You are on page 1of 12

Suicide and Life-Threatening Behavior 40(2) April 2010 181

 2010 The American Association of Suicidology

The Internal Suicide Debate Hypothesis:


Exploring the Life versus Death Struggle
Keith M. Harris, PhD, John P. McLean, PhD, Jeanie Sheffield, PhD,
and David Jobes, PhD, ABPP

Researchers and theorists (e.g., Shneidman, Stengel, Kovacs, and Beck)


hyothesized that suicidal people engage in an internal debate, or struggle, over
whether to live or die, but few studies have tested its tenability. This study intro-
duces direct assessment of a suicidal debate, revealing new aspects of suicidal ide-
ation. Results, from an online survey (N = 1,016), showed nearly all suicide-risk
respondents engaged in the debate. In addition, debate frequency accounted for
54% of the variance in suicidality scores, and showed significant associations with
other indicators of suicide risk. Likely factors of the debate, reasons for living and
dying, showed significant differences by suicidality, and most suicide-risk partici-
pants reported going online for such purposes, demonstrating a behavioral com-
ponent of the debate.

The internal suicide debate hypothesis en- hypothesis, and present characteristics of
capsulates a belief that suicidal individuals, what are likely to be the opposing sides; that
cognitively constricted though they may be, is, reasons for living (RFL) versus reasons for
are sometimes embroiled in an internal dying (RFD) ( Jobes & Mann, 1999).
struggle over whether to live or die. The hy- Shneidman (1964), based on observa-
pothesis incorporates ambivalence toward life tions of suicidal patients and suicide notes,
and death with a unique human attribute— described the ambivalent thoughts and feel-
deliberation. Although Kovacs and Beck ings he encountered as an internal debate, a
(1977) initiated empirical study of the hy- “continuum of vitality-of-consciousness, rang-
pothesis over three decades ago, there have ing between cessation and ardent living” (p.
been few researchers since who have ex- 98). Similarly, in his chapter “For and
plored this aspect of the suicidal mind. In this Against Self-Destruction,” Stengel (1964, pp.
study, we bring forward new evidence for the 85–88) discussed how the opposing sides of
self destruction and survival can coexist in
the suicidal person, but to differing degrees.
Keith Harris is Lecutrer of Psychology at Neuringer and Lettieri (1971) further pro-
James Cook University; John McLean is Associ- posed that while suicidal individuals were
ate Professor of Psychology at The University of more death than life oriented, they still re-
Queensland, Australia; Jeanie Sheffield is a Post-
doctoral Research Fellow at The University of tained a paradoxical drive toward life. Indeed,
Queensland, Australia; and David Jobes is Profes- it can be argued that indecision—ambiva-
sor of Psychology and Co-Director of Clinical lence—on whether to live versus die may be
Training at The Catholic University of Ameria. the most defining feature of the suicidal
Address correspondence to Keith Harris, mind. In The Encyclopedia of Suicide, Evans
Department of Psychology, James Cook Univer-
sity Australia, Singapore Campus, 600 Upper Thom- and Farberow (1988) refer to ambivalence as
son Rd. Singapore 574421. E-mail: keith.harris@ “perhaps the single most important psycho-
jcu.edu.sg logical concept in our understanding suicide”
182 The Internal Suicide Debate Hypothesis

(p. 12). Shneidman’s (1996) ten commonali- death by suicide first becomes an acceptable
ties of suicide include “the common purpose choice. This specific type of suicidal ideation
is to seek a solution,” and “the common cog- may be a pivot point, where a person takes
nitive state is ambivalence” (pp. 131–133). their first suicidal step by thinking seriously
These two commonalities help to define the about suicide and allowing for the possibility
debate, and suicidality in general, as an am- of a fatal outcome. Afterwards, whether sec-
bivalent struggle for a solution to seemingly onds or years later, he or she may become
insurmountable problems. These earlier ob- (more) cognitively constricted and attempt
servations have been well supported, if indi- suicide. A second scenario for an internal sui-
rectly, by research findings on aborted suicide cide debate is possible if we acknowledge that
attempts (e.g., Barber, Marzuk, Leon, & Por- cognitive constriction is not itself dichoto-
tera, 1998; Marzuk, Tardiff, Leon, Portera, mous, and that there is a spectrum of cogni-
& Weiner, 1997). Also, studies on attempters tive limitations. A range of constriction could
have demonstrated that many did not fully allow for some, albeit limited, rational thought
expect to live or die due to their most recent for those less cognitively impaired.
attempt (e.g., Freedenthal, 2007; Hjelme- The hypotheses and findings above
land, Stiles, Bille-Brahe, Ostamo, Renberg, help to encourage thoughtful discussion on
& Wasserman, 1998; Rudd, Joiner, & Rajab, ambivalence and the debate; however, empir-
1996). ical studies are needed to examine the nature
Many authors have described suicidal of the debate and its importance to suicidol-
individuals as limited in cognitive and ratio- ogy. Kovacs and Beck (1977) determined that
nal abilities. If suicidal people are cognitively 50% of patients hospitalized for suicide at-
impaired by tunnel vision, that may prevent tempts experienced an internal suicide de-
them from engaging in a rational decision- bate, based on discrepancies in wish to live
making process, such as an internal life versus (WTL) and wish to die (WTD) scores. Later
death debate. As part of his ten commonali- research (Brown, Steer, Henriques, & Beck,
ties of suicide, Shneidman (1996) stated that 2005) on a large sample of outpatients con-
“the common perceptual state of suicide is firmed that the interplay between WTL and
constriction“ (p. 133). Neuringer and Let- WTD posed a significant risk factor for
tieri (1971) found suicidal females were char- eventual suicide, even controlling for history
acterized by their dichotomous thinking; of attempts, psychiatric illness, hospitaliza-
while Weishaar and Beck (1990) found sup- tion, and demographics. If people do debate
port for suicidal individuals’ tendencies to- within themselves to live or to die, there
ward cognitive rigidity and problem-solving must be factors they consider as weights for
deficits. More recently, in a clinical sample, life and for death. Linehan, Goodstein, Niel-
those with a history of suicidal behavior sen, and Chiles (1983) introduced the Rea-
scored significantly lower than controls on sons for Living Inventory (RFLI) as a means
the Iowa Gambling Task, leading the authors of assessing reasons for not committing sui-
to conclude that impaired decision making cide. Total RFLI scores and the subscales
may be a suicide risk factor ( Jollant et al., survival and coping beliefs, responsibility to
2005). family, child-related concerns, and moral ob-
If most suicidal individuals enter a jections to suicide have been found to be sig-
state of tunnel vision, or cognitive constric- nificantly higher for nonsuicidal respondents
tion, is an internal suicide debate possible? compared with suicidal counterparts (e.g.,
There are at least two conditions in which a Dean & Range, 1999; Hirsch & Ellis, 1996;
debate might occur. Before becoming acutely Malone, Oquendo, Haas, Ellis, Li, & Mann,
suicidal, a person might, with some degree of 2000; Muehlenkamp & Gutierrez, 2007;
rationality, debate whether their life is worth Osman, Gifford, Jones, Lickiss, Osman, &
living or not. An internal debate could be Wenzel, 1993; Range & Penton, 1994).
part of early suicidal ideation, a stage where While the RFLI has demonstrated sta-
Harris et al. 183

tistically significant differences between sui- ing nonsuicidal individuals and a broader age
cide-risk and nonsuicidal groups, the differ- range.
ences are not always consistent by sex The current study was designed to fur-
(Hirsch & Ellis, 1996; Linehan et al., 1983), ther explore the internal suicide debate hy-
age (Miller, Segal, & Coolidge, 2001), cul- pothesis, as well as other unique facets of sui-
ture (Dobrov & Thorell, 2004; Oquendo et cidal ideation and behaviors. Rather than
al., 2005), and sexuality (Hirsch & Ellis, infer through circuitous means that a debate
1998). Despite these limitations, assessment had occurred, we decided to ask in a straight-
of RFL remains a potentially valuable area forward manner—both suicidal and nonsui-
for prevention and therapy; however, more cidal individuals—if they had ever engaged
research is required to understand how spe- in an internal suicide debate. It was hypothe-
cific RFL fit into the array of suicide risk and sized that a majority of suicidal participants
protective factors. would indicate experience with the debate.
Gutierrez, Osman, Kopper, Barrios, Demonstrating unique characteristics of the
and Bagge (2000) used the RFLI and other debate, RFD and RFL were hypothesized to
established measures to predict suicidality, show significant differences between suicidal
as assessed by Linehan’s Suicide Behaviors and nonsuicidal respondents. Frequency of
Questionnaire (SBQ; Linehan & Nielsen, the debate was also examined as it related to
1981). Gutierrez et al. (2000) found that two other indicators of suicidality, such as the
distinct facets, protective and risk factors, wish to live and die, as well as depressive
were related to suicidality. However, research symptoms.
comparing the utility of these polar opposites
has shown risk factors (e.g., hopelessness, sui-
METHOD
cidal ideation, and repulsion by life) remain
the better predictors of suicide risk (e.g.,
Participants
Dean & Range, 1999; Gutierrez, Osman,
Kopper, & Barrios, 2000). Therefore, while Ethics clearance was obtained from the
RFL remain important, it appears more per- School of Psychology, University of Queens-
tinent for suicidologists to assess the factors land, Australia. Volunteer participants from
inherent in the suicidal person’s attraction to 40 countries were solicited through an anon-
death. ymous online survey (N = 1,016). To attract
Work by Jobes and Mann ( Jobes, suicide-risk participants, we promoted the
2006; Jobes & Mann, 1999; Mann, 2002) has survey online as a study on the Internet and
brought forward a much needed assessment suicide. The survey (described in more detail
of reasons both for living and for dying. In in: Harris, McLean, & Sheffield, 2007) was
her study of these constructs, Mann (2002) advertised through Google, online forums,
found significant differences between suicidal and distributed through university networks.
and nonsuicidal individuals. Specifically, sui- Participants were aged 18–76, (M = 30.37,
cidal university patients who expressed gen- SD = 10.54); 62.3% were female; 75.4% had
eral feelings of hopefulness, in contrast to a university degree; and 82.2% were Cauca-
specific requirements for a hopeful future, sian and 11.7% were Asian. A total of 6.7%
were more likely to resolve their suicidal be- stated a history of psychiatric institutionaliza-
haviors. Additionally, Jobes and Mann (1999) tion, and 25.9% volunteered a history of psy-
found suicidal participants tended to endorse chiatric diagnosis.
their self, escape, and other people as RFD.
Jobes’ inclusion of RFD along with RFL rep- Materials and Instruments
resents an important step forward in the sys-
tematic evaluation of suicide risk and protec- Internal Suicide Debate and Current Life-
tive factors. However, further work is required Death Orientation. The debate question, de-
to normalize these constructs, such as includ- vised for this study to determine the presence
184 The Internal Suicide Debate Hypothesis

and frequency of internal suicidal debates, Center for Epidemiologic Studies Short
asked “Have you ever had an internal debate Depression Scale (CES-D 10). The CES-D
(in your head) about whether to live or die?” 10 (Andresen, Malmgren, Carter, & Patrick,
(1 = Never, 7 = Frequently). Prior to the on- 1994) was chosen to examine relationships
line survey, a pilot test (N = 17) included dis- between depressive symptoms and suicidality
cussions with a multicultural group of univer- measures. This 10-item version of the origi-
sity undergraduates on their comprehension of nal 20-item CES-D assesses depressive symp-
the debate question. No problems with inter- toms in the general population by examining
pretation of meaning were found. To assess the endurance and intensity of depressive
current life-death orientation, participants symptoms over the past week. Items are
were asked to respond to questions on their scored on a 4-point Likert scale (0 = rarely or
current wishes to live and die (Brown et al., none of the time [less than 1 day] to 3 = all of the
2005; Kovacs & Beck, 1977); that is, “Right time [5–7 days]). Higher scale totals indicate
now, I wish to live (or die) to the following greater depressive symptoms. Andresen et al.
extent” (1 = Not at all, 10 = Very much). found the CES-D 10 to have adequate con-
Reasons for Living and Dying. These vergent validity, internal consistency, and
items are part of the Suicide Status Form test-retest reliability. For the current study,
( Jobes, 2006). Respondents are instructed to Cronbach’s α = .90.
type in up to five RFL and RFD and to rank-
order them by importance. To elicit examples Procedure
of behavioral aspects of the debate, partici-
pants were asked “Have you ever gone online After a screening item on age, anony-
to look for, or confirm, your reasons for liv- mous respondents were asked to complete
ing (also dying)?” (1 = Never, 7 = Frequently). the above scales and respond to questions
Suicide Behaviors Questionnaire–Revised concerning suicidality, Internet usage, and
(SBQ-R). The SBQ-R (Osman, Bagge, Gu- demographics. The final page included infor-
tierrez, Konick, Kopper, & Barrios, 2001) mation on help for suicidal and other per-
was chosen to explore relationships among sonal problems.
the debate, RFL, RFD, and suicidality. It is a
4-item measure assessing current and lifetime
suicidality. The four items assess history of RESULTS
suicidal behavior (History), suicidal ideation
over the past year (Ideation), disclosing sui- We followed recommendations by Ta-
cidal plans (Disclose), and possibility of fu- bachnick and Fidell (2007) to cleanse the data
ture suicide (Predict). Item scores are summed, set and insure the overall integrity of the
with higher scores reflecting more severe sui- data. Various univariate and multivariate out-
cidality (Total). Osman et al. found that a lier adjustments were necessary and there was
cutoff total score of seven for community an overall reduction of the sample from
samples and eight for clinical samples pro- 1,058 to 1,016 complete surveys. To better
vided good sensitivity and specificity for dif- explore key variable relationships by suicidal-
ferentiating suicidal and nonsuicidal groups. ity, two extreme groups were formed by
As respondents may indicate more severe meeting criteria assessing lifetime prevalence
symptoms on computer-administered scales of suicidality and current life-death orienta-
than offline (e.g., Beck, Steer, & Ranieri, tion. Groups were formed based on Osman
1988; Joinson, 2007), the more stringent cut- et al.’s (2001) research, mentioned above, and
off score of eight was adopted for this study. Brown et al.’s (2005) study, which found a
Adequate to moderately high internal reli- combined WTL/WTD score to be a unique
ability has been reported for community and risk factor for suicide. Respondents were cat-
clinical samples (Osman et al., 2001). For the egorized as not suicidal by meeting the follow-
current study, Cronbach’s α = .83. ing criteria: minimum score on SBQ-RTotal,
Harris et al. 185

WTL = 10, and WTD = 1 (n = 192). Re- amining this item by degree of suicidality, we
spondents were grouped as high suicidal by found that 22.4% of the not suicidal (M =
meeting the following criteria: SBQ-RTotal ≥ 1.30, SD = 0.66) and 94.1% of high suicidal
8; and WTD ≥ WTL (n = 101). (M = 5.51, SD = 1.67) participants indicated
Pearson’s chi-squares were conducted to engaging in a debate, while a considerable
test for differences between the not suicidal 37.6% of the high suicidal participants re-
and high suicidal groups on sex and ethnicity. ported frequently engaging in a life-death
Only ethnicity showed a significant difference debate. To examine further relationships with
(at p < .05), with non-Caucasians overrepre- the debate item, two hierarchical multiple re-
sented in the not suicidal group (24.9%) com- gressions were run with SBQ-RTotal scores as
pared to the suicidal group (14.9%), χ2(1, N = the dependent (criterion) variable. Hierarchi-
290) = 3.93, p = .047, Cramer’s V = .12. A cal multiple regression is recommended for
one-way ANOVA showed no group differ- testing how much additional variance, in the
ences on age, p = .44. criterion, can be explained by the inclusion
Pearson correlation coefficients, con- of new predictor variables (Petrocelli, 2003).
trolling for sex, were calculated for the full In this case, due to its specificity to suicidal
sample to examine the relationships among ideation, it was hypothesized that an internal
debate, WTL, WTD, and SBQ-RTotal scores. suicide debate would be more relevant to sui-
The results showed that self-reported debate cide risk than depressive symptoms, WTL,
is significantly and positively correlated with or WTD. When conducting hierarchical re-
all measures of suicidality, and negatively gressions, variables are entered in order of
with WTL, while WTL and WTD are sig- causality or relevance (Cohen, Cohen, West,
nificantly negatively correlated with each & Aiken, 2003). Depression covers a number
other (Table 1). Also of interest are the on- of types of related disorders and is not spe-
line RFL and RFD items, both significantly cific to suicidality (e.g., Joiner, Walker, Pettit,
positively correlated with suicidality, indicat- Perez, & Cukrowicz, 2005). Wishes to live
ing that going online for RFD and RFL are and die are more likely than depressive symp-
both suicidal behaviors. toms to be relevant to suicide risk but are
Of the total sample (N = 1,016), 66.5% vague, and even a strong WTD may not in-
indicated some experience with the RFL ver- clude suicidal plans or other risky behaviors.
sus RFD debate (M = 2.64, SD = 1.78). Ex- In contrast, if we consider causality, depres-

TABLE 1
Interrelationships Between Suicidality and Key Variables (Controlling for Sex)
Correlation coefficients*

Variable Debate CES-D WTD WTL SBQ-RH SBQ-RI SBQ-RD SBQ-RP SBQ-RT

Debate — .51 .55 −.55 .67 .61 .45 .53 .69


CES-D .51 — .59 −.59 .46 .56 .34 .38 .50
WTD .55 .59 — −.80 .52 .65 .38 .49 .57
WTL −.55 −.59 −.80 — −.51 −.62 −.34 −.49 −.56
e-RFL .38 .35 .38 −.35 .29 .33 .20 .25 .30
e-RFD .46 .45 .53 −.48 .38 .46 .32 .35 .40

Note. N = 1,009 (7 cases were omitted due to missing data on sex). Debate = internal suicide de-
bate; CES-D = Center for Epidemiologic Studies short Depression scale; WTD = Wish to Die; WTL =
Wish to Live; e-RFL = going online for reasons for living; e-RFD = going online for reasons for dying;
SBQ-R = Suicidal Behaviors Questionnaire–Revised (H = history, I = ideation, D = disclose, P = predict,
T = total); *p < .0001 (all correlations).
186 The Internal Suicide Debate Hypothesis

sive symptoms are likely to precede current suicide, or death, or their consequences. The
wishes to live and die, as well as the begin- none responses were nearly always listed as
ning of a suicide debate. Therefore, a causal the first and only response and clearly indi-
model would place depressive symptoms first, cated no compelling reason to live, or die.
followed by WTL (WTL should be more RFL and RFD responses were evaluated by
natural, instinctive, than WTD), WTD, then two trained coders, with high inter-rater reli-
debate. In the first model (causal), depressive ability (RFL κ = .83; RFD κ = .81). Ques-
symptoms explain 25% of the variation in tionable responses and disagreements in cod-
suicidality scores, while WTL adds a further ing were finalized by discussion and consensus.
11%, WTD 2%, and debate an additional When response quantity was com-
16% to the variance explained by the preced- pared by degree of suicidality, not suicidal re-
ing variables (Table 2). Together, these four spondents were found to have produced sig-
variables explain 54% of the variance in nificantly more RFL (M = 3.58, SD = 1.68)
SBQ-RTotal scores. In the second model (rele- than high suicidal respondents (M = 2.97, SD =
vance), debate alone accounts for 48% of 1.68), t(291) = 4.39, p < .001, while high sui-
SBQ-RTotal variance, WTD adds an addi- cidal participants produced significantly more
tional 5%, while depressive symptoms and RFD (M = 3.58, SD = 1.61) than not suicidal
WTL contribute very little. These analyses participants (M = 1.24, SD = 1.53), t(291) =
highlight the importance, and particularly 18.03, p < .001. Of importance to life versus
the relevance, of the debate item to suicide death orientation, the high suicidal group’s
risk. number of RFD responses was significantly
greater than their RFL quantity, while the
RFL and RFD Responses not suicidal group’s number of RFL re-
sponses was significantly greater than their
Participants entered 3,787 RFL and RFD output (all p < .001).
2,394 RFD. Responses were categorized ac- First-ranked RFL and RFD were cho-
cording to Jobes’ (2006) coding manual, with sen, due to their obvious importance and
the exception of two new RFL categories, higher completion levels, for inter-group
fear and none, and one new RFD category, comparisons. Chi-square analyses were first
none. Fear includes any mention of fear of conducted to determine if there were any dif-

TABLE 2
Hierarchical Regression Models Predicting Suicidality Scores (SBQ-RTotal)
Model, Step, and Predictor Variable R2 ∆R 2 ∆F df

Model 1 (causal priority)


1. CES-D 10 .25 .25 337.64** (1, 1014)
2. Wish to live (WTL) .36 .11 167.62** (1, 1013)
3. Wish to die (WTD) .38 .02 37.74** (1, 1012)
4. Debate .54 .16 553.78** (1, 1011)
Model 2 (relevance priority)
1. Debate .48 .48 936.49** (1, 1014)
2. Wish to die (WTD) .53 .05 106.93** (1, 1013)
3. CES-D 10 .54 .01 14.62** (1, 1013)
4. Wish to live (WTL) .54 .004 8.45* (1, 1011)

Note. SBQ-R = Suicidal Behaviors Questionnaire–Revised; ∆R 2 =


change in R 2; ∆F = change in F; CES-D 10 = Center for Epidemiologic
Studies short Depression scale; Debate = internal suicide debate; *p < .01,
**p < .001.
Harris et al. 187

ferences on RFL by sex, age, or ethnicity. stark contrast, 54.5% of the high suicidal
There were no significant differences by sex, group reported going online for RFL (24.7%
χ2(9, N = 941) = 13.23, p = .15; but there frequently), while 63.4% went online for
were significant differences by age, χ2(9, N = RFD (37.6% frequently).
949) = 23.34, p = .005, Cramer’s V = .16, and
between Asians and Caucasians, χ2(9, N =
884) = 33.44, p < .001, Cramer’s V = .19. DISCUSSION
Younger participants more often reported
burdening others, friends, and future plans as In this study, anonymous Internet us-
RFL, while older respondents more often in- ers were asked to report their experiences of
dicated family as RFL. Caucasians more of- engaging in an internal suicide debate, their
ten listed burdening others, enjoyment, specific reasons for living and dying, and the
friends, hopefulness, and fear as RFL, while intensity of their current wishes to live and
Asian participants were more likely to report die. Support for the suicide debate hypothe-
responsibility, plans, and beliefs. sis was strong. We found that two-thirds of
Chi-square tests were conducted to the total sample indicated at least one occur-
determine if participants’ choices for RFL rence of debating within themselves to live
differed by level of suicidality. The not sui- or die. Further, in support of our hypothesis
cidal group was found to significantly differ that most suicidal people have engaged in an
from the high suicidal group, χ2(9, N = 273) = internal suicide debate, 94.5% of the high
46.99, p < .001, Cramer’s V = .41. The im- suicidal group indicated debate experience,
portance of family, beliefs, a strong self, and with 37.6% of these reporting often engag-
enjoyable things rank strongly as possible ing in the debate. In addition, significant pos-
protective factors. In contrast, fear (of suicide itive correlations between the debate and sui-
or death), fear of burdening others (by com- cidality and WTD, and a significant negative
pleting suicide), and responsibility appear to correlation with WTL, demonstrate the po-
be important RFL for the high suicidal tential of this one item in extrapolating sui-
group. Also of interest is that 5.1% of the cidal characteristics. While previous research
high suicidal group indicated they had no (Brown et al., 2005; Kovacs & Beck, 1977)
RFL (see Figure 1). found evidence for the internal debate through
Chi-square tests revealed even greater ambiguity in WTL and WTD scores, this is
differences (based on Cramer’s V scores) be- the first study to present direct evidence for
tween the high and not suicidal groups on the debate as a debate.
first-ranked RFD, χ2(7, N = 212) = 64.07, p < The debate factor correlated most
.001, Cramer’s V = .55. The not suicidal re- highly with respondents’ history of suicidal
spondents were likely to indicate no RFD, or behaviors and with SBQ-RTotal scores, fol-
a hopeless situation, while the high suicidal lowed by SBQ-RIdeation. Beck and colleagues
respondents were focused on escape, hope- (Beck, Brown, & Steer, 1997; Beck, Brown,
lessness, and their self. Of note, the four es- Steer, Dahlsgaard, & Grisham, 1999) found
cape categories combine for 19.3% of RFD worst point suicidal ideation to be a signifi-
for the not suicidal respondents, but appear cant predictor of suicide behaviors and even-
nearly twice as often (35.7%) for the high tual suicide, well beyond typical Scale for
suicidal group (see Figure 2). Suicide Ideation scores. The results of this
Lastly, we compared the not suicidal study indicate that the debate could be an in-
and high suicidal groups on their tendencies dicator of serious suicidal ideation, possibly
to go online to look for, or confirm, their rea- similar to worst point ideation. Further study
sons for living and dying. Not suicidal re- is required to understand where the life ver-
spondents reported very little active online sus death debate and worst points fall in the
pursuit of RFL or RFD, with only 4.7% go- spectrum of suicidal ideation.
ing online for RFL and 3.1% for RFD. In To help identify some of the specifics
188 The Internal Suicide Debate Hypothesis

Figure 1. First-ranked reasons for living (RFL) by degree of suicidality

of the debate, we asked participants to pro- vealed significant differences in the ways sui-
vide their reasons for living and dying. Re- cidal and nonsuicidal people value their lives.
spondents contributed nearly 3,800 RFL and For RFL, family oriented, positively self-
nearly 2,400 RFD. As hypothesized, and sup- focused, more enjoyable things (g.e., pets,
porting previous findings ( Jobes & Mann, nature), and personal beliefs were character-
1999; Mann, 2002), chi-square analyses re- istics of the not suicidal group. Reports of
Harris et al. 189

Figure 2. First-ranked reasons for living (RFD) by degree of suicidality


190 The Internal Suicide Debate Hypothesis

burdening others, responsibility, fear of sui- nal loneliness as their most substantive RFD,
cide/death, or having no RFL emerged as in a debate tilted toward death.
primary rationale for the high suicidal group. While this study included anonymous
In direct contrast to their RFL responses, the responses from 1,016 respondents in 40 coun-
high suicidal respondents were more self- tries, the sample was not a random selection
focused when stating RFD, and more likely of Internet users and was biased toward
than not suicidal respondents to indicate a higher educated, female, and (intentionally)
desire for escape, while the not suicidal par- suicide risk netizens. However, these demo-
ticipants were more likely to state having no graphics are not unusual; Internet users gen-
RFD or could not imagine a hopeless situa- erally show higher education levels than non-
tion. users (Estabrook, Witt, & Rainie, 2007), and
Active development of RFL and RFD the number of females online is beginning to
could be interpreted as engagement in at surpass that of males (Simpson, 2007). Al-
least one side of the debate to live or die. though the pilot study discussants indicated
While less than 5% of not suicidal respon- no problems in comprehending the debate
dents sought either RFL or RFD online, item, the nature of the internal debate and
54.5% of the high suicidal group went online the cognitive processes involved are complex
for RFL and 63.4% went online in pursuit of and not easily articulated. To be of use in de-
RFD. These results offer additional support constructing the suicidal mind, this item
for the suicide debate hypothesis, and dem- would benefit from additional work specifi-
onstrate the dynamic intentional nature of cally examining its construct validity.
the internal debate. Respondents also re- This study found substantial support
ported frequently going online to look for for the internal suicide debate hypothesis.
and confirm RFL and RFD, providing evi- Not only did nearly all high suicidal partici-
dence not only for the relevance of the de- pants report engaging in the debate, a third
bate, but for the important role the Internet indicated they do so often. The debate factor
is playing in the life of suicidal individuals. correlated significantly and positively with
Consistent with Baumeister’s (1990) items assessing suicidality, depression, and
suicide as escape from self theory, self-focus WTD, and negatively with WTL. Also note-
was a significant characteristic of suicidal in- worthy is that the debate item correlated
dividuals when presenting their RFD. Also, with going online for RFL and RFD, and was
in line with Shneidman’s (1993) theories on more strongly associated with suicidal behav-
psychache (or anguish) and one of his ten iors than suicidal ideation. Suicidal and non-
commonalities of suicide (i.e., “the common suicidal respondents were also found to differ
stimulus in suicide is intolerable psychologi- significantly on RFL, and even more strongly
cal pain”; 1996, p. 131), our research showed on RFD.
escape from anguish to be more often stated There are several clinical and research
by suicidal than not suicidal respondents. related implications we can draw from these
Combined, escape from anguish, escape from results. While maintaining a cognitive and
responsibility, escape from the past, and a behavioral focus on RFL, discussing and un-
general desire for escape accounted for derstanding a suicidal person’s RFD can be
35.7% of the first-ranked RFD of the high very useful in their treatment, as has been
suicidal group. Other categories of note were pointed out by a number of practitioners
statements of hopelessness (24.5%), negative (e.g., Ellis, 2004; Jobes, 2006; Kovacs, Beck,
statements of the self (21.4%), and state- & Weissman, 1975; Shneidman, 1999). This
ments concerning loneliness (8.2%). Based research has shown that there are important
on these results and the writings of the theo- realms of the suicidal mind left to explore.
rists above, highly suicidal individuals present Further ventures into this area will better de-
escape from the psychological pain of a nega- termine where the debate, RFL, and RFD fit
tive self-image, a hopeless future, and termi- in this psychological puzzle.
Harris et al. 191

REFERENCES

Andresen, E. M., Malmgren, J. A., Carter, trusted? Omega: Journal of Death and Dying, 55,
W. B., & Patrick, D. L. (1994). Screening for 57–70.
depression in well older adults: Evaluation of a Gutierrez, P. M., Osman, A., Kopper,
short form of the CES-D. American Journal of Pre- B. A., & Barrios, F. X. (2000). Why young people
ventive Medicine, 10, 77–84. do not kill themselves: The reasons for living in-
Barber, M. E., Marzuk, P. M., Leon, ventory for adolescents. Journal of Clinical Child
A. C., & Portera, L. (1998). Aborted suicide at- Psychology, 29, 177–187.
tempts: A new classification of suicidal behavior. Gutierrez, P. M., Osman, A., Kopper,
American Journal of Psychiatry, 155, 385–389. B. A., Barrios, F. X., & Bagge, C. L. (2000). Sui-
Baumeister, R. F. (1990). Suicide as escape cide risk assessment in a college student popula-
from self. Psychological Review, 97, 90–113. tion. Journal of Counseling Psychology, 47, 403–413.
Beck, A. T., Brown, G. K., & Steer, R. A. Harris, K. M., McLean, J. P., & Shef-
(1997). Psychometric characteristics of the scale field, J. (2009). Examining suicide-risk individu-
for suicide ideation with psychiatric outpatients. als who go online for suicide-related purposes. Ar-
Behaviour Research and Therapy, 35, 1039–1046. chives of Suicide Research, 14(3), 264–276.
Beck, A. T., Brown, G. K., Steer, R. A., Hirsch, J. K., & Ellis, J. B. (1996). Differ-
Dahlsgaard, K. K., & Grisham, J. R. (1999). Sui- ences in life stress and reasons for living among
cide ideation at its worst point: A predictor of college suicide ideators and non-ideators. College
eventual suicide in psychiatric outpatients. Suicide Student Journal, 30, 377–386.
and Life-Threatening Behavior, 29, 1–9. Hirsch, J. K., & Ellis, J. B. (1998). Rea-
Beck, A. T., Steer, R. A., & Ranieri, sons for living in homosexual and heterosexual
W. F. (1988). Scale for suicide ideation: Psycho- young adults. Archives of Suicide Research, 4, 243–
metric properties of a self-report version. Journal 248.
of Clinical Psychology, 44, 499–505. Hjelmeland, H., Stiles, T. C., Bille-
Brown, G. K., Steer, R. A., Henriques, Brahe, U., Ostamo, A., Renberg, E. S., &
G. R., & Beck, A. T. (2005). The internal struggle Wasserman, D. (1998). Parasuicide: The value of
between the wish to die and the wish to live: A suicidal intent and various motives as predictors of
risk factor for suicide. American Journal of Psychia- future suicidal behaviour. Archives of Suicide Re-
try, 162, 1977–1979. search, 4, 209–225.
Cohen, J., Cohen, P., West, S. G., & Jobes, D. A. (2006). Managing suicidal risk:
Aiken, L. S. (2003). Applied multiple regression/cor- A collaborative approach. New York: Guilford.
relation analysis for the behavioral sciences (3rd ed.). Jobes, D. A., & Mann, R. E. (1999). Rea-
Mahwah, NJ: Lawrence Erlbaum Associates. sons for living versus reasons for dying: Examin-
Dean, P. J., & Range, L. M. (1999). Test- ing the internal debate of suicide. Suicide and Life-
ing the escape theory of suicide in an outpatient Threatening Behavior, 29, 97–104.
clinical population. Cognitive Therapy and Research, Joiner, T. E., Jr., Walker, R. L., Pettit,
23, 561–572. J. W., Perez, M., & Cukrowicz, K. C. (2005).
Dobrov, E., & Thorell, L. H. (2004). Evidence-based assessment of depression in
“Reasons For Living”—Translation, psychometric adults. Psychological Assessment, 17, 267–277.
evaluation and relationships to suicidal behaviour Joinson, A. N. (2007). Disinhibition and
in a Swedish random sample. Nordic Journal of Psy- the internet. In J. Gackenbach (Ed.), Psychology
chiatry, 58, 277–285. and the internet: Intrapersonal, interpersonal, and
Ellis, T. E. (2004). Collaboration and a transpersonal implications (2nd ed., pp. 75–92). San
self-help orientation in therapy with suicidal cli- Diego, CA: Academic Press.
ents. Journal of Contemporary Psychotherapy, 34, Jollant, F., Bellivier, F., Leboyer, M.,
41–57. Astruc, B., Torres, S., Verdier, R., et al.
Estabrook, L., Witt, E., & Rainie, L. (2005). Impaired decision making in suicide at-
(2007). Information searches that solve problems: tempters. American Journal of Psychiatry, 162, 304–
How people use the internet, libraries, and gov- 310.
ernment agencies when they need help. Pew In- Kovacs, M., & Beck, A. T. (1977). The
ternet & American Life Project. Retrieved April 1, wish to die and the wish to live in attempted sui-
2008, from http://www.pewinternet.org/pdfs/Pew cides. Journal of Clinical Psychology, 33, 361–365.
_UI_LibrariesReport.pdf Kovacs, M., Beck, A. T., & Weissman, A.
Evans, G., & Farberow, N. L. (1988). The (1975). The use of suicidal motives in the psycho-
encyclopedia of suicide. New York: Facts on File. therapy of attempted suicides. American Journal of
Freedenthal, S. (2007). Challenges in as- Psychotherapy, 29, 363–368.
sessing intent to die: Can suicide attempters be Linehan, M. M., Goodstein, J. L., Niel-
192 The Internal Suicide Debate Hypothesis

sen, S. L., & Chiles, J. A. (1983). Reasons for Petrocelli, J. V. (2003). Hierarchical mul-
staying alive when you are thinking of killing tiple regression in counseling research: Common
yourself: The reasons for living inventory. Journal problems and possible remedies. Measurement and
of Consulting and Clinical Psychology, 51, 276–286. Evaluation in Counseling and Development, 36,
Linehan, M. M., & Nielsen, S. L. (1981). 9–22.
Assessment of suicide ideation and parasuicide: Range, L. M., & Penton, S. R. (1994).
Hopelessness and social desirability. Journal of Hope, hopelessness, and suicidality in college stu-
Consulting and Clinical Psychology, 49, 773–775. dents. Psychological Reports, 75, 456–458.
Malone, K. M., Oquendo, M. A., Haas, Rudd, M. D., Joiner, T., & Rajab, M. H.
G. L., Ellis, S. P., Li, S., & Mann, J. J. (2000). (1996). Relationships among suicide ideators, at-
Protective factors against suicidal acts in major de- tempters, and multiple attempters in a young-
pression: Reasons for living. American Journal of adult sample. Journal of Abnormal Psychology, 105,
Psychiatry, 157, 1084–1088. 541–550.
Mann, R. E. (2002). Reasons for living vs. Shneidman, E. S. (1964). Suicide, sleep,
reasons for dying: The development of suicidal and death: Some possible interrelations among
typologies for predicting treatment outcomes cessation, interruption, and continuous phenom-
(Doctoral dissertation, Catholic University of ena. Journal of Consulting Psychology, 28, 95–106.
America). Dissertation Abstracts International: Sec- Shneidman, E. S. (1993). Suicide as psy-
tion B: The Sciences and Engineering, 63(3-B). chache: A clinical approach to self-destructive behavior.
Marzuk, P. M., Tardiff, K., Leon, A. C., Lanham, MD: Jason Aronson.
Portera, L., & Weiner, C. (1997). The preva- Shneidman, E. S. (1996). The suicidal mind.
lence of aborted suicide attempts among psychiat- New York: Oxford University Press.
ric in-patients. Acta Psychiatrica Scandinavica, 96, Shneidman, E. S. (1999). Perturbation and
492–496. lethality: A psychological approach to assessment
Miller, J. S., Segal, D. L., & Coolidge, and intervention. In D. G. Jacobs (Ed.), The Har-
F. L. (2001). A comparison of suicidal thinking vard Medical School guide to suicide assessment and
and reasons for living among younger and older intervention (pp. 83–97). San Francisco, CA: Jos-
adults. Death Studies, 25, 357–365. sey-Bass.
Muehlenkamp, J. J., & Gutierrez, P. M. Simpson, G. (2007). Young Brit women
(2007). Risk for suicide attempts among adoles- online more than men. Business Week. Retrieved
cents who engage in non-suicidal self-injury. Ar- May 7, 2008, from http://www.businessweek.com/
chives of Suicide Research, 11, 69–82. globalbiz/content/aug2007/gb20070823_306674.
Neuringer, C., & Lettieri, D. J. (1971). htm?chan=technology_technology+index+page_
Cognition, attitude, and affect in suicidal individ- top+storiesTechnology
uals. Suicide and Life-Threatening Behavior, 1, 106– Stengel, E. (1964). Suicide and attempted
124. suicide. Oxford, England: Penguin.
Oquendo, M. A., Dragatsi, D., Harkavy- Tabachnick, B. G., & Fidell, L. S. (2007).
Friedman, J., Dervic, K., Currier, D., Burke, Using multivariate statistics. Boston: Pearson/Allyn
A. K., et al. (2005). Protective factors against sui- & Bacon.
cidal behavior in Latinos. Journal of Nervous and Weishaar, M. E., & Beck, A. T. (1990).
Mental Disease, 193, 438–443. Cognitive approaches to understanding and treat-
Osman, A., Bagge, C. L., Gutierrez, ing suicidal behavior. In S. J. Blumenthal & D. J.
P. M., Konick, L. C., Kopper, B. A., & Barrios, Kupfer (Eds.), Suicide over the life cycle: Risk factors,
F. X. (2001). The suicidal behaviors question- assessment, and treatment of suicidal patients (pp.
naire—Revised (SBQ-R): Validation with clinical 469–498). Washington, DC: American Psychiatric
and nonclinical samples. Assessment, 8, 443–454. Association.
Osman, A., Gifford, J., Jones, T., Lickiss,
L., Osman, J., & Wenzel, R. (1993). Psychomet- Manuscript Received: April 7, 2009
ric evaluation of the reasons for living inventory. Revision Accepted: July 16, 2009
Psychological Assessment, 5, 154–158.

You might also like