Psychometric properties, factor structure, and validity of the multidimensional dispositional greed Assessment with undergraduate college students

Abstract

We conducted three studies examining the reliability, factor structure, and validity of Multidimensional Dispositional Greed Assessment (MDGA) scores with samples of undergraduate college students. The MDGA is an instrument developed to measure adults’ levels of dispositional greed. In study 1 (test-retest reliability; N = 150), we assessed the stability of MDGA scores over time. In study 2, we examined the stability and factor structure of the 20-item MDGA (confirmatory factor analysis; N = 2,178), and measurement invariance across demographic groups (multiple group confirmatory factor analysis) among a sample of undergraduate college students in the United States. In study 3 (multiple linear regression; N = 486), we investigated evidence of concurrent validity through measuring the relationship between MDGA and Dispositional Greed Scale scores. Through these studies, we identified evidence for good test–retest reliability for MDGA total and subscale scores. The results also provided evidence for measurement invariance of MDGA items across gender and race/ethnicity groups. Lastly, we found evidence for concurrent validity of MDGA through identifying large positive predictive relationship between MDGA and DGS total scores, while there was a weaker positive correlation with the MDGA’s more unique factor, retention motivation. We provide an overview of the implications of this study, and suggest potential areas of future research.

Keywords:

• dispositional greed
• factor structure
• test-retest reliability
• concurrent validity

1.

Greed has become a salient subject in literature across diverse professional disciplines such as psychology (Lambie & Stickl Haugen, 2019; Zeelenberg & Breugelmans, 2022), economic anthropology (Bosco, 2014; Oka & Kuijt, 2014), criminology (Wang et al., 2019), and business management (Takacs Haynes et al., 2017). Scholars have contributed increased efforts toward understanding greed since the Great Recession of 2008, the second worst economic crisis in United States history, caused by the unregulated greed of bankers (Oka & Kuijt, 2014). Greed has been a major issue addressed through activism efforts of the 21^st century, such as the Occupy Wall Street movement, where protestors were devoted to closing the rich-poor gap and opposing corrupt systems grounded in greed and power.

Greed is most often viewed as a negative personality trait. From various religious perspectives (e.g., Christianity and Islam), greed is considered a vice and associated with sinful behaviors (Aimran et al., 2014). Christian texts have designated greed as the “love of money” which is the “root of all evil” (1 Timothy 6:10). Psychology scholars suggest that greed should be considered a fourth dimension of the current dark triad, which includes narcissism, Machiavellianism, and psychopathy (Marcus & Ziegler-Hill, 2015). Individuals’ dispositional greed levels have been positively associated with narcissism levels (Sekhar et al., 2020). Dispositional greed is conceptually related to constructs such as envy, self-interest, miserliness, and materialism; however, dispositional greed is distinct from these constructs in terms of motivation, focus, and level of desire (Lambie & Stickl Haugen, 2019).

While greed is often viewed as an entirely bad and immoral trait, some argue that it is necessary to differentiate positive and negative manifestations of greed (Wang & Murnighan, 2009). Scholars presented greed as a complex personality trait that can include positive features (Oka & Kuijt, 2014). For example, Zeelenberg and Breugelmans (2022) argued that greed is logically related to prized societal traits such as perseverance, goal achievement, and self-improvement motivations. Greed may also be understood as an adaptive trait through an evolutionary perspective, as individuals in resource deprived environments may strive to acquire excess resources to increase the likelihood of survival (Krekels & Pandelaere, 2015).

1. Defining and measuring dispositional greed

Recently, scholars have proposed various definitions and psychological conceptualizations of dispositional greed. Wang et al. (2012) described greed in terms of overly self-interested behaviors, despite how these behaviors may negatively impact others. However, others argued that dispositional greed is a distinct construct from self-interest (e.g., Lambie & Stickl Haugen, 2019; Seuntjens et al., 2015). For example, Haynes et al. (2015) describes how greed and self-interest are on a continuum, but there is a “tipping point” when extreme self-interest becomes greed. Despite inconsistent definitions, scholars have agreed that dispositional greed involves an insatiable desire for acquiring more (Krekels & Pandelaere, 2015; Lambie & Stickl Haugen, 2019; Seuntjens et al., 2015). This insatiable desire may extend to anything of value, which can include material or non-material things (e.g., money, power, sex, social status; Balot, 2001). Following an extensive review of the literature, Lambie and Stickl Haugen (2019) outlined a multidimensional definition of dispositional greed that includes: (a) desire to gain more of valuable things, not limited to material goods; (b) insatiability in the pursuit for gaining more; (c) disregard for the costs of pursuing more; and (d) retention motivations. This multidimensional definition of dispositional greed is unique because of the inclusion of a retention component—people with higher dispositional greed may refuse to let go of the things they desire.

Based on the multidimensional conceptualization of dispositional greed, Lambie et al. (2022) developed the Multidimensional Dispositional Greed Assessment (MDGA) to measure individuals’ levels of dispositional greed across all four dimensions highlighted in Lambie and Stickl Haugen’s (2019) definition. Prior to the MDGA’s development, several instruments were created to measure individual differences in greed (e.g., GR€€D Scale, Mussel & Hewig, 2016; Dispositional Greed Scale [DGS]; Seuntjens et al., 2015; Greed Trait Measure;; Mussel et al., 2015). However, these pre-existing instruments did not capture a major dimension of dispositional greed—retention motivation (i.e., the drive to retain resources obtained through greedy pursuits; Lambie et al., 2022). The MDGA initially demonstrated adequate psychometric properties, factor structure, and evidence of convergent validity among samples of adults (Lambie et al., 2022). Yet, the authors identified the need for additional research to examine evidence of stability of the MDGA over time along with additional evidence of criterion-related validity to help refine the dimensions of greed as a construct. Furthermore, there is a need to explore the MDGA among diverse samples to determine if the factor structure and psychometric properties remain intact across populations.

2. The Current research studies

To examine evidence of reliability and validity of the MDGA scores, we conducted a three-part study with samples of undergraduate college students in the United States. In study 1, we conducted a test–retest reliability investigation to examine the stability of multidimensional dispositional greed traits over time. In study 2, we administered the MDGA to a large validation sample of college students across the United States to examine the stability and factor structure of the 20-item MDGA. In study 3, we investigated evidence of concurrent validity by measuring the relationship between MDGA scores and Dispositional Greed Scale (DGS; Seuntjens et al., 2015) scores.

3. Study 1. MDGA test-retest reliability

3.1. Methods

3.1.1. Participants

To examine the stability of MDGA scores over time, we administered the MDGA to a sample of undergraduate college students at a large university in the Southeast United States. We recruited a convenience sample of undergraduate students (N = 299) via email solicitation. We set the following minimal inclusion criteria for participation in the study: (a) at least 18 years of age and (b) currently enrolled as an undergraduate college student. The incentive for completing this study was a $.50 donation to a university-based charity. Of the 299 participants who completed the initial survey, 150 participants completed the second survey (50.2%).

Demographic data indicated that most participants identified as women (n = 177; 78%), followed my men (n = 29; 19.3%) and non-binary and/or gender expansive (n = 4; 2.7%). Participants’ ages ranged from 18 to 55 years-old (M = 22.4; SD = 5.91). In terms of participants’ racial and ethnic identities, participants self-identified as White (48%; n = 72), Latinx or Hispanic (n = 41; 27.3%), Black and/or African American (n = 18; 12%), Asian (n = 9; 6%), and multiracial (n = 9; 6%). One participant identified his or her racial/ethnic background as “other,” and self-identified as Arab American. Participants reported their college-level classification as Senior (n = 57; 38.0%), Junior (n = 30; 20.0%), Sophomore (n = 35; 23.3%), Freshman (n = 19; 12.7%), unclassified (n = 2; 1.3%) or other (n = 6; 4.0%). Table 1 presents the complete demographic characteristics, including race, ethnicity, gender, college classification, and college major.

Table 1. Participant demographic characteristics

Characteristic		N	%
Gender
Women		117	78.0
Men		29	19.3
Gender Expansive/non-binary		4	2.7
Race and/or Ethnicity
White		72	48.0
Latinx or Hispanic		41	27.3
Black or African American		18	12.0
Asian		9	6.0
Multiracial		9	6.0
Other		1	0.7
College Year
Freshman		19	12.7
Sophomore		35	23.3
Junior		30	20.0
Senior		57	38.0
Unclassified		2	1.3
Other		6	4.0
College Major
Social Sciences		33	22.0
Biology		22	14.7
Health-Related Fields (e.g., nursing)		20	13.3
Business		18	12.0
Parks, recreational		8	5.3
Communication		6	4.0
Education		3	2.0
Pre-professional		3	2.0
Other		18	12.0

Note: Percentages based on total sample (N = 150)

3.1.2. Procedures

Prior to data collection, we received institutional review board (IRB) approval to ensure ethical research practices. We created a Qualtrics questionnaire that included: (a) an explanation of research and informed consent, (b) the 20-item MDGA (Lambie et al., 2022), and (c) a brief demographic survey. At the start of each survey, participants were asked to create an anonymous research ID comprised letters and numbers, which the researchers used to match participants’ Time 1 and Time 2 responses. Three weeks after completing the initial survey, participants received an email reminder to complete the survey for a second time. The Qualtrics data were exported to a Statistical Package for the Social Sciences (SPSS; Version 28) dataset file for analyses.

3.1.3. Instrumentation

3.1.3.1. Demographic questionnaire

A demographic questionnaire was included in the Qualtrics survey to gather information about participants’ characteristics, including race, gender, and ethnicity. Additionally, participants reported information related to their college major (e.g., nursing, business, psychology, etc.) and college year (i.e., freshman, sophomore, junior, and senior).

3.1.3.2. Multidimensional dispositional greed assessment

The MDGA (Lambie et al., 2022) is a 20-item instrument that measures individuals’ levels of multidimensional dispositional greed across three domains: (a) Insatiable Pursuit for More at all Costs (9 items), (b) Desire for More (7 items), and (c) Retention Motivation (4 items). The MDGA items are worded in statements such as “I will get what I want at all costs, even if I have to lie.” For each item, participants indicated their levels of agreement or disagreement with each statement on a 5-point Likert scale, ranging from 1 (“Strongly Disagree”) to 5 (“Strongly Agree”). Lambie et al. (2022) demonstrated adequate internal consistency ranges (α = .943–.956).

3.2. Data analysis

We exported the Qualtrics survey data and converted the files to an SPSS (Version 28.0) dataset. Before conducting the test-retest reliability analyses, we screened the data to ensure that statistical assumptions were met. We observed boxplots, histograms, and Shapiro-Wilk values, which provided evidence for normality (Time 1 and Time 2). Linearity was observed in the standardized residual plots. We conducted an Intraclass Correlation Coefficients (ICC; Gisev et al., 2013; Hallgren, 2012) procedure to examine test–retest reliability for the MDGA. Scholars recommend using ICC to examine test–retest reliability since they are more sensitive to statistical procedures than the Pearson correlation coefficient (Koo & Li, 2016). ICC values reflect the degree of correlation and agreement between measures, which is desirable for investigating the test–retest reliability of an instrument (Koo & Li, 2016).

4. Results

Cronbach’s alpha coefficients were computed, and the results indicated strong internal consistency across Time 1 and Time 2 scores for the MDGA total score (α = .895), Insatiable Pursuit (α = .845), Desire for More (α = .862), and Retention Motivation (α = .839). We calculated ICC values for the MDGA total and dimensional scores between Time 1 and Time 2 (See Table 2). Specifically, we computed two-way mixed effects models for the following: (a) MDGA total score; (b) Insatiable Pursuit total; (c) Desire for More total; and (d) Retention Motivation total. Aligned with evaluation guidelines (Koo & Li, 2016), results identified good test–retest reliability for the MDGA total scores (ICC = .811; 95% confidence interval [CI] = .745–.860), and good to moderate scores for the subscales including Desire for More (ICC = .758; 95% CI = .680–.819), Insatiable Pursuit (ICC = .731; 95% CI = .647–.797), and Retention Motivation (ICC = .724; 95% CI = .638–.792).

Table 2. Intraclass correlation coefficients (two-way mixed) for MDGA and subscales

95% Confidence Interval
Single Measures	Intraclass Correlation	Lower Bound	Upper Bound
MDGA Total	0.811	0.748	0.860
Insatiable Pursuit	0.731	0.647	0.797
Desire for More	0.758	0.680	0.819
Retention Motivation	0.724	0.638	0.792

5. Study 2. Factor structure

5.1. Methods

5.1.1. Participants

The researchers recruited a sample of undergraduate students through two distinct data collection procedures. The initial sample of participants (n = 1,063) were recruited via Amazon’s Mechanical Turk (MTurk). MTurk is an online labor marketing website that provides users with the opportunity to complete surveys for monetary compensation. The inclusion criteria specified that individuals must be: (a) at least 18 years of age and (b) currently enrolled as an undergraduate student. While the quality of data collection procedures through online labor portals has been confirmed (e.g., Behrend et al., 2011), we also recruited a convenient sample of undergraduate students (n = 1,307) at one university to increase the reliability of our potential results. Of the 1,307 participant surveys, 168 were removed due to full scales missing and 36 outliers were identified via boxplots and z scores and removed (N = 1,103). All participants completed a Qualtrics survey, which included the following: (a) demographic survey; (b) 20-item MDGA; (c) informed consent and a research explanation form.

For the combined University and MTurk sample of undergraduate students (N = 2,166), demographic data indicated that the largest proportion of participants identified as female (n = 1204; 55.6%), followed by male (n = 890; 41.1%) and gender expansive and/or non-binary (n = 16; .7%). Two participants preferred to self-describe their gender (“agender” and “not important to me”) and 54 participants did not provide gender demographic information. In terms of racial and ethnic backgrounds, participants self-identified as White (n = 1,147; 53%), Hispanic or Latinx (n = 399; 18.4%), Black and/or African American (n = 286; 13.2%), Asian (n = 201; 9.3%), multi-racial (n = 61; 2.8%), American Indian or Alaskan Native (n = 20; .9%), Hawaiian or Pacific Islander (n = 5; .2%). Thirty-one participants (1.4%) selected “other racial/ethnic identity” and 16 (.7%) participants did not answer the item. In terms of college classification, participants identified as Freshmen (n = 133; 6.1%), Sophomore (n = 256; 11.8%), Junior (n = 356; 16.4%), Senior (n = 704; 32.5%), Unclassified (n = 71; 3.3%), Other (n = 31; 1.4%), and 615 participants (28.4%) did not provide college classification information.

5.2. Data analysis

Since the MDGA was previously only examined among adult samples, we sought to assess whether the MDGA factor structure was stable with undergraduate college students. We computed a Confirmatory Factor Analysis (CFA) using IBM SPSS AMOS (Version 28) to examine the MDGA three-factor structure. Researchers use CFA, a type of structural equation modeling (SEM), to test hypotheses and theories based on pre-existing evidence (Brown, 2015). Consistent with findings from previous MDGA Exploratory Factor Analysis and CFA studies (Lambie et al., 2022), we tested the three-factor oblique structure, consisting of (a) Factor 1: Insatiable Pursuit for More at All Costs (items # 1, 4, 7, 10, 13, 15, 17, 19, 20), (b) Factor 2: Desire for More (items # 2, 5, 8, 11, 14, 16, 18), and (c) Factor 3: Retention Motivation (items # 3, 6, 9, 12). To examine if the model was a good fit, we used the following recommended indices (Browne & Cudeck, 1993; Hu & Bentler, 1999; Mvududu & Sink, 2013): (a) Chi-square statistic (p > .05 = good fit); (b) Standardized Root Mean Square Residual (SRMR < .08 = acceptable fit) (c) Comparative Fit Index (CFI > .95 = good fit); (d) Root Mean Squared Error of Approximation (RMSEA; values < .08. = acceptable fit; values < .06 = good fit).

6. Results

The CFA (see Figure 1) confirmed that the 20 observed indicators related to their respective latent variables (standardized regression weights ranged from .65 to .88) and the three latent variables correlated (r_s = .37, .37, .38). The chi-square statistic value was significant, χ² (167) = 1407.3, p < .001; however, significant chi-square statistics are commonly observed among larger sample sizes (Kline, 2015). Therefore, we consulted additional fit indices to determine the model’s goodness-of-fit. Fit indices were within acceptable ranges, identifying an overall good fit for the 20-item MDGA model: SRMR = .044, RMSEA = .059 [90% CI = .056–.061], and CFI = .953. For Factor 1 (Insatiable Pursuit for More at All Costs), the standardized regression weights ranged from .61 to .83. The standardized regression weights for Factor 2 (Desire for More) ranged from .65 to .81. For Factor 3 (Retention Motivation), the standardized regression weights were between .86 and .88.

Figure 1. 20-Item, three-factor model with standardized factor loadings.

To examine the need for a multidimensional dispositional greed scale, we compared the three-factor and one-factor MDGA models through computing an additional CFA of the one factor MDGA (see Figure 2). The standardized regression weights were significantly lower for the unidimensional scale (one-factor model, ranging from .22 to .82; three-factor model, ranging from .65 to .88). The chi-square statistic value of the one-factor model was significant, χ² = (170) = 12,492.5, p < .001. We consulted fit indices to determine the one-factor model’s goodness-of-fit. The RMSEA value for the one-factor model was .183 (90% CI = .180–.186), indicating that the one-factor model was not an acceptable fit. The CFI was .537, which failed to exceed the cut-off guidelines for acceptable fit (CFI > .95 = good fit). The SRMR value was .1705, providing additional evidence for unacceptable fit of the one-factor MDGA model.

Figure 2. 20-Item, one-factor model with standardized factor loadings.

6.1. Multiple Group Confirmatory Factor Analysis

6.1.1. Gender

To examine the presence of measurement invariance across demographic groups, we computed two multiple group CFA models for the three-factor MDGA. First, we generated a multiple group CFA to assess invariance across gender groups. We included two comparison groups (male, n = 890; female, n = 1204), and excluded the gender expansive and/or non-binary group (n = 16) variable because of sampling inadequacy (n < 100; Kline, 2015). The multiple group CFA (see Table 3) confirmed that the 20 observed indicators related to their respective latent variables (standardized regression weights ranges: female, .53 to .90; male, .63 to .86). The three latent variables correlated for both gender groups (female, r_s = .33, .36, .36; male, r_s = .47, .36, .36). Evidence for configural invariance was identified through model fit indices associated with overall good fit (RMSEA = .042 [90% CI = .040–.045]; SRMR = .0471; CFI = .950). After examining evidence for configural invariance, we tested for metric (weak) and scalar (strong) invariance. At the metric level, the results indicated invariance with evidence for good fit (RMSEA = 0.481; SRMR = .0481; CFI = .950). The difference in CFI between the metric and configural models was less than .01 and indicated good fit at the metric level (ΔCFI = 0). At the scalar level, there was evidence for invariance and good fit (RMSEA = .041; SRMR = .041; CFI = .949), the ΔCFI was .001 (see Table 4).

Table 3. Pearson correlation results for DGA total, MDGA total, and MDGA subscale scores

Variables	1	2	3	4	5
1. MDGA Total	1	0.842**	0.789**	0.665**	0.645**
2. MDGA Desire for More	0.842**	1	0.495**	0.348**	0.690**
3. MDGA Insatiable Pursuit	0.789**	0.495**	1	0.310**	0.542**
4. MDGA Retention Motivation	0.665**	0.348**	0.310**	1	0.179**
5. DGA Total	0.645**	0.690**	0.542**	0.179**	1

Note: ** p < .001

Table 4. Standardized factor loadings for MDGA domains for gender

Subscale and Items	Female	Male
Insatiable Pursuit
Item 1	0.75	0.78
Item 4	0.79	0.77
Item 7	0.80	0.83
Item 10	0.77	0.80
Item 13	0.74	0.79
Item 15	0.53	0.67
Item 17	0.75	0.76
Item 19	0.71	0.76
Item 20	0.76	0.79
Desire for More
Item 2	0.68	0.64
Item 5	0.72	0.64
Item 8	0.80	0.76
Item 11	0.82	0.80
Item 14	0.81	0.77
Item 16	66	0.66
Item 18	0.66	0.68
Retention Motivation
Item 3	0.87	0.84
Item 6	0.88	0.87
Item 9	0.86	0.82
Item 12	0.90	0.85

6.1.2. Race and ethnicity

To examine the presence of measurement invariance across race and ethnicity groups, we computed a multiple group CFA model for the three-factor MDGA, including the following groups (groups with at least 100 cases in current sample): (a) White, (b) Black/African American, (c) Hispanic or Latino/a/x, and (d) Asian. The multiple group CFA (see Table 5) confirmed that the 20 observed indicators related to their respective latent variables (standardized regression weights ranges, with three correlated latent variables). Evidence for configural invariance was demonstrated through model fit indices indicating overall acceptable to good fit (RMSEA = .031 [.029–.032]; SRMR = .0704; CFI = .948). At the metric level, the results indicated invariance with evidence for good fit (RMSEA = .030 [90% CI .029–.032]; SRMR = .0726; CFI = .945; ΔCFI = .003). At the scalar level, there was evidence for invariance and acceptable fit (RMSEA = .032 [90% CI .030–.033]; SRMR = .0829; CFI = .949), and the ΔCFI was .008 (see Table 6).

Table 5. Invariance CFA models for gender

Invariance Form	x^2 (df)	p	RMSEA (90% CI)	SRMR	CFI	Δ x^2 (df)	ΔCFI
Configural	1,594.47 (334)	<.001	0.042 (.040–.045)	0.0471	0.950
Metric	1,623.55 (351)	<.001	0.042 (.040–.044)	0.0481	0.950	29.18 (17)	0
Scalar	1,659.83 (368)	<.001	0.041 (.039–.043)	0.0480	0.949	36.17 (17)	0.0010

Table 6. Standardized factor loadings for MDGA domains for race and ethnicity

Subscale and Items	White	Black/African American	Hispanic or Latino/a/x	Asian
Insatiable Pursuit
Item 1	0.77	0.85	0.61	0.75
Item 4	0.78	0.84	0.74	0.74
Item 7	0.84	0.87	0.76	0.75
Item 10	0.81	0.84	0.74	0.71
Item 13	0.79	0.77	0.70	0.80
Item 15	0.59	0.71	0.36	0.58
Item 17	0.77	0.81	0.67	0.72
Item 19	0.74	0.83	0.51	0.74
Item 20	0.79	0.85	0.68	0.75
Desire for More
Item 2	0.65	0.70	0.69	0.60
Item 5	0.69	0.60	0.72	0.61
Item 8	0.79	0.73	0.82	0.73
Item 11	0.81	0.74	0.84	0.78
Item 14	0.80	0.73	0.78	0.81
Item 16	0.64	0.67	0.67	0.63
Item 18	0.69	0.61	0.64	0.66
Retention Motivation
Item 3	0.87	0.79	0.87	0.83
Item 6	0.89	0.84	0.86	0.86
Item 9	0.85	0.82	0.87	0.80
Item 12	0.88	0.84	0.92	0.85

7. Study 3: concurrent validity

In study 3, we sought to determine evidence of concurrent validity of MDGA scores. In this study, we examined the relationship between participants’ Dispositional Greed Scale (DGS; Seuntjens et al., 2015) and MDGA scores. Additionally, we further examined the relationship between DGS scores and the three MDGA subscale scores (Desire for More, Insatiable Pursuit, and Retention Motivation).

7.1. Methods

7.1.1. Participants

We recruited a sample of college students in the United States (N = 486) via Prolific (https://www.prolific.co), an online crowdsourcing platform that allows users to complete surveys for monetary compensation. For this study, participants were compensated $3.00 to complete the survey. We created a Qualtrics survey that included a demographic questionnaire, the MDGA (Lambie et al., 2022), and the DGS (Seuntjens et al., 2015). In terms of participants’ racial identities, the largest proportion of participants identified as White (n = 360; 74.1%), followed by Black and/or African American (n = 40; 8.2%), Asian (n = 37; 7.6%), multi-racial (n = 27; 5.6%), American Indian or Alaskan Native (n = 3; .6%), and Native Hawaiian and Pacific Islander (n = 1; .2%). Four participants noted that they preferred to self-describe their racial identity, and all four of these participants self-identified as Hispanic. Fourteen participants did not provide racial demographic information. In terms of ethnicity, 42 participants (8.7%) self-identified as Hispanic or Latino/a/x, including (a) Mexican, Mexican American, or Chicano (n = 23; 4.7%), Puerto Rican (n = 3; .6%), Cuban (n = 3; .6%), and “another Hispanic or Latino/a/x origin” (n = 13; 1.9%). Most participants identified as men (n = 273; 56.2%), followed by women (n = 192; 39.5%), non-binary or gender expansive (n = 13; 2.7%), and one participant self-identified as “trans man.” Seven participants did not provide gender demographic information (n = 7; 1.5%).

7.1.2. Instrumentation

7.1.2.1. Dispositional greed scale

We utilized the DGS (Seuntjens et al., 2015) to measure participants’ levels of dispositional greed. The DGS is a 7-item unidimensional self-report instrument. The DGS items were constructed based on the developers’ theoretical framework and definition of greed: “the dissatisfaction of not having enough, combined with the desire to acquire more” (Seuntjens et al., 2015, p. 928). The DGS items are worded in self-describing statements (e.g., “I always want more”), and participants report their levels of agreement or disagreement with the self-descriptions on a 5-point Likert scale, ranging from 1 (“strongly disagree”) to 5 (“strongly agree”). Participants’ DGS scores are calculated by averaging their responses to the seven items. Seuntjens et al. (2015) reported strong internal consistency reliability (α ranging .82–.90) across four adult samples, and good test–retest reliability (r = .77). To determine the predictive validity of DGS scores, the developers examined whether participants’ DGS scores predicted their behaviors within economics games. In one study, Seuntjens identified that individuals’ DGS scores predicted their behaviors within a dictator game (individuals with higher dispositional greed levels kept more money for themselves and limited resources for others). In our present study, the Cronbach’s alpha coefficient for the DGS (α = .90) indicated strong internal consistency reliability.

8. Results

To assess for evidence of convergent validity for MDGA scores, we used Pearson correlations to examine the relationship between participants’ MDGA (Lambie et al., 2022) and DGS (Seuntjens et al., 2015) scores (see Table 7). The Pearson correlation results indicated a strong positive correlation (r = .645; p < .001) between participants’ DGS scores and MDGA total scores. Next, we examined the relationship between participants’ DGS scores and MDGA subscale scores. The results identified that DGS scores correlated strongly with MDGA Desire for More (r = .690; p < .001) and Insatiable Pursuit (r = .542; p < .001) subscale scores. The results identified a weak positive correlation between DGS scores and MDGA Retention Motivation subscale scores (r = .179; p < .001).

Table 7. Invariance CFA models for race

Invariance Form	x^2 (df)	p	RMSEA (90% CI)	SRMR	CFI	Δ x^2 (df)	ΔCFI
Configural	1,961.28 (668)	<.001	0.031 (.029–.032)	0.0704	0.948
Metric	2,006.17 (719)	<.001	0.030 (.029–.032)	0.0726	0.945	44.88 (51)	0.003
Scalar	2,339.84 (776)	<.001	0.032 (.030–.033)	0.0829	0.937	333.67 (57)	0.008

9. Discussion

We conducted a three-part study to examine psychometric properties, factor structure, and evidence of validity of MDGA scores among samples of undergraduate college students in the United States. In study 1, we examined test–retest reliability to determine if MDGA scores were stable over a three-week period. Our findings indicated that undergraduate college students’ MDGA total scores had good test–retest reliability (ICC = .811; 95% CI = .745–.860). Our results also identified moderate to good test–retest reliability for the MDGA subscale scores: Desire for More (ICC = .758); Retention Motivation (ICC = .724); Insatiable Pursuit (ICC = .731). In addition to providing evidence of the MDGA’s stability over time, these findings support scholars’ conceptualizations of greed as a stable personality trait (Lambie & Stickl Haugen, 2019; Mussel et al., 2015; Seuntjens et al., 2015). While certain situations may increase individuals’ state-greed levels, dispositional greed is a relatively stable trait. In study 2, we examined the stability and factor structure of the 20-item MDGA among a large sample of undergraduate college students. The CFA results indicated that the 20-item MDGA’s three-factor oblique structure demonstrated an overall good fit. These results are consistent with findings from Lambie et al. (2022) CFA study with a validating sample of adults in the United States. Findings from the multiple group CFA demonstrated evidence for measurement invariance across gender, race, and ethnicity demographic groups.

Lastly, our results from Study 3 demonstrated evidence for concurrent validity of MDGA scores. Our results indicated a large positive relationship between MDGA total scores and DGS (Seuntjens et al., 2015) scores, providing empirical support for the MDGA’s concurrent validity. These findings align with previous research and provide additional support regarding evidence of concurrent validity for MDGA scores (Lambie et al., 2022). We also examined the relationship between participants’ individual MDGA subscale scores and DGS scores. Our results suggested strong positive relationships between DGS scores and two MDGA subscales: Desire for More and Insatiable Pursuit. The strong relationship between DGS scores and these subscale scores is consistent with Lambie and Stickl Haugen’s (2019) multidimensional definition of greed encompassed within these factors. In developing the DGS, Seuntjens et al. (2015) defined dispositional greed as “the dissatisfaction of not having enough” (relating to MDGA insatiability) paired with “the desire to acquire more” (relating to MDGA desire for more; Seuntjens et al., 2015, p. 928). Of note, the MDGA subscale of Retention Motivation demonstrated a weak correlation with DGS scores. This finding suggests that the retention motivation aspect of greed measured within the MDGA is unique. The MDGA is the only known greed scale that includes an individual retention motivation component, and further research is needed to explore the retention aspect of dispositional greed and how it relates to the construct as a whole.

10. Limitations

While the present studies provided evidence for the validity of the MDGA scores, there are notable limitations. Across the three studies, the largest proportion of participants identified as White (between 48% and 74%), followed by Latinx and/or Hispanic (8.7–27.3%). Fewer participants identified as Black and/or African American (8.2–13%), Asian (6–9.3%), and American Indian or Alaskan Native (.6–.9%). Future researchers should examine the validity and reliability of MDGA scores using racially and ethnically representative validating samples. It would be beneficial to explore additional evidence of test–retest reliability and concurrent validity with a broader national sample. Additionally, there are limitations associated with self-report measures, especially when measuring negatively viewed traits (e.g., greed). Future researchers may examine participants’ levels of dispositional greed through observational methods (e.g., examining stability of greed-related behaviors across conditions, using behavioral coding system).

In summary, we sought to examine the stability, factor structure, and validity of MDGA scores among validating samples of undergraduate college students in the United States. Our results identified strong test–retest reliability, which confirmed the stability of MDGA scores over time, and provided further evidence that dispositional greed is a stable personality trait (Mussel et al., 2015; Seuntjens et al., 2015). Results in Study 2 indicated that the three-factor oblique structure demonstrated an overall good fit, consistent with Lambie et al. (2022) CFA study with a validating U.S. adult sample. Results from Study 3 provided further evidence for concurrent validity of MDGA scores through positive correlations with DGS scores (Seuntjens et al., 2015). Our findings provided evidence of stability and confirmed the factor structure of the MDGA. Clinicians and researchers may review these results while considering the utility of the MDGA for assessing individuals’ levels of multidimensional dispositional greed.

Preregistration

This study was not preregistered

Acknowledgements

Funding to support these studies were provided by the Robert N. Heintzelman Eminent Scholar Endowment.

Data availability statement

The data that support the findings of this study are available from the corresponding author, upon reasonable request.

Disclosure statement

We have no known conflicts to disclose.

Additional information

Funding

The work was supported by the Robert N. Heintzelman Eminent Scholar Endowment.

Notes on contributors

Glenn W. Lambie

Dr. Glenn Lambie serves as Senior Associate Dean and The Robert N. Heintzelman Eminent Scholar Endowed Chair. He is a Professor of Counselor Education and a fellow of the American Counseling Association.

Caitlin Frawley

Dr. Caitlin Frawley is a visiting lecturer in the Counselor Education program at the University of Central Florida.

Jaimie Stickl Haugen

Dr. Jaimie Stickl Haugen is a clinical assistant professor in the Counselor Education program at William & Mary.