An Examination of the Relationship Between Primal World Beliefs and Suicidal Thoughts and Behaviors in Three Samples of U.S. Adults

ABSTRACT

Objectives

Primal world beliefs (primals) are fundamental assumptions about the world that may serve as suicide risk factors. This study examined relationships between primals and suicidal thoughts and behaviors in three samples of adults with recent suicidal ideation (SI).

Method

Study 1 analyzed cross‐sectional data (N = 304 undergraduates; N = 390 Prolific participants) involving measures of primals, SI, wish to live (WTL), wish to die (WTD), and coping self‐efficacy. Study 2 included Prolific participants (N = 174) who completed measures of primals and SI at baseline and 1‐month follow‐up. Cross‐lagged panel analyses examined temporal relationships between primals and SI.

Results

Positive primals were associated with lower SI, greater WTL, lower WTD, and higher suicidal coping self‐efficacy. Believing the world is meaningful, needs me, and good at baseline predicted lower SI at 1‐month follow up, whereas SI at baseline did not predict primals at follow up. Primals were inconsistently related to past suicide attempts.

Conclusion

Findings support a relationship between primals and SI both cross‐sectionally and over time. Positive primals may play a protective role against SI over time and may serve as a promising treatment target for adapting existing interventions.

Suicidal thoughts and behaviors (STBs) are prevalent in the United States, with 13.1 million adults reporting serious thoughts of suicide, a plan for suicide, or attempted suicide in 2023 (Substance Abuse and Mental Health Services Administration ²⁰²⁴). Contemporary research on the phenomenology of suicidal thinking demonstrates the importance of considering multiple facets and experiences (Jobes et al. 2024). For instance, suicidal ideation (SI) can be passive (e.g., “wish to go to sleep and not wake up”) or active (e.g., “thinking about killing myself”), with research demonstrating large relationships between both dimensions and suicide death (Liu et al. 2020). Ambivalence, or the struggle between a wish to live (WTL) and a wish to die (WTD), is another aspect of SI uniquely associated with suicide risk. Research has shown that an individual's WTL is associated with decreased suicidal behaviors (Brown et al. 2005), and change patterns of WTL and WTD distinguished higher and lower suicidal risk states (Bryan et al. 2025). Further, prior suicide attempts are a strong predictor of future suicide death (Franklin et al. 2017). Although, individuals at high‐risk for suicide were less likely to experience a suicidal event in the near future if they had higher levels of suicide‐related coping self‐efficacy (i.e., belief in one's ability to utilize strategies to manage STBs; Interian et al. ²⁰²¹). These multidimensional aspects of STBs emphasize the importance of comprehensive approaches to understanding suicide risk.

Research points to the importance of understanding vulnerability to cognitive styles that enhance risk for STBs (Bryan et al. 2022). Multiple cognitive styles have been linked to SI and suicidal behavior, including hopelessness, viewing oneself as unlovable or a burden, as well as perceptions of defeat and entrapment (Cha et al. 2019). Commonly investigated cognitive styles typically focus on an individual's beliefs about themselves, their emotional experiences, and their futures. Less commonly investigated are cognitive styles about the world. Beliefs about the world, including the social and natural environment, are one aspect of Beck's cognitive triad (Beck 1979) thought to influence behavior. Just as negative beliefs about the self and future have been linked with STBs, beliefs about the world as dangerous, unpredictable, or lacking meaning may shape whether individuals perceive life as worth living. In this way, world beliefs may contribute to SI by altering perceptions of safety, reward, and purpose in one's environment. Initial support for this perspective comes from research on social axioms, which has shown that certain generalized worldviews are linked with SI and can buffer against the impact of negative life events on SI (Lam et al. 2010).

Until recently, there has not been a cohesive framework defining world beliefs or the mechanisms through which they impact well‐being (Koltko‐Rivera ²⁰⁰⁴). In response, Clifton et al. (2019) systematically defined world beliefs and created the Primals Inventory (PI‐99). According to the authors, primal world beliefs, or primals, are beliefs about overall characteristics of the world (e.g., the world is beautiful or dangerous) that are conceptualized to fall along continuums anchored by opposing ends. Factor analysis has proposed three levels, with the primary level conceptualized as Good versus Bad; the secondary level involving three factors: Safe versus Unsafe, Enticing versus Dull, and Alive versus Mechanistic; and the tertiary level involving 22 subfactors, including Pleasurable versus Miserable (under Safe), Meaningful versus Meaningless (under Enticing), and Needs Me versus Doesn't Need Me (under Alive). Primals have been shown to meet three benchmarks: they are stable across a 19‐month period, vary considerably from person to person, and are highly predictive of clinical, demographic, wellbeing, religious, and political variables (Clifton et al. 2019).

To date, the only investigation into the relationship between primals and STBs utilized a cross‐sectional design and a single item from the Suicidal Behaviors Questionnaire‐Revised (Osman et al. 2001), specifically asking non‐treatment seeking community and college student samples “have you ever thought about or attempted to kill yourself”. Clifton and Meindl (2022) found that maladaptive primals (e.g., viewing the world as bad) were associated with an increased history of STBs. Specifically, in the community sample, believing the world is dangerous, declining, or fragile was each correlated with a history of STBs (rs = 0.25 to 0.35, ps < 0.05). In the student sample, 14 different primals significantly correlated with a history of STBs, with the strongest correlations involving beliefs in the world as bad, dangerous, miserable, dull, or meaningless (rs = 0.26 to 0.37, ps < 0.0001). Given early evidence suggesting primals may predict individual factors such as affect, behavior, health, and wellbeing, it is possible that primals could play a role in the development and maintenance of STBs. Examining the clinical applications of primals to STBs and the longitudinal nature of this association is worthwhile to inform treatment development with the ultimate goal of improving outcomes and preventing suicide.

1. Current Study

Two studies were conducted to examine the degree to which primals are related to STBs. Study 1 involved two cross‐sectional samples of adults and examined associations of primary and secondary primals with SI severity, ambivalence, past suicidal behavior, and coping self‐efficacy. Study 2 involved a longitudinal design to examine prospective associations of primary and select tertiary primals with SI severity over 1 month. Given the preliminary work of Clifton and Meindl (2022) demonstrating relationships between several primals with STBs (including Good, Safe, Enticing, Pleasurable, and Meaningful), it was hypothesized that primals would be significantly associated with suicide‐related outcomes such that more negative beliefs about the world would be related to greater SI severity, WTD, prior suicidal behavior, and lower coping self‐efficacy. It was also hypothesized there would be a temporal relationship such that primals would precede SI. However, given the limited research to date, we did not hypothesize a priori which specific primals would be most strongly associated with STBs.

2. Study 1 Methods

2.1. Participants

2.1.1. Sample 1A

Participants included 257 undergraduate students recruited from a large, public southern university's SONA system who correctly answered all three validation checks on the study survey. This sample was largely female, heterosexual, and Caucasian/White, with a mean age of 19.3 years (see Table 1). Participants were eligible if they endorsed SI in the past 2 weeks as measured by a non‐zero response on the first item of the Depressive Symptom Inventory‐Suicidality Subscale (DSI‐SS; Joiner et al. ²⁰⁰²).

TABLE 1.

Demographic and clinical characteristics of study 1 participants.

Demographic	No. (%) [except as noted]
	Sample 1A (N = 257)	Sample 1B (N = 348)
Age in years, mean (SD) [range]	19.3 (1.4) [18–30] a	34.6 (10.4) [18–69]
Sex assigned at birth
Female	205 (79.8%)	192 (55.2%)
Male	52 (20.2%)	155 (44.5%)
Other not listed	0 (0.0%)	1 (0.3%)
Gender identity
Man	50 (19.5%)	147 (42.2%)
Non‐binary	8 (3.1%)	27 (7.8%)
Transgender	3 (1.2%)	4 (1.1%)
Woman	196 (76.3%)	168 (48.3%)
Other not listed	0 (0.0%)	2 (0.6%)
Sexual orientation
Bisexual	61 (23.7%)	99 (28.4%)
Gay	6 (2.3%)	7 (2.0%)
Lesbian	7 (2.7%)	14 (4.0%)
Straight	153 (59.5%)	188 (54.0%)
Not sure	17 (6.6%)	4 (1.1%)
Other not listed	13 (5.1%)	35 (10.1%)
Did not respond	0 (0.0%)	1 (0.3%)
Racial identity
African‐American or Black	54 (21.0%)	35 (10.1%)
Asian or Asian‐American	27 (10.5%)	32 (9.2%)
Biracial	23 (8.9%)	18 (5.2%)
Caucasian or White	137 (53.3%)	233 (67.0%)
Hispanic or Latino	15 (5.8%)	25 (7.2%)
Other not listed	1 (0.4%)	2 (0.6%)
Past‐week suicidal ideation
Active	113 (44%)	271 (78%)
Passive	112 (43.5%)	77 (22%)
Absent	32 (12.5%)	0 (0%)
Suicide index score group
Ambivalent	143 (56%)	220 (63%)
Life‐oriented	130 (43%)	80 (23%)
Death‐oriented	4 (2%)	48 (14%)
Suicide attempt history b
None	173 (77%)	225 (65%)
One	33 (13%)	52 (15%)
Two or more	52 (15%)	69 (20%)

^a A single participant in sample 1A did not provide age data (N = 256).

^b Two participants in sample 1B did not report on their suicide attempt history (N = 346).

2.1.2. Sample 1B

Participants included 348 adults residing in the United States recruited via Prolific, an online research platform, who endorsed past‐week SI as measured by a single item on the Self‐injurious Thoughts and Behaviors Interview—Revised (SITBI‐R; Nock et al. ²⁰⁰⁷) and correctly answered all three validation checks in the study survey. This sample was also largely Caucasian/White and heterosexual, slightly skewed toward females, with an average age of 34.6 years (see Table 1).

2.2. Measures

2.2.1. Primals Inventory 18‐Item (PI‐18)

The PI‐18 (Clifton and Yaden 2021) is a reliable, 18‐item self‐report measure that indexes the world as Good vs. Bad (15 items), Safe vs. Dangerous (6 items), Enticing vs. Dull (7 items), and Alive vs. Mechanistic (5 items). It demonstrated good internal consistency for the overall primal, Good, among both samples (α = 0.85). Each subscale also demonstrated adequate‐to‐good reliability across samples, respectively (Safe: α = 0.74, α = 0.86; Enticing: α = 0.78, α = 0.85; Alive: α = 0.80, α = 0.88).

2.2.2. Depressive Symptom Index‐Suicidality Subscale (DSI‐SS)

The DSI‐SS (Joiner et al. 2002) is a 4‐item self‐report measure of SI over the past 2 weeks with good psychometric properties. Total scores on the DSI‐SS range from 0 to 12 and higher scores reflect greater SI severity. In the present study, the total score demonstrated good reliability in sample 1A (α = 0.85) and adequate reliability in sample 1B (α = 0.74).

2.2.3. Self‐Injurious Thoughts and Behaviors Interview (SITBI)‐Cognitions List

Two items were administered from the SITBI, a structured interview (Nock et al. 2007). First, participants were asked: “In the past week, have you ever had any of the following thoughts for more than a few minutes? Choose all that you have experienced.” Past‐week passive or active SI was coded as “no past‐week SI” if no thoughts were endorsed, “passive SI” if only passive thoughts were endorsed (“I wish I could disappear and not exist,” “I wish I was never born,” “My life is not worth living,” “I wish I could go to sleep and never wake up”, or “I wish I were dead”), or “active SI” if any active thoughts were endorsed (“Maybe I should kill myself,” “I should kill myself,” or “I am going to kill myself”). Participants were also asked: “Have you ever made an actual attempt to kill yourself” with response options of “No” and “Yes (how many times)”. Based on these responses, history of suicide attempts was coded as “no suicide attempt history,” “single suicide attempt,” or “multiple suicide attempts.”

2.2.4. Suicide Status Form (SSF)

Items assessing participants' WTL and WTD were excerpted from the SSF utilized in the Collaborative Assessment and Management of Suicidality (Jobes 2012). Each item asks participants to rate their current WTL/WTD on a scale from 0 (“not at all”) to 8 (“very much”). Ratings were converted into a three‐point scale (0–2 = 1; 3–5 = 2; 6–8 = 3), and then the converted WTD was subtracted from the converted WTL to generate a suicide index score (SIS; Lento et al. ²⁰¹³). The SIS distinguishes individuals who are ambivalent toward suicide (score = 0) from those who are more resolutely life‐ (score = 1 or 2) or death‐oriented (score = −1 or −2).

2.2.5. Suicide‐Related Coping Scale (SRCS)

The SRCS (Stanley et al. 2017) is a 17‐item measure originally developed to assess an individual's perceived ability to divert suicidal behavior when experiencing SI using the safety planning intervention. Items assess participant's ability to use specific strategies on their safety plan (e.g., contact a professional agency for support) as well as their global beliefs about keeping themselves safe (e.g., “I am at the mercy of my suicidal thoughts”). Only the 5 items related to global coping self‐efficacy were administered to reduce participant demand and since safety planning was not provided. The 5‐item SRCS demonstrated adequate reliability in sample 1A (α = 0.75) and excellent reliability in sample 1B (α = 0.91).

2.3. Procedures

Across both samples, participants completed self‐report questionnaires (including demographics) via an online survey (median time to completion was 21.33 min), were compensated with course credit (sample 1A) or monetarily with $8 (sample 1B), and provided with a list of mental health resources.

2.4. Ethics Approval and Informed Consent

Procedures were approved by the Institutional Review Board at Louisiana State University (protocols IRBAM‐23‐0391 and IRBAM‐24‐0001). All participants provided informed consent prior to participation.

2.5. Statistical Analyses

Given differences in recruitment and demographic characteristics between samples 1A and 1B, analyses were conducted separately for each sample. First, descriptives of all measures were calculated and variables were examined for normality, which were satisfied (see Table 2). Next, we identified potential covariates by examining associations between demographic and study variables. For sample 1A, DSI‐SS and primals significantly differed by gender and sexual orientation, such that those identifying as transgender/nonbinary (DSI‐SS: p = 0.013, η ² = 0.034; Alive: p = 0.016, η ² = 0.032) or non‐heterosexual (DSI‐SS: p < 0.001, d = 0.637; Safe: p = 0.008, d = 0.337; Alive: p = 0.003, d = 0.386) endorsed greater SI severity and lower belief that the world is Safe or Alive. For sample 1B, DSI‐SS scores were negatively correlated with age (r = −0.119, p = 0.027) and positively correlated with Safe (r = 0.137, p = 0.010), such that older participants endorsed lower SI severity and greater belief that the world is Safe. Additionally for sample 1B, all primals differed by race/ethnicity (ps < 0.01), such that nonwhite participants endorsed significantly less positive primals. Thus, analyses were conducted with and without the following covariates: gender (sample 1A), sexual orientation (sample 1A), age (sample 1B), and race/ethnicity (sample 1B). With few exceptions, results were consistent with and without covariates and are reported without covariates for simplicity, except when noted.

TABLE 2.

Descriptives of study 1 outcome measures by sample.

Study measure	Sample 1A (N = 257)					Sample 1B (N = 348)
	Mean	SD	Range	Skew (SE)	Kurtosis (SE)	Mean	SD	Range	Skew (SE)	Kurtosis (SE)
DSI‐SS total	2.90	2.28	0–10	0.58 (0.15)	−0.21 (0.30)	4.97	1.97	0–12	0.56 (0.13)	1.22 (0.26)
Wish to live	5.59	1.63	0–8	−0.31 (0.15)	−0.24 (0.30)	4.16	2.00	0–8	0.02 (0.13)	−0.68 (0.26)
Wish to die	2.37	1.85	0–8	0.58 (0.15)	−0.22 (0.30)	3.57	2.18	0–8	0.10 (0.13)	−0.89 (0.26)
PI‐18
Good	2.54	0.68	0.73–4.20	−0.08 (0.15)	−0.42 (0.30)	2.39	0.72	0.00–4.13	−0.36 (0.13)	0.19 (0.26)
Safe	1.81	0.76	0.00–3.83	0.07 (0.15)	−0.35 (0.30)	1.87	0.87	0.00–4.33	0.08 (0.13)	−0.16 (0.26)
Enticing	3.16	0.80	1.14–4.86	−0.20 (0.15)	−0.52 (0.30)	3.09	0.88	0.00–4.86	−0.54 (0.13)	0.08 (0.26)
Alive	2.75	1.06	0.00–5.00	−0.08 (0.15)	−0.48 (0.30)	1.74	1.15	0.00–4.60	0.38 (0.13)	−0.79 (0.26)
SRCS a	15.29	3.61	4–20	−0.74 (0.15)	0.08 (0.30)	12.39	5.12	0–20	−0.50 (0.13)	−0.57 (0.26)

Abbreviations: DSI‐SS, Depressive Symptom Index‐Suicidality Subscale, PI‐18, 18‐Item Primals Inventory, SRCS, Suicide‐Related Coping Scale (5‐item version).

^a Two participants in sample 1A did not provide SRCS data (N = 255); Six participants in sample 1B did not provide SRCS data (N = 342).

To assess the relationship between primals and suicidal thinking, bivariate correlations were conducted between the PI‐18 subscales with DSI‐SS, WTL, and WTD ratings. Additionally, eight one‐way analyses of variance (ANOVAs) were conducted to examine differences in each PI‐18 subscale across passive/active groups as well as SIS groups. To assess how primals are associated with past suicidal behavior, four ANOVAs compared PI‐18 subscales by attempt group. To assess how primals are associated with coping self‐efficacy, bivariate correlations were conducted between the PI‐18 subscale and SRCS scores. Finally, regression models with all three secondary primals predicting DSI‐SS, WTL, WTD, and SCRCS scores were also conducted to explore whether any one primal uniquely predicted suicide‐related outcomes. With 28 tests (four primals × eight dependent variables) per sample excluding post hoc comparisons and exploratory analyses, significance was adjusted using False Discovery Rate (FDR) for each sample.

3. Study 1 Results

Demographic and clinical characteristics for samples 1A and 1B are displayed in Table 1. Descriptive statistics for study 1 measures are reported by sample in Table 2. Both samples endorsed elevated SI severity as indicated by average DSI‐SS scores greater than 2 (Von Glischinski et al. 2016). For sample 1A, there was an equal distribution of passive and active past‐week SI; for sample 1B, most endorsed active SI. The majority of both samples endorsed ambivalence toward suicide, and most participants had never made a suicide attempt.

3.1. Primals and SI Severity and Ambivalence

Bivariate correlations of primals subscales with DSI‐SS, WTL, and WTD are presented in Table 3. Across both samples, each secondary primal as well as the overall primal Good were significantly, negatively correlated with SI severity and WTD and positively with WTL, except for a non‐significant association between Alive and DSI‐SS for sample 1B. Effect sizes were small‐to‐moderate, with the strongest correlations involving Good. When all three secondary primals were included in a single regression model, Enticing emerged as a consistent, unique predictor across both samples, while Safe also emerged as a significant predictor for sample 1B only (see Tables S1 and S2).

TABLE 3.

Bivariate correlations by study 1 sample.

Study measure	Wish to live	Wish to die	DSI‐SS total	PI‐18 good	PI‐18 safe	PI‐18 enticing	PI‐18 alive	SRCS
Wish to live	—	−0.70***	−0.44***	0.44***	0.24***	0.44***	0.44***	0.37***
Wish to die	−0.70***	—	0.51***	−0.42***	−0.29***	−0.40***	−0.33***	−0.31***
DSI‐SS total	−0.47***	0.56***	—	−0.32***	−0.26***	−0.26***	−0.24***	−0.29***
PI‐18 good	0.50***	−0.42***	−0.28***	—	0.81***	0.87***	0.68***	0.35***
PI‐18 safe	0.37***	−0.40***	−0.29***	0.80***	—	0.46***	0.46***	0.18**
PI‐18 enticing	0.41***	−0.31***	−0.15**	0.84***	0.42***	—	0.49***	0.37***
PI‐18 alive	0.27***	−0.13*	−0.11	0.49***	0.26***	0.27***	—	0.29***
SRCS a	0.47***	−0.50***	−0.45***	0.34***	0.25***	0.30***	0.11	—

Note: Correlations on the top half reflect sample 1A (N = 257), while correlations on the bottom half reflect sample 1B (N = 348). Significant correlations after FDR adjustment for multiple comparisons in bold: *p < 0.05, **p < 0.01, ***p < 0.001.

Abbreviations: DSI‐SS, Depressive Symptom Index‐Suicidality Subscale, PI‐18, 18‐Item Primals Inventory, SRCS, Suicide‐Related Coping Scale (5‐item version).

^a Two participants in sample 1A did not provide SRCS data (N = 255); Six participants in sample 1B did not provide SRCS data (N = 342).

For sample 1A, ANOVAs were conducted to examine whether primals differed when past‐week SI was categorized as passive, active, or absent. For each ANOVA, the assumption of homogeneity of variances was met with non‐significant Levene's tests (ps > 0.05). Results indicated significant effects for past‐week SI group on each primal (see Table 4). Post hoc comparisons using Bonferroni tests revealed that participants reporting an absence of past‐week SI endorsed significantly higher primals compared to both passive and active SI groups (p < 0.05). Passive and active SI groups did not significantly differ on any primal except Safe, such that those with active SI endorsed believing the world is significantly less Safe (p = 0.033).

TABLE 4.

ANOVA and T‐test results by study 1 sample.

Sample	Primal	Source	Mean	SD	F‐value or t‐value	df (between groups, within groups)	p	η 2 or d
1A	Good	Active past‐week SI	2.4	0.58	9.868	2254	< 0.001	0.072
		Passive past‐week SI	2.54	0.74
		No past‐week SI	3.0	0.63
1A	Safe	Active past‐week SI	1.63	0.68	5.165	2254	0.006	0.039
		Passive past‐week SI	1.88	0.80
		No Past‐week SI	2.24	0.69
1A	Enticing	Active past‐week SI	3.1	0.72	9.501	2254	< 0.001	0.070
		Passive past‐week SI	3.13	0.86
		No Past‐week SI	3.6	0.75
1A	Alive	Active past‐week SI	2.6	1.04	10.905	2254	< 0.001	0.079
		Passive past‐week SI	2.73	1.02
		No Past‐week SI	3.5	1.0
1B	Good	Active past‐week SI	2.35	0.75	0.734	346	0.464	0.095
		Passive past‐week SI	2.53	0.57
1B	Safe	Active past‐week SI	1.82	0.90	0.782	346	0.435	0.101
		Passive past‐week SI	2.07	0.73
1B	Enticing	Active past‐week SI	3.1	0.91	2.263	346	0.024	0.292
		Passive past‐week SI	3.2	0.76
1B	Alive	Active past‐week SI	1.72	1.14	2.885	346	0.023	0.254
		Passive past‐week SI	1.83	1.2
1A	Good	Death oriented	2.0	0.59	24.005	2254	< 0.001	0.159
		Ambivalent	2.32	0.62
		Life oriented	2.9	0.65
1A	Safe	Death oriented	1.4	0.50	20.241	2254	< 0.001	0.137
		Ambivalent	2.03	0.71
		Life oriented	2.03	0.77
1A	Enticing	Death oriented	2.64	0.91	8.546	2254	< 0.001	0.063
		Ambivalent	2.91	0.75
		Life oriented	3.5	0.74
1A	Alive	Death oriented	1.7	0.87	24.118	2254	< 0.001	0.160
		Ambivalent	2.41	0.97
		Life oriented	3.23	1.0
1B	Good	Death oriented	1.74	0.74	7.568	2345	< 0.001	0.042
		Ambivalent	2.4	0.65
		Life oriented	2.81	0.60
1B	Safe	Death oriented	1.17	0.80	23.399	2345	< 0.001	0.119
		Ambivalent	1.9	0.83
		Life oriented	2.3	0.77
1B	Enticing	Death oriented	2.54	1.1	26.801	2345	< 0.001	0.134
		Ambivalent	3.04	0.79
		Life oriented	3.55	0.73
1B	Alive	Death oriented	1.25	1.01	41.550	2345	< 0.001	0.194
		Ambivalent	1.74	1.14
		Life oriented	2.05	1.18
1A	Good	No suicide attempts	2.6	0.71	1.655	2254	0.193	0.013
		One suicide attempt	2.7	0.61
		Multiple suicide attempts	2.3	0.60
1A	Safe	No suicide attempts	1.9	0.77	4.245	2254	0.015	0.032
		One suicide attempt	1.9	0.76
		Multiple suicide attempts	1.6	0.70
1A	Enticing	No suicide attempts	3.2	0.80	3.017	2254	0.051	0.023
		One suicide attempt	3.4	0.73
		Multiple suicide attempts	2.9	0.81
1A	Alive	No suicide attempts	2.8	1.1	4.567	2254	0.011	0.035
		One suicide attempt	3.02	0.92
		Multiple suicide attempts	2.6	1.03
1B	Good	No suicide attempts	2.5	0.67	0.656	2343	0.520	0.004
		One suicide attempt	2.25	0.80
		Multiple suicide attempts	2.25	0.77
1B	Safe	No suicide attempts	1.94	0.83	4.195	2343	0.016	0.061
		One suicide attempt	1.89	0.92
		Multiple suicide attempts	1.65	0.95
1B	Enticing	No suicide attempts	3.2	0.81	2.856	2343	0.059	0.016
		One suicide attempt	2.84	1.1
		Multiple suicide attempts	2.97	0.92
1B	Alive	No suicide attempts	1.74	1.18	3.566	2343	0.029	0.020
		One suicide attempt	1.6	1.17
		Multiple suicide attempts	1.84	1.1

For sample 1B, t‐tests were conducted given that no participant endorsed an absence of past‐week SI. The assumption of homogeneity of variances was met with non‐significant Levene's tests (ps > 0.05) for all primals except Good (F = 7.884, p = 0.005). There were no significant effects of SI group on Alive or Enticing, but there was a significant effect on Safe and Good (see Table 4). The passive SI group reported significantly higher belief in the world as Safe and Good compared to the active SI group. However, these effects were no longer significant when covariates were included (Safe: F(1, 344) = 3.880, p = 0.050, η ² = 0.011; Good: F(1, 344) = 3.616, p = 0.058, η ² = 0.010).

ANOVAs were also conducted to examine whether primals differed by SIS group. Assumptions of homogeneity of variances were met with non‐significant Levene's tests (ps > 0.05) except for sample 1B Good (Levene's test: F(2, 345) = 3.050, p = 0.049). Results indicated a significant effect of SIS group on each primal across samples (see Table 4). Post hoc comparisons for sample 1A using Bonferroni tests revealed a significant difference between the ambivalent and life‐oriented groups across all primals (p < 0.001) and between the life‐oriented and death‐oriented groups on Alive (p = 0.007) and Good (p = 0.025) but no differences between the ambivalent and death‐oriented groups. Post hoc comparisons for sample 1B using Games‐Howell tests on the primal Good revealed significant differences (ps < 0.001) between all SIS groups. Significant differences (ps < 0.001) were also found using Bonferroni tests for all secondary primals with the exception of a non‐significant comparison between life‐oriented and ambivalent on Alive (p = 0.102). Overall, those with a life‐orientation endorsed more adaptive primals, and those with a death‐orientation endorsed more maladaptive primals.

3.2. Primals and Past Suicidal Behavior

ANOVAs were conducted to examine whether primals differed by suicide attempt history. Assumptions of homogeneity of variances were met with non‐significant Levene's tests (ps > 0.05) except for sample 1B Enticing (Levene's test: F(2, 343) = 5.075, p = 0.007). Results indicated a significant effect of suicide attempt history on Enticing and Good but not Safe or Alive (see Table 4). However, when controlling for age and race/ethnicity for sample 1B, the effect of suicide attempt history on Good was no longer significant after adjusting for multiple comparisons (F(2, 241) = 3.262, p = 0.039, η ² = 0.019). Post hoc comparisons for sample 1A using Bonferroni test revealed significant differences between participants who had made multiple attempts compared to those with no prior attempts on Enticing (p = 0.042) and Good (p = 0.026) and between one versus multiple attempts on Enticing (p = 0.026) and Good (p = 0.024). Based on sample 1A, participants who reported multiple attempts endorsed less adaptive primals. However, none of the post hoc comparisons for sample 1B were significant. Primals appear to be inconsistently related to past suicidal behavior.

3.3. Primals and Coping Self‐Efficacy

As seen in Table 3, all four primals were each significantly, positively correlated with SRCS scores among both samples, with the exception of sample 1B Alive. When all three secondary primals were included in a single regression model, Enticing emerged as a unique predictor of SCRS scores across samples (see Tables S1 and S2). Overall, viewing the world in a positive light may be associated with one's perceived ability to cope with SI.

4. Study 2 Methods

While we found initial support for associations between primals and STBs based on two cross‐sectional samples, we were limited in the ability to examine primals across the three levels (Clifton et al. 2019) because of our use of the PI‐18. Thus, a second study was conducted using the PI‐99 to examine associations between primals at the tertiary level with SI over a 1‐month period to begin to establish a temporal relationship.

4.1. Participants

Study 2 participants included 174 adults living in the United States who were recruited through Prolific. Eligibility criteria included being 18 years or older and endorsing past‐month SI based on the Columbia Suicide Severity Rating Scale screener (C‐SSRS; Posner et al. ²⁰¹¹). Seventy‐five percent (n = 131) also completed the follow‐up survey (T2), which was administered 1 month after the baseline survey (mean days between surveys = 28.53, SD = 0.87). The study 2 sample was on average 38.73 years old, largely Caucasian/White, heterosexual, and slightly skewed toward females. A third of the sample reported a lifetime suicide attempt (see Table 5).

TABLE 5.

Demographic characteristics for sample 2 at baseline (N = 174).

Demographic	No. (%) [except as noted]
Age in years, mean (SD) [range]	38.79 (13.16) [18–83]
Sex assigned at birth
Female	96 (55.2%)
Male	78 (44.8%)
Gender identity
Agender	2 (1.2%)
Androgynous	2 (1.2%)
Man	72 (41.4%)
Non‐binary	2 (1.2%)
Third gender	1 (0.6%)
Transgender	2 (1.2%)
Trans masculine	1 (0.6%)
Woman	91 (52.3%)
Sexual orientation
Asexual	7 (4.0%)
Bisexual	30 (17.2%)
Gay	7 (4.0%)
No label	3 (1.72%)
Pansexual	4 (2.3%)
Queer	4 (2.3%)
Straight	153 (59.5%)
Other	1 (0.58%)
Decline to answer	1 (0.58%)
Racial identity
African‐American or Black	33 (19.0%)
Asian (including Asian Desi)	8 (4.6%)
American Indian/Alaska Native	2 (1.15%)
Caucasian or White	119 (68.4%)
Native Hawaiian	9 (5.2%)
Other	1 (0.6%)
Prefer not to answer	2 (1.2%)
Employment
Employed full‐time	87 (50.0%)
Employed part‐time	27 (15.5%)
Self‐employed	19 (10.9%)
Unemployed	21 (12.1%)
Retired	6 (3.5%)
Student	8 (4.6%)
Under employed (wage below industry standard)	1 (0.6%)
Other	4 (2.3%)
Prefer not to answer	1 (0.6%)
Suicide attempt history
None	95 (55.2%)
Previous attempt	63 (36.2%)
Prefer not to answer	16 (8.6%)

4.2. Measures

4.2.1. PI‐99

The PI‐99 is a 99‐item self‐report questionnaire that assesses all 26 primal subscales (Clifton et al. 2019). The current study focused on the following subscales: Good (71 items), Pleasurable (5 items), Meaningful (4 items), and Needs Me (4 items) (see Figures S2 and S3 for exploratory analyses involving secondary primals). These tertiary subscales were identified as the most correlated with lifetime STBs (Clifton et al. 2019). Internal reliability on the PI‐99 for the current study at T1 and T2, respectively, was excellent for Good (α = 0.97, α = 0.97) and Needs Me (α = 0.93, α = 0.92), and good for Meaningful (α = 0.83, α = 0.87) and Pleasurable (α = 0.86, α = 0.87).

4.2.2. DSI‐SS

The DSI‐SS (Joiner et al. 2002) was used to assess the severity of SI over the past 2 weeks. In the current sample, the internal reliability was excellent at T1 (α = 0.91) and good at T2 (α = 0.84).

4.2.3. Suicide Attempt History

History of suicide attempts was measured using one question from the SITBI (Nock et al. 2007), which was coded as binary (no suicide attempt history versus one or more prior suicide attempts) for analyses.

4.3. Procedures

Participants provided informed consent before completing the online Time 1 survey (including a demographics questionnaire, PI‐99, DSI‐SS, and suicide attempt history), and then were eligible through Prolific to complete the Time 2 survey (including the PI‐99 and DSI‐SS) 1 month later. The median time to complete the surveys was 25.48 min (T1) and 23.38 min (T2). Participants were compensated monetarily through the Prolific platform ($4 for T1, $3.50 for T2) as long as 75% of each survey was completed and no more than one attention check question was missed. After survey completion, participants were provided with a list of mental health resources.

4.4. Ethics Approval and Informed Consent

All study procedures were approved by the Institutional Review Board at the Catholic University of America (protocol 22‐0019). All participants provided informed consent prior to participation.

4.5. Statistical Analysis

Descriptive analyses were conducted for demographic characteristics and study variables, and assumptions of normality were assessed and met according to the criteria of skewness ≦ ±2 and kurtosis ≦ ±7 (Curran et al. 1996; Hair et al. ²⁰¹⁰). Due to the 25% drop from T1 to T2, an attrition analysis was completed. No significant differences were found based on demographic or study variables between those who did and did not complete the T2 survey. Potential covariates were identified by examining relationships between demographic and study variables at both time points. Differences were found by sex assigned at birth for scores on Pleasurable (T1: t(170) = −2.12, p = 0.035; T2: t(125) = −2.75, p = 0.007), Good (T1: t(156) = −2.58, p = 0.011; T2: t(115) = −2.68, p = 0.008), and Meaningful (T2: t(127) = −2.32, p = 0.022), such that females reported lower scores than males. Additionally, T2 scores on DSI‐SS, Meaningful and Needs Me significantly differed by sexual orientation, such that heterosexual participants endorsed lower SI (t(129) = −2.08, p = 0.039) and lower belief in the world as Meaningful (t(127) = 2.17, p = 0.032), but higher on Needs Me (t(124) = 2.48, p = 0.014) compared to non‐heterosexuals. Thus, analyses were conducted with and without sex and sexual orientation included as covariates. Analyses remained consistent with these covariates with one exception, and thus are presented without covariates except when noted.

Cross‐lag panel analyses (CLPA) were conducted to explore relationships between the primals Good, Meaningful, Needs Me, and Pleasurable with DSI‐SS at T1 and T2. Autocorrelations, cross‐lag correlations, and cross‐sectional correlations were specified to examine the stability of each variable over time, directionality, as well as concurrent associations, respectively. Separate CLPAs were run adding suicide attempt history as a moderator of the cross‐lag path from primals at T1 to DSI‐SS at T2. Exploratory analyses also examined secondary primals and multiple primals simultaneously (see Figures S2 and S3). Since attrition analyses did not find any significant differences between completers and dropouts, the full information maximum‐likelihood method (FIML) for dealing with missing data was used to retain the full sample (N = 174) for all analyses. Each CLPA was saturated (χ ²(0) = 0.000, p = 1.000; CFI = 1.000, TLI = 1.000), meaning it perfectly reproduced the observed data. As a result, fit indices such as RMSEA and SRMR were not informative (RMSEA = 0.000, SRMR = 0.000), and parameter estimates rather than model fit were interpreted. With 28 tests (four primals × six paths, plus four interactions), excluding exploratory analyses, significance for parameter estimates was adjusted using False Discovery Rate (FDR).

5. Study 2 Results

Descriptive statistics for study measures at T1 and T2 are presented in Table 6. On average, the sample endorsed elevated SI at both time points, with SI severity decreasing over time. Figure 1 displays the results of the four CLPAs.

TABLE 6.

DSI‐SS and PI‐99 subscale scores at baseline and follow‐up for study 2.

Study measure	Baseline (T1) (N = 174)					Follow up (T2) (N = 131)
	Mean	SD	Range	Skew (SE)	Kurtosis (SE)	Mean	SD	Range	Skew (SE)	Kurtosis (SE)
DSI‐SS total	4.00	3.72	0–12	0.27 (0.18)	0.25 (0.37)	3.00	3.27	0–12	0.45 (0.21)	0.21 (0.42)
PI‐99
Needs Me	1.98	1.39	0–5	0.39 (0.19)	0.19 (0.37)	1.97	1.41	0–5	0.22 (0.21)	−0.99 (0.42)
Meaningful	2.47	1.20	0–5	0.15 (0.19)	−0.72 (0.37)	2.54	1.29	0–5	0.08 (0.22)	−0.80 (0.43)
Pleasurable	2.05	1.05	0–4.6	0.25 (0.19)	−0.54 (0.37)	2.14	1.10	0–4.60	0.17 (0.21)	−0.55 (0.42)
Good	2.25	0.80	0.3–4.2	0.02 (0.19)	−0.47 (0.38)	2.33	0.79	0.31–4.10	0.01 (0.22)	−0.30 (0.44)

Abbreviations: DSI‐SS, Depressive Symptom Index‐Suicidality Subscale, PI‐99, 99‐Item Primals Inventory.

FIGURE 1.

Cross‐lag panel analyses of primals at baseline (T1) predicting suicidal ideation at follow‐up (T2) 1 month apart among sample 2 (N = 174). DSI‐SS = Depressive Symptom Index—Suicidality Subscale. Bolded paths are significant. *p < 0.05, **p < 0.01, ***p < 0.001.

As seen in Figure 1, the autoregressive paths were significant for each primal as well as DSI‐SS, indicating stability over time. Notably, coefficients for primals were numerically larger than for the DSI‐SS (e.g., Meaningful T1 → T2: β = 0.78; DSI‐SS T1 → T2: β = 0.65), suggesting greater stability in primals than SI over 1 month. Regarding cross‐lagged effects, Meaningful, Needs Me, and Good (but not Pleasurable) at T1 predicted lower DSI‐SS scores at T2, whereas DSI‐SS at T1 did not significantly predict primals at T2, providing initial support for the directionality of this relationship. The cross‐sectional paths also supported negative correlations between all four primals and DSI‐SS at T1, but only Meaningful was significantly correlated with DSI‐SS at T2 and this was no longer significant when covariates (sex, sexual orientation) were included in the model after adjusting for multiple comparisons (r = −0.250, p = 0.036). These results suggest that believing the world is meaningful, needs me, and good may be a protective factor contributing to lower SI.

Adding attempt history to each CLPA as a moderator of the cross‐lag path from primals at T1 to SI at T2 did not result in any significant interaction (Meaningful: β = −0.091, p = 0.148; Needs Me: β = −0.075, p = 0.248; Pleasurable: β = −0.088, p = 0.168; Good: β = −0.084, p = 0.190). Thus, prior suicidal behavior did not change the relationship of primals as a predictor of SI.

In an exploratory analysis that included all three tertiary primals in a single CLPA, only Needs Me at T1 remained a significant predictor of DSI‐SS at T2 when Pleasurable and Meaningful were also in the model, but this cross‐lag path was no longer significant when covariates were included (see Figure S1).

6. Discussion

Beliefs about the nature of the world as fundamentally good (vs. bad) have been linked to mental health outcomes and may influence how people interpret their own distress and evaluate life's worth, thus shaping suicide risk (Clifton and Meindl 2022). The present study involved the first comprehensive examination of the role of primal world beliefs in STBs. Specifically, we examined associations between primals and multiple facets of STBs, including SI severity, WTL, WTD, suicidal ambivalence, passive versus active SI, coping self‐efficacy, and suicide attempt history among three samples of adults with recent SI. Additionally, we investigated the overall primal, Good, along with secondary (study 1) and specific tertiary (study 2) primals to guide the appropriate level of assessment and inform potential targets for intervention.

Results from two cross‐sectional samples showed that primals were broadly associated with facets of SI, though less consistently with past suicidal behavior. Using the PI‐18 (Clifton and Yaden 2021), the primary primal, Good, consistently demonstrated the largest associations across suicide‐related outcomes (SI severity, WTL, WTD, SRCS) across samples. For college students, this association seemed to be driven by Enticing, while among general adults it was driven by both Enticing and Safe. Greater opportunities for experiencing traumatic or stressful events as well as greater financial, career, or caregiver responsibilities may have contributed to the emergence of world beliefs surrounding safety as an additional schema relevant to SI among general adults. Whereas, safety‐related beliefs may not be as central to college students' SI when living in a relatively protected, structured college setting. Notably, Enticing (and Good) further distinguished between suicide attempt history as expected among college students, such that those with more suicide attempts endorsed believing the world is dull and bad. Thus, holding the belief that the world is rich with beauty, intrigue, and purpose may be especially protective against suicide.

Findings regarding the relationship between primals with passive versus active SI highlight the importance of a different secondary primal: Safe was the only primal that distinguished between passive versus active SI. Perceiving the world as harmful, unstable, and unjust may contribute to the emergence of thoughts about killing oneself rather than merely wishing to be dead. It is possible that for individuals who perceive the world as unsafe, suicide may be construed as a means of avoiding or preempting anticipated environmental threats. In this way, suicide may be viewed as a form of escape from external harm or as an attempt to exert control in an environment perceived as dangerous, consistent with the escape theory of suicide (Dean et al. 1996).

Primals also differed by SIS such that those with a life‐orientation endorsed believing the world is generally good, while those with a death‐orientation endorsed believing the world is generally bad. Interestingly, those with suicidal ambivalence displayed moderate primals among the Prolific sample but did not significantly differ from those with a death‐orientation among the student sample. While a death‐orientation was found to increase risk for death by suicide (Stanley and Brown 2012), associations of ambivalence with STBs have been more mixed, with some studies finding ambivalence comparable to death‐orientation and others showing lower risk (Teismann et al. 2024).

Importantly, primals assessed at baseline significantly predicted SI 1 month later, suggesting these beliefs may serve as a stable cognitive vulnerability for future suicidal thinking. Specifically, this temporal relationship was evident for the primary primal, Good, along with tertiary primals Meaningful (under Enticing) and Needs Me (under Alive). Thus, across both the cross‐sectional and longitudinal analyses, Enticing‐related primals emerged as key predictors of SI. Prior research has similarly found the presence of meaning in life inspires hope and suppresses SI (Dogra et al. 2011; Kleiman and Beaver ²⁰¹³). It is worth noting that in our exploratory analysis involving all three tertiary primals, Needs Me was the only unique predictor (without covariates), suggesting additional attention may be warranted to studying people's perceptions of their role in their environment. While Needs Me less clearly distinguishes world beliefs from self‐focused beliefs, it captures self‐perceptions within a larger context which can be clinically important.

While these analyses don't establish causality, they do support a temporal relationship with beliefs preceding SI, which is important evidence to inform intervention efforts. Early identification of maladaptive primals can encourage targeted interventions such as cognitive restructuring, behavioral activation, or mindful awareness of one's environment. For example, a client may be encouraged to keep a journal of examples in daily life that support adaptive beliefs, such as “My presence makes a difference in my community” or “Things matter in the world”. Positive psychology interventions, such as gratitude journals, are recommended to impact primals and well‐being (Stahlmann and Ruch 2023) and demonstrated initial acceptability and feasibility in psychiatric inpatients experiencing SI (Huffman et al. 2014).

Understanding the relationship between primal world beliefs and other established suicide‐related cognitions is essential to incorporating primals into contemporary theories of STBs. Fowler et al. (2026) examined associations between primals and suicide cognitions using an overlapping dataset with study 1 and found small‐to‐moderate sized associations (ranging from r = −0.21 to r = −0.49) across samples. Hunt et al. (2026) also found small‐to‐medium sized associations between various primals and dysfunctional attitudes (rs = −0.23 to −0.39) and large effect‐sized associations with hopelessness (rs = −0.50 to −0.62) using a separate Prolific dataset. These initial findings suggest that primals are related to and yet distinct from other suicide‐related cognitive risk factors, but the extent to which primals serve as distal vulnerabilities or more proximal risk factors still needs to be explored. Certain beliefs about the world may contribute to greater hopelessness or feelings of entrapment.

The current study provides initial support for the role of primals in STBs based on three diverse samples of adults with recent SI; however, there are several notable limitations. First, recruitment for all three samples relied on self‐selection using online platforms (SONA, Prolific), thus generalizability may be limited. Second, the treatment status of participants was not collected, so the impact of primals on STBs in the context of treatment remains unknown. Third, this study relied on self‐report questionnaires to assess all constructs of interest, which can lead to recall bias, social desirability bias, and be influenced by poor insight or misunderstanding of what is being assessed. Thus, future research is needed to replicate these findings using alternative methods such as qualitative interviews or behavioral observations. Finally, study 2 utilized a brief, 1‐month follow‐up period to examine the temporal relationship, which limits the ability to describe patterns over longer or shorter time frames. Future research could use ecological momentary assessments to determine the dynamic nature of primals in relation to suicide risk. Additionally, research should also examine the role of primals in the emergence of SI and suicidal behavior, consistent with ideation‐to‐action theories such as the Three‐Step Theory (Klonsky et al. 2021; Klonsky and May ²⁰¹⁵).

7. Conclusion

Primals are relatively stable cognitions associated with multiple facets of STBs. Specific primals predicted future SI, including the belief that the world is bad (vs. good), meaningless (vs. meaningful), and doesn't need me (vs. needs me). Integrating primals into models of suicide may offer more nuanced understandings of how beliefs shape risk. For example, those who believe the world is meaningless or boring may experience greater emotional distress, isolation, and hopelessness, amplifying the intensity or persistence of SI and eroding protective processes like problem solving and perceived reasons for living. Future research needs to examine primal world beliefs in the context of current theories of suicide and evaluate interventions that target primals to decrease suicide risk.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: Cross‐lag panel analysis with three tertiary primals and DSI‐SS scores over 1‐month.

Figure S2: Three cross‐lag panel analyses with each secondary primal individually predicting DSI‐SS scores over 1‐month.

Figure S3: Cross‐lag panel analysis with three secondary primals and DSI‐SS scores over 1‐month.

Table S1: Standardized coefficients (β) from multiple regression models that each included three secondary primals predicting suicide‐related outcomes.

Table S2: Standardized coefficients (β) from multiple regression models that each included three secondary primals predicting suicide‐related outcomes with covariates.

Data Availability Statement

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