Dimensional Structure of the Arabic Positive Affect and Negative Affect Scale Adapted From its English Form

Abstract

The purpose of this study was to examine the dimensional structure of the Arabic version of the Positive Affect and Negative Affect Schedule using a sample of undergraduate students from a private university in the Sultanate of Oman. Confirmatory factor analysis was conducted to test four preconceptualized item-fit models: a one-factor structure model, a two-factor model using a factor structure of items converging on Positive Affect and Negative Affect, a correlated two-factor model, and finally the correlated three-factor model. Strongest support was found for the correlated two-factor model. A recent study provided further evidence of the robust structure of the Positive Affect and Negative Affect Schedule using the two-factor model. This study tested the model in a non-Western culture and a population that was very different from that in previous studies. The implications of these findings and recommendations are discussed herein.

Keywords

Dimensional structure positive and negative affect emotions translation factor structure validity

Introduction

In recent years, the topic of affect, emotions, and feelings has gained increasing interest. One can just peruse the literature on emotion and find several studies done in the last few decades. The concept of emotion is a complex set of feelings that could result in behavioral changes whether physical or psychological affect is the experience of these feelings. Research on feelings has been approached from different theoretical perspectives. For example, Ekman (1993) looked at discrete emotions and focused on facial expression and emotions. His research across many decades found that different aspects of facial expression of emotion can be universal and can also be culture-specific. Other researchers like Plutchik (1994) recognized that the structure of emotions has basic and distinctive categories such as anxiety or depression. The predominant work on emotions has looked at the outcomes, specifically whether they exhibited negative or positive affect and ways to measure these affects (e.g., Schimmack & Grob, 2000; Watson, Clark, & Tellegen, 1988). The development of psychometric properties is to assess emotions. Predominantly, researchers have viewed the bipolarity of emotions as a persistent disposition of human behavior. Although there is some convergence in many of these approaches, there still remain many issues that are subject to debate and many unanswered questions on the structure of emotion, especially in non-Western contexts.

Historically, research on the structure of affect has led to strong debate. Wundt’s (1896) structuralism suggested that affective experiences are a mixture of six basic feelings and fit in three dimensions. Wundt’s mentee, Titchener (1908) suggested that pleasure–displeasure is the basic dimension of affect, and therefore, the structure of affect should be one-dimensional. Schimmack and Grob (2000) further argued how these disagreements on the structure of affect persist to this day. They explained that these differences in the structure of affect could be due to many factors, including cultural differences. They also suggested that these differences could be due to the methodological approaches taken to measure emotions and the statistical methods used to assess the structure of affect.

Positive Affect and Negative Affect

The Positive Affect (PA) and Negative Affect (NA) Schedule (PANAS) model of PA and NA was developed by Watson and Tellegen (1985). PANAS model of emotion proposes a two-factor model (Watson & Tellegen, 1985; Watson et al., 1988). This model looks at emotions as a measure of affect and is the most widely used instrument to assess the levels of NA and PA. It is a self-reported scale that is easy to use and that measures basic human affect (Watson & Tellegen, 1985). The affect model is one of the most influential and well-tested models of affect and has been widely and globally used in assessing PA and NA. Watson and Tellegen (1985) found that human affect could be categorized into 20 dispositions, both positive and negative. Thus, the PANAS scale is comprised of 20 adjective emotions; 10 indicators of the PANAS have an NA connotation and 10 have a PA connotation.

The NA and PA reflect dispositional dimensions, and the attributes of NA can be described as the absence of feelings. A high NA suggests distress and unpleasant feelings. By contrast, PA represents the level of pleasure that an individual experiences within an environment (Crawford & Henry, 2004). Although the model is seen as bipolar, empirical evidence does not reflect the “perpendicularities” of this model. For instance, “emotions such as enthusiasm and alertness are indicative of high PA, while lethargy and sadness characterize low PA” (Crawford & Henry, 2004, p. 246). Hence, while negative emotions are not seen as dispositions due to high NA, they may, to some extent, reflect low ratings on the PA.

The results have been mixed, and empirical evidence among different researchers still remains contentious. Mehrabian (1997) was one of the earliest researchers who questioned this two-factor structure of the PANAS. In his research using exploratory factor analysis, he found that the two-factor and three-factor models were both possible, though the three-factor model seemed to have a better fit. Furthermore, Tuccitto, Giacobbi, and Leite (2010) and Crawford and Henry (2004) found that the best-fitting model was the two-factor model. The relationship between the main factors of NA and PA was correlated according to Tuccitto et al. (2010) and independent according to Crawford and Henry (2004). Thus, the PANAS is a well-recognized score of emotion, but it still has some interpretative issues in explaining NA and PA. The psychometric properties of the PANAS have demonstrated consistently high reliability and validity measures among the preconceptualized negative and positive dimensions. Many other researchers have since tested the factor structure and found that the three-factor and two-factor models are both tenable. The review of the literature below addresses these issues.

Cross-cultural studies of the PANAS

Emotions, whether conceived as discrete emotions or as affect, are also contextually and individually felt (Perunovic, Heller, & Rafaeli, 2007). An individual’s emotions and feelings are not easily differentiated due to language, culture, and affect. Thus, it would be interesting to see how individuals living in cultures different from Western cultures differ in how they express their emotions and feelings (e.g., Lazarus, 1999).

The PANAS scale has been widely used in many cultures and ethnic groups to assess its generalizability across groups. It has been translated into many languages; these studies found that both the three-factor and two-factor structures seem to be suitable. Studies have been done using the PANAS on Italian, German, Spanish, Turkish, Swedish, and in many other languages and cultures (e.g., Galinha, Pereire, & Esteves, 2013; Heubeck & Wilkinson, 2019; Leue & Beauducel, 2011; Seib-Pfeifer, Pugnaghi, Beauducel, & Leue, 2017; Terracciano, McCrae, & Costa, 2003). Some of these studies have found that the original two-factor model is valid, while others have found the three-factor model to be more superior than the two-factor structure. In one study done in the United Kingdom (Crawford & Henry, 2004) with a nonclinical sample, the structure of the scale emerged as two-correlated factors mainly preconceived as the PA and NA. Gaudreau, Sanchez, and Blondin (2006) conducted another study to look at the alternative factorial structures of the French-Canadian version on a sample of athletes. They found that a modified three-factor model had a better fit than the modified two-factor models. Terracciano et al. (2003) looked at the Italian version of the scale and found that the Italian version was a reliable and valid version of the original two-factor structure and had achieved the best fit. The dimensional structure of the Spanish version (Ortuño-Sierra, Santarén-Rosell, De Albéniz, & Fonseca-Pedrero, 2015) supported a two-factor model and a three-factor model of PA, NA-Upset, and NA-Afraid; PA and NA were uncorrelated.

Leue and Beauducel (2011), using confirmatory factor analysis (CFA), conducted two studies that looked at the dimensional structure of the German version of the PANAS. In their first study, they found support for an uncorrelated two-factor model of PA and NA and the three-factor model of PA, NA-Afraid, and NA-Upset as factors. Furthermore, a study that examined the factor structure of a Portuguese version of the scale (Galinha et al., 2013) showed that the Portuguese model consisted of two independent factors, PA and NA, with the cross loading of the item “excited” between PA and NA. Pandey and Srivastava (2008) studied the psychometric properties of the scale on an Indian sample with a Hindi version of the PANAS scale. Exploratory factor analysis revealed a two-factor structure with a low negative correlation between the two-factor structure. Table 1 presents a brief overview of some prior studies, including the language used and the kind of statistical analysis and type of tested model.

Table 1.

Characteristics of studies investigating the structure of the Positive and Negative Affect Schedule.

Author(s)	Language	Preferred factor model	Factor	Type of analysis
Beck et al. (2003)	English	Uncorrelated FoF	PA, NA-1, NA-2	EFA
Crawford and Henry (2004)	English	Correlated FoF; CE permitted	PA, NA	CFA
Galinha et al. (2003)	Portuguese	Uncorrelated	PA, NA	CFA
Crocker (1997)	English	Correlated FoF	PA, NA	CFA
Gaudreau et al. (2006)	French	Correlated FoF	PA, Upset, Afraid	CFA
Hillerås, Jorm, Herlitz, and Winblad (1998)	Canadian	Uncorrelated FoF	PA, NA (10 items)	EFA
Joiner et al. (1997)	Swedish	Correlated FoF	PA, NA	CFA
Kercher (1992)	Spanish	Correlated FoF	PA, NA (10 items)	CFA
Killgore (2000)	English	Correlated FoF	PA, Afraid, Upset	EFA
Krohne, Egloff, Kohlmann, and Tausch (1996)	English	Uncorrelated FoF	PA, NA	EFA
Leue and Beauducel (2011)	German	Uncorrelated	PA, NA PA, NA-1,NA-2	CFA
Lonigan, Hooe, David, and Kistner (1999)	German	Uncorrelated FoF	PA, NA	CFA
Mackinnon et al. (1999)	English	Correlated FoF	PA, NA (10 items)	CFA
Mehrabian (1997)	English	1 FoF, 1 correlated higher order factor	PA, NA (Afraid, Upset)	CFA
Melvin and Molloy (2000)	English	Uncorrelated FoF	PA, NA	EFA, CFA
Ortuño-Sierra et al. (2015)	Spanish	Uncorrelated	PA, NA-1, NA-2	CFA
Molloy, Pallant, and Kantas (2001)	English	Correlated FoF	PA, NA	CFA
Pandey and Srivastava (2008)	Hindi	Correlated FoF	PA,NA	EFA
Shapiro, Roberts, and Beck (1999)	English	Uncorrelated FoF	PA, NA	EFA, CFA
Terracciano et al. (2003)	Italian	Correlated FoF	PA, NA	CFA
Watson et al. (1988)	English	Uncorrelated FoF	PA, NA	EFA

CFA: confirmatory factor analysis; EFA: exploratory factor analysis; PA: positive affect; NA: negative affect; FoF: first-order factor(s); CE: correlated errors (Leue & Beauducel, 2011).

To summarize, the dimensional structure of the PANAS is still under debate, and many questions still remain unanswered. First, in many studies, both the two-factor and three-factor models seemed to be a good fit. Second, there have been mixed results regarding the independence of these factors; some studies have found the factors correlated, but others have found that they are independent.

There may be cross-cultural differences between the way Western and Eastern cultures react to negative and positive events. In particular, there may be differences in the way both cultures perceive, experience, and regulate their emotions. For example, some studies in Asia have examined how culture can affect the relation between pleasant emotions and unpleasant emotions in the Asian context. Schimmack, Oishi, and Diener (2002) suggested that Asians have a dialectical way of thinking and therefore can experience emotions of the opposite valence (e.g., happy, sad) as compatible with each other. This way of thinking creates a substantively different psychometric structure of the PANAS than what was assessed from a Western perspective. Some research has also demonstrated that in Western cultures, individuals may engage more in self-enhancement, but in contrast, individuals from Eastern cultures may engage more in self-reflection (Chang & Asakawa, 2003; Kitayama, Markus, Matsumoto & Norasakkunkit, 1997). In this study, we draw on these aspects of culture to conjecture that with an Arab sample, a positive correlated factor structure of the NA with PA might appear.

Other studies have also shown that Eastern cultures have more frequent negative/pessimistic outcome cognitions, and Western cultures may be more likely to have positive/optimistic outcome cognitions (Chang & Asakawa, 2003). This is consistent with some of the research done by some of the present authors on locus of control, where there was a tendency for more fatalism in Eastern cultures compared to a tendency for more self-deterministic thinking in Western cultures (Narayanan, Menon, & Spector, 1999a, 1999b; Nasser & Abouchedid, 2002).

The Present Study

The purpose of this study is to understand the dimensional structure of the PANAS-Arabic (PANAS-A) scale among a population that has not been studied previously. Most of the previous research on the PANAS was focused on Western cultures and, in some cases, in a few countries in Asia. There is little research on the PANAS in the Middle Eastern context. We looked at the dimensional structure of the PANAS with a sample of Omani university students in the Middle East. To test the factor structure of the PANAS-A scale model, we used previous models that assessed the factor structure of the PANAS using CFA.

Method

Participants

The study sample comprised undergraduate university students whose ages ranged between 18 and 45 years (mean = 31.12, SD = 4.25). The sample consisted of 340 university students from a private university in the Sultanate of Oman (155 males and 185 females). They all completed the 20-item PANAS-A scale (Watson & Tellegen, 1988). The participants majored in different subjects and were selected from the College of Arts and Applied sciences, the College of Business, and the College of Engineering.

Measure

The PANAS has been used in several studies (Crawford & Henry, 2004; DePaoli & Sweeney, 2000; Ortuño-Sierra et al., 2015) to assess positive and negative affect of adults. The PANAS consists of 20 words that describe the different types of affect. Ten items of the PANAS assess PA (e.g., attentive) and 10 items assess NA (e.g., hostile). In this study, participants responded to each item on a five-point Likert-type scale: 1 = “very slightly or not at all,” 2 = “a little,” 3 = “moderately,” 4 = “quite a bit,” and 5 = “extremely.” The higher the score on the PANAS, the greater the affect was. The time frame adopted for students to respond was open. The scores range from 10 to 50 for both PA and NA items.

Translation of the instrument

The PANAS was translated from English to Arabic using the committee translation method, where a panel of experts translated from a source language (English) to a target language (Arabic). If all the translations were the same, then the translation could be considered valid in the target language (Nasser, 2005). The committee was made up of three bilingual academics. Each translated the 20 items and the scale directions independently, the translators being blind to one another’s translations. The three blind translations were then compared by a bilingual translator, who did not take part in the translation procedure, to check the convergence of the translations. A challenge arose with one adjectival term of the PA, “excited,” which was translated into the literal term “aroused.” However, two other translators believed an appropriate and adequate Arabic translation was “Farhan.” This also has another meaning in Arabic, and in specific contexts, can be read and understood as “happy” rather than “excited.” In the final stage of the translation, one bilingual judge made the “best” translation by conferencing and reconferencing with the three bilingual translators until agreement was reached that the translations were adequate, appropriate, and applicable for all items.

The PANAS-A instrument

Evidence of internal consistency of the PANAS scores ranged from 0.86 to 0.90 for PA and from 0.84 to 0.87 for NA (Watson et al., 1988). Other researchers have found alpha reliabilities within these ranges (Huebner & Dew, 1995; Melvin & Molloy, 2000).

The internal consistency of the PANAS-Arabic (PANAS-A) scale was examined using Cronbach’s alpha. The PANAS-A subscales showed satisfactory homogeneity. Cronbach’s alpha values for PA and NA subscales were moderate in magnitude (α = 0.72 and 0.79, respectively).

Procedure

The PANAS-A scale was administered to the participants in a single evaluation session. Instructions were given to all participants to specifically let them know about the response format and debrief them about the project. Instructions were given to all participants to specifically report this as a current “state” measure.

Data analysis

The factorial validity of the Arabic translated PANAS using CFA (robust maximum likelihood) was conducted. With the aim of studying the internal structure of the PANAS-A with a Middle Eastern sample, four CFAs were conducted. A one-factor model, two-factor model, correlated two-factor model of the PANAS, and correlated three-factor model were tested with the current sample. The model’s direct paths from each affect factor (i.e., PA and NA) to their respective items were specified and the extent to which the bifurcated scale model corresponds to the collected data. The fit of the CFA models was assessed using the Satorra-Bender scaled chi-square statistic, standard chi-square, the robust comparative fit index, comparative fit index (CFI), the standardized root mean squared residual, and the root mean squared error of approximation (RMSEA; Steiger, 2000; Yu, 2002).

First, the CFA was conducted on Model 1 to confirm the one-factor model for PANAS-A. This model is the most parsimonious, for testing the hypothesis of a single dimension underlying the PANAS-A scores. Second, a CFA was conducted to assess the two-factor PA and NA model. A third CFA was conducted to test the hypothesis that PANAS-A assesses the two correlated factors NA and PA. Last, a three-factor correlated model was assessed through a CFA hierarchical model conceptualized by Mehrabian (1997).

Results

Prior to performing CFAs on the data, we tested the normality of the distributions for both the PA and NA data. The distribution of the PA and NA was normal, and the kurtosis and the skewness were near to zero. We used the Kolmogorov–Smirnov test to investigate the normality of the data for PA and NA. The results are shown in Table 2.

Table 2.

Means, standard deviation, and test of normality.

Variable	Number	Mean	Standard deviation	Kolmogorov–Smirnov (Z)	Significance
PA	340	33.012	11.13	0.035	0.067
NA	340	25.59	11.35	0.027	0.076

PA: positive affect; NA: negative affect.

As seen above, the distribution for both PA and NA was normal. Furthermore, for the missing values, the researchers removed the responses of the students who had several missing values having five missing values for a case or for a respondent. If the respondent had less than five missing values we filled the missing by taking the average of the near values. In this way, 37 responses were eliminated leaving 340 respondents for the analyses.

The goodness-of-fit indices for the PANAS models were estimated using the CFAs presented in Table 3. The CFA showed that the two-correlated factor model displayed better goodness-of-fit indices than the other models tested. Chi-square differences showed that the two-correlated factor model had a better fit than the one-factor model, the two uncorrelated factor model, and the three-factor model. The CFA used the maximum likelihood estimates for the statistics of CFI. Tucker–Lewis index (TLI) reached the meritorious 0.9 level, and the RMSEA was near an acceptable 0.08 level.

Table 3.

The one-factor model, the two uncorrelated factor model, and the three-factor model.

Model	χ²	DF	χ²/DF	CFI	TLI	RMSEA
One-factor model	868.97	170	5.11	0.846	0.828	0.110
Two uncorrelated factors model	930.12	170	5.475.47	0.833	0.813	0.115
Two-factor correlated model	312.3	169	1.85	.903	0.902	0.078
Three-factor correlated model	1118.41	167	6.7	0.804	0.777	0.130

DF: degrees of freedom; CFI: comparative fit index; TLI: Tucker–Lewis index; RMSEA: root mean square error of approximation.

The tested one-factor model or Model 1 showed a poor fit model. This model reflects prior empirical evidence that the one-factor model is not suitable and the 20 items of the PANAS-A of this Arabic speaking sample data does not fit in a one variable structure. The fit statistics of the one-factor model for the CFA model are presented in Figure 1. There were no coefficients above 0.90 or less than 0.50, which indicated a strong relationship between the items and subdimensions of the PANAS scale. However, the RMSEA indicator was above the meritorious 0.08 level. The CFI was at 0.846, and the TLI was 0.828; neither reached the acceptable and meritorious 0.9 level. The standard chi-square was 5.112. These results and indicators show that the one-factor model is not a good fit in this study.

Figure 1.

Graphical representation of the one-factor model of the PANAS. PANAS: Positive Affect and Negative Affect Schedule; RMSEA: root mean squared error of approximation.

The PANAS-A model which we call Model 2 expressed variants of the hypothesis that the PANAS-A measures two independent factors, PA and NA. Model 2 represented the original conception of the dimensionality of the PANAS-A in that the 10 PA items were indicators of a PA factor and the 10 NA items were indicators of an NA factor (Buz, Perez Arechaederrs, Fernandez-Pulido, & Urchaga, 2015; Watson et al., 1988). This model reflects prior empirical evidence that the two independent factors model is not suitable and that the comparative fit indices between the recent sample data and the hypothesis of the 20-item scale of PANAS-A fit the two independent variables model, as the fit statistics for the CFA model was inadequate and is presented in Figure 2. It indicates that the PANAS-A scale is devoid of any unwarranted association. The coefficients exhibited a strong correlation of which indicated a good association between the items and subdimensions of the PANAS-A scale. The RMSEA indicator was at a high 0.115, and the CFI was 0.833. The TLI was at 0.813, and the standard chi-square was at 5.47. These results and indicators show that the two independent factor model generally provides less than acceptable model fit for the PANAS-A scale (see Figure 2 for further details).

Figure 2.

Graphical representation of the two independent-factors model of the PANAS. PA: Positive Affect; NA: Negative Affect; RMSEA: root mean squared error of approximation.

The third CFA or Model 3 was conducted with two factors of NA and PA of PANAS-A. Ten items were indicators of a PA factor, and 10 items were indicators of an NA factor (Schmuckle, Egloff, & Burns, 2002). The NA and PA factors as independent variables were correlated in the model. The results of this model reflect the goodness-of-fit indices that the PA and NA were negatively correlated at a significant and moderate level. The fit statistics for the CFA model were presented in Figure 3. The correlations between the two factors were significant and negatively correlated at −0.74. The covariation among the item uniqueness terms reflects measurement or method variance in this regard (see Crawford & Henry, 2004; Merz & Roesch, 2011). It should be noted that the model used in this study partitions measurement error away from the variance in the latent independent factors as NA and PA and thus the variance accounted for is independent of the measurement error.

Figure 3.

Graphical representation of the two dependent-factors model of the PANAS. PA: Positive Affect; NA: Negative Affect; RMSEA: root mean squared error of approximation.

The structure of the PANAS-A scale based on CFA indicated saturation of the items on the factor or dimension to which the items belong to. The conceptualized correlated two-factor correlated model of PANAS-A has good fit indices. Figure 3 shows the loading of the items on the two preconceptualized fitting dimensions.

The last and fourth CFA analysis tested the fit for a three-factor model for PANAS-A according to Mehrabian (1997). This model had shown variants in that the PANAS measures three dependent factors: PA, NA-Upset (Upset, Distressed, Hostile, and Irritable), and NA-Afraid (Guilty, Scared, Ashamed, Nervous, Jittery, and Afraid). This model reflects prior empirical evidence of it having a better fit than the two-factor model; however, generally it reflects a more complex emotional concoction. A comparative fit indices hypothesis of the 20-item scale of PANAS-A fitting the three correlated factor model is presented in Figure 4. The RMSEA indicator was at 0.130, the CFI was 0.804, and the TLI was 0.805. The standard chi-square was at 6.70. These statistics and indicators show that the three-factor independent factor model does not fit in this study.

Figure 4.

Graphical representation of the three dependent-factors model of the PANAS. PA: Positive Affect; NA: Negative Affect; RMSEA: root mean squared error of approximation.

The structural validity of the scale, which consists of two related factors, was verified. The values of the loadings were less than 0.9 and higher than 0.5. This model analysis also eliminates the common loadings of items on both dimensions of the scale as two related factors. Cohabitation covariance results in less than the obtained covariation.

In the last analysis, we attempted a post hoc analysis. The approach was to relate the CFA-related factors with the expectation of a stronger fit of items in the two related restricted variables. Based on the modification indices, the post hoc analysis is presented in Figure 5. It indicates that the PANAS has some unwarranted correlation relationships, which exceed 0.9. The RMSEA indicator was at a high 0.113, the CFI was 0.842, and the TLI was at 0.865. The standard chi-square was 5.37. These results and indicators show that the two independent factor model generally provides less than acceptable model fit for the PANAS-A scale (see Figure 5 for post hoc analysis results).

Figure 5.

Graphical representation of the two dependent-factors model of the PANAS with correlation between errors. PA: Positive Affect; NA: Negative Affect; RMSEA: root mean squared error of approximation.

Discussion

The objective of this study was to have a reliable and valid Arabic version of the PANAS. A random sample of Omani students at a private higher education establishment was administered the PANAS-A. A CFA method assessed the instrument’s factorial validity, seeking adequate factorial stability, reliability indices, and goodness-of-fit. Four consecutive CFAs were conducted. The first CFA was conducted on a single-factor structure, the second was conducted on a two-factor structure of NA and PA, the third was conducted on a correlated two-factor structure, and the fourth was conducted on the PA, NA-Upset (Upset, Distressed, Hostile, and Irritable), and NA-Afraid (Guilty, Scared, Ashamed, Nervous, Jittery, and Afraid) correlated factors as a counter model to the bipolar structure originally conceptualized by Watson et al. (1988). The finding suggests that the bifurcated correlated two-factor model provides the strongest structural integrity and is the best fit out of the four models tested. Surprisingly, the post hoc analyses did not provide an evidence of goodness-of-fit.

Results of the CFA of the original uncorrelated two-factor model (Watson & Clark, 1997) indicated that it was a poor fit for the PANAS-A. The correlation of the main factors implies that there are common variances in the two factors that commonly contribute to indicators fitting in both. Hence, adding correlations between pairs of uniqueness can significantly improve the model by the process of fitting the model. This would also point to the need to isolate those indicators that uniquely contribute to the NA and PA independently. Although inconsistent with the original conceptualization of bifurcated model of independent PA and NA, the study does negate the item fit structure of the three-factor model (Crawford & Henry, 2004; Mehrabian, 2009; Melvin & Molloy, 2000). The correlated two-factor model has some items that could be misspecified and some items that are redundant or not fitting within each factor than by the pattern conceptualized. Thus, by fixing the patterns, we could draw on the indicators or items removed from the model to the fit, but that would be an over specification of the model than by chance alone.

Another significant finding suggests that emotional structure in a Middle Eastern sample as measured by PANAS-A does not significantly differ from that of Western samples. Although Eastern cultures such as those of the Middle East have a dialectical self in expressions of anger and calmness, negative emotions are not necessarily contradictory of positive emotions (Schimmack et al., 2002). However, this study shows that PANAS-A can significantly identify emotions that are negative and positive, particularly when these two factors are negatively correlated. Nevertheless, caution should be exercised because this was a one-time measurement of emotions through the PANAS-A, and a measurement over time could further validate the bifurcation of the scale.

Although our cross-cultural study lent support to the two-factor model of the PANAS, we must be cautious about making assumptions about the cultural generalizability of the two-factor structure of the PANAS. As pointed out earlier in this paper, emotions, whether conceived as discrete or as affect, are not easily differentiated because there still may be many cultural differences in the experiences of affect. Many of these variables have been discussed in this paper (Chang & Asakawa, 2003; Kitayama et al., 1997). Future studies should be conducted on more diverse samples of other Middle Eastern cultures to examine how these variables are related to NA and PA structures.

The PANAS-A was administered just once; this might be a limitation to validating the PANAS-A. As moods and feelings change over time, measuring emotions over time can be a better measure of the factorial validity of PANAS-A scale. In addition, the previous literature (see Mackinnon et al., 1999; Thompson & Meyer, 2007) showed that a shorter version of the PANAS can be a better measure for identifying positive and negative emotions. Recognizing those indicators with the lowest loading and refining the current PANAS-A would provide evidence of a better measuring tool for the PANAS-A for a Middle East sample and thus a roadmap for further investigation. More studies should be conducted on other Asian and Middle Eastern cultures to further validate the findings of this study and others that may be similar to it. Finally, we also recommend that research in the future should look at other evidence for validity, this study only looked at factorial structure, there are many other ways to validate scales, and these should be explored.

Footnotes

Article Notes

References

Beck

J. G.

Novy

D. M.

Diefenbach

G. J.

Stanley

M. A.

Averill

P. M.

Swann

A. C.

(2003). Differentiating anxiety and depression in older adults with generalized anxiety disorder. Psychological Assessment, 15, 184–192. doi:10.1037/0022-3514.79.1.131

Buz

Perez-Arechaederra

Fernandez-Pulido

Urchaga

(2015). Factorial structure and measurement invariance of the PANAS in Spanish older adults. The Spanish Journal of Psychology, 18(e3), 1–11. doi:10.1017/sjp.2015.6

Chang

E. C.

Asakawa

(2003). Cultural variations on optimistic and pessimistic bias for self versus a sibling: Is there evidence for self-enhancement in the West and for self-criticism in the East when the referent group is specified? Journal of Personality and Social Psychology, 84(3), 569–581. doi:10.1037/0022-3514.84.3.569

Crawford

J. R.

Henry

J. D.

(2004). The Positive and Negative Affect Schedule (PANAS): Construct validity, measurement properties and normative data in a large non-clinical sample. British Journal of Clinical Psychology, 24, 245–265. doi:10.1348/0144665031752934

Crocker

P. R.

(1997). A confirmatory factor analysis of the positive affect and negative affect Schedule PANAS with a youth sport sample. Journal of Sport & Executive Psychology, 19, 91–97. doi:10.1123/jsep.19.1.91

DePaoli

L. C.

Sweeney

D. C.

(2000). Further validation of the Positive and Negative Affect Schedule. Journal of Social Behavior and Personality, 15(4), 561–568.

Ekman

(1993). Facial expression and emotion. American Psychologist, 48(4), 384–392. doi:10.1037/0003-066X.48.4.384

Galinha

I. C.

Pereira

C. R.

Esteves

F. G.

(2013). Confirmatory factor analysis and temporal invariance of the Positive and Negative Affect Schedule (PANAS). Psicologia: Reflexão e Crítica, 26(4), 671–679. doi:10.1590/S0102-79722013000400007

Gaudreau

Sanchez

Blondin

J. P.

(2006). Positive and negative affective states in a performance-related setting: Testing the factorial structure of the PANAS across two samples of French-Canadian participants. European Journal of Psychological Assessment, 22, 240–249. doi:10.1027/1015-5759.22.4.240

10.

Heubeck

B. G.

Wilkinson

(2019). Is all fit that glitters gold? Comparisons of two, three and bi-factor models for Watson, Clark & Tellegen’s 20-item state and trait PANAS. Personality and Individual Differences, 144, 132–140. doi:10.1016/j.paid.2019.03.002

11.

Hillerås

P. K.

Jorm

A. F.

Herlitz

Winblad

(1998). Negative and positive affect among the very old: A survey on a sample age 90 years or older. Research on Aging, 20, 593–610. doi:10.1177/0164027598205003

12.

Huebner

E. S.

Dew

(1995). Preliminary validation of the Positive and Negative Affect Schedule with adolescents. Journal of Psychoeducational Assessment, 13, 286–293. doi:10.1177/073428299501300307

13.

Joiner

T. E.

Sandin

Chorok

Lostao

Marqurina

(1997). Development and factor analytic validation of the SPANAS among women in Spain: (More) cross-cultural convergence in the structure of mood. Journal of Personality Assessment, 68, 600–615. doi:10.1207/s15327752.jpa6803_8

14.

Kercher

(1992). Assessing subjective well-being in the old-old: The PANAS as a measure of orthogonal dimensions of positive and negative affect. Research on Aging, 14, 131–168. doi:10.1177/0164027592142001

15.

Killgore

W. D.

(2000). Evidence for a third factor on the Positive and Negative Affect Schedule PANAS in a college student sample. Perceptional and Motor Skills, 90, 147–152. doi:10.2466//pms.2000.90.1.147

16.

Kitayama

Markus

H. R.

Matsumoto

Norasakkunkit

(1997). Individual and collective processes in the construction of the self: Self-enhancement in the United States and self-criticism in Japan. Journal of Personality and Social Psychology, 72(6), 1245–1267. doi:10.1037/0022-3514.72.6.1245

17.

Krohne

H. W.

Egloff

Kohlmann

C. W.

Tausch

(1996). Investigations with a German version of the Positive and Negative Affect Schedule (PANAS). Diagnostica, 42, 139–156.

18.

Lazarus

R. S.

(1999). The cognition–emotion debate: A bit of history. In Dalgleish

Power

(Eds.), Handbook of cognition and emotion (pp. 3–19). New York, NY: John Wiley.

19.

Leue

Beauducel

(2011). The PANAS structure revisited: On the validity of a bifactor model in community and forensic samples. Psychological Assessment, 23(1), 215–225. doi:10.1037/a0021400

20.

Lonigan

C. I.

Hooe

E. S.

David

C. F.

Kistner

J. A.

(1999). Positive and negative affectivity in children: Confirmatory factor analysis of a two-factor model and its relation to symptoms of anxiety and depression. Journal of Consulting and Clinical Psychology, 67(3), 374–386. doi:10.1037/0022.006x.67.3.374

21.

Mackinnon

Jorm

Christensen

Korten

Jacomb

Rodgers

(1999). A short form of the Positive and Negative Affect Schedule: Evaluation of factorial validity and invariance across demographic variables in a community sample. Personality and Individual Differences, 27, 405–416. doi:10.1016/S0191-8869(98)00251-7

22.

Mehrabian

(1997). Comparison of the PAD and PANAS as models for describing emotions and for differentiating anxiety from depression. Journal of Psychopathology and Behavioural Assessment, 19, 331–357. doi:10.1007/BF02229025

23.

Mehrabian

(2009). “Silent messages”—A wealth of information about nonverbal communication (body language). Personality & Emotion Tests & Software: Psychological Books & Articles of Popular Interest. Los Angeles, CA: Self-Published. Retrieved April 6, 2010.

24.

Melvin

G. A.

Molloy

G. N.

(2000). Some psychometric properties of the Positive and Negative Affect Schedule among Australian youth. Psychological Reports, 86, 1209–1212. doi:10.2466/pr02000.86.3c.1209

25.

Merz

E. L.

Roesch

S. C.

(2011). Modelling trait and state variation using multilevel factor analysis with PANAS daily diary data. Journal of Research in Personality, 45, 2–9. doi:10.1016/j.jrp.2010.11.003

26.

Molloy

G. N.

Pallant

J. F.

Kantas

(2001). A psychometric comparison of the Positive and Negative Affect Schedule across age and sex. Psychological Reports, 88(3 Pt 1), 861–862. doi:10.2466/pr0.2001.88.3.861

27.

Narayanan

Menon

Spector

P. E.

(1999a). Stress in the workplace: A comparison of gender and occupations. Journal of Organizational Behavior, 20, 63–73. doi:10.1002/(SICI)1099-1379(199901)20:1<63::AID-JOB873>3.0.CO;2-J

28.

Narayanan

Menon

Spector

P. E.

(1999b). A cross-cultural comparison of job stressors and reactions among employees holding comparable jobs in two countries. International Journal of Stress Management, 6(3), 197–212. doi:10.1023/A:1021986709317

29.

Nasser

(2005). A method for social scientists to adapt instruments from one culture to another: The case of the job descriptive index. Journal of Social Science, 1(4), 232–237. doi:10.3844/jssp.2005.232.237

30.

Nasser

Abouchedid

(2002). Attributions for the causes of peace and locus of control: Their relation in a country where there is conflict. Current Research in Social Psychology, 7(7), 94–113. Retrieved from http://www.uiowa.edu/∼grpproc/crisp/crisp.7.7.htm

31.

Ortuño-Sierra

Santarén-Rosell

De Albéniz

A. P.

Fonseca-Pedrero

(2015). Dimensional structure of the Spanish version of the Positive and Negative Affect Schedule (PANAS) in adolescents and young adults. Psychological Assessment, 27(3), e1–e9. doi:10.1037/pas0000107

32.

Pandey

Srivastava

(2008). Psychometric evaluation of a Hindi version of positive-negative affect schedule. Industrial Psychiatry Journal, 17, 59–54.

33.

Perunovic

W. Q.

Heller

Rafaeli

(2007). Within-person positive emotion culture 5 changes in the structure of emotion: The role of cultural identification and language. Psychological Science, 18, 607–613. doi:10.1111/j.1467-9280.2007.01947.x

34.

Plutchik

(1994). The psychology and biology of emotion. New York, NY: HarperCollins.

35.

Schimmack

Grob

(2000). Dimensional models of core affect: A quantitative comparison by means of structural equation modeling. European Journal of Personality, 14, 325–345. doi:10.1002/1099-0984(200007/08)14:4::AID-PER3803.0.CO;2-I

36.

Schimmack

Oishi

Diener

(2002). Cultural influences on the relation between pleasant emotions and unpleasant emotions: Asian dialectic philosophies or individualism-collectivism? Cognition and Emotion, 16(6), 705–719. doi:10.1080/02699930143000590

37.

Schmuckle

S. C.

Egloff

Burnz

I. R.

(2002). The relationship between positive and negative affect in the Positive and Negative Affect Schedule. Journal of Research in Personality, 36(5), 463–475. doi:10.1016/s0092-6566(02)00007-7

38.

Seib-Pfeifer

L. E.

Pugnaghi

Beauducel

Leue

(2017). On the replication of factor structures of the Positive and Negative Affect Schedule (PANAS). Personality and Individual Differences, 107, 201–207. doi:10.1016/j.paid.2016.11.053

39.

Shapiro

A. M.

Roberts

J. E.

Beck

J. G.

(1999). Differentiating symptoms of anxiety and depression in older adults: Distinct cognitive and affective profiles? Cognitive Therapy and Research, 23, 53–74. doi:10.1023/A:1018710707960

40.

Steiger

J. H.

(2000). Point estimation, hypothesis testing, and interval estimation using the RMSEA: Some comments and reply to Hayduk and Glasser. Structural Equation Modelling, 7, 149–162. doi:10.1207/S15328007SEM0702_1

41.

Terracciano

McCrae

R. R.

Costa

P. T.

(2003). Factorial and construct validity of the Italian Positive and Negative Affect Schedule (PANAS). European Journal of Psychological Assessment, 19(2), 131–141. doi:10.1027//1015-5759.19.2.131

42.

Thompson, R. A., & Meyer S. (2007). Socialization of emotion regulation in the family. In Gross

J. J.

(Ed.), Handbook of emotion regulation (pp. 249–268). New York, NY: Guilford Press.

43.

Titchener

E. B.

(1908). The psychology of feeling and attention. New York, NY: Macmillan.

44.

Tuccitto

D. E.

Giacobbi

P. R.

Jr. Leite

W. L.

(2010). The internal structure of positive and negative affect: A confirmatory factor analysis of the PANAS. Educational and Psychological Measurement, 70, 125–141. doi:10.1177/0013164409344522

45.

Watson

Clark

L. A.

(1997). The measurement and mismeasurement of mood: Recurrent and emergent issues. Journal of Personality Assessment, 86, 267–296. doi:10.1207/s15327752jpa6802_4

46.

Watson

Clark

L. A.

Tellegen

(1988). Development and validation of brief measures of positive and negative affect: The PANAS scales. Journal of Personality and Social Psychology, 54(6), 1063–2. doi:10.1037/0022-3514.54.6.1063

47.

Watson

Tellegen

(1985). Toward a consensual structure of mood. Psychology Bulletin, 98, 219–235. doi:10.1037/0033-2909.98.2.219

48.

Wundt

(1896). Grundriss der Psychologic [Floor plan of psychologic]. (Trans, by C. Judd of 1896 and 1907 editions as Outlines of psychology.) Leipzig, Germany: Engelmann.

49.

C. Y.

(2002). Evaluating cutoff criteria of model fit indices for latent variable models with binary and continuous outcomes (Doctoral dissertation). University of California, Los Angeles.