- Email: [email protected]
Development and validation of the Kernis-Goldman authenticity inventory-short form (KGAI-SF)
Personality and Individual Differences 134 (2018) 1–6
Contents lists available at ScienceDirect
Personality and Individual Differences journal homepage: www.elsevier.com/locate/paid
Development and validation of the Kernis-Goldman authenticity inventoryshort form (KGAI-SF)☆
T
⁎
Melissa J. Bond , Nicole E. Strauss, Robert E. Wickham Palo Alto University, USA
A B S T R A C T
Psychological authenticity has long been of interest to social thinkers, and empirical research on this topic has steadily increased since the 1970s. The present work contributes to this body of knowledge by describing the development of a short form for the Kernis-Goldman Authenticity Inventory (KGAI), a widely used self-report instrument. Thirteen of the 45 items from the original KGAI were removed on the basis of semantic ambiguity or redundancy, and the remaining 32-items were subjected to cross-validated ordinal confirmatory factor analysis (N = 1252) in an effort to identify psychometrically redundant or unreliable items. The resulting 20-item shortform (KGAI-SF) exhibited good reliability across the ranges of the four underlying dimensions of authenticity (awareness, unbiased processing, behavior, relational orientation), as well as good discriminant and convergent validity.
Authenticity has been conceptualized from a multitude of philosophical and psychological perspectives as central to healthy functioning and well-being. Aristotle suggested that the key to authentic living is self-realization by way of discovery of one's soul, and that behaving in accordance with such self-realization leads one to live authentically and with a sense of purpose (Hutchinson, 1995; Kernis & Goldman, 2006). Subsequent to Aristotle's proposition, prominent philosophers such as Descartes, Hume, and Kierkegaard expanded on the concept of authenticity, suggesting that it involves engaging in conscious awareness of one's sensory experiences (Kernis & Goldman, 2006), virtuous social interaction, even in the face of societal conventions (Kierkegaard, 2004), and harmonious behavior with one's intrapsychic experiences (Markus & Nurius, 1986). Kernis and Goldman's (2006) multicomponent conceptualization of authenticity provides a psychological operationalization of these historical perspectives. Kernis and Goldman defined the construct as “the unobstructed operation of one's true- or core-self in one's daily enterprise” (p. 294) and proposed that authenticity is comprised of four distinct but interrelated psychological processes: awareness, unbiased processing, behavior, and relational orientation. Awareness involves attention to, and willingness to learn more about one's intrapsychic processes, including sensory perception, feelings, thoughts, and desires. ☆
Unbiased processing refers to an individual's ability to view themselves honestly and accurately, without the use of defense mechanisms to maintain a positive self-image. The third component, behavior, refers to outwardly engaging in accordance with one's internal processes; in essence, practicing what one preaches. Finally, relational orientation refers to the way one engages in the context of close relationships. Relationally authentic individuals tend to prefer that close others see the ‘real them’ (Kernis & Goldman, 2006). Although a number of self-report instruments measuring felt authenticity have been reported in the literature (e.g., Lopez & Rice, 2006; Wickham, Reed, & Williamson, 2015; Wood, Linley, Maltby, Baliousis, & Joseph, 2008), the Kernis-Goldman Authenticity Inventory Version 3 (KGAI-3; Kernis & Goldman, 2006) is one of the more widely used authenticity measures. Between 2006 and 2018, the KGAI-3 has appeared in approximately fifty published studies and more than a dozen dissertations, and has been translated into multiple languages, as identified through an extensive review of the literature in PsycINFO.1 Though the composite reliability of the scale is high (α = 0.90), the subscale reliabilities demonstrate great variability (α between 0.64 and 0.90), which calls the scale's overall psychometric properties into question.2 Upon investigation of the KGAI item pool, it appears that although the subscale constructs are broadly defined, the items seem to
Note: Raw unidentified data and item characteristic curves have been provided as supplemental materials using Open Science Framework. Direct URL: https://osf.io/b2dqs/. Corresponding author at: Palo Alto University, 1791 Arastradero Rd., Palo Alto, CA 94304, USA. E-mail address: [email protected] (M.J. Bond). 1 An exhaustive search was conducted using subject terms: KGAI, Kernis-Goldman, Kernis-Goldman Authenticity, Kernis-Goldman Authenticity Inventory, Kernis & Goldman, 2006, Kernis & Goldman, Authenticity, Authenticity Inventory, AI3, and AI. 2 There is a common misconception that a high overall alpha points to unidimensionality of a scale. However, multidimensional scales can still exhibit high overall internal consistency, leading to a high alpha (Sijtsma, 2009). ⁎
https://doi.org/10.1016/j.paid.2018.05.033 Received 14 March 2018; Received in revised form 16 May 2018; Accepted 25 May 2018 0191-8869/ © 2018 Elsevier Ltd. All rights reserved.
Personality and Individual Differences 134 (2018) 1–6
M.J. Bond et al.
point Likert scale from “Strongly Disagree” to “Strongly Agree.” The awareness domain contains 12 items and measures the level of selfawareness into one's motives, values, and beliefs (“I have a very good understanding of why I do the things I do.”). The unbiased processing domain contains 10 items, all reverse coded, and measures the degree to which a person does not deny their self-understanding (“I find it very difficult to critically assess myself.”). The behavior domain consists of 11 items measuring the degree to which someone acts in accordance with their personal beliefs (“I try to act in a manner that is consistent with my personally held values, even if others criticize or reject me for doing so.”). Lastly, the relational orientation domain, consisting of 12 items, measures the level of openness within relationships (“I want close others to understand the real me rather than just my public persona or ‘image.’”). The item content of original 45-item version of the measure was evaluated in an effort to eliminate redundant and confusing items. Some items could be characterized as rephrasings of the same expression, which contribute little to the breadth of the construct, and could artificially increase internal consistency estimates. For example, item 14 (“I am not in touch with my deepest thoughts and feelings.”) was removed due to semantic redundancy with item 19 (“I prefer to ignore my darkest thoughts and feelings.”). The original form also contained “double barreled” questions, such as item 8 (“I've often used my silence or head-nodding to convey agreement with someone else's statement or position even though I really disagree.”), which was eliminated due to ambiguous wording. The initial items retained for the current analysis are shown in Table 2.3
hold a great deal of redundancy. Moreover, the existing measure could be considered long, consisting of forty-five items. The considerable length and redundancy of the existing KGAI measure may impede further dissemination of research on authenticity. Though the KGAI-3 may useful for thorough assessment of personality and decision-making, when included as one of several measures in a study, participants are likely to be reluctant to complete the lengthy measure, and thus a shortform is a more appropriate approach when examining correlations (Ziegler, Kemper, & Kruyen, 2014). Research has posited that when used in studies with large and representative samples, short forms yield data which is equivalent to that which may be produced using their longer counterparts (e.g., Rammstedt & Beierlein, 2014; Ziegler et al., 2014). As a result, the present study sought to develop a reliable and valid short-form of the KGAI-3 and selected a 20- rather than 16-item version in an effort to strike a balance between brevity and breadth. 1. Method 1.1. Participants A total of 1252 participants (Mage = 35.73, SDage = 12.48) provided responses. Slightly more than half of the participants identified as female (54.42%) and were mostly of European descent (72.57%). Participants were randomly allocated into two sub-samples (N = 629 and 623) in order to apply a cross-validation approach (Brown, 2006), and the sub-samples did not differ significantly on gender (χ2(4) = 3.4, p = .49), age (t(1185) = 0.12, p = .90), or race (χ2(7) = 5.3, p = .62). Demographics for each sample are presented in Table 1.
1.3.2. Authenticity in Relationships Scale - Short Form (AIRS-SF) The AIRS-SF (Wickham et al., 2015) is a 12-item short form measure based on the original 22-item AIRS (Lopez & Rice, 2006). The measure consists of two subscales: unacceptability of deception (UOD; 6 items, α = 0.84) and intimate risk taking (IRT; 6 items, α = 0.93). The UOD subscale measures a dislike of deception within relationships (“I'd rather my partner have a positive view of me than a completely accurate one”). The IRT subscale the level of comfort one feels disclosing core aspects of one's self to another person (“I am totally myself when I am with my partner”).
1.2. Procedure Data were collected from four studies as part of ongoing projects in the research lab, which were unrelated to the goal of the current report. Three of four studies (N = 1182) included were conducted online via Amazon Mturk and Facebook. Participants that participated through Amazon Mturk received $0.50 for completing the survey, while participants recruited through Facebook were offered results of their personality data, as well as the opportunity to participate in a raffle for a $25 Amazon gift card. The remaining 70 participants were university students, recruited as part of a study about relationships.
1.3.3. Experiences in Close Relationships - Short Form (ECR-S) The ECR-S (Wei, Russell, Mallinckrodt, & Vogel, 2007) is a 12-item measure of attachment styles. The avoidance subscale (6 items, α = 0.84) measures the degree to which one fears dependency in relationships (“I am nervous when partners get too close to me”), while the anxiety subscale (6 items, α = 0.78) measures the degree to which one fears abandonment in relationships (“I need a lot of reassurance that I am loved by my partner”).
1.3. Measures 1.3.1. Kernis-Goldman Authenticity Inventory Version 3 (KGAI-3) The KGAI-3 (Kernis & Goldman, 2006) is a 45-item measure of authenticity, which is comprised of four domains, and answered on a 5-
1.3.4. Positive and Negative Affect Scale (PANAS) The PANAS (Watson, Clark, & Tellegen, 1988) is a 20-item scale that measures a person's present mood state. The scale consists of ten positive affect words (i.e. excited; α = 0.87) and ten negative affect words (i.e. scared; α = 0.92) for the participant to rate the extent to which they feel that affect currently.
Table 1 Demographic characteristics for samples 1 and 2. Characteristic
Gender Male Female Other Race/ethnicity European African Hispanic/Latino Asian descent Indigenous Other/multiethnic Age M (SD)
All
Sample 1
Sample 2
(n = 1252)
(n = 629)
(n = 623)
n = 1108 497 (44.86%) 603 (54.42%) 8 (0.72%) n = 1170 849 (72.57%) 97 (8.29%) 81 (6.92%) 85 (7.26%) 8 (0.69%) 50 (4.27%) n = 1187 35.73 (12.48)
n = 558 256 (45.88%) 296 (53.05%) 6 (1.07%) n = 590 436 (73.90%) 40 (6.78%) 40 (6.78%) 47 (7.97%) 4 (0.68%) 23 (3.90%) n = 599 35.77 (12.51)
n = 550 241 (43.82%) 307 (55.82%) 2 (0.36%) n = 580 413 (71.21%) 57 (9.83%) 41 (7.07%) 38 (6.55%) 4 (0.69%) 27 (4.65%) n = 588 35.69 (12.46)
1.3.5. Rosenberg Self-Esteem Scale (RSE) The RSE (Rosenberg, 1965) is a 10-item measure that assesses a person's self-esteem (“I feel that I have a number of good qualities”; α = 90). 1.3.6. Mini International Personality Item Pool (Mini-IPIP) The Mini-IPIP (Donnellan, Oswald, Baird, & Lucas, 2006) is a 203 Within Sample 1, 148 participants completed the full 45-item KGAI and 481 completed the 32-item version. Within Sample 2, 150 participants completed the full KGAI while 473 completed the 32-item version.
Note: Differences between groups were all non-significant. 2
Personality and Individual Differences 134 (2018) 1–6
M.J. Bond et al.
Table 2 Factor loadings for KGAI items within Sample 1 (32-item) and Sample 2 (20-item). Item
Awareness
Unbiased processing
Behavior
Relational orientation
Factor loading
1. I am often confused about my feelings. 4. I understand why I believe the things I do about myself. 6. I actively try to understand which of my self-aspects fit together to form my core- or true-self. 9. I have a very good understanding of why I do the things I do. 14. am not in touch with my deepest thoughts and feelings. 20. I am aware of when I am not being my true-self. 34. I frequently am not in touch with what's important to me. 36. I often question whether I really know what I want to accomplish in my lifetime. 13. I find it very difficult to critically assess myself. 16. I tend to have difficulty accepting my personal faults, so I try to cast them in a more positive way. 30. I'd rather feel good about myself than objectively assess my personal limitations and shortcomings. 35. I try to block out any unpleasant feelings I might have about myself. 37. I often find that I am overly critical about myself. 39. I often deny the validity of any compliments that I receive. 41. I find it difficult to embrace and feel good about the things I have accomplished. 42. If someone points out or focuses on one of my shortcomings I quickly try to block it out of my mind and forget it. 2. I frequently pretend to enjoy something when in actuality I really don't. 10. I am willing to change myself for others if the reward is desirable enough. 25. I try to act in a manner that is consistent with my personally held values, even if others criticize or reject me for doing so. 27. I've often done things that I don't want to do merely not to disappoint people. 28. I find that my behavior typically expresses my values. 32. I rarely if ever, put on a “false face” for others to see. 33. I spend a lot of energy pursuing goals that are very important to other people even though they are unimportant to me. 45. I am willing to endure negative consequences by expressing my true beliefs about things. 5. I want people with whom I am close to understand my strengths. 15. I make it a point to express to close others how much I truly care for them. 17. I tend to idealize close others rather than objectively see them as they truly are. 23. It is important for me to understand my close others' needs and desires. 24. I want close others to understand the real me rather than just my public persona or “image.” 26. If a close other and I are in disagreement I would rather ignore the issue than constructively work it out. 43. The people I am close to can count on me being who I am regardless of what setting we are in. 44. My openness and honesty in close relationships are extremely important to me.
Sample 1
Sample 2
0.582 0.748 0.398 0.695 0.636 0.462 0.753 0.475 0.736 0.661 0.427 0.687 0.379 0.547 0.675 0.673 0.546 0.331 0.714
0.681 0.668 – – 0.687 – 0.697 0.570 0.747 – – 0.517 – 0.580 0.655 0.603 0.676 – 0.722
0.307 0.750 0.388 0.418 0.538 0.585 0.610 0.397 0.642 0.718 0.598 0.752 0.811
– 0.807 0.470 – 0.615 – 0.644 – 0.659 0.704 – 0.825 0.828
Note: Bolded items were kept in the 20-short form. Numbers correspond to item numbers from original 45-measure.
as well as correlations with related constructs.
item scale that measures personality along the Big-5 traits. The scale comprises of 4 items under each trait: extraversion (“I am the life of the party”; α = 0.80), openness (“I have a vivid imagination”; α = 0.70), conscientiousness (“I get chores done right away”; α = 0.71), neuroticism (“I have frequent mood swings”; α = 0.77), and agreeableness (“I feel others' emotions”; α = 0.79).
2. Results 2.1. Confirmatory factor analysis Fit indices for the sample 1 (32-item measure; χ2(458) = 3967.34, p < .001, RMSEA = 0.110 [90%CI: 0.107, 0.114], p < .01) and sample 2 (20-item measure: χ2(164) = 1555.36, p < .001; RMSEA = 0.117, p < .001) suggest that the solution provided a marginal fit for the data.4 Factor loadings from the CFA applied to both samples are provided in Table 2. Factor loadings from Sample 1 were variable, ranging from 0.307 to 0.811. Items 6 and 20 were eliminated from the awareness factor due to low reliability and item 9 was removed due to psychometric redundancy with item 4. In unbiased processing, items 30 and 37 were eliminated due to poor factor loadings, and test information curves revealed that item 16 did not increase overall reliability of the subscale. For the behavior subscale, items 10, 27, 32, and 33 all exhibited low factor loadings. When examining test information curves, 32 exhibited the best reliability across the latent trait and was retained in the final measure. For relational orientation, item 17 was eliminated due to a poor loading, and test information curves with items 5 and 26 revealed that they did not provide
1.3.7. Analysis plan A cross-validation approach (Brown, 2006) was used to identify candidate item sets based on the results from Sample 1, which were then confirmed in Sample 2. Confirmatory Factor Analysis (CFA) was first conducted on the 32-item version of the KGAI using the data from Sample 1. Analyses were conducted in Mplus 8.1 (Muthén & Muthén, 2017) using the weighted least squares means and variance adjusted (WLSMV) estimator, which has been found to be the most robust for categorical data because it does not assume distribution normality (Wirth & Edwards, 2007). Unlike traditional factor analytic approaches that assume continuous indicator variables, the ordinal factor analytic approach used here allows the analyst to derive not only model fit indices, but also item and test information curves, which illustrate the reliability of an item across the range of the latent trait. However, unlike ordinary continuous-indicator CFA in which model χ2 and fit indices reflect the correspondence between the observed and modelimplied mean and covariance matrices, the ordinal CFA fit indices are based on the correspondence between the observed and model implied categorical response frequencies and polychoric correlations (Muthén & Muthén, 2017). Items were selected for retention through careful examination of item information and item characteristic curves. The 20item short form was then subjected to CFA using data from Sample 2. Reliability and validity for the original and short form were examined,
4 Alternative model specifications were examined to improve model fit. For example, the matrix of standardized residuals suggested that including several covariances among disturbance terms would improve the fit, and although these additional parameters resulted in lower χ2 and RMSEA values, they did not impact the interpretation of the results and were therefore not included in the formal analysis. The analyses were also re-run using robust maximum likelihood estimation, which resulted in an identical pattern of findings as well as better fitting models (RMSEA < 0.07).
3
Personality and Individual Differences 134 (2018) 1–6
M.J. Bond et al.
Fig. 1. Test information curves for awareness (top-left), unbiased processing (top-right), behavior (bottom-left), and relational orientation (bottom-right) for short form item sets in Sample 2.
Furthermore, the short-form reliability estimates were, on average, 0.02 below the 32-item estimates, providing additional evidence that the short-form and full-scale measures exhibited similar levels of internal consistency.
improvement to the reliability of this dimension. Once the candidate items from Sample 1 were identified, the data from Sample 2 was subjected to a subsequent CFA. Factor loadings from Sample 2 ranged from 0.470 to 0.828, though all loadings were within range of what is generally considered acceptable (Brown, 2006). As illustrated in Fig. 1, test information curves for all four latent traits suggest that the items retained in the 20-item scale capture high levels of information across a representative range of each factor, in standardized (Z-score) units, suggesting that the measure successfully differentiates between individuals at varying levels of each subscale. Unbiased processing (Fig. 1, top-left panel) and behavior (Fig. 1, bottomright panel) exhibit high information curves throughout the middle of the latent trait but less so above or below the mean. Awareness (Fig. 1, top-right panel) and relational orientation (Fig. 1, bottom-left panel) demonstrate high information curves throughout the low and middle levels of the trait, but the curves slope downward near the positive end of the trait, suggesting that the subscales are not differentiating as well among individuals who are high in the trait. The majority of factor loadings increased from the 32-item to the 20-item measure.
2.2.2. Validity Table 3 displays correlations of the four dimensions and total score of the KGAI-SF with measures of domain-specific authenticity (AIRSSF), romantic attachment style (ECR-S), global self-esteem (RSE), and Big-5 personality traits (Mini-IPIP). As expected, the KGAI-SF exhibited moderate-to-strong correlations with both dimensions of the AIRS-SF (r = 0.55 to 0.67), which represent the level of authenticity an individual experiences within relationships. Similarly, the KGAI-SF subscales and total score exhibited moderate-to-strong correlations with attachment avoidance (r = −0.36 to −0.60) and anxiety (r = −0.18 to −0.46), which is consistent with the idea that individuals who have low attachment avoidance and anxiety exhibit greater authenticity. The weak correlations (r = 0.04 to 0.24) between trait positive affect (PANAS-Positive) and KGAI-SF scores were expected and reflect the fact that some aspects of authenticity may lead individuals to confront aspects of themselves that are not flattering. Similarly, the modest correlations between KGAI-SF and trait negative affect (r = −0.21 to −0.41) and global self-esteem (r = 0.14 to 0.32) are consistent with Kernis' (2003) conceptualization that authenticity isn't about simply feeling good. As expected, the KGAI-SF was moderately correlated with conscientiousness (r = 0.29 to 0.46), agreeableness (r = 0.30 to 0.54), and neuroticism (r = −0.21 to −0.47), suggesting that individuals reporting greater authenticity tended to be more task-oriented and friendly, but less emotionally reactive. Additionally, the KGAI-SF exhibited modest positive correlations with imagination (r = 0.25 to
2.2. Short form reliability and validity 2.2.1. Reliability Subscale and composite reliabilities from both the 32-item and 20item versions were computed, as well as the differences between the two. Internal consistency estimates (Chronbach's α) for the 20-item subscale (0.72, 0.71, 0.70, and 0.81, for Awareness, Unbiased Processing, Behavior, Relational Orientation, respectively) and composite (0.87) versions were comparable to the 32-item subscale (0.75, 0.77, 0.71, and 0.78, respectively) and composite (0.90) α estimates. 4
Personality and Individual Differences 134 (2018) 1–6
M.J. Bond et al.
The moderate-to-strong correlations between KGAI-SF scores and relationship authenticity, attachment, and some aspects of the Big-5 provide good evidence of convergent validity. The modest correlations between the KGAI-SF subscales and global indices of well-being (i.e., self-esteem, positive/negative affect) are consistent with Kernis' (2003) argument that authenticity represents a form ‘adaptively optimal’ selfesteem that tends to remain relatively stable in response to the ups and downs of daily life. The weak correlations between KGAI-SF scores and extraversion provide evidence of discriminant validity since preferring more (or less) social interaction should have little-to-no bearing on the expression of an individual's true-self. The concept of psychological authenticity is inherently broad, which in part accounts for the large number of items generated by Kernis and Goldman (2006). The present study sought to further refine this construct by identifying items that exhibited both conceptual and psychometric redundancy. The final 20-item short form aims to capture the subtle nuances of each factor while reducing participant burden, by retaining only the items that reflect the core aspects of authenticity. It is important to note that to the best of the authors' knowledge, an item-factor analysis has never been published for any version of the KGAI. The primary validation of the KGAI-3 (Kernis & Goldman, 2006) reported a continuous indicator CFA based on item parcels, which provided a weak test of the psychometric properties of the measure. This modest specification, which treats the parceled items as parallel indicators (likely an untenable assumption), provided a middling fit to the data (χ2(48) = 159.39, p < .01, RMSEA = 0.075; Kernis & Goldman, 2006, Table 11, pp. 306). As such, it is hardly surprising that the rigorous approach applied here on both the long and short forms exhibited a marginal fit when using the ordinal CFA approach and corresponding WLSMV estimator.
Table 3 Correlations between KGAI-SF subscales and relevant measures. Measure
Awareness
Behavior
Relational
Processing
Total
AIRS – IRT AIRS – UOD AIRS – total ECR-S – anxiety ECR-S – avoidance PANAS – positive affect PANAS – negative Affect RSE – total Mini-IPIP – extraversion Mini-IPIP – agreeableness Mini-IPIP – conscientiousness Mini-IPIP – neuroticism Mini-IPIP – imagination M SD N
0.36 0.48 0.52 −0.46 −0.49 0.14 −0.40
0.48 0.38 0.52 −0.32 −0.44 0.24 −0.32
0.62 0.31 0.55 −0.18 −0.56 0.24 −0.21
0.26 0.51 0.48 −0.47 −0.36 0.04 −0.34
0.56 0.55 0.67 −0.46 −0.60 0.21 −0.41
0.29 0.14
0.28 0.17
0.14 0.14
0.27 0.21
0.32 0.21
0.37
0.35
0.54
0.30
0.49
0.46
0.32
0.29
0.37
0.46
−0.44 0.33 3.60 0.79 1252
−0.39 0.27 3.65 0.71 1252
−0.21 0.25 3.84 0.75 1251
−0.43 0.28 3.31 0.80 1252
−0.47 0.36 3.60 0.59 1252
Note. All correlations shown are Pearson correlations. Correlations above 0.06 are significant at the p < .01 level.
0.36), which suggests that more authentic individuals show a slight tendency towards unconventional thinking. Finally, the weak correlations between KGAI-SF scores and extraversion (r = 0.14 to 0.21) make sense, given that authenticity is not conceptually related to an individual's preference for social interaction. 3. Discussion
3.2. Limitations and future directions
The idea that expression of an individual's “true” or authentic self imparts psychological benefit has long been recognized by philosophers and social thinkers. More recently, behavioral scientists have come to rely on self-report instruments to assess differences in the extent to which individuals feel authentic in their daily life (Kernis & Goldman, 2006; Lopez & Rice, 2006; Wickham et al., 2015; Wood et al., 2008). Using content- and empirically-driven methods, the present study provides refinement to the KGAI, a widely-used measure of psychological authenticity, by reducing the number of total items from forty-five to twenty, while retaining good reliability and validity. The development of the KGAI-SF should help encourage additional research in this area by reducing participant burden in the assessment of authenticity.
The present study utilized a large sample of North American adults and college students predominantly of European heritage, and future studies should examine potential cross-cultural and linguistic differences in measurement properties with the KGAI-SF. Although the present study featured state-of-the art psychometric analyses, the designs of the studies were cross-sectional, which precludes any direct testing of the test-retest validity. Despite not being able to make any direct conclusions regarding the temporal stability of the short form, the similar alpha estimates for the short and long forms indicate that the items retained are more reliable, which suggests that the test-retest validity coefficient could only grow larger as a result of the modifications. Future research should utilize multi-wave panel designs to more precisely determine the extent to which the KGAI-SF is able to detect meaningful fluctuations in authenticity over time. Future studies should also strive to provide additional evidence for the validity of the KGAI-SF by examining the extent to which individuals reporting higher levels on this measure exhibit more adaptive responses to self-esteem threats, such as interpersonal conflicts (Wickham, Williamson, Beard, Kobayashi, & Hirst, 2016). More specifically, these studies should strive to incorporate measures of theoretically relevant constructs into subsequent analyses to provide evidence of the incremental validity of the KGAI-SF.
3.1. Summary of analyses Cross-validated ordinal CFA revealed a number of items for each authenticity dimension that exhibited poor reliability. Specifically, 2 items from the awareness dimension, 2 items from the unbiased processing component, 3 items from the behavior subscale, and 1 item from the relational orientation dimension were eliminated on the basis of weak factor loadings. Examination of item and test information curves provided by the ordinal CFA also allowed us to eliminate items that were psychometrically redundant or unreliable, which reduced the number of items for each dimension to five, while still maintaining comparable reliability and validity indices. Since Cronbach's α is directly influenced by the number of items within a scale (Sijtsma, 2009), it is especially notable that the differences between the alphas for the 20- and 32-item measures were negligible. The test information curves for the short form indicated that the subscales were less efficient in differentiating between individuals at the low or high ends of the trait. However, this property of the measure is not concerning, given that the KGAI is used exclusively for research (rather than diagnostic) and the extreme ends of the distribution are of less interest.
References Brown, T. A. (2006). Confirmatory factor analysis for applied research. New York, NY, US: Guilford Press. Donnellan, M. B., Oswald, F. L., Baird, B. M., & Lucas, R. E. (2006). The mini-IPIP scales: Tiny-yet-effective measures of the big five factors of personality. Psychological Assessment, 18(2), 192–203. Hutchinson, D. S. (1995). Ethics. In J. Barnes (Ed.). The Cambridge companion to Aristotle (pp. 195–232). Cambridge, UK: The Press Syndicate of the University of Cambridge. Kernis, M. (2003). Toward a conceptualization of optimal self-esteem. Psychological Inquiry, 14, 1–26. Kernis, M., & Goldman, B. (2006). A multicomponent conceptualization of authenticity:
5
Personality and Individual Differences 134 (2018) 1–6
M.J. Bond et al.
measures of positive and negative affect: The PANAS scales. Journal of Personality and Social Psychology, 54(6), 1063–1070. Wei, M., Russell, D. W., Mallinckrodt, B., & Vogel, D. L. (2007). The experiences in close relationship scale (ECR)-short form: Reliability, validity, and factor structure. Journal of Personality Assessment, 88(2), 187–204. Wickham, R., Reed, D., & Williamson, R. (2015). Establishing the psychometric properties of the self and perceived-partner authenticity in relationships scale-short form (AIRSSF): Measurement invariance, reliability, and incremental validity. Personality and Individual Differences, 77, 62–67. Wickham, R., Williamson, R., Beard, C., Kobayashi, C., & Hirst, T. (2016). Authenticity attenuates the negative effects of interpersonal conflict on daily well-being. Journal of Research in Personality, 65, 109–119. Wirth, R. J., & Edwards, M. C. (2007). Item factor analysis: Current approaches and future directions. Psychological Methods, 12(1), 58–79. Wood, A. M., Linley, P. A., Maltby, J., Baliousis, M., & Joseph, S. (2008). The authentic personality: A theoretical and empirical conceptualization and the development of the authenticity scale. Journal of Counseling Psychology, 55(3), 385–399. Ziegler, M., Kemper, C. J., & Kruyen, P. (2014). Short scales – Five misunderstandings and ways to overcome them. Journal of Individual Differences, 35(4), 185–189.
Theory and research. In Mark P. Zanna (Ed.). Advances in Experimental Social Psychology (38, pp. 284–357). New York, NY: Academic Press. Kierkegaard, S. (2004). From fear and trembling problema I. In G. Marino (Ed.). Basic writings of existentialism (pp. 7–24). New York: The Modern LibraryNew Jersey: Princeton University Press Excerpts From Søren Kierkegaard, Fear and Trembling, V. Howard & E. H. Hong, Trans., 1983. Lopez, F. G., & Rice, K. G. (2006). Preliminary development and validation of a measure of relationship authenticity. Journal of Counseling Psychology, 53(3), 362–371. Markus, H., & Nurius, P. (1986). Possible selves. American Psychologist, 41, 954–969. Muthén, L., & Muthén, B. (1998–2017). Mplus User's Guide (Eighth Ed.). Los Angeles, CA: Muthén & Muthén. Rammstedt, B., & Beierlein, C. (2014). Can't we make it any shorter? The limits of personality assessment and ways to overcome them. Journal of Individual Differences, 35(4), 212–220. Rosenberg, M. (1965). Society and the adolescent self-image. Princeton, NJ: Princeton University Press. Sijtsma, K. (2009). On the use, the misuse, and the very limited usefulness of Cronbach's alpha. Psychometrika, 74(1), 107. Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and validation of brief
6
Contents lists available at ScienceDirect
Personality and Individual Differences journal homepage: www.elsevier.com/locate/paid
Development and validation of the Kernis-Goldman authenticity inventoryshort form (KGAI-SF)☆
T
⁎
Melissa J. Bond , Nicole E. Strauss, Robert E. Wickham Palo Alto University, USA
A B S T R A C T
Psychological authenticity has long been of interest to social thinkers, and empirical research on this topic has steadily increased since the 1970s. The present work contributes to this body of knowledge by describing the development of a short form for the Kernis-Goldman Authenticity Inventory (KGAI), a widely used self-report instrument. Thirteen of the 45 items from the original KGAI were removed on the basis of semantic ambiguity or redundancy, and the remaining 32-items were subjected to cross-validated ordinal confirmatory factor analysis (N = 1252) in an effort to identify psychometrically redundant or unreliable items. The resulting 20-item shortform (KGAI-SF) exhibited good reliability across the ranges of the four underlying dimensions of authenticity (awareness, unbiased processing, behavior, relational orientation), as well as good discriminant and convergent validity.
Authenticity has been conceptualized from a multitude of philosophical and psychological perspectives as central to healthy functioning and well-being. Aristotle suggested that the key to authentic living is self-realization by way of discovery of one's soul, and that behaving in accordance with such self-realization leads one to live authentically and with a sense of purpose (Hutchinson, 1995; Kernis & Goldman, 2006). Subsequent to Aristotle's proposition, prominent philosophers such as Descartes, Hume, and Kierkegaard expanded on the concept of authenticity, suggesting that it involves engaging in conscious awareness of one's sensory experiences (Kernis & Goldman, 2006), virtuous social interaction, even in the face of societal conventions (Kierkegaard, 2004), and harmonious behavior with one's intrapsychic experiences (Markus & Nurius, 1986). Kernis and Goldman's (2006) multicomponent conceptualization of authenticity provides a psychological operationalization of these historical perspectives. Kernis and Goldman defined the construct as “the unobstructed operation of one's true- or core-self in one's daily enterprise” (p. 294) and proposed that authenticity is comprised of four distinct but interrelated psychological processes: awareness, unbiased processing, behavior, and relational orientation. Awareness involves attention to, and willingness to learn more about one's intrapsychic processes, including sensory perception, feelings, thoughts, and desires. ☆
Unbiased processing refers to an individual's ability to view themselves honestly and accurately, without the use of defense mechanisms to maintain a positive self-image. The third component, behavior, refers to outwardly engaging in accordance with one's internal processes; in essence, practicing what one preaches. Finally, relational orientation refers to the way one engages in the context of close relationships. Relationally authentic individuals tend to prefer that close others see the ‘real them’ (Kernis & Goldman, 2006). Although a number of self-report instruments measuring felt authenticity have been reported in the literature (e.g., Lopez & Rice, 2006; Wickham, Reed, & Williamson, 2015; Wood, Linley, Maltby, Baliousis, & Joseph, 2008), the Kernis-Goldman Authenticity Inventory Version 3 (KGAI-3; Kernis & Goldman, 2006) is one of the more widely used authenticity measures. Between 2006 and 2018, the KGAI-3 has appeared in approximately fifty published studies and more than a dozen dissertations, and has been translated into multiple languages, as identified through an extensive review of the literature in PsycINFO.1 Though the composite reliability of the scale is high (α = 0.90), the subscale reliabilities demonstrate great variability (α between 0.64 and 0.90), which calls the scale's overall psychometric properties into question.2 Upon investigation of the KGAI item pool, it appears that although the subscale constructs are broadly defined, the items seem to
Note: Raw unidentified data and item characteristic curves have been provided as supplemental materials using Open Science Framework. Direct URL: https://osf.io/b2dqs/. Corresponding author at: Palo Alto University, 1791 Arastradero Rd., Palo Alto, CA 94304, USA. E-mail address: [email protected] (M.J. Bond). 1 An exhaustive search was conducted using subject terms: KGAI, Kernis-Goldman, Kernis-Goldman Authenticity, Kernis-Goldman Authenticity Inventory, Kernis & Goldman, 2006, Kernis & Goldman, Authenticity, Authenticity Inventory, AI3, and AI. 2 There is a common misconception that a high overall alpha points to unidimensionality of a scale. However, multidimensional scales can still exhibit high overall internal consistency, leading to a high alpha (Sijtsma, 2009). ⁎
https://doi.org/10.1016/j.paid.2018.05.033 Received 14 March 2018; Received in revised form 16 May 2018; Accepted 25 May 2018 0191-8869/ © 2018 Elsevier Ltd. All rights reserved.
Personality and Individual Differences 134 (2018) 1–6
M.J. Bond et al.
point Likert scale from “Strongly Disagree” to “Strongly Agree.” The awareness domain contains 12 items and measures the level of selfawareness into one's motives, values, and beliefs (“I have a very good understanding of why I do the things I do.”). The unbiased processing domain contains 10 items, all reverse coded, and measures the degree to which a person does not deny their self-understanding (“I find it very difficult to critically assess myself.”). The behavior domain consists of 11 items measuring the degree to which someone acts in accordance with their personal beliefs (“I try to act in a manner that is consistent with my personally held values, even if others criticize or reject me for doing so.”). Lastly, the relational orientation domain, consisting of 12 items, measures the level of openness within relationships (“I want close others to understand the real me rather than just my public persona or ‘image.’”). The item content of original 45-item version of the measure was evaluated in an effort to eliminate redundant and confusing items. Some items could be characterized as rephrasings of the same expression, which contribute little to the breadth of the construct, and could artificially increase internal consistency estimates. For example, item 14 (“I am not in touch with my deepest thoughts and feelings.”) was removed due to semantic redundancy with item 19 (“I prefer to ignore my darkest thoughts and feelings.”). The original form also contained “double barreled” questions, such as item 8 (“I've often used my silence or head-nodding to convey agreement with someone else's statement or position even though I really disagree.”), which was eliminated due to ambiguous wording. The initial items retained for the current analysis are shown in Table 2.3
hold a great deal of redundancy. Moreover, the existing measure could be considered long, consisting of forty-five items. The considerable length and redundancy of the existing KGAI measure may impede further dissemination of research on authenticity. Though the KGAI-3 may useful for thorough assessment of personality and decision-making, when included as one of several measures in a study, participants are likely to be reluctant to complete the lengthy measure, and thus a shortform is a more appropriate approach when examining correlations (Ziegler, Kemper, & Kruyen, 2014). Research has posited that when used in studies with large and representative samples, short forms yield data which is equivalent to that which may be produced using their longer counterparts (e.g., Rammstedt & Beierlein, 2014; Ziegler et al., 2014). As a result, the present study sought to develop a reliable and valid short-form of the KGAI-3 and selected a 20- rather than 16-item version in an effort to strike a balance between brevity and breadth. 1. Method 1.1. Participants A total of 1252 participants (Mage = 35.73, SDage = 12.48) provided responses. Slightly more than half of the participants identified as female (54.42%) and were mostly of European descent (72.57%). Participants were randomly allocated into two sub-samples (N = 629 and 623) in order to apply a cross-validation approach (Brown, 2006), and the sub-samples did not differ significantly on gender (χ2(4) = 3.4, p = .49), age (t(1185) = 0.12, p = .90), or race (χ2(7) = 5.3, p = .62). Demographics for each sample are presented in Table 1.
1.3.2. Authenticity in Relationships Scale - Short Form (AIRS-SF) The AIRS-SF (Wickham et al., 2015) is a 12-item short form measure based on the original 22-item AIRS (Lopez & Rice, 2006). The measure consists of two subscales: unacceptability of deception (UOD; 6 items, α = 0.84) and intimate risk taking (IRT; 6 items, α = 0.93). The UOD subscale measures a dislike of deception within relationships (“I'd rather my partner have a positive view of me than a completely accurate one”). The IRT subscale the level of comfort one feels disclosing core aspects of one's self to another person (“I am totally myself when I am with my partner”).
1.2. Procedure Data were collected from four studies as part of ongoing projects in the research lab, which were unrelated to the goal of the current report. Three of four studies (N = 1182) included were conducted online via Amazon Mturk and Facebook. Participants that participated through Amazon Mturk received $0.50 for completing the survey, while participants recruited through Facebook were offered results of their personality data, as well as the opportunity to participate in a raffle for a $25 Amazon gift card. The remaining 70 participants were university students, recruited as part of a study about relationships.
1.3.3. Experiences in Close Relationships - Short Form (ECR-S) The ECR-S (Wei, Russell, Mallinckrodt, & Vogel, 2007) is a 12-item measure of attachment styles. The avoidance subscale (6 items, α = 0.84) measures the degree to which one fears dependency in relationships (“I am nervous when partners get too close to me”), while the anxiety subscale (6 items, α = 0.78) measures the degree to which one fears abandonment in relationships (“I need a lot of reassurance that I am loved by my partner”).
1.3. Measures 1.3.1. Kernis-Goldman Authenticity Inventory Version 3 (KGAI-3) The KGAI-3 (Kernis & Goldman, 2006) is a 45-item measure of authenticity, which is comprised of four domains, and answered on a 5-
1.3.4. Positive and Negative Affect Scale (PANAS) The PANAS (Watson, Clark, & Tellegen, 1988) is a 20-item scale that measures a person's present mood state. The scale consists of ten positive affect words (i.e. excited; α = 0.87) and ten negative affect words (i.e. scared; α = 0.92) for the participant to rate the extent to which they feel that affect currently.
Table 1 Demographic characteristics for samples 1 and 2. Characteristic
Gender Male Female Other Race/ethnicity European African Hispanic/Latino Asian descent Indigenous Other/multiethnic Age M (SD)
All
Sample 1
Sample 2
(n = 1252)
(n = 629)
(n = 623)
n = 1108 497 (44.86%) 603 (54.42%) 8 (0.72%) n = 1170 849 (72.57%) 97 (8.29%) 81 (6.92%) 85 (7.26%) 8 (0.69%) 50 (4.27%) n = 1187 35.73 (12.48)
n = 558 256 (45.88%) 296 (53.05%) 6 (1.07%) n = 590 436 (73.90%) 40 (6.78%) 40 (6.78%) 47 (7.97%) 4 (0.68%) 23 (3.90%) n = 599 35.77 (12.51)
n = 550 241 (43.82%) 307 (55.82%) 2 (0.36%) n = 580 413 (71.21%) 57 (9.83%) 41 (7.07%) 38 (6.55%) 4 (0.69%) 27 (4.65%) n = 588 35.69 (12.46)
1.3.5. Rosenberg Self-Esteem Scale (RSE) The RSE (Rosenberg, 1965) is a 10-item measure that assesses a person's self-esteem (“I feel that I have a number of good qualities”; α = 90). 1.3.6. Mini International Personality Item Pool (Mini-IPIP) The Mini-IPIP (Donnellan, Oswald, Baird, & Lucas, 2006) is a 203 Within Sample 1, 148 participants completed the full 45-item KGAI and 481 completed the 32-item version. Within Sample 2, 150 participants completed the full KGAI while 473 completed the 32-item version.
Note: Differences between groups were all non-significant. 2
Personality and Individual Differences 134 (2018) 1–6
M.J. Bond et al.
Table 2 Factor loadings for KGAI items within Sample 1 (32-item) and Sample 2 (20-item). Item
Awareness
Unbiased processing
Behavior
Relational orientation
Factor loading
1. I am often confused about my feelings. 4. I understand why I believe the things I do about myself. 6. I actively try to understand which of my self-aspects fit together to form my core- or true-self. 9. I have a very good understanding of why I do the things I do. 14. am not in touch with my deepest thoughts and feelings. 20. I am aware of when I am not being my true-self. 34. I frequently am not in touch with what's important to me. 36. I often question whether I really know what I want to accomplish in my lifetime. 13. I find it very difficult to critically assess myself. 16. I tend to have difficulty accepting my personal faults, so I try to cast them in a more positive way. 30. I'd rather feel good about myself than objectively assess my personal limitations and shortcomings. 35. I try to block out any unpleasant feelings I might have about myself. 37. I often find that I am overly critical about myself. 39. I often deny the validity of any compliments that I receive. 41. I find it difficult to embrace and feel good about the things I have accomplished. 42. If someone points out or focuses on one of my shortcomings I quickly try to block it out of my mind and forget it. 2. I frequently pretend to enjoy something when in actuality I really don't. 10. I am willing to change myself for others if the reward is desirable enough. 25. I try to act in a manner that is consistent with my personally held values, even if others criticize or reject me for doing so. 27. I've often done things that I don't want to do merely not to disappoint people. 28. I find that my behavior typically expresses my values. 32. I rarely if ever, put on a “false face” for others to see. 33. I spend a lot of energy pursuing goals that are very important to other people even though they are unimportant to me. 45. I am willing to endure negative consequences by expressing my true beliefs about things. 5. I want people with whom I am close to understand my strengths. 15. I make it a point to express to close others how much I truly care for them. 17. I tend to idealize close others rather than objectively see them as they truly are. 23. It is important for me to understand my close others' needs and desires. 24. I want close others to understand the real me rather than just my public persona or “image.” 26. If a close other and I are in disagreement I would rather ignore the issue than constructively work it out. 43. The people I am close to can count on me being who I am regardless of what setting we are in. 44. My openness and honesty in close relationships are extremely important to me.
Sample 1
Sample 2
0.582 0.748 0.398 0.695 0.636 0.462 0.753 0.475 0.736 0.661 0.427 0.687 0.379 0.547 0.675 0.673 0.546 0.331 0.714
0.681 0.668 – – 0.687 – 0.697 0.570 0.747 – – 0.517 – 0.580 0.655 0.603 0.676 – 0.722
0.307 0.750 0.388 0.418 0.538 0.585 0.610 0.397 0.642 0.718 0.598 0.752 0.811
– 0.807 0.470 – 0.615 – 0.644 – 0.659 0.704 – 0.825 0.828
Note: Bolded items were kept in the 20-short form. Numbers correspond to item numbers from original 45-measure.
as well as correlations with related constructs.
item scale that measures personality along the Big-5 traits. The scale comprises of 4 items under each trait: extraversion (“I am the life of the party”; α = 0.80), openness (“I have a vivid imagination”; α = 0.70), conscientiousness (“I get chores done right away”; α = 0.71), neuroticism (“I have frequent mood swings”; α = 0.77), and agreeableness (“I feel others' emotions”; α = 0.79).
2. Results 2.1. Confirmatory factor analysis Fit indices for the sample 1 (32-item measure; χ2(458) = 3967.34, p < .001, RMSEA = 0.110 [90%CI: 0.107, 0.114], p < .01) and sample 2 (20-item measure: χ2(164) = 1555.36, p < .001; RMSEA = 0.117, p < .001) suggest that the solution provided a marginal fit for the data.4 Factor loadings from the CFA applied to both samples are provided in Table 2. Factor loadings from Sample 1 were variable, ranging from 0.307 to 0.811. Items 6 and 20 were eliminated from the awareness factor due to low reliability and item 9 was removed due to psychometric redundancy with item 4. In unbiased processing, items 30 and 37 were eliminated due to poor factor loadings, and test information curves revealed that item 16 did not increase overall reliability of the subscale. For the behavior subscale, items 10, 27, 32, and 33 all exhibited low factor loadings. When examining test information curves, 32 exhibited the best reliability across the latent trait and was retained in the final measure. For relational orientation, item 17 was eliminated due to a poor loading, and test information curves with items 5 and 26 revealed that they did not provide
1.3.7. Analysis plan A cross-validation approach (Brown, 2006) was used to identify candidate item sets based on the results from Sample 1, which were then confirmed in Sample 2. Confirmatory Factor Analysis (CFA) was first conducted on the 32-item version of the KGAI using the data from Sample 1. Analyses were conducted in Mplus 8.1 (Muthén & Muthén, 2017) using the weighted least squares means and variance adjusted (WLSMV) estimator, which has been found to be the most robust for categorical data because it does not assume distribution normality (Wirth & Edwards, 2007). Unlike traditional factor analytic approaches that assume continuous indicator variables, the ordinal factor analytic approach used here allows the analyst to derive not only model fit indices, but also item and test information curves, which illustrate the reliability of an item across the range of the latent trait. However, unlike ordinary continuous-indicator CFA in which model χ2 and fit indices reflect the correspondence between the observed and modelimplied mean and covariance matrices, the ordinal CFA fit indices are based on the correspondence between the observed and model implied categorical response frequencies and polychoric correlations (Muthén & Muthén, 2017). Items were selected for retention through careful examination of item information and item characteristic curves. The 20item short form was then subjected to CFA using data from Sample 2. Reliability and validity for the original and short form were examined,
4 Alternative model specifications were examined to improve model fit. For example, the matrix of standardized residuals suggested that including several covariances among disturbance terms would improve the fit, and although these additional parameters resulted in lower χ2 and RMSEA values, they did not impact the interpretation of the results and were therefore not included in the formal analysis. The analyses were also re-run using robust maximum likelihood estimation, which resulted in an identical pattern of findings as well as better fitting models (RMSEA < 0.07).
3
Personality and Individual Differences 134 (2018) 1–6
M.J. Bond et al.
Fig. 1. Test information curves for awareness (top-left), unbiased processing (top-right), behavior (bottom-left), and relational orientation (bottom-right) for short form item sets in Sample 2.
Furthermore, the short-form reliability estimates were, on average, 0.02 below the 32-item estimates, providing additional evidence that the short-form and full-scale measures exhibited similar levels of internal consistency.
improvement to the reliability of this dimension. Once the candidate items from Sample 1 were identified, the data from Sample 2 was subjected to a subsequent CFA. Factor loadings from Sample 2 ranged from 0.470 to 0.828, though all loadings were within range of what is generally considered acceptable (Brown, 2006). As illustrated in Fig. 1, test information curves for all four latent traits suggest that the items retained in the 20-item scale capture high levels of information across a representative range of each factor, in standardized (Z-score) units, suggesting that the measure successfully differentiates between individuals at varying levels of each subscale. Unbiased processing (Fig. 1, top-left panel) and behavior (Fig. 1, bottomright panel) exhibit high information curves throughout the middle of the latent trait but less so above or below the mean. Awareness (Fig. 1, top-right panel) and relational orientation (Fig. 1, bottom-left panel) demonstrate high information curves throughout the low and middle levels of the trait, but the curves slope downward near the positive end of the trait, suggesting that the subscales are not differentiating as well among individuals who are high in the trait. The majority of factor loadings increased from the 32-item to the 20-item measure.
2.2.2. Validity Table 3 displays correlations of the four dimensions and total score of the KGAI-SF with measures of domain-specific authenticity (AIRSSF), romantic attachment style (ECR-S), global self-esteem (RSE), and Big-5 personality traits (Mini-IPIP). As expected, the KGAI-SF exhibited moderate-to-strong correlations with both dimensions of the AIRS-SF (r = 0.55 to 0.67), which represent the level of authenticity an individual experiences within relationships. Similarly, the KGAI-SF subscales and total score exhibited moderate-to-strong correlations with attachment avoidance (r = −0.36 to −0.60) and anxiety (r = −0.18 to −0.46), which is consistent with the idea that individuals who have low attachment avoidance and anxiety exhibit greater authenticity. The weak correlations (r = 0.04 to 0.24) between trait positive affect (PANAS-Positive) and KGAI-SF scores were expected and reflect the fact that some aspects of authenticity may lead individuals to confront aspects of themselves that are not flattering. Similarly, the modest correlations between KGAI-SF and trait negative affect (r = −0.21 to −0.41) and global self-esteem (r = 0.14 to 0.32) are consistent with Kernis' (2003) conceptualization that authenticity isn't about simply feeling good. As expected, the KGAI-SF was moderately correlated with conscientiousness (r = 0.29 to 0.46), agreeableness (r = 0.30 to 0.54), and neuroticism (r = −0.21 to −0.47), suggesting that individuals reporting greater authenticity tended to be more task-oriented and friendly, but less emotionally reactive. Additionally, the KGAI-SF exhibited modest positive correlations with imagination (r = 0.25 to
2.2. Short form reliability and validity 2.2.1. Reliability Subscale and composite reliabilities from both the 32-item and 20item versions were computed, as well as the differences between the two. Internal consistency estimates (Chronbach's α) for the 20-item subscale (0.72, 0.71, 0.70, and 0.81, for Awareness, Unbiased Processing, Behavior, Relational Orientation, respectively) and composite (0.87) versions were comparable to the 32-item subscale (0.75, 0.77, 0.71, and 0.78, respectively) and composite (0.90) α estimates. 4
Personality and Individual Differences 134 (2018) 1–6
M.J. Bond et al.
The moderate-to-strong correlations between KGAI-SF scores and relationship authenticity, attachment, and some aspects of the Big-5 provide good evidence of convergent validity. The modest correlations between the KGAI-SF subscales and global indices of well-being (i.e., self-esteem, positive/negative affect) are consistent with Kernis' (2003) argument that authenticity represents a form ‘adaptively optimal’ selfesteem that tends to remain relatively stable in response to the ups and downs of daily life. The weak correlations between KGAI-SF scores and extraversion provide evidence of discriminant validity since preferring more (or less) social interaction should have little-to-no bearing on the expression of an individual's true-self. The concept of psychological authenticity is inherently broad, which in part accounts for the large number of items generated by Kernis and Goldman (2006). The present study sought to further refine this construct by identifying items that exhibited both conceptual and psychometric redundancy. The final 20-item short form aims to capture the subtle nuances of each factor while reducing participant burden, by retaining only the items that reflect the core aspects of authenticity. It is important to note that to the best of the authors' knowledge, an item-factor analysis has never been published for any version of the KGAI. The primary validation of the KGAI-3 (Kernis & Goldman, 2006) reported a continuous indicator CFA based on item parcels, which provided a weak test of the psychometric properties of the measure. This modest specification, which treats the parceled items as parallel indicators (likely an untenable assumption), provided a middling fit to the data (χ2(48) = 159.39, p < .01, RMSEA = 0.075; Kernis & Goldman, 2006, Table 11, pp. 306). As such, it is hardly surprising that the rigorous approach applied here on both the long and short forms exhibited a marginal fit when using the ordinal CFA approach and corresponding WLSMV estimator.
Table 3 Correlations between KGAI-SF subscales and relevant measures. Measure
Awareness
Behavior
Relational
Processing
Total
AIRS – IRT AIRS – UOD AIRS – total ECR-S – anxiety ECR-S – avoidance PANAS – positive affect PANAS – negative Affect RSE – total Mini-IPIP – extraversion Mini-IPIP – agreeableness Mini-IPIP – conscientiousness Mini-IPIP – neuroticism Mini-IPIP – imagination M SD N
0.36 0.48 0.52 −0.46 −0.49 0.14 −0.40
0.48 0.38 0.52 −0.32 −0.44 0.24 −0.32
0.62 0.31 0.55 −0.18 −0.56 0.24 −0.21
0.26 0.51 0.48 −0.47 −0.36 0.04 −0.34
0.56 0.55 0.67 −0.46 −0.60 0.21 −0.41
0.29 0.14
0.28 0.17
0.14 0.14
0.27 0.21
0.32 0.21
0.37
0.35
0.54
0.30
0.49
0.46
0.32
0.29
0.37
0.46
−0.44 0.33 3.60 0.79 1252
−0.39 0.27 3.65 0.71 1252
−0.21 0.25 3.84 0.75 1251
−0.43 0.28 3.31 0.80 1252
−0.47 0.36 3.60 0.59 1252
Note. All correlations shown are Pearson correlations. Correlations above 0.06 are significant at the p < .01 level.
0.36), which suggests that more authentic individuals show a slight tendency towards unconventional thinking. Finally, the weak correlations between KGAI-SF scores and extraversion (r = 0.14 to 0.21) make sense, given that authenticity is not conceptually related to an individual's preference for social interaction. 3. Discussion
3.2. Limitations and future directions
The idea that expression of an individual's “true” or authentic self imparts psychological benefit has long been recognized by philosophers and social thinkers. More recently, behavioral scientists have come to rely on self-report instruments to assess differences in the extent to which individuals feel authentic in their daily life (Kernis & Goldman, 2006; Lopez & Rice, 2006; Wickham et al., 2015; Wood et al., 2008). Using content- and empirically-driven methods, the present study provides refinement to the KGAI, a widely-used measure of psychological authenticity, by reducing the number of total items from forty-five to twenty, while retaining good reliability and validity. The development of the KGAI-SF should help encourage additional research in this area by reducing participant burden in the assessment of authenticity.
The present study utilized a large sample of North American adults and college students predominantly of European heritage, and future studies should examine potential cross-cultural and linguistic differences in measurement properties with the KGAI-SF. Although the present study featured state-of-the art psychometric analyses, the designs of the studies were cross-sectional, which precludes any direct testing of the test-retest validity. Despite not being able to make any direct conclusions regarding the temporal stability of the short form, the similar alpha estimates for the short and long forms indicate that the items retained are more reliable, which suggests that the test-retest validity coefficient could only grow larger as a result of the modifications. Future research should utilize multi-wave panel designs to more precisely determine the extent to which the KGAI-SF is able to detect meaningful fluctuations in authenticity over time. Future studies should also strive to provide additional evidence for the validity of the KGAI-SF by examining the extent to which individuals reporting higher levels on this measure exhibit more adaptive responses to self-esteem threats, such as interpersonal conflicts (Wickham, Williamson, Beard, Kobayashi, & Hirst, 2016). More specifically, these studies should strive to incorporate measures of theoretically relevant constructs into subsequent analyses to provide evidence of the incremental validity of the KGAI-SF.
3.1. Summary of analyses Cross-validated ordinal CFA revealed a number of items for each authenticity dimension that exhibited poor reliability. Specifically, 2 items from the awareness dimension, 2 items from the unbiased processing component, 3 items from the behavior subscale, and 1 item from the relational orientation dimension were eliminated on the basis of weak factor loadings. Examination of item and test information curves provided by the ordinal CFA also allowed us to eliminate items that were psychometrically redundant or unreliable, which reduced the number of items for each dimension to five, while still maintaining comparable reliability and validity indices. Since Cronbach's α is directly influenced by the number of items within a scale (Sijtsma, 2009), it is especially notable that the differences between the alphas for the 20- and 32-item measures were negligible. The test information curves for the short form indicated that the subscales were less efficient in differentiating between individuals at the low or high ends of the trait. However, this property of the measure is not concerning, given that the KGAI is used exclusively for research (rather than diagnostic) and the extreme ends of the distribution are of less interest.
References Brown, T. A. (2006). Confirmatory factor analysis for applied research. New York, NY, US: Guilford Press. Donnellan, M. B., Oswald, F. L., Baird, B. M., & Lucas, R. E. (2006). The mini-IPIP scales: Tiny-yet-effective measures of the big five factors of personality. Psychological Assessment, 18(2), 192–203. Hutchinson, D. S. (1995). Ethics. In J. Barnes (Ed.). The Cambridge companion to Aristotle (pp. 195–232). Cambridge, UK: The Press Syndicate of the University of Cambridge. Kernis, M. (2003). Toward a conceptualization of optimal self-esteem. Psychological Inquiry, 14, 1–26. Kernis, M., & Goldman, B. (2006). A multicomponent conceptualization of authenticity:
5
Personality and Individual Differences 134 (2018) 1–6
M.J. Bond et al.
measures of positive and negative affect: The PANAS scales. Journal of Personality and Social Psychology, 54(6), 1063–1070. Wei, M., Russell, D. W., Mallinckrodt, B., & Vogel, D. L. (2007). The experiences in close relationship scale (ECR)-short form: Reliability, validity, and factor structure. Journal of Personality Assessment, 88(2), 187–204. Wickham, R., Reed, D., & Williamson, R. (2015). Establishing the psychometric properties of the self and perceived-partner authenticity in relationships scale-short form (AIRSSF): Measurement invariance, reliability, and incremental validity. Personality and Individual Differences, 77, 62–67. Wickham, R., Williamson, R., Beard, C., Kobayashi, C., & Hirst, T. (2016). Authenticity attenuates the negative effects of interpersonal conflict on daily well-being. Journal of Research in Personality, 65, 109–119. Wirth, R. J., & Edwards, M. C. (2007). Item factor analysis: Current approaches and future directions. Psychological Methods, 12(1), 58–79. Wood, A. M., Linley, P. A., Maltby, J., Baliousis, M., & Joseph, S. (2008). The authentic personality: A theoretical and empirical conceptualization and the development of the authenticity scale. Journal of Counseling Psychology, 55(3), 385–399. Ziegler, M., Kemper, C. J., & Kruyen, P. (2014). Short scales – Five misunderstandings and ways to overcome them. Journal of Individual Differences, 35(4), 185–189.
Theory and research. In Mark P. Zanna (Ed.). Advances in Experimental Social Psychology (38, pp. 284–357). New York, NY: Academic Press. Kierkegaard, S. (2004). From fear and trembling problema I. In G. Marino (Ed.). Basic writings of existentialism (pp. 7–24). New York: The Modern LibraryNew Jersey: Princeton University Press Excerpts From Søren Kierkegaard, Fear and Trembling, V. Howard & E. H. Hong, Trans., 1983. Lopez, F. G., & Rice, K. G. (2006). Preliminary development and validation of a measure of relationship authenticity. Journal of Counseling Psychology, 53(3), 362–371. Markus, H., & Nurius, P. (1986). Possible selves. American Psychologist, 41, 954–969. Muthén, L., & Muthén, B. (1998–2017). Mplus User's Guide (Eighth Ed.). Los Angeles, CA: Muthén & Muthén. Rammstedt, B., & Beierlein, C. (2014). Can't we make it any shorter? The limits of personality assessment and ways to overcome them. Journal of Individual Differences, 35(4), 212–220. Rosenberg, M. (1965). Society and the adolescent self-image. Princeton, NJ: Princeton University Press. Sijtsma, K. (2009). On the use, the misuse, and the very limited usefulness of Cronbach's alpha. Psychometrika, 74(1), 107. Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and validation of brief
6