The Crisis Migration Experience Scale: Developing and Validating a Tool for Venezuelan Youth and Adults Residing in Colombia

Abstract

Objective:

Although prior research has shown that an array of distinct experiences related to crisis migration are associated with mental health, there is a pressing need for a theory-driven, multidimensional measure to assess the broad spectrum of crisis migration experiences. As such, the present study focused on developing and validating the Crisis Migration Experiences Scale (CMES) with a sample of Venezuelan migrants in Colombia.

Method:

Participants were adolescent (ages 12–17; n=430) and adult migrants from Venezuela (ages 18+; n=569). Randomly splitting the adolescent and adult samples in half, exploratory factor analysis and confirmatory factor analyses were conducted with 26 original items. After identifying a satisfactory factor structure to generate a 16-item scale, we examined the associations of the CMES-16 with mental health outcomes.

Results:

We provide evidence for reliability, factorial validity, and concurrent validity of scores generated by the CMES-16 in a sample of Venezuelan crisis migrants in Colombia. Whereas our a priori conceptualization included seven domains, the exploratory and confirmatory factor analyses indicated that four are especially salient: material hardship, desperation, danger, and unplanned departure.

Conclusions:

Crisis migration is an increasingly important construct frequently referenced in the literature on migrant health and by international humanitarian organizations. The number of crisis-migrant groups worldwide is increasing, with Ukrainians and Afghans recently added to the list of such groups, along with Venezuelans, Syrians, South Sudanese, Iraqis, and Central Americans. Developing and validating the CMES-16 with Venezuelan crisis migrants opens up important avenues of research, including work that incorporates other crisis migrant populations.

Introduction

Crisis migration is a major worldwide challenge. The term crisis migrant includes asylum seekers and refugees but is also inclusive of other migrant populations that have emigrated from their home country for reasons of safety or survival, typically involving an uncertain and unplanned departure (Martin et al., 2014). The United Nations High Commissioner for Refugees (UNHCR, 2023) estimates that more than 100 million people worldwide have been forced to flee their homes—escaping economic collapse and insecurity in Venezuela, widespread violence in Central America’s “Northern Triangle,” uncertainty and terrorism in Afghanistan, and armed conflict in countries such as Ukraine, the Democratic Republic of the Congo, and Sudan.

Below, we provide an overview of our conceptualization of the essential aspects of crisis migration and the experiences of crisis migrants—with a particular emphasis on the Venezuelan diaspora. Subsequently, we argue that there is a pressing need for a theory-driven, multidimensional measure of crisis migration experiences for use with children and adults. With that as our foundation, we describe the development and validation of the Crisis Migrant Experiences Scale (CMES) with a sample of Venezuelan crisis migrants in Colombia. Our overall objective is to generate a cutting-edge measure of crisis migration experiences that can be utilized with the Venezuelan diaspora in Colombia and elsewhere to understand crisis exposure and its links with mental health, and that has the potential to be adapted and refined for use with crisis migrants worldwide.

Defining Crisis Migration

We understand crisis migration as the large-scale relocation of individuals due to catastrophic military, social, political, economic, and/or environmental events (Vos et al., 2021). Crisis migration is characterized by the often uncertain and unplanned emigration of children, adults, and families from their home country for reasons of safety or survival. The emotional resonance of crisis migration can be captured by situations in which persons or families look at their immediate context and conclude that “we have to go now” (Salas-Wright et al., 2021).

Two important definitional points can be made here: First, crisis migrants are distinct from economic or “voluntary” migrants who leave their homelands by choice in search of employment, economic opportunities, marriage, or family unification (Steiner, 2023). Although Martin et al. (2014) rightly note that “few migrants are wholly voluntary or wholly forced,” an essential characteristic of crisis migration is that pressing contextual factors (e.g., violence, hardship, natural disasters) constrain choice and compel crisis migrants to flee. Second, here, the term crisis migrant is not limited only to refugees and asylum seekers—terms defined by the United Nation’s (UN) 1951 Refugee Convention relating to people who have fled “owing to well-founded fear of being persecuted for reasons of race, religion, nationality, or membership of a particular group or social opinion.” Rather, the terms crisis migrant and crisis migration are inclusive of refugees, asylum-seekers, and other migrants who have fled their homes due to disaster, including those who may not fit refugee/asylum criteria, such as climate migrants, those fleeing gang violence, and individuals escaping despotic or unstable governments.

The Venezuelan Diaspora—A Prime Example of Crisis Migration

The Venezuelan diaspora is emblematic of crisis migration for several reasons. First, it is a large out-migration (more than seven million Venezuelans have relocated since 2015, representing roughly 20% of the country’s population; R4V, 2023). Second, the Venezuelan out-migration is not due to one singular factor but rather is the result of the confluence of several macro-level factors (i.e., economic instability, hyperinflation, political repression, and widespread violence) that have made life untenable for many in Venezuela (Salas-Wright et al., 2022). As explained in more detail below, crisis migrations may be triggered by a particular event (e.g., a catastrophic inflationary increase or serious food shortage), but typically, the triggering event prompts a large migration only in the presence of other ongoing severe economic, social, or political conditions. Third, the Venezuelan out-migration was termed an “overnight migration” (Salas-Wright et al., 2022) because it occurred rapidly. Finally, the status of Venezuelans is “blurry” in the sense that some Venezuelans have obtained asylum status, some are refugees, and some have relocated as international migrants (UNHCR, 2023). The broader construct of crisis migration allows us to focus not on these distinctions but on emphasizing the relocation of millions amid a crisis.

Crisis Migration Experiences—A Multidimensional Construct

As depicted in Figure 1, our conceptualization of the crisis migration experience is a multidimensional construct that we have organized a priori along three dimensions: pre-migration, departure, and aftermath. These dimensions and their corresponding constructs were derived from prior research and theory related to crisis migration, including our own work with crisis migrants in the Americas detailed below.

Figure 1.

Multidimensional Conceptualization of Crisis Migration Experience

We suggest the presence of three constructs within the pre-migration dimension that relate to an individual’s pre-migration experiences: material hardship (i.e., economic scarcity, inability to meet basic needs), physical danger (i.e., lack of personal/family safety), and psychological desperation (i.e., intense and emotion-laden drive to escape). Consistent with our theorizing and that of others (see Martin et al., 2014; Salas-Wright et al., 2021; Vos et al., 2021), these three constructs are fundamental aspects of life amid or in the aftermath of a crisis event and, by definition, one or more of these must be experienced by all crisis migrants. Mejía-Trujillo et al. (2023) describe in depth how hardship and danger amid Venezuela’s social and political instability led many parents to go to great lengths to relocate and remain as expatriates despite missing their homeland and struggling to resettle.

Within the departure dimension, we also posit three constructs. The two primary constructs are unplanned departure and abrupt departure, both of which distinguish crisis migration from less-pressured, voluntary migration processes in which circumstances allow for careful planning and intentional parting/farewell. The third construct is a discrete “tipping point” in which something took place—potentially a frightening or dangerous event or an acute economic shock—in which the decision to flee was triggered. These departure dimensions align with McAdam’s (2012, 2014) conceptualization in which a crisis hastens migration amid pre-existing stressors such as poverty, weak social or political institutions, or changing climate conditions. The tipping point construct is also consistent with our research indicating that social vulnerability profoundly shaped the experiences of climate migrants displaced by Hurricane Maria (Clark-Ginsberg et al., 2023; Salas-Wright et al., 2021).

Finally, we postulate one construct within the aftermath dimension: guilt and worry. The constructs of guilt and worry are rooted in the notion of “survivor’s guilt,” which has been identified as a salient construct in research on refugee populations (Stotz et al., 2015). Survivor’s guilt relates to feelings of guilt for leaving friends and family behind, whereas worry relates to concern about loved ones who remain in the crisis context and who may experience danger or hardship (Hutson et al., 2015).

There are several compelling arguments for the need to develop a multidimensional measure. For one, the term crisis migration is an important construct without an established measure—it is used frequently by scholars and humanitarian organizations with a reasonable degree of consensus as to the general constructs, but no systematically developed measures exist to assess self-reported crisis migration experiences at the individual level. This lack of measurement instruments impedes our understanding of specific crisis migration experiences among adolescents and adults. There also may be a tendency to conceptualize international migrants dichotomously as either crisis migrants or not rather than considering the degrees to which individuals can endorse or embody specific dimensions or criteria representing crisis migration. As a key example, some individuals fleeing Venezuela, Ukraine, Afghanistan, or any other crisis situation may strongly endorse criteria for crisis migration, whereas other individuals from these same groups may endorse these criteria far less strongly. In turn, these endorsements may relate to or predict mental health outcomes such as symptoms of depression or anxiety. Examining these research questions is not feasible without psychometrically valid and reliable measures of the various crisis migration criteria enumerated here and elsewhere in the literature.

The Present Study

In the present study, we aimed to develop and validate the CMES with a sample of Venezuelan crisis migrant adolescents and adults in Colombia. Establishing such a measure can open new avenues for research with Venezuelan crisis migrants and, critically, lays the groundwork for adapting a measure for use within and across crisis migration groups. Importantly, a measure that can be used with adolescents and adults opens up important possibilities for practice-relevant research.

Method

Recruitment and Sample

Participants were recruited using a referral system where initial seed participants are referred through community partners, and participants refer others to the study in exchange for additional compensation (Heckathorn, 2002). Participants were compensated for completing the study measures, with adolescents receiving 40,000 Colombian Pesos (COP; approximately $10 US dollars [USD]) and adults receiving 70,000 COP (approximately USD 18). Participants were eligible to refer up to five additional participants and could receive 20,000 COP (approximately USD 4) for each successful referral (i.e., the referred participant who joined the study). Most participants received only one or two secondary referral incentives, and fewer than 3 percent of our participants referred all five additional participants.

Study participants were Venezuelan adolescents (ages 12–17; 217 males, 212 females; mean age 14.0) and Venezuelan adults (ages 18+; 104 males, 462 females; mean age 35.1, range 18–76, 91% under age 50) who migrated from Venezuela to Colombia. Nearly half of the adolescent sample migrated from Venezuela to Colombia in 2018 or 2019 (49.5%; 12% arrived prior to 2018, and 38.5% arrived in 2020 or later), and a narrow majority of adults migrated from Venezuela to Colombia in 2018 or 2019 (54.0%; 18% arrived prior to 2018 and 28.01% arrived in 2020 or later). The sample included those residing in two Colombian cities: Bogotá (52.1% of adolescents, 42.6% of adults) and Medellín (47.9% of adolescents, 57.4% of adults).

Procedures

Data were collected between April and July 2023. Participants were recruited in partnership with our community partner organizations and via peer referrals from existing participants. Research team members, along with assessors working in the local community, distributed hard-copy flyers and recruited eligible participants to enroll in the study. Anyone who was a Venezuelan migrant living in Bogotá or Medellín and who arrived after 2015 was eligible to take part in the study. We selected 2015 as a cutoff as this was the year that the number of Venezuelan migrants in Colombia increased dramatically (Salas-Wright et al., 2022). Exclusion criteria included having plans to return to Venezuela in the near future or intoxication from alcohol or drugs—no individuals met these exclusion criteria. All data were collected in person, in office spaces managed by the Colombia partner organization. Individuals who agreed to participate were consented/enrolled by a research staff member (adolescents also needed a caregiver to provide consent) and then independently completed the surveys on tablets using the Qualtrics software in a private room, with project staff available to troubleshoot or answer questions. The survey administration process was identical for adolescents and adults. In cases where a participant had limited literacy or proficiency with a tablet, the questions were read aloud to the participant and recorded on the tablet by a staff member. All participants completed the survey in Spanish using a survey that had been translated, back-translated, and evaluated for interpretability via cognitive interviews with Venezuelan migrant adolescents and adults.

Measures

Crisis Migration Experiences

Following McAdam’s (2014) conceptualization and our theoretical work (see Salas-Wright et al., 2021; and Vos et al., 2021), we operationalized crisis migration experiences as relating to several domains: material hardship, desperation, danger, abrupt departure, unplanned departure, tipping point, and guilt and worry. Although the de facto approach—our own included—is typically one rooted in a yes/no categorization of who is or is not a crisis migrant (or tacit assumptions regarding crisis migrant status) (see Ertanir et al., 2023), here we attempted to assess how crisis migration experiences manifested to varying degrees in the lives of migrants who are part of a major crisis out-migration—in this case, Venezuelans relocated to Colombia.

In developing a measure of crisis migration experiences, a panel of three academic experts in crisis migration (with doctoral degrees in developmental psychology, human development and family studies, and social work) and a pre-doctoral anthropologist articulated items across salient domains with the goal of generating four unique items per domain. Only items that received consensus support from all three experts were retained. All items were positively phrased to avoid pitfalls with negatively phrased items (see Yu et al., 2013; and Salas-Wright et al., 2015). Response options included strongly disagree (“totalmente en desacuerdo”), disagree (“en desacuerdo”), neutral (“neutral”), agree (“de acuerdo”), and strongly agree (“totalmente de acuerdo”). Following item generation and fine-tuning (i.e., minor edits to improve interpretability), items were further refined (i.e., minor changes based on social context)—in Spanish, following preliminary translation—in conversation with community partners working with Venezuelan crisis migrants (two in Colombia, two in the United States). After two items in the “tipping point” category were discarded due to concerns regarding interpretability, a total of 26 items were utilized. A complete list of the initial 26 items is presented in English in Table 1.

Table 1.

Exploratory Factor Loadings for Crisis Migration Experience Items among Youth and Adults

		Material	Desperate.	Danger	Departure	Guilt
	Item	(Y | A)	(Y | A)	(Y | A)	(Y | A)	(Y | A)
1	I had to leave because there was no way to meet my needs.	.65 | .63
2	I had to leave because members of my family were suffering a great deal.	.63 | .62	.31 | .33
3	I had to leave because I couldn’t get everyday basics like food, water, and clothes.	.75 | .80
4	Life was so hard that I was no longer able to stay.	.64 | .87	.33 | .00
5	At a certain point, life in Venezuela was no longer feasible for me.	.65 | .66
6	I was so desperate to leave Venezuela that I didn’t care where I would wind up.		.75 | .44		.00 | .66
7	I was willing to take whatever kind of work I needed to in order to get out.	.00 | .34	.72 | .68
8	I was willing to go anywhere or do anything if it meant being able to leave Venezuela.		.83 | .76		.00 | .30
9	As long as I was able to leave Venezuela, I didn’t care what happened after that.		.71 | .73
10	I had to leave because I was in serious danger.			.51 | .70
11	In order to stay alive and safe, I had to get out of Venezuela.			.65 | .75
12	It is not safe for me or my family to return to Venezuela at this time.			.61 | .67
13	I was not safe in Venezuela.			.72 | .72
14	Something terrible or frightening happened that convinced me that I could not stay in Venezuela.			.48 | .58
15	I had to leave abruptly and could not take time to plan.	.40 | .32			.33 | .70
16	I never thought I would leave Venezuela until very shortly before I left.	.32 | .47			.00 | .38
17	I left Venezuela all of the sudden.				.44 | .77
18	I never planned to leave Venezuela.	.36 | .34			.35 | .51
19	I left without knowing for sure where I was going to live.				.61 | .69
20	I did not have time to plan how things would be after I arrived.				.61 | .69
21	When I left Venezuela, I wasn’t even sure where I would wind up.		.34 | .00		.65 | .69
22	I wish I could have taken more time to decide whether I really wanted to leave Venezuela.				.00 | .44	.49 | .00
23	I feel guilty for leaving Venezuela.					.45 | .62
24	Part of me feels as though I should have stayed in Venezuela and helped my family and friends there.					.62 | .63
25	I feel bad for leaving my family members or friends behind in Venezuela.					.58 | .58
26	I worry that my friends and family in Venezuela are suffering or cannot make ends meet.	.00 | .43				.36 | .44

Note: Y = youth. A = adult. Items loading below .30 are represented as blank values. Loading values in bold represent items that loaded clearly (only one item above .40) for both youth and adults at .50 or greater with rounding. Loading values in italics represent items that did not load on a retained factor. Exploratory factor analysis conducted with samples divided in half while accounting for age, sex, and city of residence in Colombia.

Mental Health

Mental health was assessed in terms of depression, anxiety, and posttraumatic stress disorder (PTSD) symptoms. For depressive symptoms (α = .79 for adolescents, .91 for adults), we used the Boston Form of the Center for Epidemiological Studies Depression Scale (Kohout, 1993). This instrument taps into symptoms such as listlessness, anhedonia, and lack of interest in activities during the week before assessment. Participants responded to each item using a 4-point Likert scale ranging from 0 (Rarely or None of the Time) to 3 (All of the Time). This measure has been used in prior research with Spanish language samples (Gonzalez et al., 2017) and has been shown to be valid with an array of Latin American migrant samples, including Venezuelans (Schwartz et al., 2018).

For anxiety symptoms (α = .83 for adolescents, .87 for adults), we used the 7-item Generalized Anxiety Disorder (GAD-7) scale (Spitzer et al., 2006). This measure assesses anxiety symptoms, such as excessive worrying, tension, irritability, and difficulty sleeping, during the two weeks before the assessment. Sample items include “feeling nervous, anxious, or on edge” and “having difficulty relaxing.” Participants responded to each item using a 4-point Likert scale ranging from 0 (Not at all) to 3 (Nearly Every Day). This measure is valid with an array of Latin American migrant samples, including Venezuelans in Colombia and the United States (Schwartz et al., 2018).

For PTSD symptoms, we used separate measures for adolescents and adults. Among adults, PTSD symptoms were measured using an adaptation of the Short Screening Scale for the Diagnostic and Statistical Manual of Mental Disorders (α = .81; Breslau et al., 1999). This seven-item scale, with responses of yes or no, assesses symptoms of PTSD related to diagnostic criteria domains, including avoidance/numbing and arousal related to traumatic events. This scale has been used in prior research with Venezuelan migrants in Colombia (Vos et al., 2022).

Among children, we used the Children’s Revised Impact of Event Scale (α = .92; Perrin et al., 2005). This scale is oriented around the following prompt, which we adapted—based on feedback from local experts in behavioral health—by adding an opening sentence to specifically focus the measure on pre-migration traumatic experience: “When you lived in Venezuela, did you ever have an experience that was very scary, horrible, or uncomfortable? Below is a list of comments from people who have experienced stressful events. Indicate how frequently each statement has been true for you in the last seven days.” Comments include “Do you think about it even when you don’t mean to?” and “Do you stay away from things that remind you of it (like places or situations)?” Response options include never, rarely, sometimes, and often.

Demographic Factors

Demographic factors used in the analysis include self-reported age in years, gender (male female, other), and city of residence in Colombia (Bogotá or Medellín).

Results

Analysis Plan

The present analyses proceeded in three main phases using Stata 18. In the first phase, we conducted exploratory factor analyses (using a randomly selected half-sample) with the 26 crisis migration experiences items, separately for adolescents and adults. In the second phase, we conducted confirmatory factor analyses (using the second randomly selected half-sample) including items that loaded cleanly/strongly onto factors retained in the exploratory factor analysis. We also report coefficient alphas for scores on the final, 16-item Crisis Migration Experiences Scale (CMES-16) and on its corresponding subscales. Then, in the third phase, having identified a satisfactory factor structure to generate a scale, we examined the associations of the CMES-16 (and its subscales) with key mental health outcomes. Below we describe each of these phases in greater detail.

Exploratory Factor Analysis

For the exploratory and confirmatory factor analyses, the adolescent and adult samples were each randomly split in half—balancing for age, gender, and city of residence—using Stata’s splitsample command. For both adolescents and parents, exploratory factor analyses were conducted on the first half-sample (Adolescent n = 223; Adult n = 287), and confirmatory factor analyses were conducted on the second half-sample (Adolescent n = 207; Adult n = 282).

Following Conway and Huffcutt (2003), we used a multi-stage process to decide on the number of factors to extract. First, we examined the scree plot, where the “leveling-off-point” on the scree line represents the last factor that should be extracted. Second, we considered only factors with eigenvalues above 1.00, where the eigenvalue represents the product of the number of items entered into the analysis and the percentage of variability accounted for by the factor. Finally, as recommended by Tabachnick et al. (2013), we extracted only factors characterized by at least two loadings above .70 or by at least three loadings above .60. Items that did not load cleanly onto one factor for either adolescents or adults (more than one factor loading above .40) or that failed to load at .50 or greater (with rounding) on one factor for both adolescents and adults were dropped from the scale. We used orthogonal (varimax) rotation, the most common method which provides maximal factor isolation and simplifies the resulting factor structure (Cudeck & MacCallum, 2007). Eigenvalues reported are therefore those derived after the solution was rotated.

Based on this multi-stage decision process, we extracted and retained four factors: material hardship, desperation, danger, and unplanned departure. For adolescents, desperation emerged as the strongest factor (eigenvalue 3.24, 26.05% variability explained [VE]) followed closely by material hardship (eigenvalue 3.16, 25.45% VE), danger (eigenvalue 2.24, 18.07% VE), and unplanned departure (eigenvalue 2.09, 16.80% VE). Among adults, unplanned departure (eigenvalue 4.21, 28.24% VE) emerged as the strongest factor, followed closely by material hardship (eigenvalue 3.79, 25.38% VE) and then danger (eigenvalue 2.79, 18.68% VE) and desperation (eigenvalue 2.37, 15.89% VE). Although the order of factors was different for adolescents/adults, the factors are presented side-by-side in Table 1 to facilitate interpretation.

Among both adolescents and adults, a fifth factor—guilt and worry—had an eigenvalue greater than 1.00 (adolescents: eigenvalue 1.58, 12.75% VE; adults: eigenvalue 1.56, 10.45% VE), but the corresponding items failed to meet the factor loading threshold of least two loadings above .70 or by at least three loadings above .60. Additionally, supplemental parallel analysis, which compares the eigenvalues from the data entered into analysis to the eigenvalues obtained from a random-number dataset with the same number of cases and variables (Thompson & Daniel, 1996), suggested that only four classes should be retained. As such, items loading on the guilt and worry factor (#23–26) were dropped, as were other cross-loading and weak-loading items (#6, 15–18, 22) (see Table 1 for prompts that correspond with these numbers and Table 2 for the full items that were retained in English and Spanish).

Table 2.

Final CMES-16 Items with Spanish Translation

1	I had to leave because there was no way to meet my needs.
	Tuve que irme porque no había forma de satisfacer mis necesidades.
2	I had to leave because members of my family were suffering a great deal.
	Tuve que irme porque los miembros de mi familia estaban sufriendo mucho.
3	I had to leave because I couldn’t get everyday basics like food, water, and clothes.
	Tuve que irme porque no podía conseguir las cosas cotidianas básicas como comida, agua, y ropa.
4	Life was so hard that I was no longer able to stay.
	La vida era tan dura que ya no podía quedarme.
5	At a certain point, life in Venezuela was no longer feasible for me.
	En cierto punto, la vida en Venezuela ya no era factible para mí.
7	I was willing to take whatever kind of work I needed to in order to get out.
	Estaba dispuesto/a a tomar cualquier tipo de trabajo que necesitara con el fin de salir de Venezuela.
8	I was willing to go anywhere or do anything if it meant being able to leave Venezuela.
	Estaba dispuesto/a a ir a cualquier parte o hacer cualquier cosa si eso significaba poder salir de Venezuela.
9	As long as I was able to leave Venezuela, I didn’t care what happened after that.
	Mientras que pudiera salir de Venezuela, no me importó lo que sucediera después de eso.
10	I had to leave because I was in serious danger.
	Tuve que irme porque estaba en grave peligro.
11	In order to stay alive and safe, I had to get out of Venezuela.
	Con el fin de mantenerme vivo/a y seguro/a, tuve que salir de Venezuela.
12	It is not safe for me or my family to return to Venezuela at this time.
	No es seguro para mí o mi familia regresar a Venezuela en este momento.
13	I was not safe in Venezuela.
	No estaba seguro/a en Venezuela.
14	Something terrible or frightening happened that convinced me that I could not stay in Venezuela.
	Sucedió algo terrible o aterrador que me convenció de que no podía quedarme en Venezuela.
19	I left without knowing for sure where I was going to live.
	Me fui sin saber con seguridad dónde iba a vivir.
20	I did not have time to plan how things would be after I arrived.
	No tuve tiempo de planificar cómo serían las cosas después de llegar.
21	When I left Venezuela, I wasn’t even sure where I would wind up.
	Cuando salí de Venezuela, ni siquiera estaba seguro/a de dónde terminaría.

Confirmatory Factor Analysis

Using the second half-sample, and separately for adolescents and adults, we entered this four-factor solution into a second-order confirmatory factor model (see Figure 2). This parameterization was based on the conceptualization that the four factors (material hardship, desperation, danger, and unplanned departure) represent components of an underlying crisis migration experience construct. Each item was attached to the factor onto which it patterned in the exploratory analysis, and no cross-loadings were estimated in the initial model. The factor solution was evaluated according to standard structural equation modeling criteria, using two incremental fit indices (comparative fit index [CFI] and Tucker-Lewis index [TLI]) and two absolute fit indices (root mean square error of approximation [RMSEA] and standardized root mean square residual [SRMR]). Incremental fit indices evaluate the extent to which the specified model provides improved fit over a null model with no paths or latent variables, whereas absolute fit indices evaluate the extent to which the covariance structure implied by the specified model deviates from the covariance structure observed in the data (Kline, 2023). The chi-square index is reported, but not used in interpretation, because it tests the null hypothesis of perfect model fit (which is rarely tenable in even moderately complex models or reasonably sized samples). Using Kline’s suggested criteria, excellent fit was characterized as comparative fit index (CFI) ≥ .95, Tucker Lewis index (TLI) ≥ .95; root mean square error of approximation (RMSEA) ≤ .08; and standardized root mean square residual (SRMR) ≤ .06. Adequate fit is characterized as CFI ≥ .90; TLI ≥ .90; RMSEA ≤ .10; and SRMR ≤ .08. Nonetheless, there is some controversy concerning how model fit should be evaluated, and therefore a model that satisfies most (but not all) of the criteria should not necessarily be rejected (Marsh et al., 2004).

Figure 2.

Second-Order Confirmatory Factor Model for Youth and Adults. Loading on the left represent adolescents ages 12–17 and loadings on the right represent adults age 18+. All loadings significant at p < .001.

The initial confirmatory factor analysis model fit the data adequately for both adolescents, χ²(100) = 165.871, p < .001; CFI = .943; TLI = .931; RMSEA = .061 (90% CI = .044–.077); SRMR = .067; and adults, χ²(100) = 207.934, p < .001; CFI = .947; TLI = .937; RMSEA = .068 (90% CI = .044–.079); SRMR = .068. Based on modification indices, we allowed the error terms for #12 (“Unsafe to return”) and #13 (“Not safe in Venezuela”) to covary, along with those for #19 (“No place to live”) and #21 (“Uncertain destination”) among both adolescents and adults. With these modifications, model fit improved to excellent for both adolescents, χ²(98) = 150.653, p < .01; CFI = .954; TLI = .944; RMSEA = .055 (90% CI = .037–.072); SRMR = .064; and adults, χ²(98) = 155.358, p < .001; CFI = .972; TLI = .966; RMSEA = .049 (90% CI = .034–.063); SRMR = .054.

CMES-16 Reliability and Mean Scores

The reliability coefficient alpha for CMES-16 total scores was .86 for adolescents and .88 for adults. Mean scores for the full CMES-16 were 2.95 (SD = 0.78) for adolescents and 3.51 (SD = 0.82) for adults. Among adolescents, the CMES subscales provided the following reliability coefficients and mean scores: hardship (α = .85; M = 3.57, SD = 1.01), desperation (α = .85; M = 2.81, SD = 1.18), danger (α = .80; M = 2.42, SD = 0.95), and unplanned departure (α = .78; M = 2.93, SD = 1.14). Among adults, the CMES subscales provided the following alpha coefficients and mean scores: hardship (α = .89; M = 4.26, SD = 0.91), desperation (α = .87; M = 3.61, SD = 1.19), danger (α = .84; M = 2.73, SD = 1.32), and unplanned departure (α = .84; M = 3.43, SD = 1.32).

OLS Regression to Examine Concurrent Validity

Table 3 presents the results of regression analyses examining the associations between the CMES-16 and key mental health outcomes. Higher CMES-16 scores were significantly associated with greater severity of depression, anxiety, and PTSD symptoms among adolescents and adults both at the bivariate level (i.e., OLS regression with no control variables) and when controlling for age, gender, and city of residence in Colombia.

Table 3.

Regression Analyses of the Relationships between Crisis Migration Experience and Mental Health Outcomes

Mental Health Outcome	Youth Sample (ages 12–17)								Adult Sample (18 and older)
	Bivariate Only				Demographic Controls				Bivariate Only				Demographic Controls
	B	SE B	β	p value	B	SE B	β	p value	B	SE B	β	p value	B	SE B	β	p value
Depression
16-Item Scale	0.142	0.038	.180	< .001	0.130	0.040	.163	.001	0.130	0.036	.150	< .001	0.131	0.036	.150	< .001
Subscales
Hardship	0.071	0.030	.117	.017	0.061	0.030	.099	.045	0.033	0.032	.042	.314	0.009	0.033	.011	.791
Desperation	0.097	0.025	.191	< .001	0.094	0.025	.183	< .001	0.123	0.023	.221	< .001	0.115	0.024	.206	< .001
Danger	0.080	0.032	.124	.012	0.071	0.034	.110	.035	0.037	0.024	.065	.126	0.054	0.025	.094	.028
Unplanned	0.104	0.026	.196	< .001	0.093	0.027	.172	.001	0.064	0.022	.125	.003	0.073	0.022	.143	.001
Anxiety
16-Item Scale	0.157	0.045	.170	< .001	0.156	0.047	.168	.001	0.160	0.040	.166	< .001	0.166	0.040	.173	.001
Subscales
Hardship	0.060	0.035	.085	.082	0.055	0.036	.077	.128	0.075	0.036	.086	.040	0.050	0.037	.058	.174
Desperation	0.099	0.030	.164	.001	0.100	0.031	.165	.001	0.130	0.028	.197	< .001	0.120	0.028	.182	.001
Danger	0.110	0.037	.144	.004	0.113	0.040	.148	.005	0.083	0.028	.126	.003	0.107	0.028	.160	.001
Unplanned	0.078	0.031	.123	.014	0.077	0.033	.121	.020	0.062	0.025	.106	.014	0.079	0.025	.134	.002
Posttraumatic Stress
16-Item Scale	0.225	0.058	.187	< .001	0.226	0.061	.186	< .001	0.096	0.016	.238	< .001	0.093	0.017	.230	.001
Subscales
Hardship	0.073	0.045	.079	.109	0.084	0.046	.091	.071	0.021	0.015	.058	.167	0.024	0.015	.068	.115
Desperation	0.147	0.039	.188	< .001	0.147	0.040	.187	< .001	0.073	0.011	.265	< .001	0.082	0.016	.298	.001
Danger	0.260	0.047	.266	< .001	0.263	0.051	.268	< .001	0.051	0.012	.182	< .001	0.045	0.012	.161	.001
Unplanned	0.061	0.041	.075	.135	0.052	0.043	.063	.225	0.059	0.010	.237	< .001	0.056	0.010	.226	.001

Note: B = unstandardized coefficient; SE B = standard error of unstandardized coefficient; β = standardized coefficient Demographic controls include age, sex, and city of residence in Colombia.

We also examined associations of the CMES-16 subscale scores with mental health. Among adolescents, the hardship subscale was associated with symptoms of depression, but not of anxiety or PTSD; the desperation and danger subscales were uniformly associated with all mental health outcomes; and unplanned departure was associated with symptoms of depression and anxiety but not of PTSD. Among adults, the hardship subscale was only associated with anxiety at the bivariate level, whereas the desperation, danger, and unplanned departure subscales were associated with all outcomes while controlling for demographic factors.

Discussion

The present study provides initial evidence for the construct of crisis migration experiences. Specifically, we provide evidence for reliability, factorial validity, and concurrent validity of scores generated by the CMES-16 and its subscales (hardship, desperation, danger, unplanned departure) in a sample of Venezuelan crisis migrant adolescents and adults in Colombia. Below we highlight the study’s key findings and consider how these findings fit within the broader field of crisis migration research.

Components of Crisis Migration Experiences

The first key finding involves evidence supporting the core postulate that crisis migration experiences manifest across multiple, distinct dimensions for adolescents and adults. Whereas our a priori conceptualization included seven domains, the exploratory and confirmatory factor analyses indicated that four are especially salient: material hardship, desperation, danger, and unplanned departure. Even though these constructs may be unique to the Venezuelan diaspora or Venezuelans in Colombia, we note that these constructs are closely connected to prior descriptive and conceptual work related to large-scale migration across an array of crises (Clark-Ginsberg et al., 2023; Hodges et al., 2023; Salas-Wright et al., 2020). For instance, Clark-Ginsberg et al. (2023) identified, in a qualitative examination of the experiences of climate migrants from Puerto Rico to the U.S. mainland, clear instances of how hardship (i.e., major economic shock and lack of access to food, water, and medicine) and danger (i.e., increases in crime and looting) lead to a sense of desperation that resulted in a profoundly unplanned departure after Hurricane Maria.

The constructs that were eliminated based on psychometric evaluation included abrupt departure, tipping point, and guilt and worry. Abrupt departure is a construct that we observed as overlapping with unplanned departure, but abrupt departure items loaded relatively weakly on this departure dimension. This finding suggests that the crisis aspect of departure relates more to the lack of a clear plan for the post-migration phase than to a rapid exit from the home country per se. Tipping point items did not fall away entirely from the crisis migration construct, but rather—somewhat to our surprise—were absorbed into the material hardship and danger factors. This finding seems to indicate that these tangible crisis-drivers (hardship and danger) involve a decisive moment, which is in keeping with prior theorizing and research. The guilt and worry items were eliminated because a sufficiently robust factor was not identified. Although this construct—which we understood as part of the crisis migration aftermath—does not seem to fit under the crisis migration experiences umbrella, it seems likely that it is of relevance to future research with crisis migrants and may be further developed independently.

Crisis Migration Experiences and Mental Health

Beyond modeling the components of crisis migration experience, a second key finding involves identifying a consistent pattern of associations of the CMES-16 with symptoms of depression, anxiety, and PTSD among both adolescents and adults. This finding is important because these associations suggest concurrent validity for CMES-16 scores vis-à-vis two of the most common mental health conditions among immigrants (depression and anxiety; Salas-Wright et al., 2014) and a serious challenge among individuals in high-stress and conflict settings (PTSD). Not only does this pattern of findings support the validation of our measure, but it also suggests that—moving forward—the CMES-16 has strong potential to help scholars seeking to understand the relationship between migration-related factors and mental health.

We should also note that the pattern of results regarding the relationship between crisis migration experiences subscales and mental health outcomes emerged largely as expected. Among youth and adults, emerging evidence suggests that greater scores in terms of desperation, danger, and unplanned departure were associated with greater mental health symptom scores. These findings suggest that these subscales may serve as candidates for use independently or as part of an overarching crisis migration experience construct. We can also note that the hardship subscale was associated with symptoms of depression among adolescents but did not evidence any other significant associations with mental health symptoms among adolescents or adults. This pattern is understandable given that hardship alone is not always a consistent predictor of behavioral health among migrants (Salas-Wright & Schwartz, 2019)—additionally, this finding should be interpreted in light of the fact that most adults in our sample (93.8%) reported that their economic situation had improved since migrating to Colombia.

Limitations

Findings from the present study should be interpreted considering several limitations. First, all data analyzed were cross-sectional. As such, we cannot determine the direction of the association between crisis migration experiences and mental health outcomes. In future research, we will examine these associations using a prospective design. We also plan to examine the reliability and validity of this measure among age subgroups (e.g., young adults and middle-aged adults) and by gender, which was not feasible in this formative project. Conducting analyses across gender is especially important given the overrepresentation of women in our adult sample. It would also be beneficial to systematically examine discriminant validity, which we could not do in the present study because we did not assess variables that would be expected to be unrelated to crisis migration. Second, our findings are specific to the experiences of Venezuelan crisis migrants in two major cities in Colombia, thereby limiting our ability to generalize to the broader population of Venezuelans in Colombia. That being said, among Colombian cities, Bogotá and Medellín have the largest concentrations of Venezuelan migrants. It is nevertheless important to examine the psychometric properties of the CMES-16 with Venezuelan migrants in other countries (e.g., the United States) in future research.

Conclusions

The preliminary validation of a multidimensional measure of crisis migration experiences is important for several reasons. At the most basic level, the CMES-16 provides a way to move the field beyond the predominant yes/no conceptualization of crisis migration. As noted above, in practice, crisis migration tends to be conceptualized as a dichotomous variable—either one is a crisis migrant, or one is not. However, this conceptualization fails to capture the severity gradient that exists among crisis migrants that, as we have demonstrated here, is (a) evident in this population and (b) related to mental health burdens. Additionally, the CMES-16 captures varied and distinct aspects of crisis migration. Not only is there a severity gradient regarding crisis migration exposure, but we see clearly that crisis migration represents a multidimensional construct, including at least four subconstructs.

Another contribution of the present study is providing a theory-driven measure for a field that desperately needs one. Crisis migration is an increasingly important construct that is frequently referenced in the literature on migrant health/well-being as well as by international humanitarian organizations. The number of crisis-migrant groups worldwide is increasing, with Ukrainians and Afghanis recently added to the list of such groups along with Venezuelans, Syrians, South Sudanese, Iraqis, and Central Americans. Developing and validating the CMES-16 with Venezuelan crisis migrants opens up important avenues of research, including work that incorporates other crisis migrant populations. Finally, our measure permits us to understand the experiences of both youth and adults. Although crisis migrants are often youth, crisis migrant research frequently focuses on adults. Validating a measure for youth and adults facilitates much-needed youth/adult (including child-caregiver dyads) comparisons, and we encourage others to explore such comparisons and examine differences by gender and across the developmental life course.

Clinical Impact Statement.

Here we detail the development and validation of the 16-item Crisis Migration Experiences Scale (CMES-16) with Venezuelan migrants living in Colombia. This novel scale promises to move the field forward in understanding the experiences of individuals exposed to varying levels and types of crisis migration experiences. Developing and validating the CMES-16 with Venezuelan migrant youth and adults opens up important avenues of research, including work that incorporates other migrant populations from other countries and regions experiencing catastrophic military, social, political, economic, and/or environmental events.