系統的嘗試。通過正式承認這一重要維度的存在，精神病學分類將從一個更強大的立場運作，並將在其最終的病理學目標上顯著向前推進。

Akiskal, H. S. (1990). Toward a clinical understanding of the relationship of anxiety and depressive disorders. In J. Maser & R. Cloninger (Eds.), Comorbidity in anxiety and mood disorders (pp. 597-607). Washington, DC: American Psychiatric Press.
Akiskal, H. S. (1990)。朝向對焦慮與抑鬱症之關係的臨床理解。在 J. Maser 和 R. Cloninger（編輯），焦慮與情緒障礙的共病（第 597-607 頁）。華盛頓特區：美國精神醫學出版社。
American Psychiatric Association. (1980). Diagnostic and statistical manual of mental disorders (3rd ed.). Washington, DC: Author.
美國精神醫學會。(1980)。精神疾病診斷與統計手冊（第 3 版）。華盛頓特區：作者。
American Psychiatric Association. (1987). Diagnostic and statistical manual of mental disorders (Rev. 3rd ed.). Washington, DC: Author.
美國精神醫學會。(1987)。精神疾病診斷與統計手冊（修訂第三版）。華盛頓特區：作者。
Aylard, P. R., Gooding, J. H., McKenna, P. H., & Snaith, R. P. (1987). A validation study of three anxiety and depression self-assessment scales. Journal of Psychosomatic Research, 31, 261-268.
Aylard, P. R., Gooding, J. H., McKenna, P. H., & Snaith, R. P. (1987)。三種焦慮和抑鬱自我評估量表的驗證研究。《心理生理研究雜誌》，31，261-268。
Barlow, D. H., DiNardo, P. A., Vermilyea, B. B., Vermilyea, J., & Blanchard, E. (1986). Co-morbidity and depression among the anxiety disorders. Journal of Nervous and Mental Disease, 174, 63-72.
Barlow, D. H., DiNardo, P. A., Vermilyea, B. B., Vermilyea, J., & Blanchard, E. (1986)。焦慮症中的共病與抑鬱。神經與精神疾病雜誌，174，63-72。
Beck, A. T. (1972). The phenomena of depression: A synthesis. In D. Offer & D. X. Freedman (Eds.), Modern psychiatry and clinical research (pp. 136-158). New York: Basic Books.
貝克，A. T.（1972）。抑鬱現象：一個綜合。在 D. Offer 和 D. X. Freedman（編），現代精神醫學與臨床研究（第 136-158 頁）。紐約：基本書籍。
Beck, A. T., Epstein, N., Brown, G., & Steer, R. (1988). An inventory for measuring clinical anxiety: Psychometric properties. Journal of Consulting and Clinical Psychology, 56, 893-897.
貝克，A. T.，艾普斯坦，N.，布朗，G.，& 斯蒂爾，R.（1988）。用於測量臨床焦慮的清單：心理測量特性。《諮詢與臨床心理學期刊》，56，893-897。
Beck, A. T., Steer, R. A., & Garbin, M. G. (1988). Psychometric properties of the Beck Depression Inventory: Twenty-five years of evaluation. Clinical Psychology Review, 8, 77-100.
貝克，A. T.，斯蒂爾，R. A.，& 加賓，M. G.（1988）。貝克抑鬱量表的心理測量特性：二十五年的評估。《臨床心理學評論》，8，77-100。
Beck, A. T., Ward, C. H., Mendelson, M., Mock, J. E., & Erbaugh, J. K. (1961). An inventory for measuring depression. Archives of General Psychiatry, 4, 561-571.
貝克，A. T.，沃德，C. H.，門德爾森，M.，莫克，J. E.，& 埃爾巴赫，J. K.（1961）。用於測量抑鬱的清單。《一般精神病學檔案》，4，561-571。
Benshoof, B. B., Moras, K., DiNardo, R., & Barlow, D. (1989). A comparison of symptomatology in anxiety and depressive disorders. Unpub-
Benshoof, B. B., Moras, K., DiNardo, R., & Barlow, D. (1989)。焦慮症和抑鬱症的症狀比較。未出版-