Measuring Humility in an Indonesian Sample

Humility has been a neglected virtue in the social and psychological sciences. Owing to the nature of the construct itself, humility poses some special challenges to researchers wishing to assess individual differences. Although there have been significant strides in research in humility, especially in intellectual humility, this research has yet to test measures of humility in non-Western, religious populations. The present study builds on previous research by testing Indonesian versions of humility scales. We measure humility construct based on eight domains identified by Davis and Hook (2014), namely openness/lack of superiority, other-oriented/unselfish, admit mistakes/teachable, interpersonal modesty, accurate view of self, global humility, spiritual humility, and regulate need for status. The data was analyzed using Confirmatory Factor Analysis (CFA) (N=390). Among 44 items theorized to measure Humility, 8 items do not fit the model and are deemed as not valid, while the other 36 items are valid. The recommendation for future research is to further develop the instrument of humility as some factors still have very few items.


Introduction
Until recently, humility has been a neglected virtue in the social and psychological sciences. Yet psychology potentially could have a great deal to say about the meaning, functions, and implications of humility. There are long and rich traditions of research on self-esteem, on self-enhancement biases, and more recently on narcissism that are relevant to an understanding of humility. Owing to the nature of the construct itself, humility poses some special challenges to researchers wishing to assess individual differences in this classical source of strength. Although there have been significant strides in research in humility, especially in intellectual humility, this research has yet to test measures of humility in non-Western country. The present study builds on previous research by testing Indonesian versions of humility scales.

Defining Humility
CONTRASTING CONCEPTIONS OF HUMILITY. One obstacle in humility research involves the varying definitions of the construct. As Tangney (2000) has pointed out, humility lacks a consensus of definition in the field of psychology. For many, including both lay people and many psychologists, humility simply means holding oneself in low regard (Funk & Wagnall, 1963;Klein, 1992;Knight & Nadel, 1986;Langston & Cantor, 1988;Webster, 1960;Weiss & Knight, 1980). However, others argue that humility involves such components as having an accurate view of the self (Emmons, 1998), having wisdom, recognizing the abilities of others, not being self-focused, and not being arrogant (Hailing, Kunz, & Rowe, 1994;Means, Wilson, Sturm, Bion, & Bach, 1990;Templeton, 1997).
In summary, the theological, philosophical, and psychological literatures portray humility as a rich, multifaceted construct, in sharp contrast to dictionary definitions that emphasize a sense of unworthiness and low self-regard. Specifically, Tangney (2000) mention the key elements of humility appear to include the: accurate assessment of one's abilities and achievements (not low self-esteem, self-deprecation) -ability to acknowledge one's mistakes, imperfections, gaps in knowledge, and limitations (often visa-vis a "higher power") -openness to new ideas, contradictory information, and advice keeping of one's abilities and accomplishments one's place in the world in perspective (e.g., seeing oneself as just one person in the larger scheme of things). -relatively low self-focus, a "forgetting of the self," while recognizing that one is but one part of the larger universe. -appreciation of the value of all things, as well as the many different ways that people and things can contribute to our world.
WHAT HUMILITY IS NOT. In developing a theoretical conceptualization of complex constructs such as humility, it's important to specify how the focal construct differs from other related, but distinct concepts, especially in the development of measures (Campbell & Fiske (1959). Humility is not low self-esteem (Ryan, 1983), nor is it an underestimate of one's abilities, accomplishments, or worth. Humility is related to, but distinct from, modesty and narcissism. The concept of modesty focuses primarily on a moderate estimate of one's skills, merits or achievements and propriety of behavior and dress. It does not capture key aspects of humility such as a "forgetting of the self" and an appreciation of the gifts of others. The construct of narcissism is perhaps most closely related to humility. Narcissism, according to social psychologists involves an exaggerated sense of self-importance and an overestimate of one's abilities, whereas clinical theorists typically use the term "narcissism" to refer to a distinctly pathological form of self-focus and fluctuating self-regard, which stems from fundamental defects in the self-system (e.g., Kohut, 1971). Nonnarcissists may or may demonstrate other aspects of humility such as making accurate assessments of their abilities and achievements, understanding their place in the world, and recognizing the gifts of others. For many, and important for the present study, there is also a religious dimension to humility. This religious dimension involves the recognition that "God infinitely exceeds anything anyone has ever said of Him, and that He is infinitely beyond human comprehension and understanding" (Templeton, 1997, p. 30; see also Schimmel, 1997). The idea here is not that humans are sinful and inadequate but that the human perspective is always limited when considering the perspective of a higher power. Humility, in this view, carries with it an open-mindedness, a willingness to admit mistakes and seek advice, and a desire to learn (Hwang, 1982;Templeton, 1997).
The present study acknowledges this lack of consensus and focuses on the definition provided by Peterson and Seligman (2004) as a point of departure for exploring humility within the context of research on religion. These investigators maintain that humility "... involves a non-defensive willingness to see the self accurately, including both strengths and limitations. Humble individuals will not willfully distort information in order to defend, repair, or verify their own image" (Peterson and Seligman 2004, p. 463). The opposite of humility is pride. And pride is consistently viewed as a destructive trait in the literature (Lewis, 1942).

Prior Studies
In recent years a number of scholars have built on the research of Peterson and Seligman (2004), key figures in the area of positive psychology, who identified humility as an important virtue and character strength, embedded in a broader religious tradition reaching back to early Christianity (Richardson, 1996). As mentioned previously, humility has been defined in various ways and several scales have been developed (Davis et al. 2016;Hoyle, Davisson, Diebels, & Leary, 2016;Krumrei-Mancuso, Haggard, M. C., LaBouff, J. P., & Rowatt, W. C., 2020;Kumrei-Mancuso & Rouse, 2016;Leary et al., 2017). In addition, some researchers have examined a specific type of humility, intellectual humility among religious participants from Christian backgrounds (Hook et al., 2015;McElroy et al., 2014).
MEASUREMENT OF HUMILITY: TWO LEVELS OF INTEREST, TWO LEVELS OF QUESTIONS. By its very nature, the construct of humility poses some special challenges to researchers, particularly in the area of measurement. Thus, it is not surprising that at this early stage in the scientific study of virtues, we do not have a widely recognized, well-validated measure of humility. Theoretically, humility could be assessed at two distinct levels the situational and dispositional. At this stage, most researchers probably would be most interested in the latter in assessing stable, individual differences in humility. In this context, humility would be viewed as a component of one's personality, as a relatively enduring disposition which a person brings to many different kinds of situations.
Humility might also vary depending on situation. Personality and individual differences aside, most of us have humility in some situations, but not in others. It would be useful to have a well-articulated, theoretically derived set of communications or behaviors agreed upon by experts as reflecting humility "in the moment." Currently, we have no theoretically consistent scheme for assessing situation-specific humility. But several research groups have made forays into related areas. For example, Farh, Dobbins, and Cheng (1991) and Yu and Murphy (1993) operationalized workers' modesty by comparing self-ratings to ratings by knowledgeable others (e.g., supervisors and co-workers). Those who rated themselves lower than their supervisors were viewed as showing a "modesty bias." Such "self vs. knowledgeable other" comparisons may provide a useful measure of one component of humility, but given that this virtue theoretically entails an accurate assessment of one's abilities, high humility should be indexed by high levels of agreement between self and other, not self-deprecating discrepancies.
Regarding dispositional humility, a few options presently are available to researchers. Several earlier studies have operationalized humility as low self-esteem (e.g., Weiss & Knight, 1980), but this is clearly inconsistent with the broader, non-dictionary definition of humility. In fact, theoretically, self-esteem measures such as the Rosenberg (1965) and Janis and Fields (1956) scales should be positively correlated with (although not identical to) individual differences in humility. Consider the types of items included on the Rosenberg Self-Esteem Scale (rated on a 1 to 5 scale, from "Always False" to "Always True"): "I feel that I'm a person of worth, at least on a equal plane with others," "I feel I have a number of good qualities," "On the whole, I'm satisfied with myself," "I take a positive attitude toward myself." The person with a true sense of humility would be expected to endorse such items positively, not negatively. Psychologists generally rely heavily on self-report methods for assessing personality traits, but that is especially problematic in measuring humility. To the degree that a key component of humility is a "forgetting of the self," self-reflection, and self-report of one's level of humility may be oxymoronic. What do we make of a person who views him or herself as someone with "unusually high humility?" Traits like humility, however, are not easily inferred from quick observation. Systematic behavioral observational methods are cumbersome and time-consuming; thus, there is a strong preference for paper-and-pencil questionnaires that require little time and training to administer and score. But humility may represent one of those personality constructs that is simply unamenable to self-report methods.

The Importance of This Study
IMPLICATIONS OF HUMILITY FOR PSYCHOLOGICAL, SOCIAL, AND PHYSICAL WELL-BEING: INDIRECT EVIDENCE at this point, we cannot say with great certainty what the psychological or social benefits of humility might be. Little research has addressed directly this construct, and scientists have yet to develop a theory-based, reliable, and valid index. A brief review of several closely related literatures, however, may provide some clues about what lies ahead.
Numerous studies underscore the benefits of modesty especially "moderate" modesty (Baumeister & Ilko, 1995;Bond, Leung, & Wan, 1982;Forsyth, Berger, & Mitchell, 1981;Jones & Wortman, 1973;Robinson, Johnson & Shields, 1995). People like others who are modest about their achievements, and feel less threatened by their accomplishments. Boastful, arrogant behavior does not appeal to most, often resulting in social disapproval. The benefits of modesty seem to extend beyond mere positive social evaluation.
Humility not only implies an accurate assessment of oneself (not unduly favorable nor unfavorable), but it also entails a "forgetting of the self," an outwardly directed orientation toward a world in which one is "just one part." This process of becoming "unselved" may have significant psychological and physical benefits. Clinicians have long noted the links between excessive self-focus and a broad range of psychological symptoms, including anxiety, depression, social phobias, and so forth.

Participants
The sample employed in this study consisted of Indonesian social media users (N = 390). A demographic questionnaire was also administered, with questions about gender, age, economic status, education, and ethnic identity.

Measurement
Humility was operationalized as the magnitude of difference between individuals' evaluations of self and other (Rowatt, et.al., 2002). Overvaluing the self in relation to others or undervaluing others in relation to the self was considered evidence of less humility. We developed a Humility scale based on the eight subdomains identified by Davis and Hook (2014 (2017). The selected items are translated from English language to Indonesia and then back-translated into English and Indonesian final translation, we also checked for discrepancies languages and context which are then resolved through discussion.

Subdomain
Indicator Openness/lack of superiority • Open-minded. Does not see self as perfect • all-knowing, or superior • Open rather than superior stance towards the values and perspectives of other individuals and groups Otheroriented/unselfish • Focuses more on others than self in interpersonal interactions.
• Has interpersonal qualities such as empathy, compassion, and generosity.
• Gives others the credit they deserve.
• Does not try to manipulate or control others for personal gain or benefit Admit mistakes/teachable • Able to recognize a particular mistake, flaw, or limitation within oneself.
• Willing to receive feedback and learn from it.
• Not defensive when others note mistakes, flaws, or limitations and give feedback. Interpersonal modesty • Does not show off, boast, or brag. Does not call attention to self, possessions, or accomplishments. • involves sharing credit fairly and moderating attention that could lead to envy or jealousy. Accurate view of self • Has a desire to know true self.
• Has an awareness of their strengths and weaknesses Global humility • refer explicitly to "humility." Spiritual humility • Recognizes one's place in relation to the Sacred.
• Recognizes the existence of something greater than themselves. Regulate need for status • Able to regulate need for having and demonstrating social status.
• Not overly concerned with others recognizing their status or being impressed by them.

Data Analysis
This study used Confirmatory Factor Analysis (CFA) to measure construct dimensionality to find the aspects and indicators that reflect the latent variable.
To find a model that fits with the data, fit index must be examined. CFA method indicates several fit indexes to find whether the model fits with the theory. This study used four fit model indexes to not only depend on one index in testing the model. The four fit indexes are as follows: 1. Chi-square ( ) Chi-square ( ) index is the most common compared to other indexes. To find a model fit, Chi-square value should show a 'not significant' value (p-value > .05), indicating there is no difference between data and model (Joreskog & Sorbom, 1993, Wang & Wang, 2012. However, Chi-square is very sensitive towards sample size. Big sample size will most likely cause a significant chi-square showing a model to always be not fit (Umar & Nisa, 2020 (Browne & Cudeck, 1993, Wang & Wang, 2012.

CFI (Comparative Fit Index)
CFI analyses the model fit by examining the discrepancy between data and model with CFI values ranging from 0 to 1. Larger values indicate better fit with recommended values of more than .90 (Hu & Bentler, 1999, Hooper, Coughlan, & Mullen, 2008.

TLI (Tucker Lewis Index)
TLI is an incremental index intended to analyze a model fit by comparing the tested model with baseline model. TLI values higher than .09 are recommended (Hu & Bentler, 1999, Brown, 2003. All four fit indexes are used to test the hypothesis formulated as = ∑ or usually noted as null hypothesis ( 0 ) = − ∑ = 0. Hypothesis testing refers to the value of 2 , when 2 is found to be not significant (p > .05), or RMSEA, CFI, and TLI are in the cutoff, the null hypothesis (H 0 ) − ∑ = 0 is not rejected, indicating the one factor model is fit with the data. Thus, the items are valid in measuring only one factor, namely Humility.

Results and Discussion
This study aims to interpret the construct validity testing results of each dimension of Humility Scale. The CFA method is used to analyze the unidimensional model and item significance, producing path diagram and item significance table to describe the model fit. MPlus 8.4 was used to operate Confirmatory Factor Analysis (Muthen & Muthen, 2017).

Construct Validity Test of Openness/Lack of Superiority
The CFA results of one-factor model on four items constructing the Openness/lack of superiority dimension showed the Chi-square= 0.003, df= 2, P-Value = 0.9986, RMSEA = 0.000, CFI= 1.000 and TLI=1.000. Based on the result, the one-factor model (unidimensional) is fit with the data. The path diagram of CFA results above is as follows: Openness/lack of superiority as intended or not, by testing the null hypothesis of the coefficient of the item's factor loading. The testing was done by examining z-score of each factor loading, as shown in the table below.

Construct Validity Test of Other-Oriented/Unselfish
The CFA results of one-factor model on eight items constructing the Other-oriented/unselfish dimension showed the values of Chi-square= 289.727, df= 20, P-Value < .01, RMSEA = 0.184, CFI= 0.857 and TLI=0.800. Based on the result, the one-factor model (unidimensional) is not fit with the data. Therefore, modifications by allowing item correlation are necessary. Modification result shows values of Chi-square= 25.600, df= 14, P-Value = 0.0291, RMSEA = 0.046, CFI=0.994 and TLI=0.988. The path diagram of CFA results above is as follows: The path diagram above shows a one factor (unidimensional) model fit. The results fulfil criterion of RMSEA ≤ 0.05, CFI and TLI > 0.90, meaning the one-factor (unidimensional) model is accepted even when Chi-square's issue of sample size sensitivity shows a significant model (not fit). Based on the results, the model is accepted and all items only measure one factor/construct, namely Other-oriented/unselfish. After obtaining the overall model fit, the researchers tested the significance of each item to find out whether all items measure the construct of Other-oriented/unselfish as intended or not, by testing the null hypothesis of the coefficient of the item's factor loading. The testing was done by examining the z-scores of each factor loading, as shown in the table below.   < 1.96). Thus, the result shows only seven items are valid.

Construct Validity Test of Admit Mistakes/Teachable
The CFA results of the one-factor model on the eight items constructing the admit mistakes/teachable dimension showed the values of Chi-square= 276.387, df= 20, P-Value < .01, RMSEA = 0.179, CFI= 0.767 and TLI=0.674. Based on the result, the one-factor model (unidimensional) does not fit with the data. Therefore, modifications by allowing item correlation are necessary. The modification result shows values of Chi-square= 16.738, df= 11, P-Value = 0.1159, RMSEA = 0.036, CFI=0.995 and TLI=0.987. The path diagram of CFA results above is as follows: The path diagram above shows a one-factor (unidimensional) model fit. Based on the results, all items only measure one factor/construct, namely admit mistakes/teachable. After obtaining the overall model fit, the researchers tested the significance of each item to find out whether all items measure the construct of Admit mistakes/teachable as intended or not, by testing the null hypothesis of coefficient of item's factor loading. The testing is done by examining z-score of each factor loading, as shown in the table below.  Based on the information in the Table above, only five out of eight items are valid in measuring the construct of admit mistakes/teachable with z-value > 1.96. The following is the specification of each item that are not valid: item 15 has negative factor loading of -.297 and z-value of -5.319, item 18 has negative factor loading of -.136 and z-value of -2.165, item 19 has negative factor loading of -.226 and z-value of -4.501. The mentioned three items have z-value < 1.96 and negative factor loading. Thus, the result shows only five items are valid.

Construct Validity Test of Interpersonal Modesty Dimension
The CFA results of the one-factor model on the five items constructing the Interpersonal modesty dimension showed the values of Chi-square= 15.333, df= 5, P-Value < .01, RMSEA = 0.72, CFI= 0.935 and TLI=0.870. Based on the result, the one-factor model (unidimensional) does not fit with the data. Therefore, modifications by allowing item correlation are necessary. The modification result shows values of Chi-square= 7.351, df= 4, P-Value = 0.1185, RMSEA = 0.046, CFI=0.979 and TLI=0.947. The path diagram of CFA results above is as follows: The path diagram above shows a one-factor (unidimensional) model fit. Based on the results, all items only measure one factor/construct, namely Interpersonal modesty. After obtaining the overall model fit, the researchers tested the significance of each item to find out whether all items measure the construct of Interpersonal modesty as intended or not, by testing the null hypothesis of the coefficient of the item's factor loading. The testing was done by examining the z-score of each factor loading, as shown in the table below.

Construct Validity Test of Accurate View of Self
The CFA results of the one-factor model on the seven items constructing the Accurate view of self-dimension showed the values of Chi-square= 232.751, df= 14, P-Value < .01, RMSEA = 0.198,CFI= 0.642 and TLI=0.464. Based on the result, the one-factor model (unidimensional) is not fit with the data. Therefore, modifications by allowing item correlation are necessary. The modification result shows values of Chi-square= 9.881, df= 6, P-Value = 0.1297, RMSEA = 0.040, CFI=0.994 and TLI=0.978. The path diagram of CFA results above is as follows: The path diagram above shows a one-factor (unidimensional) model fit. Based on the results, all items only measure one factor/construct, namely Accurate view of self. After obtaining the overall model fit, the researchers tested the significance of each item to find out whether all items measure the construct of Accurate view of self as intended or not, by testing the null hypothesis of coefficient of item's factor loading. The testing was done by examining the z-score of each factor loading, as shown in the table below.  Based on the information in the Table above, only four out of seven items are valid in measuring the construct of Accurate view of self with z-value > 1.96. The following is the specification of each item that are not valid: item 28 has negative factor loading of -.892 and z-value of -13.787, item 29 has negative factor loading of -.668 and z-value of -13.383, item 30 has negative factor loading of -.302 and z-value of -5.957. The mentioned three items have z-value < 1.96 and negative factor loading. Thus, the result shows only four items are valid.

Construct Validity Test of Global Humility
The CFA results of one-factor model on four items constructing the Global humility dimension showed no Chi-square, RMSEA, CFI, and TLI. The researchers decided to drop item 33 with a very small factor loading of .027. After dropping one item, the CFA results of one-factor model on three items show values of Chi-square= 0.000, df= 0, P-Value = <0.01, RMSEA = 0.000, CFI=1.000 and TLI=1.000. Based on the result, the one-factor model (unidimensional) is fit with the data. The path diagram of CFA results above is as follows: The path diagram above shows a one factor (unidimensional) model fit. The results fulfil criterion of RMSEA ≤ 0.05, CFI and TLI > 0.90, meaning the one-factor (unidimensional) model is accepted even when Chi-square's issue of sample size sensitivity shows a significant model (not fit). Based on the results, the model is accepted and all items only measure one factor/construct, namely Global humility. After obtaining the overall model fit, the researchers tested the significance of each item to find out whether all items measure the construct of Global humility as intended or not, by testing the null hypothesis of coefficient of item's factor loading. The testing was done by examining the z-score of each factor loading, as shown in the table below.  Based on the information in the Table above, all items show positive factor loading and z-value > 1.96, indicating that all items are statistically significant (p < .05). This means that all three items truly measure what is being theorized, namely the Global humility.

Construct Validity Test of Spiritual Humility
The CFA results of one-factor model on three items constructing the Spiritual humility dimension showed the values of Chi-square= 0.000, df= 0, P-Value = <0.01, RMSEA = 0.000, CFI=1.000 and TLI=1.000. Based on the result, the one-factor model (unidimensional) is fit with the data. The path diagram of CFA results above is as follows: The path diagram above shows a one factor (unidimensional) model fit. The results fulfil criterion of RMSEA ≤ 0.05, CFI and TLI > 0.90, meaning the one-factor (unidimensional) model is accepted even when Chi-square's issue of sample size sensitivity shows a significant model (not fit). Based on the results, the model is accepted and all items only measure one factor/construct, namely Spiritual humility. After obtaining the overall model fit, the researchers tested the significance of each item to find out whether all items measure the construct of Spiritual Humility as intended or not, by testing the null hypothesis of coefficient of item's factor loading. The testing was done by examining the z-score of each factor loading, as shown in the table below.

Construct Validity Test of Regulate Need for Status
The CFA results of one-factor model on five items constructing the Regulate need for status dimension showed the values of Chi-square= 38.357, df= 5, P-Value = <0.01, RMSEA = 0.129, CFI= 0.929 and TLI=0.857. Based on the result, the one-factor model (unidimensional) does not fit with the data. Therefore, modifications by allowing item correlation are necessary. Modification result shows values of Chi-square= 2.773, df= 4, P-Value = 0.5965, RMSEA = 0.000, CFI=1.000 and TLI=1.000. The path diagram of CFA results above is as follows:

Figure 8. Path Diagram of Regulate Need for Status
The path diagram above shows a one-factor (unidimensional) model fit. Based on the results, all items only measure one factor/construct, namely Regulate need for status. After obtaining the overall model fit, the researchers tested the significance of each item to find out whether all items measure the construct of Regulate need for status as intended or not by testing the null hypothesis of coefficient of item's factor loading. The testing was done by examining the z-score of each factor loading, as shown in the table below.

Conclusion
This study aimed to find a valid instrument to measure humility consisting of eight dimensions, namely openness/lack of superiority, other-oriented/unselfish, admit mistakes/teachable, interpersonal modesty, accurate view of self, global humility, spiritual humility, and regulate need for status using Confirmatory Factor Analysis (CFA). The CFA results show all one-factor (unidimensional) models fit with the data after administering modifications by allowing item correlation. Among 44 items theorized to measure Humility, 8 items do not fit the one-factor (unidimensional) model and are deemed as not valid, while the other 36 items are valid. The CFA result in this study shows only a few modifications indicating model fit with slight biases. The recommendation for future research is to further develop the Indonesian version of the instrument of humility as some factors still have very few items.
The present studies had several limitations. First, participants provided the only source of ratings, and in every case, were rating another individual. Using a single source of data is subject to biases such as rater bias, person perception heuristic, and assumed similarity between self and other (Lee et al., 2009). Furthermore, each target was only rated by one individual and we were therefore unable to calculate interrater reliability. The gold standard for research on personality judgments involves triangulation of selfreport, other report, and behavioral observation (Roberts & Ilardi, 2003