Research Article: Moral expansiveness short form: Validity and reliability of the MESx

Date Published: October 18, 2018

Publisher: Public Library of Science

Author(s): Daniel Crimston, Matthew J. Hornsey, Paul G. Bain, Brock Bastian, Geoffrey Wetherell.


Moral expansiveness refers to the range of entities (human and non-human) deemed worthy of moral concern and treatment. Previous research has established that the Moral Expansiveness Scale (MES) is a powerful predictor of altruistic moral decision-making and captures a unique dimension of moral cognition. However, the length of the full MES may be restrictive for some researchers. Here we establish the reliability and validity of a reduced moral expansiveness scale, the MESx. Consistent with the full version, the MESx is strongly associated with (but not reducible to) theoretically related constructs, such as endorsement of universalism values, identification with all humanity, and connectedness to nature. The MESx also predicted measures of altruistic moral decision-making to the same degree as the full MES. Further, the MESx passed tests of discriminant validity, was unrelated to political conservatism (unlike the full MES), only mildly associated with the tendency to provide socially desirable responses, and produced moderate reliability over time. We conclude that the MESx is a psychometrically valid alternative for researchers requiring a short measure of moral expansiveness.

Partial Text

Moral expansiveness refers to the extent to which a range of entities (human and non-human) are deemed worthy of moral concern and treatment [1,2]. An individual low in moral expansiveness typically restricts concern to “close” entities; those within more traditional bounds of consideration like kin and ingroup. In contrast, a morally expansive individual extends concern beyond traditional boundaries to entities more “distant” (e.g., strangers, animals, plants).

The purpose of the pilot study was to construct the shortened scale by identifying single-entity exemplars within each of the ten original MES categories, thereby reducing the number of MES entities from 30 to 10. The chosen analytical approach used to identify the MESx items was guided by the approach taken in the construction of the full MES [1]. As the original MES categories do not represent statistically unique sub-dimensions, our goal was simply to identify a single exemplar entity from each of the pre-existing categories. In doing so, we intended to capture the broad spread of human and non-human entities that made up the full version of the MES. Given that factor loadings reflect the correlation between a variable and the underlying factor [9], responses to the complete list of 30 MES entities were gathered and these loadings were used to identity the 10 category exemplars.

One-hundred and forty-five U.S. participants (57.90% male; Mage = 33.76, SD = 15.80) accepted the invitation to complete the survey online after reading a popular science article on the topic of moral decision-making. After reading the MES instructions participants completed the full MES containing all 30 of the original entities. The complete instructions are presented in S1 Appendix; additional files for those wishing to use the MESx can be found on the Open Science Framework at DOI 10.17605/OSF.IO/NC5F6.

Two participants were removed because they failed an attention check, leaving 143 for analysis. Individual factor analyses (principal axis factoring) were conducted on each of the 10 MES categories. Each category produced a single factor with generally very high entity loadings within each of the categories. Factor loadings used to identify the best individual targets are presented in Table 1. Based on these results, the ten entities selected within each category were: family member, citizen of your country, somebody with different religious beliefs, charity/aid worker, mentally challenged individual, murderer, dolphin, fish, old-growth forest, and apple tree.

In Study 1, our primary goal was to examine the validity of the MESx against the original scale. In assessing convergent and predictive validity, we examined responses to the MESx against key constructs and outcome measures that were used to establish the full MES [1]. Constructs used to test for convergent validity were universalism values [11], connectedness to nature [12], and identification with all humanity [13]. Constructs used to test predictive validity were two measures of altruistic moral decision-making: prosociality and willingness to self-sacrifice.

Five hundred and forty-nine U.S. participants (55.20% female; Mage = 39.06, SD = 11.74) were sourced through Amazon’s Mechanical Turk. After reading the MES instructions participants were randomly assigned to either complete the full MES or the shortened MESx. Following this, all participants completed the selected measures of convergent validity, criterion measures of altruistic moral decision-making, and demographic variables.

Thirty-one participants (5.65%) were removed because they failed an attention check, leaving 518 for analyses. Mean moral standings across both scales are presented in Table 2. The pattern of MESx largely mirrors that of the full MES, with human targets worthy of the greatest standing and concern (unless they have committed an act to lower their standing, i.e., “murderer”), followed by non-human entities and elements of the natural environment.

Study 2 examined three elements of scale validation previously unexplored with regard to the MES: discriminant validity, social desirability and test-retest reliability. Convergent validity of the full MES was thoroughly explored in the original examination of the construct [1], yet discriminant validity was not investigated. Likewise, an examination of the extent to which MES responses were tied to impression management, and their stability versus variability over time, were novel lines of inquiry for the moral expansiveness construct.

Study 2 was conducted over two waves. In the first wave, participants completed the MESx and measures of discriminant validity. In the second wave (undertaken 5 weeks after the initial testing) follow-up participants completed only the MESx to examine test-retest reliability. In line with recent research [15] we estimated a follow-up response rate on Mechanical Turk of approximately 40%. Therefore, to ensure an adequately sized wave two sample (at least 200 based on sample size conventions for an estimated effect size between 0.6 and 0.8; [16]), we aimed for a sample of approximately 600 in wave one. Six hundred and nine U.S. participants (51.90% female; Mage = 36.85, SD = 12.20) sourced through Amazon’s Mechanical Turk participated in the initial wave of the study. These participants completed the MESx, followed by measures of discriminant validity and social desirability.

Twenty-five participants (4.11%) were removed because they failed attention checks, leaving 584 for analyses. In line with predictions concerning discriminant validity, individual differences in moral expansiveness as captured by the MESx were not significantly associated with trait levels of extraversion (r = .02, p = .588), emotional stability (r = .02, p = .710) or overall self-esteem (r = .04, p = .355). In addition, scores on the MESx had a significant, yet very low association with responses on the Social Desirability Scale (r = .09, p = .036).

The current manuscript outlines the feasibility of a reduced moral expansiveness scale, the MESx. The MESx produced high internal consistency, and responses were not tied to measures of political conservatism. Further, the MESx was strongly associated with a sense of identification with all humanity, connectedness to nature, and endorsement of universalism values, though not to the point of redundancy. Most crucially, the MESx predicted key criterion measures of altruistic moral decision-making as strongly as did the full MES. Finally, the MESx was not connected to theoretically unrelated constructs (self-esteem, extraversion, or emotional stability), was only mildly associated with the tendency to provide socially desirable responses, and produced moderate reliability over time.




0 0 vote
Article Rating
Notify of
Inline Feedbacks
View all comments