SUICIDE RISK, SELF-INJURY RISK, AND EXPECTED
INTENTIONALITY FOR A POPULATION AND ITS
KARL EUGENE WILSON
A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
KARL EUGENE WILSON
The data utilized in this study were collected for research sup-
ported in part by National Institute of Mental Health Grant MH-18911.
I wish to acknowledge the cooperation of Dr. George Murphy and Dr.
Richard Wetzel in permitting me use of the data for this study, and
Ms. Marcia Lange for her invaluable statistical and computer consul-
I wish to thank Dr. Richard McGee, Chairman of my Committee, for
his advice and guidance in this work and throughout my graduate career.
He has been an exceptional professional model, mentor, and friend. I
also wish to express appreciation to Dr. Benjamin Barger, Dr. Hugh Davis,
Dr. Everett Hall, Dr. Richard Swanson, and Dr. Joe Wittmer, who served
on the Committee, for the interest they have shown and the advice which
they have offered.
I thank Mr. Douglas Freeman for rating the reliability sample and
Dr. Joe Thigpen for his hospitality and facilitative efforts on my
I reserve special acknowledgments and loving gratitude to my wife,
Jan, for her support and understanding.
TABLE OF CONTENTS
ACKNOWLEDGMENTS . . . . . . . . . . . .
ABSTRACT . . . . . . . . . . . . . .
CHAPTER I INTRODUCTION . . . . . . .
Suicide in the United States . . . . .
Toward an Understanding of Suicide . . . .
Attempted Suicide . . . . . . . .
Intention-to-Die . . . . . .....
The Role of Intention-to-Die in Self-Injurious
Assessing Intention-to-Die . . . . .
The Relationship of Suicide Risk Components: A
Theoretical System . . . . . .
The Study: Purpose and Hypotheses . . .
. . .
CHAPTER II METHODS . . . .
The Population. . . . .
The Sample . . . . .
Assessing Intention-to-Die . .
Revision of the Probability of
Reliability . . . .
Probability of Dying . . .
Correlates of Intention-to-Die.
Risks and Expected Intentionality
Hypotheses Testing . . . .
Hypothesis 1 . . . . .
Hypothesis 2 . . . .
Hypothesis 3 . . . . .
Hypothesis 4 . . . . .
Hypothesis 5 . . . .
Hypothesis 6 . . . . .
. . . . . .
. . . . ..
. . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
CHAPTER III RESULTS . . . . .
Intention-to-Die Matrix . . . . .
Characteristics of Self-Injurers . .
Multivariate Analyses of Intention-to-Die
Self-Injury Risk, Suicide Attempt Risk, and
Hypotheses Testing . . . . . .
Hypotnesis 1 . . . . . . .
Hypothesis 2 . . . . . . .
Hypothesis 3 . . . . . . .
Hypothesis 4 . . . . . . .
Hypothesis 5 . . . . . . .
Hypothesis 6 . . . . . . .
. . .
CHAPTER IV DISCUSSION . . . . . . . . . 64
Hypotheses Testing . . . . . . . . .. . 64
Expected Intentionality . . . . . . . . .. 65
Assessing Intention-to-Die . . . . . . . . 65
Predicting Intention-to-Die . . . . . . . . 67
Developmental Life Crises. . . . . . . . . .. 68
Suicide Prevention . . . . . . . . . . . 71
Formulating Suicide Prevention Strategies . . . .. 71
Evaluating Suicide Prevention Services. . . . . ... 74
Parasuicide . . . . . .. . ......... 75
Significance and Prospects. .. . . . . . . . 76
CHAPTER V SUMn1ARY . . . . . . . . ... ..... 79
APPENDIX A REVERSIBILITY OF METHOD SCALE. . . . . ... 80
APPENDIX B PROBABILITY OF INTERVENTION SCALE. . . . ... 81
APPENDIX C PROBABILITY OF INTERVENTION REVISED SCALE. .... . 82
APPENDIX D INTENTION-TO-DIE MATRIX. . . . . . . ... 83
REFERENCES. . . . . . . . . ... ....... 84
BIOGRAPHICAL SKETCH . . . . . . . .... .... 91
Abstract of Dissertation Presented to the Graduate Council
of the University of Florida i- Pdrtial Fulfillment of the Requirements
for the Degree of Doctor of Philosophy
SUICIDE RISK, SELF-INJURY RISK, AND EXPECTED
IHTENTIONALITY FOR A POPULATION AND ITS
Karl Eugene Wilson
Chairman: Richard K. McGee
Major Department: Psychology
Although a number of good studies have been able to delineate
factors associated with suicide, only recently have researchers begun
to make epidemiological investigations of self-injurious acts which did
not directly result in the death of the victim. The investigation of
factors associated with self-injurious acts, both suicides and suicide
attempts, was the major focus of this study.
A schematization which was proposed earlier by Wilson (1974) and
Freeman, Wilson, Thigpen, and McGee (1974) was applied to self-injuries
reported for a general population and component sub-populations.
Specifically, the self-injuries were rated for intention-to-die, which
is the probability of dying given a self-injury. Suicide risk, suicide
attempt risk, self-injury risk, and expected intentionality were then
ascertained. Expected intentionality is a prediction statement of the
probability of dying of an individual given a future act of self-injury
and is derived from known average levels of intention-to-die of sub-
The subject population for this study consisted of all attempted
suicides (n = 692) and suicides (n = 115) reported by police for a two
year period in a majority of municipalities of St. Louis County, Missouri
These 807 reports were analyzed and rated utilizing a revised version
of the Intention-to-Die Scales developed by Freeman et al. (1974). The
expected intentionalities associated with the cells of the Intention-
to-Die Matrix were ascertained through Stepwise Regression Analysis of
suicide death as a function of scores on the scales. Suicide attempt
risk, suicide risk, and their sum, self-injury risk, were obtained for
each sex X social-status sub-population of interest by dividing inci-
dence by sub-population size times 2 years. For the overall sample
and for each sex, multivariate analyses were utilized to determine the
relationship of a number of variables, including age, health, living
conditions, marital status, and education, to intention-to-die.
A number of hypotheses were tested. The majority of self-injurious
behaviors were found to be associated with essentially no intention-
to-die. Self-injury risk and suicide attempt risk decreased signifi-
cantly over the life span for most sub-populations while suicide risk
increased for both males and females. The hypothesis that moderate
intention-to-die self-injuries would increase in age ranges associated
with developmental life crises was rejected. Intention-to-die was
found to increase steadily as a strongly significant function of age.
Living alone was found to lower intention-to-die in older females.
Although age and sex were the strongest factors relating to
intention-to-die, other personal factors were found to make minor, but
significant, contributions to the multiple relationships, which were
multiple r = .54 for the overall population, multiple r = .43 for
females and multiple r = .56 for males. Although social-status did
not relate to intention-to-die, lower social-status for both males and
females was found to increase self-injury risk, suicide attempt risk
and suicide risk. Actuarial type tables for these risks were reported
for each sex X social-status sub-population.
This study was the first to ascertain the role of expected inten-
tionality and self-injury risk in suicide risk. Findings are discussed
in terms of evaluating suicide prevention centers, formulating suicide
prevention strategies, and in the ongoing reconceptualization evolving
in the literature of acts presently labeled as suicide attempts.
The subject of this study is the role of the victims' intention-
to-die in self-injurious behaviors, suicide and suicide attempts, for
a general population and relevant component sub-populations.
Suicide in the United States
Suicide has been a prevalent factor in civilized cultures through-
out history (Choron, 1972). Within the United States it ranks, from
year to year, as the ninth to eleventh leading cause of death: eighth
for white males, third for the 15-24 age group, and fourth in the 25-
44 age group. Officially reported suicides in the United States now
average between 22,000 and 25,000 annually, about one percent of all
deaths. Most investigations estimate suicide to be underreported by a
factor of one fourth to one third, increasing incidence to perhaps
30,000 annually. The overall suicide rate has remained fairly stable
in this country, ranging from 10-17 per 100,000 since the beginning of
the century, and averaging approximately 11 per 100,000 in each of the
years of the last decade. However, regional differences can range from
less than 7 for Newark or Providence, to over 17 for the Tampa, San
Francisco, and Los Angeles areas. Even larger differences in rates
between nations demonstrate both the effect of culture upon suicide,
and possible differential biases in the reporting of these phenomena
(World Health Organization, 1967; Massey, 1967; United Nations, 1967).
In the Western World, men commit suicide at a higher rate than
women. For the United States a consistent ratio of 3:1, men to women,
is found, although the female rate has been steadily increasing relative
to the male rate since 1950 (Dublin, 1963; flassey, 1967). The average
age of suicides is usually found to be around 50 while rates increase
steadily with age from 6.0 for the 15-24 age group, to 20.5 for the
45-54 age group, to 23.9 for the 75-84 age group (Tuckman & Lavell,
1958). Female rates, however, peak around age 50 and decrease steadily
into old age, while white male rates increase monotonically. Non-white
suicide rates are significantly lower than rates for whites. Negro
males commit suicide at a rate similar to that of white males until
the mid-thirties and their rates decline gradually, rising again to a
smaller peak in old age. However, non-white suicide rates, especially
among young males, have been increasing dramatically (Hendin, 1969;
Massey, 1967). Marital status hes also been found to significantly
effect suicide rates: for persons over age 15 rates are 11.9 for married,
20.9 for single, 23.8 for widowed, and 39.9 for divorced persons.
These patterns are consistent over age, with the exception that being
widowed increases suicide risk dramatically in younger groups, and
becomes less of a factor with increasing age (Massey, 1967; Durkheim,
1951; MacMahon & Pugh, 1965).
MacMahon, Johnson and Pugh (1963) found striking parallels between
suicide rates for white males in the United States and level of unem-
ployment, especially among 45-55 year olds. Sociological studies from
Durkheim's classic study of 1897 (1951) have concentrated upon
additional factors in differential suicide rates that include
occupation, socioeconomic status, and religion (Farberow, Shneidman,
and Neuringer, 1966). For example, Kennedy, Kreitman, and Ovenstone
(1974) found both suicice and "parasuicide" (i.e., suicide attempt)
rates to be highest in impoverished, socially disorganized slum areas.
Maris found that, for males, ". . the social-status hierarchy is
inversely related to the suicide rate" (1967, p. 249). He utilized
occupational-status as an indicator of social-status in one analysis,
and also utilized an index of socice:coomic tatuss based on occupation,
income, and educational level. Breed (1963, 1967) found decreasing
income and downward occupational mobility to be associated with in-
creased suicide risk. Loss of position, especially among men; loss
of another person, especially among women; and loss of mutuality, the
weakening of mutual social relationships over time, have all been found
to contribute to increased suicide risk (Creed, 1967, 1966). Humphry
(1974) reports that loss of social roles increases vulnerability to
suicide. In the case histories of 160 suicides he found prevalent
patterns of role-disturbances from childhood, through chaotic marriages,
to later loss of occupational roles.
Broken homes in childhood, loss or prolonged absence of at least
one parent before age 15, increase both the probability of suicide
attempts and suicide (Stengel & Cook, 1958; Batchelor & Napier, 1953/54;
Dorpat, Jackson, and Ripley, 1965). Loss of some kind is strongly
associated with suicide. Dorpat and Ripley (1960) and Murphy and
Robins (1968) found 27 and 26 percent respectively of suicides in a
consecutive series had suffered a recent loss of a love object. These
studies als) report 49 and 43 51 percent, respectively, to have a
medical or surgical illness. Murphy and Robins caution, however, that
base rates for these phenomena have to be taken into account: They note
that Burnight found 64 percent of a ". . random sample of noninstitu-
tionalized married white urban men aged 60-64 years (among the highest
suicide risk sub-population) reported the presence of one or more
chronic medical and/or surgical conditions" (Murphy & Robins, 1967,
Many psychological studies of suicide utilize case histories of
patients who subsequently commit suicide. Farberow, Shneidman, and
Neuringer describe patients who later commit suicide as exhibiting a
"characteristic pattern . the 'dependent-dissatisfied' person . ."
(1966, p. 42). These patients are more complaining, demanding, in-
sisting, controlling, inflexible, and unadapting. They would turn to
the staff for support but proceed in alienating them with insatiable
demands for special attention. Continual strokes were needed for self
esteem, which resulted in a "bind" where increased demands would exhaust
sources of gratification, leading to an acceleration of demands, despite
their negative effect. Prior to suicide, the most frequent behavioral
manifestations were: agitation; depression; withdrawal; sleep troubles;
complaints about health; and difficulty in thinking and concentrating.
In a longitudinal study, Fawcett (1969) found four characteristics
which differentiated high from moderate and low suicidal risk patients:
"interpersonal capacity;" marital isolation; distorted communication;
and "help negation."
In studies of consecutive series of suicides in general populations,
Robins, Murphy, Wilkinson, Gassner, and Kayes (1959b) found 21.6 percent
and Dorpat and Ripley (1960) found 33.3 had made a prior suicide attempt.
Studies with more bias in the selection of the sample report up to 62.5
percent prior suicide attempts in suicides (Dorpat & Ripley, 1967).
Most suicides have communicated their intent to kill themselves
(Dorpat & Ripley, 1962). Robins, Gassner, Kayes, Wilkinson, and Murphy
(1959a) found over two-thirds had made some, usually several, attempts
to communicate their concerns with death or suicide, usually for the
first time, recently, directly, and to more than one other person.
One-fifth of these were to physicians, although a higher percentage,
half of the suicides, had seen a physician within the previous year,
a frequent finding in the literature (Motto, 1958; Murphy, 1972; Robins
et al., 1959b; Dorpat & Ripley, 1960; McCarthy and Walsh, 1966).
The most frequently found psychiatric diagnosis in suicide is a
depressive illness, usually affective psychosis. Dorpat and Ripley
(1960) found 30 percent while Murphy and Robins (1968) found 45 percent
of suicides in this category. Both of these studies report a fourth
of the suicides as chronic alcoholics. Psychiatric disorders that
rarely occurred, and distinguished suicides from suicide attempters,
were sociopathy and hysteria. Within a patient population, Pokorny
(1960) found 26.5 percent of suicides to be schizophrenics. Farberow,
Shneidman and Neuringer (1966) had similar findings, but a difficulty
again occurs with base rates. Both patient populations were from V.A.
hospitals, where a high proportion of psychiatric patients are labeled
Although depression has been found to be a prevalent factor in
suicide, most studies have imposed diagnoses retroactively. Some
studies have retrospectively diagnosed as many as 94 percent of
suicides as psychiatrically ill (Robins et al., 1959b), but are
methodologically weakened through the absence of psychiatric postmortem
of a control group. Patients diagnosed as manic-depressive are at great
suicidal risk, however, as most studies find that 15 percent will
eventually suicide (Diggory, 1967).
In summary, suicidal factors, although isolated in most studies,
work together to form patterns of suicide risk. For example, a fairly
typical suicide would involve a number of the following characteristics:
white male over 40 years of age; retired, unemployed or suffering
recent job troubles; living alone; divorced; having a history of
suicide attempts; somatic complaints; alcoholism, where the drinking
recently has become more uncontrolled; and, a recent loss or threat
of loss of a close relationship (Murphy, 1969). The picture isn't
completely clear, however, as young people and women also kill them-
selves and the majority of people with even the most suicidogenic
characteristics will go on living, and eventually die of other causes.
Toward an Understanding of Suicide
Until recently suicide attempters have been viewed as failed, or
bungled, suicides. In other words, attempters were viewed as belonging
to the same, or at least a similar, population as completed suicides.
While sociologists and anthropologists concerned themselves almost
exclusively with incidence and characteristics of suicides, psychologists
and psychiatrists were presented with methodological difficulties in
garnering generalizable information concerning the motives, dynamics,
and personality characteristics of suicides. Several approaches evolved,
including: retrospective psychiatric diagnoses of consecutive series
of suicides through perusal of public and medical records and interviews
with families (Robins et al., 1959b); the "method of residuals," the
study of notes and other clues which are left by approximately a third
of suicides (Shneidman & Farberow, 1957a); longitudinal follow up of
highly selected, at risk, psychiatrically hospitalized populations
(Farberow, Shneidman, and Neuringer, 1966); and psychological studies
of people who have attempted suicide and survived or who have threatened
suicide (Freud, 1925; Menninger, 1938; Stengel & Cook, 1958; Shneidman,
1963). The difficulty with the retrospective diagnostic approach is
one of base rates and the use of retrospective data. The difficulty
with longitudinal studies of selected populations or with the "method
of residuals" is in generalizing from what is known about a biased
population to all suicides. The difficulty with the last approach is
within the assumption that suicide attempters, threateners, and com-
pleters are interchangeable in characteristics (Neuringer, 1962).
Following the conceptual lead of Stengel (1952), more recent studies
have begun to find that suicide attempters and completers represent
two different, but overlapping populations (Dorpat & Ripley, 1967).
Sociological and psychological factors can be shown to covary
with suicide rates. However, they serve only as correlates of, rather
than explanations of the phenomena. Although an increase in unemploy-
ment is associated with increases in white male suicides, the vast
majority of the unemployed will not commit suicide. What does unemploy-
ment, or any other correlate of suicide, entail that it should raise
suicide rates? Or, why should increased unemployment lower a group's
collective immunity to this behavior, leaving a small but significant
Freud (1925) outlined the dynamics of depressive suicide as the
turning inward of sadistic impulses, where the ego perceives itself as
deserted by the superego and permits its own demise. Suicide is seen
as an outcome of a strong ambivalent dependence on a sadistic superego
and the necessity of ridding oneself of an unbearable guilt tension at
any cost. Jackson (1957) sees motivational theories of suicide as
stressing suicide as a symptomatic act, and not a discrete entity,
which is motivated by either: self-directed aggression (thanatos); re-
birth and restitution, the doing away of the "bad me" to permit a new
beginning; or, loss and despair, loss of self esteem or a real or
imagined love object, or loss of health, prestige, or resources.
Menninger (1938) expanded on Freud and his postulation of a "death
instinct," or "thanatos." He proposes a triad of lethal wishes: to
kill; to be killed; and to die.
"Psychologists refer to social factors in their theoretical for-
mulations of the causation of suicide" (Farberow, Shneidman, and Neu-
ringer, 1966, 32). Economic, religious or political conflicts, de-
pressions in reaction to the loss of status, blows to the self-concept,
and dependency frustration in reaction to loss of a love object are
all seen as interactive factors in the dynamics of suicide. For
example, Jackson (1957) describes a typical suicide scenario advanced
by Davison: The victim, having reached the limit of his resources
(following a crisis situation) loses sight of his life goals. The
immediate situation acts as a "dominant" which restricts the field
of consciousness, resulting in an inattention to life and depression.
Higher brain centers are unable to comply with and control incoming
impulses to choose an action. The victim, losing perspective of his
problems, gives way to imagination, and loses the normal ability to
inhibit unhealthy impulses, resulting in suicide.
Neuringer (1964) and Neuringer and Lettieri (1971) stressed the
role of cognition in suicide. They found high risk groups to be
distinguished by rigid (i.e., simplistic and dogmatic) cognitive styles.
Life would have to be categorized as black or white, with greater and
greater discrepancy between extremes.
Shneidman and Farberow (1957) coined the term "catalogic" to
describe the fallacious thinking patterns in most suicides, especially
mistaking or fallaciously interchanging non-equivalent concepts. For
example, they outline two different concepts subsumed by the word "I":
I-self, or our conception of ourselves; and I-other, or our c ;'ceptions
of other persons' perceptions of ourselves. Through the study of
suicide notes they found a high incidence of concern on the part of
the victim with the I-other; of getting even with other people, etc.,
as if the I-self would be around to see these reactions: "They'll
feel different about me then."
Common to most theories of the causes of suicide is the inability
of the victim to cope with some perceived personal failure and to make
necessary decisions to resolve a crisis (Farberow, 1967). The crisis
can result in suicide as a function of sociological, interpersonal,
intrapersonal, and/or developmental factors. For example, one's in-
ability to solve an interpersonal crisis, such as threatened loss of
a love object, can be aggravated by the poor resolution of what
Erikson (1950) describes as a developmental life crisis. Of concern
to the study of suicide are: the "identity versus role confusion"
crisis of adolescence; the crisis of "intimacy versus isolation"
within the young adult; the crisis of "generativity versus stagnation"
of the young adult, and the crisis of "ego integrity versus despair"
of middle age and maturity. Specifically, the young adult, the middle
aged person and the older person face the life paradox of giving some-
thing up in order to get something of much greater value: the chance
to continue to grow, accommodate to life, and receive life's bounties.
Those who fail at these tasks are particularly vulnerable to other
factors, interpersonal and sociological, which correlate with suicide.
In these crises, what is given up are the old cognitions and accommo-
dative patterns. The cost of not rescinding them is increased vul-
nerability to failure of adaptation and, possibly, the final regression
from reality, suicide.
However, most investigators do not see suicide as a uni-dimensional
concept or discrete psychologic entity, but as a number of different
syndromes under one rubric. A number of different schemes for
categorizing suicide have appeared, beginning with Durkheim's
schematization of 1897 (1951). Durkheim classified suicides as a
function of sociological factors: "egoistic suicide," where the
individual is not sufficiently integrated into his society; "altruistic
suicide," where overintegration of the individual with society leads
to the individual's self-sacrifice, as with kamikaze pilots; "anomic
suicide," where the individual's adjustment to society is suddenly
disrupted (e.g., through a loss of wealth) combined with a lack of
sympathetic acceptance of the individual by his social group; and
"fatalistic suicide" as a reaction to oppressive authoritarian con-
Shneidman and Farberow (1957) list: "surcease suicides," where
the individual is making a rational choice to escape pain and no
reasonable future can be anticipated; "catalogic suicides," which have
already been discussed and which make up the majority of suicides;
"cultural suicide," which is similar to Durkheim's classification of
altruistic suicides; and "schizophrenic suicide," where the victim
utilizes "paleologic," making identifications in terms of predicates
rather than subjects, which may result in the cutting out of the "bad
me," although no intention-to-die exists.
Neuringer (1962), in reviewing classification schemes, found the
following different categories of suicidal acts: 1) intentional suicide,
which includes altruistic and surcease suicides, and Camus' classifica-
tion of "existential suicide," for reasons of the basic absurdity of
life; 2) psychotic suicide; 3) automatization suicide, where the victim
is a habitual abuser of sedatives and alcohol, and ingests one sedative
after another in order to reach an unobtainable desired effect, which
results in an accidental death; 4) accidental suicide, especially in
cases of "contra-intentional" attempts where the intention is not to
die, but to elicit a response from some significant other, and the
victim accidentally dies; 5) manipulative suicidal act, or a suicide
attempt where the motivation is a warning or plea, a "cry for help;"
6-9) chronic, neglect, probability, or self-destructive suicides, which
involve killing oneself slowly or increasing the probability of death
occurring early, as is more or less the case with smokers, drug-addicts,
race-drivers, or overeaters; 10) suicidal threats; 11) suicidal thinking,
and 12) test suicide, or persons giving suicidal or depressive responses
on psychological tests.
Shneidman notes the semantic confusion in the field and notes that
present concepts of death and suicide are too ambiguous to be either
scientifically or clinically useful. He defines suicide as ". . the
human act of self-inflicted, self-intentioned cessation." Further,
intention is defined as: ". . the role of the victim in his own
demise" (1969, p. 225). Only category 1 of those categories of suicidal
acts Neuringer found can clearly fit this definition. Category 2 is
also traditionally included as suicide in the literature, while
categories 3 and 4 are usually considered accidental deaths, categories
6-9 are considered self-destructive life styles rather than suicide,
and category 5 is considered a suicide attempt.
Stengel defined a suicide attempt as: ". . any non-fatal act of
self-damage inflicted with self-destructive intention, however vague
and ambiguous. Sometimes this intention has to be inferred from the
individual's behavior" (1968, p. 172). He justifies this usage, even
when there is no risk of death, because, from the victim's point of
view, ". . those attempts are risk-taking acts whose outcome is
uncertain" (1968, p. 173). He also justifies this usage from the
standpoint of the victim's higher future suicidal risk: ". . People
who tend to react to stressful situations with suicidal gestures are
more likely sooner or later to commit suicidal acts than people who
make no such gestures" (1968, p. 173). This view stems from an
assumption that suicides and suicide attempts are part of some continuum
of behaviors. Stengel (1960) had noted earlier that human behavior
usually is a function of multiple motivations, and both suicides and
suicide attempts are a mixture of many motives, including wanting to
live and not live.
Many investigators agree that: ". .. even the suicide gesture
should be considered a 'cry for help' which, if ignored, may later
lead to more serious and lethal self-destructive behavior" (Dorpat &
Ripley, 1967, p. 77). Freeman, Wilson, Thigpen, and McGee (1974) argue
that the use of the terms suicide gesture or suicide attempt in cases
of low probability of dying, is inaccurate, pejorative, and leads
treatment staff at hospitals and essential gatekeeper intervenors,
such as police, to deal with the victim in non-facilitative ways. They
cite studies of the negative attitudes and sometimes open hostility
of these essential caregivers toward low-intentioned suicide attempts.
They propose the use of the term "self-injurious behaviors" for both
suicide attempts and suicide. Similarly, Kessel and McCulloch (1966)
had utilized the terms "deliberate self-poisoning" and "deliberate
Kennedy, Kreitman, and Ovenstone (1974) suggest that omission of
the reference to the term suicide neglects the very real association
between attempted and completed suicide. They propose the term
"parasuicide" while Choron (1972) proposes "protosuicide." As a sub-
stitute for the term suicide gesture, where there is no intention-to-
die, Lennard-Jones and Asher (1959) propose the term "pseudo-suicide."
Kessel and Lee (1962) and Clendenin and Murphy (1971) utilize opera-
tional definitions for their studies. For the purposes of consistency,
this study will utilize the term suicide attempt as it has evolved
in the literature and as defined earlier by Stengel (1968). However,
implications of the empirical findings of this study will be discussed
in terms of a classificatory schematization of self-injurious behaviors,
acts presently labeled as suicides and suicide attempts.
"Compared with groups that commit suicide, those who attempt
suicide are younger, use less lethal methods, include more women than
men, and more often include impulsive self-destructive behavior per-
formed in the presence of other people" (Dorpat & Ripley, 1967, p. 74;
Dorpat & Boswell, 1963). Although Dublin (1963) advances the notion
that women outnumber men in suicide attempts by 3:1, a lower ratio has
been found in most studies which utilize other than populations
selected towards particular sub-populations. Women usually are found
to outnumber mcn by ratios of from 2.1:1 to 2.5:1 (Shneidman & Farberow,
1961; Dorpat & Boswell, 1963; Edwards & Whitlock, 1968; Murphy, Clendenin,
Darvish, and Robins, 1971). In other woros, about two-thirds of suicides
are men while over two-thirds of attempters are found to be women. The
majority of suicide attempters are women under 40 (Hopkins, 1937;
Ettlinger & Flordh, 1955; Dahlgren, 1955; Gold, 1965; Sclare & Hamilton,
1963; Whitlock & Schapira, 1967). Non-white women have a high incidence
of suicide attempts in the United States (Davis, 1967). Dorpat and
Boswell (1963) found the average age of their Seattle sample of
attempters to be 35.1 years while 51.3 was the average age of completed
suicides. For Los Angeles, Shneidman and Farberow (1961) found modal
age of suicides to be 42, with age of attempters peaking at 32 for
males and 27 for females. Depressive and alcoholic psychiatric
diagnoses are represented frequently among suicide attempters, but the
high frequency diagnoses, which appear less often among completed
suicides, include hysteria, sociopathy, or character disorder, and
anxiety neurosis (Robins et al., 1959b). Suicide notes are left less
often by attempters than suicides, especially in low lethality attempts
labeled "gestures" (Dorpat & Boswell, 1963).
Tuckman, Youngman, and Bleiberg (1962) found a higher rate of
attempts in health districts of Philadelphia characterized by poor
housing, low income, high morbidity, and delinquency; factors associated
with social disorganization. They found interpersonal motives given
by over half while the most frequent reason given was disturbed family
relations. Shneidman and Farberow (1961) found marital difficulties
and depression given as reasons for both sexes with financial and
employment difficulties added for men. Ill health was given as a
reason less frequently than it was among suicides. Psychodynamic
studies reveal a higher proportion of "auto-plastic," or inner directed
motives in suicides, with more "alloplastic" motives in suicide
attempters ". . involving an appeal for help and efforts to manipulate
others in order to be rescued from their suffering" (Dorpat & Ripley,
1967, p. 74). Maris (1969) compared suicides and suicide attempters
in New Hampshire and found attempters were most frequently young females
who were divorced or separated, had problems with work, changed jobs
frequently, were not very successful and accomplished few life goals.
They were more likely to be from broken homes and were characterized as
more dependent personalities. They were more socially involved than
the suicides who were characterized as more often being socially
isolated, independent, held jobs longer, had accomplished more life
goals and were regarded by others as more successful.
In a study which matched hospitalized female suicide attempters
with depressed patients, Weissman, Fox, and Kerman (1973) found the
attempters were distinguished by manifest hostility, pervasive and
overly hostile relationships, poor long term work history, antisocial
behavior, and were demanding and hostile during the interview.
Lukianowicz (1973) studied a similar population of female attempters
and found that, with the exception of the psychotic and psychopathic,
they were "goal" and "gain" directed and the attempts were aimed at
changing the environment to the attempter's benefit. He found a
dramatic increase in suicide attempts in the last decade, as did
Weissman (1974) who reports hospital admissions for attempted suicide
rising, especially among youth. She speculates on a delayed increase
in suicide rates as this group increases in age.
Dorpat and Ripley (1967) reviewed 15 follow-up studies of suicide
attempters which report incidence of suicide and found that suicide
risk is highest in the first two years following an attempt. They
estimate the incidence of suicide among suicide attempters to be 10 to
20 percent. Tuckman and Youngman (1963a; 1963b) followed 1112
attempters for one year and found suicide rate to be 140 times the rate
of the general population for this period. Later follow up revealed
two percent suicide the first year and one percent the second year
following the suicide attempt (Tuckman and Youngman, 1968). Motto
(1965) estimated eventual suicide among attempters to be 80 to 100
times the rate of the general population. Therefore, although
differences can be shown when comparing suicide and suicide attempt
populations, it is important to any strategy of suicide prevention to
understand the overlap, or similarities in suicide and suicide attempt,
and to differentiate levels of risk among suicide attempters (Segal &
Humphry, 1970). Tuckman and Youngman (1963a) found that suicide rates
(per 1000) for attempters in the year following the attempt was a
compound function of three demographic characteristics: age, sex, and
race. Characteristics associated with higher suicide risk were: being
white, male, and/or over the age of 45 years. Those with none of these
characteristics had a 0.0 rate of suicide; those with one characteristic
had a rate of 8.55; those with two characteristics had a rate of 16.21;
and those with all three had a 44.12 suicide rate. The rate of the
general population was 0.14 per 1000 people. They conclude: ".
among the attempted suicides the more closely individuals approximate
completed suicides with respect to sex, race, or age, the higher their
suicide risk. The data also suggest that risk is accentuated by the
compounding or cumulative effect of the three characteristics" (1963a,
p. 587). In a follow-up study they found fourteen factors which were
combined into a scale. A cut-off score of four yielded two groups with
suicide rates of 0 and 35.20 per 1,000 population. The two most
differentiating characteristics besides sex and age were living
arrangements and method employed in the attempt (Tuckman and Youngman,
1958a and 1968b).
The incidence of suicide is a matter of public record; but the
incidence and prevalence of attempted suicide is difficult to ascertain.
Most studies of suicide attempters have utilized selective samples,
such as a consecutive series of medically treated attempters in a single
private hospital. One indication of the proportion of the population
of suicide attempters such a sample obtains is to take the ratio of
the sample to the number of suicides from the general population for
the same period. In the literature, estimates have been made of the
"true" ratio as being anywhere from 5:1 to 15:1 (Ruegseggar, 1963),
with most estimates around 10:1. Utilizing such estimates, prevalence
of suicide attempts is extrapolated by Dublin (1963) as being about
one percent. Stengel (1968) estimated the number of attempts in the
United States annually as up to 166,000 by 120,000 people, with per-
haps 2 million people having attempted suicide at some time in their
lives. Mintz (1970) estimated a prevalence as high as 2 1/2 percent,
or 5 million. Mintz bases his estimates on survey data from Los Angeles.
Other population sample survey studies have found high rates of claimed
suicide attempts, but too small absolute numbers to make generalizations
(Paykel, Myers, and Lindenthal, 1971; Schwab, Worheit, and Holzer, 1972).
Schwab et al. found approximately 12:1, suicide attempts to suicides,
in a sample survey of one Florida county. However, this is based on
only 10 positive responses in his sample of 1645 people. They found
2.7 percent claimed to have made an attempt sometime during their
lives, a prevalence similar to that which Mintz (1970) reports.
Locating suicide attempts is more difficult than determining
incidence through surveys, however. The first major effort to locate
all suicide attempts in one area over a period of time was made by
Shneidman and Farberow (1961). They obtained all public hospital
emergency room medical records for one year and also sent question-
naires to all physicians and osteopaths in Los Angeles County. From
their responses it was determined that approximately 5906 suicide
attempts had taken place in this community of 5 million,-where 768
suicides had occurred. This yields a ratio of 7.7:1, suicide attempts
for each completed suicide for this population. However, the total
number of attempts for which they had sufficient information to make
generalizations was 2652, yielding a ratio of 3.3:1. Parkin and
Stengel (1965) used all hospital admissions, both private and public,
and reports from general practitioners to find 820 attempts in two
years in Sheffield, England, a population of approximately half a
million. During the same period 86 suicides occurred, for a ratio of
9.5:1. However, 639 attempts had sufficient information for study, a
ratio to suicides of 7.4:1, the highest reported in the literature.
Both of the above studies concluded that the true ratio would be
higher by an unknown factor if attempts that did not come to medical
attention could be included. Other studies which utilized hospital
admissions include Gold who reports a ratio of 3.7:1 (1965) and Edwards
and Whitlock who found a ratio of 4.2:1 (1968). One study (Bergstrand
and Otto, 1962) of a sub-population of attempters found a ratio of
16:1 among adolescents. However, adolescents are known to have a
higher incidence of attempts to committed suicides than the general
population (Jacobziner, 1965).
Murphy et al., (1971; Clendenin & Murphy, 1971) took a different
strategy in compiling information on suicide attempts in a general
population. They devised a standardized police report form for suicides
and suicide attempts which was utilized in St. Louis County. For the
year 1968 they report 336 attempts and 58 suicides, a ratio of 5.8:1.
A similar ratio was found by Freeman et al. (1974) for a smaller county
in Florida over a 30 month period utilizing police reports and records
of the Suicide and Crisis Intervention Service. Tuckman, Youngman, and
Bleiberg (1962) found information for 1251 suicide attempts in Phila-
delphia, a ratio of 3.2:1, utilizing police reports. This is also the
only study which reports suicide attempt rates. The total population
of attempters was broken down by sex and race and suicide attempt rates
given per 100,000 population. These decreased as age increased for all
groups, and in a steep monotonic gradient for all but white males.
The Role of Intention-to-Die in Self-Injurious Behaviors
From the above literature review one can conclude that suicide
and attempted suicides come from two separate, but overlapping, popu-
lations (Wilkins, 1967; Stengel, 1964; Freeman et al., 1974). Within
each category, researchers have attempted to further differentiate
these behaviors (Stengel & Cook, 1958; Herdin, 1950). The concept of
suicide attempt can lead to confusion when it is applied to widely
discrepant behaviors. For example, both of the following hypothetical
cases are presently labeled as suicide attempts: a case where the vic-
tim ingested ten aspirin in the presence of their spouse; and, a case
where the victim drove to an isolated spot in the country, severely
wounded himself in the chest with a firearm, and was subsequently
rescued through the chance intervention of a passing hunter. McGee
and Hegert (1966) underscore this conceptual confusion and conclude:
It is evident . that suicide is not a
dichotomous behavior by which the participants
in the act can be meaningfully separated into
categories denoting whether or not they actually
expired . It is important to note that the
populations of people who participate in various
types of suicidal behavior are in fact different
populations, which are graduated along a continuum.
Even within the total group that expires, there
are still degrees of 'suicidality' based upon
method of injury, and on demographic variables
of age and sex (1966, p. 9).
Stengel (1968) utilizes the concept of suicidal intent in evalua-
ting these behaviors. He concludes that many suicide attempts have no
intent-to-die and many have ambiguous motivation:
Many suicidal attempts and quite a few suicides
are committed in the mood of 'I don't care whether
I live or die,' rather than with a definite and
unambiguous determination to end life. Most
people, in committing a suicidal act, are just
as ambivalent and muddled as they are whenever
they do anything of importance under emotional
stress. This is why many people who honestly
deny that they really wanted to kill themselves
admit that they did not care whether they lived
(1968, p. 172).
Stengel further delineates factors which determine whether a
suicide attempt becomes a suicide: chance factors due to intervention
or the breakdown of the plan; the method employed; and the "social
constellation" at the time of the attempt. This uncertainty of outcome
is labeled the "gamble with life" resulting from the multiple, and
many times contradictory, motivations of the victim. Stengel labels
this ambivalence the "double vector" in suicide attempts (1968).
Shneidman and Farberow (1961) distinguish between: those who really
want to die; those who leave survival to chance; and those who definitely
expect to be saved. They were able to classify the attempters in their
study as being almost equally divided among the three groups. Shneidman
(1968) labels these three orientations towards one's own death as:
intentioned; subintentioned; and contraintentioned or unintentioned.
In relation to suicidal or self-injurious behavior, Freeman et al.
. Persons who make self-inflicted injuries do
so within a set of specially contrived circumstances
which they have deliberately -- perhaps not
consciously -- created for the purpose of either
providing for, permitting, or preventing their
own rescue. Thus, persons who provide for their
own rescue have low intentionality, those who
permit a rescue have moderate intentionality,
and those who seek to prevent a rescue may be
seen as having high intentionality (1974, p. 23).
Tuckman and his associates (Tuckman and Lavell, 1958; Tuckman and
Youngman, 1963a; 1968a; 1968b) have found that the self-report of the
attempter as to his or her intent-to-die is actually related negatively,
albeit weakly, to suicide risk. In other words, the investigator into
suicide attempts cannot rely on the self-report of the victim to
differentiate attempters in any meaningful way.
An alternative strategy is to utilize judged seriousness of the
attempt in order to categorize attempts. Dorpat and Boswell (1963)
developed a five-point rating scale to evaluate the seriousness of the
attempt. Ratings of "1" represented a suicide "gesture," "3" an
ambivalent suicide attempt, and "5" a serious suicide attempt. "Suicide
gesture was defined as behavior indicating a pretense of suicide in
which there was no intent-to-die" (1963, p. 117). They utilized both
the statements of the patient and an evaluation of the method in their
judgments and found 20 percent suicide gestures, 60 percent ambivalent
attempts and 20 percent serious attempts. When these groups and a group
of suicides were compared, the average age and the sex ratio of male:
female increased monotonically with increased seriousness. The "gesture"
group was described in relation to the serious group as containing fewer
isolated individuals, demonstrating less premeditation, and as being
with someone much more often at the time of the attempt:
In the gesture group the action was directed
almost entirely at effecting some change in
others. More serious self-destructive motiva-
tion was observed in the ambivalent group whose
action was meant to bring not only suffering
to the patient but rescue and help from others
. a kind of gamble with death . The
serious suicide attempt and completed suicide
groups showed little concern about rescue or
directing change in others (1963, p. 123).
This, and further studies (Dorpat & Ripley, 1967), led Lester (1970)
to his succinct conclusion that suicidal behaviors fall on a continuum
of seriousness and that extrapolations can be made on the basis of this
Weisman (1970) and Weisman and Worden (1972) developed a procedure
for assessing the "lethality of implementation" in a suicide attempt
which is a function of two ratings: the degree of self-inflicted damage,
or "risk;" and the resources for "rescue" in the environment. Ratings
of these two dimensions were utilized in an arbitrary formula of risk-
rescue scores which ranged from 17 to 83. This total score is seen as
a representation of the continuum of lethality possible in suicide
Weisman proposes that any suicidal event, a self-injury regardless
of whether the result is death, is composed of 1) ideation, 2) imple-
mentation, and 3) intervention.
Implementation refers to more than just the
instrument or agent that a suicidal patient
uses . We should be able to recognize the
options open to him, his style of communication
with others, and his available and accessible
rescuers within the inner sphere of his rela-
tionships . Consequently, the edge of life
and death that a person inserts in his suicide
attempt should express a singular relation
between the risk of death and the potential
rescue operations (1970, p. 17).
Freeman et al. (1974) developed a scale designed to assess the
intention-to-die of the person in a suicide or suicide attempt. Their
assumption, which was discussed earlier, was that the victim chooses
the circumstances surrounding the event in order to provide for, permit,
or prevent his own rescue. "The specially contrived circumstances
which are of interest in making this assessment are: 1) the reversibility
of the method of self-destruction, and 2) the . probability of
intervention by others in the victim's environment" (1974, p. 23).
Intention-to-die, then, is advanced as the continuum upon which
self-destructive acts fall which is necessary for any understanding of
these events. Although seriousness of the attempt is a correlate of
intention-to-die, intention-to-die involves the state of the individual
immediately prior to the act and is inferred directly from behaviors
on the part of the victim over which the victim has some control. These
are rated on two separate scales, the Reversibility of Method Scale
(Appendix A) and the Probability of Intervention Scale (Appendix B),
which are each ". . 5-point ordinal scales designed to accommodate
and represent the circumstances surrounding an individual's suicide
attempt" (1974, p. 25). The Reversibility of Method Scale measures the
probability of stopping, or reversing, the action once set into motion.
An example of a method of "complete" reversibility is the ingestion of
small amounts of commercial drugs, while an example of a method of
"remote" probability of reversibility is a self-inflicted gunshot
wound to a vital area. Intermediate ranges include "probable,"
"questionable," and "improbable" reversibility. Both the method and
the degree to which it is employed are accounted for in the scale. The
Probability of Intervention Scale takes into account the degree to
which the ". .. victim can expect someone to become aware of the
event, to recognize it as an attempt, and to intercede . ." (1974,
p. 26), and is a function of the proximity and expected proximity of
other people. An example of a rating of "certain" intervention is
when the act is committed in the presence of one's spouse, while an
example of "remote" chance of intervention is when the victim makes
his attempt in an isolated, non-populated area where communication
with the rest of the world would be difficult. Intermediate ranges
include "probable," "ambiguous chance of,' and "improbable" intervention.
These two scales were combined as axes of the Intention-to-Die
Matrix (Appendix D) which is utilized in determining degree of intention-
to-die. A major contribution of the authors in this study was the
empirical validation and quantification of the concept of intention-to-
die. While Weisman and Worden (1972) arrived at an arbitrary index of
intention-to-die in order to quantify the concept, Freeman, Wilson,
Thigpen, and McGee defined intention-to-die as . the probability
that death will occur as a consequence of the circumstances in which a
self-injury event occurs" (1974, p. 39). Their sample of 243 suicide
attempts and 34 completed suicides were all rated and multivariate
analysis employed to ascertain the probability of dying, or intention-
to-die, given the ratings of the two scales. Each cell of the Intention-
to-Die Matrix was associated with a quantified index of intention-to-die,
the probability of dying given those circumstances. Low intention-to-
die was associated with cells with probability of dying approximately
0 (65 percent of the attempts and no suicides). Moderate intention-to-
die was defined for this study as between .05 and .30 probability of
dying (26 percent of the attempts and 21 percent of the suicides), while
high intention-to-die was associated with cells having greater than
.30 probability of dying (9 percent of the attempts and 79 percent of
All completely reversible methods were associated with .low intention-
to-die and all irreversible methods were associated with high intention-
to-die (Appendix D). In other words, at the extremes of the Reversi-
bility of Method Scale probability of intervention did not differentiate
as to overall intention-to-die. The probability of intervention score
was found to be important in differentiating levels of intention-to-die
at intermediate ranges of reversibility of method.
The Relationship of Suicide Risk Components: A Theoretical System
Wilson (1974) and Freeman et al. (1974) suggest that the intention-
to-die of self-inflicted injuries is an important variable in the
assessment of suicide risk. They suggest that an epidemiological study
of the role of intention-to-die in self-injurious acts would be an
important contribution toward developing an understanding of self-injury
and a technology of suicide prevention. They advance quantifiable con-
ceptualizations of the components of suicide risk and suggest these be
applied in epidemiological studies to ascertain incidence within sub-
Many epidemiological studies of suicide rate have been reported.
These have led to an operational definition of suicide risk (SR) as:
1. SR = frequency of suicide death for the population + chance factors
population size X time
A future event, suicide risk, is projected through the known suicide
rate. Only recently were good epidemiological studies of suicide
attempt rate accomplished. Only one, Tuckman, Youngman, and Bleiberg
(1962), goes beyond the description of incidence to utilize an opera-
tional definition for suicide attempt rate which can serve as an indica-
tor of suicide attempt risk (SAR):
2. SAR = frequency of suicide attempts for the population + chance
population size X time factors
If self-injuries are defined as any self-destructive act, then
self-injury rate can be determined by summing suicide rate and suicide
attempt rate. Therefore, self-injury risk (SIR) is the sum of suicide
risk (SR) and suicide attempt risk (SAR):
3. SIR = SR + SAR
frequency of self-injury behavior for the
4. SIR = population + chance factors
population size X time
Intention-to-die was previously operationally defined as the
probability of dying as the result of a self-injury. Expected
intentionality is a prediction statement of the intention-to-die of the
victim given a future act of self-injury. As suicide risk can be
ascertained through epidemiological use of appropriate data for suicide
occurrence within groups, so cculd the expected intentionality of sub-
populations of self-injurers be ascertained (Wilson, 1974). Expected
intentionality (Ex.In.) is projected directly from the average intention-
to-die found for a group, just as suicide risk is projected directly
from suicide rate. Freeman et al. (1974) propose:
frequency of suicide death for the
5. Ex.In. = population + chance factors
frequency of self-injury behavior for
Expected intentionality and self-injury risk are each partial
statements, or components, of suicide risk:
6. SR = SIR X Ex.In.
frequency of self-injury behavior
7. SR = population size X time X
frequency of suicide death + chance factors
frequency of self-injury behavior
1. SR = frequency of suicide death + chance factors
population size X time
This theoretical system has important implications both in dealing
with groups of people and in dealing with individuals. As all of its
major elements (suicide risk, self-injury risk, and expected intentionality)
are construed as probability events, it lends itself to more precise
prognostications of future behaviors based on past events. For example,
the clinician can reformulate his more general predictions of "high"
or "low" suicide risk to more precise probability statements of not
only suicide risk, but its components, self-injury risk and expected
intentionality. If the probability that a self-injury event of any
kind will take place for the individual "A" is 1 in 4 and the probability
that death will occur as a result of the occurrence of the event
(expected intentionality) is 1 in 5, then suicide risk is 0.05.
Individual "B,' with the same suicide risk of 0.05, may have very
different degrees of risk of self-injury and expected intentionality.
For example, he could have a self-injury risk of 0.10 and an expected
intentionality of 0.50. The clinician would base these predictions on
clinical and population base rates and would have a better conceptuali-
zation of the dimensions of risk associated with each individual.
Similarly, a program planner may find that different strategies
of intervention and prevention need to be employed with populations
that have been lumped together in the past as having near equivalent
suicide risks but which demonstrate very different degrees of intention-
to-die and rates of self-injury. Hypothetically, for the given locality
it may be found that middle-age Negro males and adolescent white
females have similar suicide rates. However, the former group may
have a very low self-injury rate and display a very high degree of
average intentionality, while the latter group had 100 times the
self-injury rate but with very low average intention-to-die. Given
these added dimensions, very different strategies would be called for
in any preventative programming.
The Study: Purpose and Hypotheses
Many studies have examined factors that differentiate suicide
attempters and suicides. The purpose of this study is: to assess the
levels of intention-to-die of the self-injurious acts of a general
population; to determine the expected intentionality, self-injury risk
and suicide risk for relevant component sub-populations; and to apply
multivariate analyses to assess the role of intention-to-die in self-
injurious acts as a function of demographic and personal variables.
Dorpat and Ripley recommended that: ". multivariate abstract
variance analysis methods be used to determine the patterns of attempted
suicide behavior that are related to suicide risk . (as previous)
research on attempted suicides has used only single-variable dimensions"
(1967, p. 78).
Hypotheses are derived from the cumulative suicidology literature
and through the application of the schematization of suicide risk,
self-injury risk, and expected intentionality which was reviewed and
The first hypothesis of this study is:
1) The majority of self-injurious behaviors found in a general
population will have essentially no intention-to-die associated with
these acts. Dorpat and Ripley (1967) and Shneidman and Farberow (1961)
judged intention-to-die of attempters to be distributed approximately
equally between low, moderate, and high levels. However, the only
study of quantified intention-to-die in a general population found nearly
two-thirds of attempters in their sample to have essentially no intention-
to-die associated with their attempts (Freeman et al., 1974).
2) Although suicide risk will increase with age, self-injury risk
and suicide attempt risk will both decrease as a function of age within
each sex X social-status subpopulation. Only one study has assessed
suicide attempt risk in a general population (Tuckman, Youngman, and
Bleiberg, 1962). They found that suicide attempt risk decreases with
age monotonically, with the exception of males, where the decrease is
less dramatic and less even. From the literature, self-injurious acts
contain a communicative function. This function should be especially
important in self-injurious acts of younger people.
3) Self-injurious behaviors, both suicide and suicide attempts,
which evince moderate intention-to-die will increase for each sex-age
group as a function of developmental life crises. This pattern has
not been demonstrated previously as no study has attempted to control
for the confounding factors of different types of self-injurious be-
haviors. Suicide is seen as a multi-modal concept within the literature.
However, the majority of suicides within this culture fall within
Shneidman's category of catalogical suicides, where the individual is
neither suffering from a schizophrenic state, nor is he "rationally"
committing a surcease suicide to avoid an inevitable and unrelieved
future of excruciation, nor is he committing an altruistic or cultural
suicide where the culture recognizes his suicide as an heroic and
socially beneficial act. It is hypothesized that developmental life
crises will increase the risk of suicide in an individual. High in-
tentioned self-injurious acts should more reflect all types of suicide;
moderately intentioned acts should reflect the ambivalence associated
with life crises; while low intentioned acts should most reflect inter-
personal crises and a breakdown in communication.
4) Average intention-to-die for each sex X social-status subpopu-
lation will demonstrate peaks in the under 25, and in age groups over
45. Overall increases will occur as a function of age. Related to the
third hypothesis is the effect of life crises on average intention-to-
die. Although confounding factors will obliterate some of the effects,
developmental life crises should be more reflected in changes in
average intention-to-die over the life span than in changes in suicide
rates. The cumulative effect of poor adaptation to developmental life
crises should increase average intention-to-die of self-injurious be-
haviors, with peaks immediately following the ages when they mostly
5) Subpopulations which are similar on one factor, either suicide
risk, self-injury risk, or expected intentionality, will be differen-
tiated by the other two factors as a result of the different functions
of each factor for each subpopulation. For example, it is expected
that middle-aged white males of different social-status will be found
to have similar levels of average intention-to-die, but will be
differentiated by self-injury risk, and as a result, suicide risk.
Although suicide is not taken as an option as often by middle-aged
and upper social-status white males, if they do decide to make a self-
injury it will be demonstrated by high intention-to-die, as would be
a middle-aged lower-social-status white male's, as this act serves
little communicative function for either group.
6) Living alone will increase expected intentionality, especially
in all but the older age ranges for each sex X social-status subpopu-
lation. Besides age, sex, and social-status, multivariate analysis
will reveal the living arrangement of the individual to be related to
suicide risk. This has been documented in the literature. However,
it will more effect intentionality than self-injury risk because of
the greater communication function of the latter.
The subject population for this study consisted of all attempted
suicides and suicides for a two year period in a majority of the muni-
cipalities of St. Louis County, Missouri that came to the attention of
the police. "These reports include not only the cases the police are
called upon to handle but also those reported to them by hospitals"
(Clendenin & Murphy, 1971). This consecutive series of systematic
police reports were analyzed and rated utilizing a revised version of
the Intention-to-Die Scales developed by Freeman et al. (1974).
The expected intentionalities associated with each cell of the
Intention-to-Die Matrix were ascertained through Stepwise Regression
Analysis of suicide death as a function of scores on the Probability
of Intervention Revised Scale and the Reversibility of Method Scale
of all self-injury cases in the sample. Suicide attempt risk, suicide
risk, and their sum, self-injury risk, were obtained for each subpopu-
lation of interest by dividing the incidence found by the subpopulation
size times two years.
The subpopulations of interest were sex X social-status groups
over the age span. The non-white subpopulations were too small to
analyze. Expected intentionality and risk scores were derived for each
of these subpopulations for 5 year age groups over the life span. For
the overall sample and for each sex, multivariate analyses were utilized
to determine the relationship of a number of different variables to
intention-to-die. These independent variables included age, health,
living conditions, marital status, and education.
St. Louis County is comprised of nearly a hundred municipalities
and had a population of approximately 910,000 people during the period
of this study. It is an urban and suburban area of 406 square miles
which borders the three land sides of the City of St. Louis, which is
politically and administratively separate. St. Louis County has its
own commercial and governmental center and its population is diverse
in social class makeup, although it has an overrepresentation of the
upper end of the socio-economic spectrum. From the U.S. Bureau of
Census Classification of Occupational Status, 31 percent of the employ-
ment of county residents is in upper status occupations, 57 percent
middle status, and 13 percent lower status (1962, 1972). Just under
5 percent of the county is black.
The standardized police report form was developed for the St.
Louis County Coroner's Office by Murphy et al. (1971) and adopted in
1967 for use by police in the investigation of attempted and completed
suicides. The form was designed to include social and personal items
that are known to be associated with suicide (Murphy et al., 1971).
In 1968, 408 reports, and in 1969, 459 reports, were forwarded from
the police (Clendenin & Murphy, 1971). Suicide attempts for the two
year period totaled 714 by 686 different people, while 126 completed
suicides and 25 suicide threats were reported. Two reports concerned
deaths judged as accidents by the author. Nearly all the attempters
and threateners were seen by a physician: For 1968, ". . 53 percent
(were seen) at a private hospital, 43 percent at a public hospital, and
2.3 percent at a . private office" (Murphy et al., 1971, p. 100).
These reports were earlier used in one article on the demographic
differences between wrist cutters and other attempters (Clendenin &
Murphy, 1971) and in an article describing the police report form, with
a demographic descriptive breakdown of the 336 attempters reported in
1968 (Murphy et al., 1971).
Not all municipalities in the county cooperated and consistently
reported self-injuries. Seven municipalities with a total population
of 23,210 did not adopt the standardized report form and did not report
any self-injuries during the two year period of the study. An additional
ten municipalities with a total population of 152,324 reported only
suicides, or grossly under-reported suicide attempts. Therefore, the
20 suicide attempts and eleven suicides reported from these municipalities
and two suicide attempts from the City of St. Louis were dropped from
The final sample consisted of 807 reports of self-injuries, 115
suicides and 692 suicide attempts, from a population of 738,904 for a
two year period. The ratio of reported suicide attempts to suicides
was 6.0:1. The area represented comprises 81 percent of the population
of St. Louis County and is approximately 3.7 percent black.
Revision of the Probability of Intervention Scale
Freeman et al. (1974) noted that the Probability of Intervention
Scale contributes less than the Reversibility of Method Scale to the
variance of Intention-to-Die scores. After a sample of the reports
were rated for reliability purposes it was noted that several categories
seemed to arbitrarily combine behaviors that might be related very
differently to the probability of death given these behaviors. For
example, the third category contained cases where the victim called by
telephone to report the self-injury during or immediately following the
act, and cases where the victim was alone, but anticipated the arrival
of someone who could intervene. The former involves a more active role
in one's own rescue than the latter.
The second category contained cases where the victim was in his
own basement with the family upstairs asleep but did not initiate his
own rescue, and cases where the victim walked out of his bathroom to
announce to the family he had just attempted "suicide." These cases
demonstrate differences in the activity level of the victim which are
not reflected adequately in a scale that so heavily emphasizes actual
or potential proximity of others. Therefore, it was decided to revise
the scale to attempt to increase the variance attributable to the
probability of intervention in the self-injury.
The Probability of Intervention Revised Scale is a 6 point ordinal
scale designed to emphasize both the actual or potential proximity of
possible intervenors and the activity or passivity on the part of the
victim in mobilizing this intervention (Appendix C).
One month of reports were picked for the reliability sample. An
additional person independently rated these 40 reports (8 suicides and
32 suicide attempts) without knowledge of the ratings of the primary
rater. Pearson correlations for agreement were obtained. For the
Reversibility of Method Scale, r = .938, while r = .898 for the
Probability of Intervention Scale, and r = .850 for the Probability
of Intervention Revised Scale. The latter two reliability scores
exceeded the r = .80 for the Probability of Intervention Scale reported
by Freeman et al. (1974). However, the revised scale demonstrated a
slight sacrifice in reliability from the original in this study.
Probability of Dying
Stepwise Regression analysis was utilized to ascertain the
probability of dying associated with each cell of the Intention-to-
Die Matrix. The occurence of death was the dependent variable while
the rating for each scale, their interaction, and quadratics were
submitted as potential independent variables. These were selected in
a stepwise fashion while overall F and F to add or delete remained
significant. The resultant regression equation was utilized to generate
values for each cell.
Correlates of Intention-to-Die
Stepwise Regression Analysis was utilized to determine the
relationship of dependent variables individually and as a group to
intention-to-die. Thirty dependent variables came directly from the
police reports and were recorded in the forms of ordinal or binary
variables. The nominal variable of marital status was reduced to three
values which each became a binary variable: single, never married;
divorced, separated, or widowed; and married, living with spouse.
Occupational-status was utilized as an indicator of social-economic-
status (i.e., social-status). This variable was derived from four
separate variables which were not independently included in the regres-
sion analysis: current occupational status; former occupational status;
current occupational status of the principal wage earner; and former
occupational status of the principal wage earner. The highest value
among these four derivative variables became the value of social-status.
Social-status could take three values: high, middle, and lower. These
were determined utilizing the U.S. Bureau of Census occupation categories
and occupational-status classification system (1962, 1972).
The Stepwise Regression analyses with intention-to-die as the
dependent variable were run for the entire population and for white
females (n = 522) and white males (n = 266). The dependent variable
list was reduced to exclude four variables with excessively small
standard deviations. These were race, where 97.6 percent of cases
were white, and three variables concerning types of police records.
Separate runs were accomplished excluding the variables of education,
occupational status, and living conditions. These variables each
had missing data which reduced the sample a total of 37 percent when
they were all included.
Risks and Expected Intentionality of Subpopulations
Subpopulations of interest were sex X social-status groups over
the age range. For each subpopulation the average (i.e., mean)
intention-to-die was determined at each age level. Suicide attempt
risk, suicide risk, and their sum, self-injury risk, were obtained
for each subpopulation of interest by dividing the incidence formed
by the subpopulation size times 2 years.
The percentage of self-injurious behaviors in the sample found to
have low intention-to-die (i.e. less than .05) was determined to see
if it was a majority of cases.
For each subpopulation the Kolmogorov-Smirnov One Sample Test was
applied to determine if self-injury risk and suicide attempt risk de-
creased significantly as a function of age for age levels over 15 years.
Number of moderate intention-to-die self-injuries from two age
groups which should not be as typified by developmental life crises
(i.e. ages 30-34 and 35-39) were compared to the frequency among two
age groups that should more be typified by developmental life crises
(i.e. ages 20-24 and 40-44) using the X2 One-Sample Test. Moderate
intention-to-die was defined as greater than or equal to .05 and less
than .35 probability of dying.
For each subpopulation average intention-to-die was plotted and
the curve investigated for peaks before age 25 and after age 45. The
correlation of age and intention-to-die was determined for each sex.
Similar levels of factors which were demonstrated by different sub-
populations were noted and the groups compared on other factors. For
example, age groups among different subpopulations with similar levels
of self-injury risk were compared for average intention-to-die and
Analyses of Variance were executed for each sex to determine
whether living alone increased intention-to-die and whether any con-
tribution of living alone was significant in conjunction with the
factor of age. In addition Stepwise Regression Analyses with intention-
to-die as the independent variable and living alone and age as dependent
variables were executed for each sex.
The frequencies of ratings corresponding to each cell of the
Intention-to-Die Matrix are shown separately in Figure 1 for the 115
suicide cases and the 692 suicide attempt cases. For example, cell
3-6 demonstrates that 5 suicide attempt cases and 6 suicides were
rated 3 on the Reversibility of Method Scale and 6 on the Probability
of Intervention Revised Scale. Figure 2 demonstrates the percentage
of suicides among the self-injuries in each cell. For example, for
the previously mentioned cell 3-6 there were 55 percent suicides, or
6 of 11 self-injuries.
Stepwise Regression analysis yielded a formula from the ratings
for the probability of death, or intention-to-die. Figure 3 demon-
strates the resulting intention-to-die associated with each cell of
the Intention-to-Die Matrix. Cells are grouped as to whether they
represent high, moderate, or low intention-to-die. High intention-to-
die cells are defined as those which demonstrate a probability of
death greater than or equal to .35, low intention-to-die cells demon-
strate probability less than .05, and moderate Intention-to-die cells
are greater than or equal to .05 and less than .35 probability of dying.
Over 60 percent of the self-injuries were classified as low intention-
to-die, confirming hypothesis 1. Over 47 percent of the total
C 1- ))
C:) 0 CD
C C) C) C:) -
SpoCa m LLq
POL|4@ 0 4.[!qLSJ.A0a
u 11 0
u II 0
- .y *r-
Probability of Intervention
1 2 3 4 5 6
1 0 0 0 0 0
-' ~ i ~
S 2 0 0 .02 0 .33
0 0 .02 .19 .20 .55
o ~ o~--1-- ; ~ "~
4 0 .25 0 .55 .88 .67
50 .25 .75 .94 91 .88
Figure 2. Percentage of suicide death for each cell
of the Intention-to-Die Matrix.
Probability of Intervention
Figure 3. Probability of dying for each cell of
the Intention-to-Die Matrix.
3 4- 0 I
J -0 4-1 4-)
self-injuries were in cells with 0 probability of dying. Only 1 suicide
was among the 490 low intention-to-die self-injuries.
Characteristics of Self-Injurers
Women made up 66.0 percent of the self-injuries: accounting for
70.4 percent of suicide attempts and 40.0 percent of suicide deaths.
Blacks, who made up 3.7 percent of the sample, accounted for only 2.3
percent of suicide attempts and 1.7 percent of suicides. Of the 65.6
percent of the sample where social-status could be determined, 32.7
percent of self-injurers were lower, 40.1 percent were middle, and
27.2 percent were upper social-status. This compares to base rates,
respectively, of 12.5, 56.8, and 30.7 percent. In other words, lower
social-status individuals have a greater probability of self-injury
than either middle or upper social-status individuals. This difference
is significant and is reported later with other results concerning
The following factors significantly differentiated suicide
attempters from suicides: being female; being younger; having some
high school, but no degree; having very acute problems; being lower
or middle social-status; and drinking at the time of the incident
(all X2 less than .025 probability). Being single approached signi-
ficance as did being employed for males, not being under recent
physician's care, and not having a recent hospitalization. These
factors non-significantly increased the likelihood of living given
The simple correlations of each of the dependent variables to
intention-to-die are given in Table 1. These are ordered as to their
Table 1. Simple correlations of dependent variables and intention-to-
die for the total population and by sex.
Dependent Variable Overall Male Female
Age .411** .461** .386**
Single, never married -.129** -.197** -.184**
Acuteness of problem -.100** -.051 -.136**
Drinking at time -.096** -.228** -.058
Under drug therapy -.087* -.104 .005
Education level .069 .000 .079
Recency: Visit to physician .066 .070 .094*
Physician's care within month .065 .057 .120**
Married, living with spouse .064 .135* .094*
Divorced, separated, widowed .058 .063 .069
Living alone .057 .058 -.057
Social-status .057 .106 .035
Prior attempt .050 -.099 .019
Hospitalization within month .044 -.027 .081
Prior attempt or threat -.043 -.111 .040
Acute depression -.043 -.028 -.045
Recency: hospitalization .041 -.037 .103*
Police record: drunkenness -.030 -.081 -.042
Police record -.028 -.100 -.074
Recency: prior attempt -.027 -.106 .061
Recency: prior threat -.023 -.095 .040
Depression .023 .080 .017
Prior threat .022 -.090 .046
Police record: conduct -.018 -.096 -.052
Nervous condition -.006 .017 .020
magnitude for the overall sample. In addition, correlations are shown
for white males (n = 266) and white females (n = 522). Age and sex
demonstrate the greatest magnitude of correlations. Other variables
differentially relate to intention-to-die according to the sex of the
victim. The vast majority of factors demonstrate very low order re-
lationships with intention-to-die.
Multivariate Analyses of Intention-to-Die
Multiple relationships of the dependent variables with intention-
to-die were assessed utilizing a series of Stepwise Regression Analyses
for the overall population, white females, and white males. Table 2
shows the summary table of the Stepwise Regression Analysis for the
overall population. Although six variables each maintained significant
Fs to add or delete and a significant overall f6,800 = 54.92, two
variables contributed the greatest portion of the variance. Knowing
the sex and age of the victim accounts for 24.8 percent of the variance
of intention-to-die in a self-injury. The other four variables con-
tribute an additional 4.4 percent for a total of 29.17 percent explained
variance, which is the sum of r2 change. Being male, older, not
drinking at the time of the incident, not being under drug therapy, not
having a police record for a conduct offense, and having seen a physician
within a month increased probability of dying in the self-injury. The
additional contributions of other variables were insignificant. Table
1 shows a significant negative relationship for being single, never
married and degree of intention-to-die. Its relationship with other
variables, including age, make its contribution insignificant when
considered in a multiple correlation. In a separate Stepwise Regression
Table 2. Multiple correlations stepwise with
total population (n = 807).
intention-to-die for the
Dependent Variable Mult. r r2 Change
Age .411 .169
Sex .498 .079
Drinking at time .524 .026
Under drug therapy .534 .011
Police record: conduct .538 .004
Physician's care within month .540 .003
Table 3. Multiple correlations stepwise with intention-to-die for the
white males (n = 266).
Dependent Variable Mult. r r2 Change
Age .461 .212
Drinking at time .511 .049
Under drug therapy .539 .029
Recency: hospitalization .545 .007
Recency: visit to physician .559 .015
Table 4. Multiple correlations stepwise with intention-to-die for
white females (n = 522).
Dependent Variable Mult. r r2 Change
Age .386 .149
Drinking at time .399 .010
Recency: prior attempt .409 .008
Under drug therapy .414 .005
Acuteness of problem .421 .005
Police record .426 .004-
Analysis where only demographic variables were used, the three factors
of age, sex and single, never married were significant, explaining 25.22
percent of the variance, although being single and never married con-
tributed only 0.43 percent to the total.
Table 3 shows the summary table of the Stepwise Regression Analysis
for white males. Five variables each maintained significant Fs to add
or delete and a significant overall F5,260 = 23.60. The variance ex-
plained was 31.22 percent, although age accounted for over two-thirds
of this total. Being older, not drinking at the time of the incident,
not being under drug therapy, and not having been in the hospital
recently but having recently visited a physician, increased the prob-
ability of dying in the self-injury of a white male in this sample.
Table 4 shows the summary table of the Stepwise Regression Analysis
for white females. Six variables each maintained significant Fs to add
or delete and a significant overall 56,515 = 18.98. The variance ex-
plained was 18.11 percent while age alone accounted for over 82 percent
of this total. Being older, not drinking at the time of the incident,
having a more recent prior attempt, not being under drug therapy, prob-
lems being less acute, and not having a police record increased the
probability of dying in the self-injury of a white female in this
Self-Injury Risk, Suicide Attempt Risk, and Suicide Risk
Although social-status did not relate significantly to intention-
to-die overall or for either sex (Table 1), it was retained as a factor
in determining subpopulations because of its significant relationship
to self-injury risk, suicide attempt risk, and suicide risk. Table 5
Table 5. Self-injury risk (SIR), suicide attempt
suicide risk (SR), per hundred thousand
sex by social-status level.
risk (SAR), and
population for each
SIR SAR SR
Overall 39.20 29.33 9.87
Upper 35.11 20.25 14.85
Middle 25.58 17.91 7.67
Lower 110.89 89.38 21.52
X2 92.83** 95.87** 10.09*
Overall 68.44 62.54 5.91
Upper 60.29 52.42 7.86
Middle 50.37 48.24 2.13
Lower 170.30 155.84 14.46
X2 109.29** 97.59** 16.67**
shows the levels of these risks for males and females by social-status
level per hundred thousand population. The statistic X2 for differences
between social-status levels was significant in all cases. Actual
frequencies were used in computing this statistic, rather than the rates
shown in Table 5.
The significantly higher risks for lower social-status males and
females over either middle or upper social-status groups would be masked
if only raw frequencies were reported and subpopulation sizes were not
considered. Although lower social-status self-injury risk was demon-
strated as 3.4 times as great as that for middle social-status, more
middle social-status self-injuries occurred. However, middle social-
status individuals made up the majority of the base population, and
this base group was 4.5 times the size of the lower social-status base
The other factors in determining subpopulations were sex and age.
Kolmogorov-Smirnov One Sample Tests were utilized to analyze the rela-
tionship of age with each type of risk, for age levels of 15 and over,
which are reported in Table 6. The difference between the cumulative
percentage of cases by age level was compared to the expected cumulative
percentage for each group. It was hypothesized that both self-injury
risk and suicide attempt risk would decrease over the age span for both
males and females (i.e. hypothesis 2). This is demonstrated in Table
6. However, the relationship was not significant for upper and middle
social-status males for self-injury risk, and for upper social-status
males for suicide attempt risk. In each instance, the relationship
was stronger for females, for the lower social-status, and with suicide
attempt risk. It was hypothesized from the literature (i.e. hypothesis
2) that suicide risk would increase, however. This was the case in all
instances except for upper social-status males, where suicide risk
actually decreased after middle age, but not significantly. When
analyzed by social-status groups, changes in suicide risk over age
groups were not found to be significant, partially as a function of the
smaller n of suicides.
Sex was also a significant factor in determining subpopulations.
The X2 for differences between male and female frequencies were 66.84
(p<.001) for self-injuries, 97.02 (p<.001) for suicide attempts, and
6.26 (p<.025) for suicides. Females demonstrated higher rates of
self-injuries and suicide attempts, but a lower rate of suicide.
Self-injury risk, suicide attempt risk, suicide risk, and expected
intentionality (average intention-to-die) by age level for each sex are
shown in Table 7. Self-injury risk, suicide attempt risk, and suicide
risk are expressed as rates per hundred thousand persons.
For males, expected intentionality, or probability of dying, in-
creases steadily and with increasing slope into old age. Only the 15-19
and 50-54 age ranges demonstrated slight declines over the previous age
range. For females, expected intentionality also rises steadily into
old age, but more slowly than for males. The slope increases in the
early forties. Only the 60-64 age group demonstrates a slight decline
over the previous age range. In the oldest age group, expected inten-
tionality approaches a one in three probability of dying for. females,
while it surpasses a three in four probability of dying for males.
For females, self-injury risk and suicide attempt risk are bimodal
curves with their greatest peaks in the 20-24 age range and a similar
peak in the 35-39 and 40-44 age ranges. The low point between the peaks
Table 6. Kolmogorov-Smirnov differences of expected and found cumulative
frequency percentages by age for self-injury risk (SIR),
suicide attempt risk (SAR), and suicide risk (SR) for each
sex by social-status level.
SIR SAR SR
Overall .21** .23** .17*
Upper .14 .18 .15
Middle .10 .20* .23
Lower .23** .32** .28
Overall .13** .24** .28**
Upper .22** .23** .25
Middle .28** .29** .25
Lower .24** .32** .30
10 k0 01 CO
m C -
co 00 C) C'
IT) ^t -
)0 0c r0 0
C C)O C) In m
o. C:) n -.
CM Cr r-- r-
LI LI l) U!
)I 'I LI U-
Ui Zj- M* 0
c0 r~~ C.
cM C an
r- 0L CM
-o c- Co
Kr CO Co
C0 L.0 CM!
C) U) C'
is the 30-34 age group, with lower levels of self-injury risk and
suicide attempt risk that are not again reached until about age 50. In
other words, for both these factors there is a sharp rise into the
early twenties which declines rapidly until the middle thirties where
it rises into the forties, and then decreases sharply. These curves
continue to decrease gradually from the fifties into old age. Suicide
risk for females is very low until the early forties, where it peaks
and decreases, with a smaller peak in the late fifties. The suicide
risk peak in the early forties for females is a function of high self-
injury risk, while the suicide risk peak in the late fifties is a
function of increased expected intentionality.
For males, self-injury risk and suicide attempt risk reach a
sharp peak in the early twenties, decreases rapidly, and then more
gradually into middle age. A slight increase in self-injury risk was
demonstrated in the late forties, after which the curve again decreases.
Suicide risk increases steadily for males until it peaks in the
late forties where it plateaus into old age. Declines over the previous
age group take place in the late twenties and early fifties. For males,
suicide risk is primarily a function of high self-injury risk in the
younger age groups. The increase of expected intentionality is counter-
vailed by a similarly sloped decrease in self-injury risk, creating a
plateau in middle and older groups in suicide risk.
Social-status affected the magnitude of the curves for risk levels
over the age span for both males and females. Tables 8 and 9 show, for
males and females respectively, self-injury risk, suicide attempt risk,
and suicide risk for social-status groups by age level.
Because of the smaller frequencies within age groups when
l- cM CM
r^ (A <-
cO (A t-.
Cn CD CM
<. 0 T
CO c- ^'
C(J i- i-
001 It r
CO *3 ^1
cc c- cc
m m ro
I. ,- CM
CSJ I- I'-
3 Co CM)
1-- tT (
0 ) 0
- c cc-
(A (A (
uO CM co
Mc LO CO
cc c (
"c cc 0
Lo LO i
CO C Mj
.r C^ ol
Co LC) C >
.,j- CM. C'M
1"- CM. (A
(A LO 0
(A A *-
^- Cc 10
Cs, ^J- Cc
cc r- (A
C) C) C)
C c0 L
C.) m Co
(A (A of (a (A
.: Q) -i Cn Q,
t/lv (/ ^ lml00(/
r- Ln MC
n ci cu
C'i 0 CMJ
c*) %D r-
i 2n C)j
co C r-
r- r- C
(A uA oA
social-status groups are considered, it would be misleading to utilize
Tables 8 and 9 as actuarial type tables, as Table 7 might be used.
Although some differences in pattern exist, it might be more accurate
to utilize items in Table 7 multiplied by a factor for the contribution
of social-status. This factor can be derived from Table 5. For example,
in Table 5 lower social-status male self-injury risk is 110.89 while
overall male self-injury risk is 39.20. Dividing yields a factor of
2.83. In order to estimate the self-injury risk of lower social-status
males in the 55-59 age range, the appropriate factor is applied to the
self-injury risk of males in this age range, 4.21 from Table 7, yielding
119.1 per hundred thousand. These factors appear in Table 10.
A drawback in the use of factors from Table 10 is that there are
some differences in patterns for the risk functions between age X
social-status groups. For lower social-status males, self-injury risk
is bimodal with a large peak from the twenties through the early
thirties, with a sharp decline to a nadir in the late thirties and
early forties age groups, and a rise to a smaller peak in the early
fifties. This pattern is most similar to female subpopulations, with
the exception that the nadir occurs earlier for middle and lower social-
status females. The other two male social-status subpopulations demon-
strate less clear cut patterns. Middle social-status males demonstrate
a peak of self-injury risk in the early twenties, declining to a nadir
in the late twenties and early thirties, which is followed by a smaller
peak and a saw-toothed pattern into old age. Upper social-status males
demonstrate a similar, but even less consistent pattern. The peaks in
upper social-status male self-injury risk are in the early twenties
and late forties.
Table 10. Factors for the contribution of social-status to self-injury
risk (SIR), suicide attempt risk (SAR), and suicide risk (SR)
for males and females.
SIR SAR SR
Upper 0.90 0.69 1.50
Middle 0.65 0.61 0.78
Lower 2.83 3.05 2.18
Upper 0.88 0.84 1.33
Middle 0.74 0.77 0.36
Lower 2.49 2.49 2.45
The first hypothesis, that the majority of self-injurious behaviors
found in a general population would demonstrate no intention-to-die, was
confirmed. Low intention-to-die cells of the Intention-to-Die Matrix
in Figure 3 were those associated with less than .05 probability of
dying. Over 60 percent of self-injuries fell in those cells.
The second hypothesis, that self-injury risk and suicide attempt
risk would decrease within each sex X social-status subpopulation
while suicide risk increased, was partially confirmed. Table 6 shows
Kolmogorov-Smirnov differences in cumulative expected from found per-
centages. Decreases over age in self-injury risk and suicide attempt
risk were significant for the overall male and female groups, and
significant for these risks for all female social-status subpopulations.
The decreases were not significant, however, for the male upper social-
status subpopulation or, for self-injury risk, for middle social-status
males. Increases in suicide risk were significant for males and females,
but not when divided into social-status subpopulations. The upper
social-status male subpopulation actually decreased, although non-
significantly, in suicide risk from middle to older age groups.
The hypothesis that moderate intention-to-die self-injuries will
increase during developmental life crises is not confirmed. To test
this hypothesis two age groups, ages 20-24 and 40-44, were picked which
were thought to be more prone to developmental life crises, and two
groups, ages 30-34 and 35-39, were picked which should be less prone.
The One Sample X2 for the difference between frequencies within each
were 1.1 for males and 0.75 for females, both non-significant. For
males, the reverse of the hypothesis was actually found, more moderate
intention-to-die self-injuries were found among the thirties age groups.
The fourth hypothesis was that: overall increases in average in-
tention-to-die for each sex X social-status subpopulation would increase
as a function of age; and peaks would occur in the early twenties and
late forties age groups. The first part of this hypothesis is confirmed
and the latter part is rejected.
The correlations of age to intention-to-diewere r = .41 (p<.001)
overall, r = .46 (p<.001) for males, and r = .39 (p<.001) for females.
No separate correlations were computed for social-status groups as
social-status did not turn out to be a factor which differentiated
intention-to-die for either sex. This was ascertained utilizing analyses
of variance which yielded F2,138 = 0.34 for males and F2,304 = .00 for
females (both p>.90) for the contribution of social-status to intention-
to-die, in two two-way analyses with age also a factor.
Intention-to-die for males increased steadily and increasingly
into old age. A small dip occurred in the early fifties age group,
but the curve was essentially smooth. For females, intention-to-die
increased for each age group until the late fifties. No peaks occurred
where hypothesized. Therefore, intention-to-die in self-injuries for
both males and females was shown to be highly and significantly age
related and not influenced by developmental life crises.
The fifth hypothesis was that groups which were similar on any
one factor among self-injury risk, expected intentionality, or suicide
risk, would be differentiated by the other two factors. This hypothesis
is largely confirmed. From Table 7, males in their early forties and
females over age 65 demonstrate respective expected intentionalities
of .29 and .31. However, self-injury risk and suicide risk for the
middle-age males is over twice that of the elderly females. Males over
age 65 and females in their early forties demonstrated similar levels
of suicide risk. But self-injury risk is over 4 times as great for the
middle-age females as for the elderly males, while expected intentionality
levels are .14 for the former and .76 for the latter. Finally, females
in their late teens and males in their early twenties demonstrated
similar levels of self-injury risk. However, expected intentionalities
were, respectively, .02 and .12 while suicide risks were 9.4 and 2.9
for the two groups.
The last hypothesis, that living alone would increase expected
intentionality, especially in all but the older age ranges, is rejected
for both males and females. For males, the correlation of intention-
to-die and living alone was non-significant, r = .06. Even this small
correlation is mostly a function of the small positive relationship of
living alone and age in males. A two-way analysis of variance yielded
a highly non-significant F1,113 = 0.001 for the contribution of living
alone to intention-to-die in males. In a Stepwise Regression Analysis,
F for living alone to be added following age was a similar F 176 =
For females, the correlation of intention-to-die and living alone
was also non-significant, r = .057. However, because of the positive
relationship of living alone and age, the contribution of living alone
to explaining intention-to-die is significant for the multiple correla-
tion including age, F to add or delete living alone was 1,330 = 3.81
(p<.05). However, the contribution of living alone, beyond that of
age, was only 1.0 percent contribution to the variance that could be
explained (i.e. change in r squared). A two-way analysis of variance
yielded a significant E1,302 = 7.30 (p<.01) for the contribution of
living alone to intention-to-die with age and the interaction also
significant. However, the relationship of living alone and intention-
to-die was found to be the reverse of that hypothesized. It was
hypothesized that the relationship would be positive and effect younger
over older people. What was found was that living alone significantly
lowered intention-to-die in older females.
The majority of self-injurious behaviors in this study were found
to be associated with essentially no intention-to-die. Self-injury risk
and suicide attempt risk were found to decrease significantly over the
life span for both males and females, and for most sex X social-status
subpopulations. Suicide risk increased significantly over the life
span for both males and females, but not when broken down into any sex
X social-status subpopulation. Expected intentionality increased
significantly over the life span for both males and females. Social-
status was not found to differentiate social-status groups for expected
Developmental life crises were not found to be a factor in expected
intentionality. The increases in intention-to-die were found to be a
function of its strong relationship with age. Nor did self-injuries
which evinced moderate intention-to-die increase in frequency for age
levels which were hypothesized as reflecting more developmental life
crises. Parenthetically, although no hypotheses were formulated for
these relationships, self-injury risk and suicide attempt risk both
demonstrated bimodal patterns for most sex X social-status subpopula-
tions which will be discussed later in tennis of possible developmental
life crises influences.
Living alone was not found to have the hypothesized positive rela-
tionship with expected intentionality. Rather, among elderly females,
those who lived alone, were found to have significantly lower expected
intentionality. This relationship, although significant, contributed
little to the overall explanation of the variance of expected intentionality
It was also hypothesized that populations that were similar for
either suicide risk, self-injury risk, or expected intentionality, would
be differentiated by the other factors, and this was demonstrated. This
hypothesis tested the utility of reconceptualizing suicide risk as a
function of self-injury risk and expected intentionality. If these
factors vary with some independence, then they add to the understanding
of suicide risk as a non-unitary phenomenon. For this sample, males
over age 65 and females in their early forties demonstrated similar
levels of suicide risk. But, for this age group of females, self-injury
risk was over 4 times as great as for the elderly males, while expected
intentionality was 5 times as great for the elderly males as for the
female group. This and similar comparisons between groups give the
researcher and program planner important insights into the varying
functions of self-injuries to different subpopulations. Important
differences between groups were demonstrated through this schematiza-
tion where previously the groups would be lumped together on the basis
of their similar suicide risks.
Freeman et al. (1974) found that about 20 percent of the variance
of intention-to-die could be explained utilizing stepwise regression
analysis. In this study the explained variance increased to over 29
percent. Over 31 percent of the variance of intention-to-die could be
accounted for in the male group, while, for the females, 18 percent was
Two factors might have increased this relationship for this study.
The first is the increased number of dependent variables and the second
might be a function of a more accurate assessment of intention-to-die.
This study utilized 26 dependent variables of which six were found to
relate significantly as a group to intention-to-die. However, age and
sex alone, with r2 of nearly .25, accounted for more total explained
variance than the total explained in Freeman et al. Having access to
data for more dependent variables than Freeman et al. added only 4.4
percent of explained variance to the total. Therefore, the increased
relationship found might be a function of increased sensitivity of the
assessment instrument. Although Freeman et al. do not report on the
correlation of intention-to-die and death for their sample, this was r
= .75 for this sample, or over 56 percent of explained variance. In
comparing the Intention-to-Die Matrices from both studies, a greater
range in scores is found in this study, especially as a function of
levels of probability of intervention. For example, from Figure 3,
intention-to-die can range from .31 to .98 as a function of probability
of intervention for self-injurers rated 5 for reversibility of method.
Freeman et al. (1974, p. 31) found a range of from .54 to .70 for these
cells, a little less than one forth of the range found for this study.
The Probability of Intervention Revised Scale therefore seems to be
more sensitive and may contribute more to overall intention-to-die than
did the original.
The strong positive relationship of age and intention-to-die for
both sexes, especially for males, was reinforced by this study. The
personal variables that significantly increased intention-to-die for
both males and females were not drinking at the time of the incident
and not being under drug therapy. Although drinking problems are known
to increase suicide risk, the findings could indicate that self-injury
risk is increased by drinking, while expected intentionality decreases
because of the increased frequency of impulsive self-injuries. Being
under drug therapy could also increase the probability of self-injuries
taking place, through increased availability of the means. However,
drugs are usually associated with lower intention-to-die self-injuries.
A combination of these two factors, increased availability of means,
but the means usually having low-lethality, could explain the relation-
While seeing a physician within a month increased intention-to-die
for males, being in a hospital recently had the opposite relationship.
Health problems are known to increase suicide risk, which is compatible
with the former but not the latter finding. Knowing why the individual
was in the hospital or seeing a physician could explain the apparent
discrepancy in the findings.
The pattern of relationships between intention-to-die and personal
variables, variables other than sex and age, do not lend themselves to
ready explanations. The dependent variables in this study could be
indicators of other factors which directly influence intention-to-die.
These variables could take many forms, but may be motivational. For
example, Dorpat and Ripley (1967) studied alloplastic, (i.e. other
directed) and autoplastic (i.e. inner directed) motives of self-injurers.
He found suicides as having more autoplastic motives while attempters
were more alloplastically motivated. Perhaps the relationships of the
personal variables in this study to intention-to-die could be a function
of their relationship to alloplastic or autoplastic motivation, which
was not studied. For example, having a police record for a conduct
offense had a small but significant relationship to intention-to-die
in the overall sample. Perhaps having this record is related to an
orientation of other directed manipulation, or alloplastic motivation.
Developmental Life Crises
Erikson (1950) conceptualized growth and development in terms of
a series of crises in life where the individual is confronted with a
shift in his basic social roles as a function of biological maturation.
Although he describes the ordering of the major developmental life
crises, he only roughly describes the timing of their onset as occurring,
for example, in the "young adult," etc. He speculates that, although
the major developmental life crises and their sequences are increased,
the culture determines, within some range, when particular crises will
occur. For example, the crises of "intimacy versus isolation" may be
usually confronted earlier in an agrarian than in an industrial culture.
As the United States, and especially its metropolitan areas, are
pluralistic in population make-up, it could be hypothesized that there
would be some variance between subpopulations and groups as to the
typical onset of these crises.
It was hypothesized that particular age groups should more repre-
sent developmental life crises and that expected intentionality would
increase as a function. This was not found to be the case. However,
Freeman et al. (1974) reported a bimodal curve for expected intentionality
among males with the smaller peak in the early twenties, a nadir in the
late thirties, and a steady and rapid increase into old age. Why the
discrepancy between the two studies? One possibility is the nature of
the population studied by Freeman et al., a small city with a major
university as the dominant industry. Student-status was found to in-
crease intention-to-die in self-injuries in their study, and a large
proportion of the male population was comprised of students in their
early twenties. It is harder to form any explanation for the discrepancy
between the .12 expected intentionality they found for males in their
late thirties, and the .25 found in this study.
Expected intentionality was hypothesized as relating most to
autoplastic motives in this study, while self-injury risk was thought
to be a function of alloplastic motives. Therefore, no hypotheses were
made as to a relationship between developmental life crises and self-
injury risk. However in both males and females and for the majority
of subpopulations, self-injury risk was bimodal with one peak either
in the early or late twenties, and a second peak anywhere from the late
thirties to the early fifties. For females the peaks were in the early
twenties and early forties with a nadir in the early thirties. There
was some difference in magnitude between the peaks for female sub-
populations and, within five years, where they would occur, but the
overall patterns were consistent. For males the first peak was in the
early twenties for all subpopulations, and was of greatest magnitude,
except for the upper social-status. The occurrence of the second peak
was not as pronounced in middle and upper social-status males as it was
for lower social-status males and all female subpopulations.
Self-injury risk varies as a function of age and generally declines.
However it rises and falls in definite patterns, which vary by sex and
social-status. These patterns could be a result of particular stresses,
which become crises, associated with maturational stages that vary as
to their onset of occurrence in different subpopulations and between
men and women. Exactly what form these stresses take could be the sub-
ject of a developmental study of normal people representative of
It is as much interest that a hiatus consistently occurs between
early adulthood and middle-age in self-injury risk as that there are
peaks of the phenomena. For female subpopulations this respite occurs
around the early thirties while for males its occurrence may be as late
as the early forties. The occurrence of this hiatus is consistent with
Erikson's theory. Evidently some difficult social role and inter-
personal adjustments occur before and after this hiatus. Once the
individual reaches a particular age within his/her culture, he/she
has made some accommodations which will permit at least minimal function-
ing until vectors for change in roles and relationships occur again in
middle age. From this study, the interpersonal and cultural environ-
mentsseem to put the greatest stresses and demands for change upon the
individual, rather than these stresses coming intrapersonally, directly
as a function of biological maturation. If the latter were the case
then expected intentionality would also demonstrate peaks and nadirs
and the relationship would not vary as a function of social-status.
It is hypothesized from the results of this study that developmental
life crises are expressed more interpersonally or alloplastically,
where the individual attempts to change his/her environment through a
cry for help. Developmental life crises are not related to autoplastic
motivation, as would be reflected in greater intention-to-die.
Formulating Suicide Prevention Strategies
Many clinically oriented authors such as Murphy and Robins (1967)
acknowledge the existence of sociological and personal factors in
suicide, but disparage their significance, in that they give us ". .
little help in predicting, and thus preventing, the individual suicide"
(1967, p. 303). Rosen (1954) discusses the difficulty in predicting
a low frequency event, such as suicide, without incurring the high
cost of identifying large numbers of false positives. Both of these
attitudes, however, are based on assumptions that the individual has
to be identified in order to mobilize individual treatment, usually
medical in nature. If suicidal individuals are being located in order
to be hospitalized, it is true that we are being confronted with a
hopeless task. There are not enough hospitals to accommodate all the
people our best screening instruments would identify as high suicide
risks. Neither has this strategy of suicide prevention been demon-
strated as effective. However, there are other strategies of suicide
prevention that are possible, and alternative approaches to medical
intervention. For example, Diggory (1969) outlines a program designed
for increasing hit rates, optimizing suicide prevention programs by
directing resources to high risk populations. Any such suicide pre-
vention effort needs two components: the identification of subpopula-
tions which are at risk, or identifying points in peoples lives or
situations which raise suicide risk; and the development of alternative
strategies which can either be directed at groups rather than individuals
or which utilize intervention strategies which incur relatively low
cost. This study was addressed to the first of these two components
with the assumption that alternative forms of suicide prevention are
One such alternative strategy of suicide prevention is postventa-
tive work with suicide attempters. We know that, despite medical inter-
vention including hospitalization, suicide attempters will have about a
140 times greater chance of dying in the year following the attempt
than other members of the general population. It is also known which
factors will increase or lower this risk (Tuckman & Youngman, 1968b).
And yet only a few programs around the country will systematically
postvene non-medically following the release of the suicide attempter
from the hospital, emergency room, or physician's office. These pro-
grams work with the individual in the community setting to mobilize
resources to work through their crises and change the factors in the
situation which led to the suicide attempt. Good follow-up studies
are needed to evaluate the effectiveness of these programs. It would
be possible to assign suicide risk factors to these individuals using
a modified scale from Tuckman and Youngman (1968b) and compare these
at follow-up points with actual death rates. We know that from 10 to
20 percent of suicides have made prior attempts (Dorpat & Ripley, 1967),
which is the maximum degree to which overall suicide rates potentially
could be lowered through this one approach.
Suicide prevention services have not adopted alternative strategies
which are aimed at high risk groups, however. We have instead developed
technologies for suicide prevention and crisis intervention which take
the passive stance of responding to self-selected people in trouble.
We need to develop programs which are actively aimed at influencing
high risk groups prior to self-injuries. One form such programs might
take is in extending the accessibility of our crisis phone centers.
Although these centers are available to the troubled individual, these
individuals may not see the center as a resource. Therefore, crisis
centers need as a first step, to target messages to high risk popula-
tions which will educate as to their appropriateness as a resource in
time of crisis. This would, hopefully, improve the accessibility of
an already available service.
Many individuals may not use an available service in a suicidal
crisis, even if they know of its existence and its mission. Part of
the phenomenon of suicide is a cognitive inflexibility on the part of
the victim where resources are not perceived, even if others near the
situation perceive the resource as appropriate and accessible to the
victim. Our programs need to be aimed at those who are close to high
suicide risk individuals. These natural caregivers need to learn to
assess suicidal risk and to mobilize our intervention services. We
need programs aimed at the police, clergy, physicians, and bartenders
to mobilize our services on behalf of high risk individuals they meet
in the normal course of their professions. A few demonstration pro-
grams of this type have recently begun. We also need to educate
spouses and children to recognize suicide risk and to know when to call
a suicide prevention services on behalf of their loved ones.
Evaluating Suicide Prevention Services
When a service has taken "suicide prevention" in its name or as
part of its mission, it has incurred an obligation to evaluate the ex-
tent to which it prevents suicides and to structure its program to
optimize its impact on the suicide rate. This evaluation can take many
forms, but this study is directly applicable to one needed procedure:
The center can evaluate itself as to whether it is being utilized by
high risk subpopulations. Although the center undertakes an obligation
to people in any type of trouble who choose to utilize it as a resource,
it also has an obligation to evaluate its accessibility to those at
high suicide risk. For this purpose the center needs to know suicide
risk, self-injury risk, and expected intentionality. In other words
to apply the schematization of Freeman et al. (1974) and Wilson (1974)
which are demonstrated in Tables 7, 8 ard 9. If the service does not
have the resources to utilize the Intention-to-Die Scales, it can use
a near equivalent method of assigning the value of number of suicides
divided by number of suicide attempts for the group of interest as a
measure of expected intentionality. When dealing with smaller numbers,
this will be less accurate and the curve should be smoothed out to form
One difficulty in using the schematization of self-injury risk,
expected intentionality, and suicide risk is in obtaining a complete
and unbiased sample from the general population. For this purpose
cooperation with hospitals, physicians and police are necessary. If
the estimate of 10 suicide attempts per suicide is accepted, then an
estimate can be made of the extent to which the true population of
self-injuries was tapped by comparing this theoretical ratio to the
ratio found. In this study a ratio of 6.02:1 was obtained. By dividing
this into the theoretical ratio, a factor of 1.66 is obtained. This
factor could be multiplied to find suicide attempt risk for an estimate
of the theoretical rates of suicide attempts. This procedure may or
may not be justified, depending on the size of the factor and the uses
of the results.
Freeman et al. (1974) advance the argument that the term suicide
attempt should be stricken from the language. They point to the in-
accuracies of the term suicide in conjunction with an act where, in
a majority of cases, no intention-to-die exists. This reconceptualiza-
tion is seen as not just a semantic exercise, but needed because of
the sometimes insidious abuses created by the label. For example, a
self-injury that is labeled a suicide attempt sometimes evokes conno-
tations of the individual being hopelessly incompetent or a liar who
is trying to "put something over" on the caregiver (1974, p. 36).
They propose the term self-injury as a substitute. However, a term
is needed which includes both suicides and what are presently labeled
suicide attempts, and self-injury is a good fit. A term which was
advanced by Kennedy, Kreitman, and Ovenstone (1974) which conceptually
includes the relationship of the behavior with suicide, without the
pejorative inaccuracies of the term suicide attempt, is parasuicide.
Of all the alternatives in the literature this one seems to have the
most promise of offering a label which is useful and accurate, without
introducing new semantic or connotative difficulties. The prefix para-
denotes that these non-fatal self-injuries resemble suicide in some
ways, but also have important functional differences. It is suggested
that parasuicide be defined utilizing Stengel's definition of suicide
attempt: ". . any non-fatal act of self-damage inflicted with self-
destructive intention, however vague and ambiguous" (1968, p. 172), or,
as any non-fatal self-injury. Within the classification of parasuicides
are suicide attempts, which are defined as high intention-to-die para-
suicides, a quantified definition. In other words, for this study only
30 of the 692 parasuicides, about 4 percent, were suicide attempts. As
can be seen, suicide attempts are a relatively rare special case of
parasuicide, occurring only about a fourth as often as suicides.
An additional conceptualization is important which involves the
issue of self-definition. In survey work a larger magnitude of para-
suicides will be found than in the best case-finding study, as self-
defined cases will be included for which there was no intervention, or
only private intervention by significant others. In this study over
98 percent of the self-injuries involved medical intervention. It
would be important for the purpose of comparing results between studies
if survey studies also asked if any type of intervention occurred. This
would sub-classify parasuicides into those with and without caregiver
intervention, whether the intervention is by police, medical, etc.
Self-defined parasuicides and intervened parasuicides need to be
identified in survey research in order for results to be compared to
studies involving other methodologies.
Significance and Prospects
Probably the most important contribution of this study is its
naturalistic-descriptive function. Data from as thorough a sampling
of a large general population are intrinsically interesting and valuable.
They give a new picture of patterns of the phenomena in an area where
there were only partial pictures available previously. There are
limitations, of course, to the study and its generalizability; but
it represents a step forward in the empirical conceptualization of self-
injurious behaviors. This study is the first to ascertain the role of
intentionality in self-injuries in relation to population base rates.
In other words, this study puts it together, and demonstrates how other
researchers or program planners can determine a more vivid picture of
the phenomena in their own community.
Hopefully, this study and the conceptual framework it employs
will have heuristic value in that it stimulates hypotheses for future
research. The reconceptualized form in which the data are presented
in this study are seen as leading to more antecedent-consequent types
of research in the area. For example, a longitudinal follow up study
such as Tuckman and Youngman's (1968a) could add the factor of previous
intention-to-die of parasuicides in assessing not only suicide risk,
but self-injury risk and expected intentionality among prior parasuicides.
The relationship of developmental life crises and self-injury risk was
a serendipitous finding of the study and suggests further research.
A number of good scales of suicide risk have been developed,
including Litman's (1971). Perhaps it would clinically be valuable
to sub-scale these suicide risk scales into a self-injury risk scale
and an expected intentionality scale. It is important to know both
the probability of the event happening and the probable consequences
of the event if it does happen. It would also be of clinical use to
be able to assess prior parasuicides of clients for intention-to-die.
Weissman (1974) found the absolute number of young parasuicides
in one hospital to be rising and speculated that it could be a precursor
of later increases in suicide rates as the group ages. A more accurate
warning might come from a replication of this study in the same
community. From this study, does the high self-injury risk found for
lower social-status males indicate increased suicide risk of this group
as it ages? A follow-up study would show any changes in patterns that
would indicate later changes in suicide risk. If a trend in increases
in self-injury risk was ascertained it might precede a subsequent rise
in suicide rate, giving earlier warning for possible prevention efforts.
Although the results of this study will be of interest to people
working in suicide prevention they may have limited generalizability
to their own community. Patterns of the phenomena will vary over
time and between localities. Hopefully, suicide prevention centers
will begin to apply the schematization in order to monitor the pheno-
mena for their own community and to differentially program and evaluate
their suicide prevention efforts.
As complete and unbiased a sample of self-injuries in a general
population were obtained and suicide risk, self-injury risk and expected
intentionality were determined and related for the population and sex
X social-status subpopulations over the age span. The 692 parasuicides
(i.e. suicide attempts) and 115 suicides were rated for intention-to-
die utilizing a revised version of the Intention-to-Die Scales developed
by Freeman et al. (1974). Although age and sex were the strongest
factors relating to intention-to-die, other personal factors were found
to make minor, but significant contributions to the multiple relation-
ship. Although social-status did not relate to intention-to-die,
lower social-status for both males and females were found to increase
each of the risk levels. Self-injury risk was found to relate to
developmental life crises for both males and females and for most sex
X social-status subpopulations. Findings are discussed in terms of
evaluating suicide prevention centers, formulating suicide prevention
strategies, and in the reconceptualization of acts presently labeled
suicide attempts as parasuicides.
REVERSIBILITY OF METHOD SCALE
1. COMPLETE REVERSIBILITY OF METHOD:
Ingestion of aspirin or other commercial drug items (such as
Excedrin, Bufferin, Midol, etc.); antihistamines, or other non-
toxic household substances (such as baking powder, mouthwashes,
etc.). Also slight cuts not requiring treatment.
2. PROBABLE AND EXPECTED REVERSIBILITY OF METHOD:
Ingestion of 10 or more tranquilizers or nonprescription
sleeping pills (such as Sominex, pep pills, etc.). Ingestion
of 10 or more stimulants (such as Serpasil, reserpine, Raudixin,
Thorazine, Compazine, Dartal, Mellaril, Permitil, Trilafon,
Stelazine, meprobamate; Librium, valium, Miltown, and Equanil).
Also wrist cuts requiring vessel and/or tendon repair.
3. QUESTIONABLE REVERSIBILITY OF METHOD:
Ingestion of 10 or more soporific medications, poisons, large
amounts or combinations of several drugs, narcotics (barbiturates:
phenobarbital, sodium butisal, Nembutal, Seconal, Sodium amytal,
tuinal; non-barbiturate hypnotics: bromides, chloral hydrate,
paraldehyde, bromural, Carbrital, halabar; narcotics: morphine,
Demoral, Darvon). Deep cuts requiring tendon or vessel repair
(except single wrist cuts) and multiple severe cuts.
4. IMPROBABLE AND UNEXPECTED REVERSIBILITY OF METHOD:
Attempted drowning, carbon monoxide suffocation, domestic gas
suffocation, suffocation. Deep cuts to the throat.
5. REMOTE OR NO CHANCE FOR REVERSIBILITY OF METHOD:
Gunshot in vital area (such as trunk of body or head). Jumping
from a high place (more than 20 feet). Hanging (feet above
PROBABILITY OF INTERVENTION SCALE
1. CERTAIN INTERVENTION
Act committed in the presence of another person.
2. PROBABLE INTERVENTION
Act committed with another person in the immediate vicinity
but not visibly present (such as in the same household).
3. AMBIGUOUS CHANCE OF INTERVENTION
Act committed by person alone, with no certainty of immediate
assistance; however, a reasonable chance for intervention exists
(such as the victim is aware of the impending arrival of others).
Telephone is available and may be used to call a significant
4. IMPROBABLE INTERVENTION
Act committed by person alone, with intervention by a passerby
possible although not expected (such as a motel room, an office
late at night, or home alone with no one expected).
5. CHANCES OF INTERVENTION REMOTE
Act committed by person in a solitary or isolated place without
access to telephone (such as a wooded area, cemetery, etc.).
PROBABILITY OF INTERVENTION
1. CERTAIN INTERVENTION
Act is committed in the presence of another person who can
recognize it as self-injurious behavior and intervene immediately.
2. PROBABLE INTERVENTION I
Act is committed while the person is alone, but a person is
contacted to intervene (e.g. by telephone) so as intervention
should be almost immediate or with short delay; or, another
person is not present but is expected immediately or with
short delay following the act.
3. PROBABLE INTERVENTION II
Act is committed with another person in the immediate vicinity
but not visibly present (such as in the same household) who
is likely to intervene with little delay or who is put in or
who finds themselves in a position of recognizing the attempt
with little delay.
4. AMBIGUOUS CHANCE OF INTERVENTION
A potential intervenor is on the premises, in the vicinity,
or expected but intervention is expected to be significantly
delayed (e.g. several hours).
5. IMPROBABLE INTERVENTION
Act is committed by the person alone. Intervention is expected
to be significantly delayed (e.g. hours).
6. CHANCES OF INTERVENTION REMOTE
Act is committed by person alone and intervention is not
expected for at least 12 hours; or the act is committed in a
solitary or isolated place without access to a telephone (such
as a cemetery late at night; an isolated wooded area, etc.).
Probability of Intervention
1 2 3 4 5
, 4 3
5 I -
Batchelor, I. R. C. and Napier, M. B. Broken homes and attempted
suicide. British Journal of Delinquency, 1953/54, 4, 99-108.
Bergstrand, C. G. and Otto, U. Suicidal attempts in adolescence and
childhood. Acta Paediatrica, 1962, 51, 17-26.
Breed, W. Occupational mobility and suicide among white males. American
Sociological Review, 1963, 28, 179-188.
Breed, W. Suicide, migration, and race: a study of cases in New Orleans.
Journal of Social Issues, 1966, 22, 30-43.
Breed, W. Male suicide: Los Angeles and New Orleans compared. Bulletin
of Suicidology, 1967, 1, 11-14.
Choron, J. Suicide. New York: Charles Scribner's Sons, 1972.
Clendonin, W. W. and Murphy, G. E. Wristcutting. Archives of General
Psychiatry, 1971, 25, 465-469.
Dahlgren, K. G. On suicide and attempted suicide. Lund, Sweden,
Daves, F. B. The relationship between suicide and attempted suicide:
a review of the literature. Psychiatric Quarterly, 1967, 41,
Diggory, J. C. The components of personal despair. In: Shneidman, E.
(Ed.), Essays in Self-Destruction. New York: Science House, 1967,
Diggory, J. C. Calculations of some costs of suicide prevention using
certain predictors of suicidal behavior. Psychological Bulletin,
1969, 71, 373-386.
Dorpat, T. L. and Boswell, J. W. An evaluation of suicidal intent in
suicide attempts. Comprehensive Psychiatry, 1963, 4, 117-125.
Dorpat, T. L., Jackson, J. K., and Ripley, H. S. Broken homes and
attempted and completed suicide. Archives of General Psychiatry,
1965, 12, 213-216.
Dorpat, T. L. and Ripley, H. S. A study of suicide in the Seattle area.
Comprehensive Psychiatry, 1960, 1, 349-359.
Dorpat, T. L. and Ripley, H. S. A study of suicide in King County,
Washington. Northwest Medicine, 1962, 61_, 655-661.
Dorpat, T. L. and Ripley, H. S. The relationship between attempted
suicide and committed suicide. Comprehensive Psychiatry, 1967,
Dublin, L. I. Suicide: A Sociological and Statistical Study. New York:
Durkheim, E. Suicide: A Study in Sociology. Glencoe, Ill.: Free Press,
Edwards, J. E. and Whitlock, F. A. Suicide and attempted suicide in
Brisbane. Medical Journal of Australia, 1968, 1, 932-928, 989-995.
Erikson, E. H. Childhood and Society. New York: Norton, 1950.
Ettlinger, R. W. and Flordh, P. Attempted suicide: experience of five
hundred cases at a general hospital. Acta Psychiatrica Scandanavia,
1955, 103, 1-45.
Farberow, N. L. Crisis, disaster, and suicide: theory and therapy. In:
Shneidrman, E. S. (Ed.), Essays in Self-Destruction. New York:
Science House, 1967.
Farberow, N. L., Shneidman, E. S. and Nouringer, C. Case history and
hospitalization factors in suicides of neuropsychiatric hospital
patients. Journal of Nervous Mental Disorders, 1966, 142, 32-44.
Fawcett, J., Leff, M., and Bunney, W. E., Jr. Suicide: clues from
interpersonal communication. Archives of General Psychiatry, 1969,
Freeman, D. J., Wilson, K. E., Thigpen, J. D., and McGee, R. K.
Assessing intention to die in self-injury behavior. In: Neuringer,
C. (Ed.), Psychological Assessment of Suicidal Risk. Springfield,
Ill.: Charles Thomas, 1974, 18-42.
Freud, S. Mourning and melancholia. In: Collected Papers, Volume IV.
London: Hogarth, 1925.
Gold, N. Suicide and attempted suicide in north-eastern Tasmania.
Medical Journal of Australia, 1965, 2, 361-364.
Hendin, H. Attempted suicide: a psychiatric and statistical study.
Psychiatric Quarterly, 1950, 24, 39-46.
Hendin, H. Black Suicide. New York: Basic Books, 1969.
Hopkins, F Attempted suicide: an investigation. Journal of Mental
Science, 1937, 83, 71-94.
Humphry, J. A., French, L., Niswander, G. D., and Casey, T. M. The
process of suicide: the sequence of disruptive events in the lives
of suicide victims. Diseases of the Nervous System, 1974, 35,
Jacobziner, H. Attempted suicide in adolescence. Journal of the
American Medical Association, 1965, 191, 7-11.
Jackson, D. D. Theories of suicide. In: Shneidman, E. S. and Farberow,
N. L. (Eds.), Clues to Suicide. New York: McGraw-Hill, 1957, 11-
Kennedy, P., Kreitman, N., and Ovenstone, I. M. The prevalence of
suicide and parasuicide ("attempted suicide") in Edinburgh.
British Journal of Psychiatry, 1974, 124, 36-41.
Kessel, N. and Lee, E. M. Attempted suicide in Edinburgh. Scottish
Medical Journal, 1962, 7, 130-135.
Kessel, N. and McCulloch, W. Repeated acts of self-poisoning and self-
injury. Proceedings of the Royal Society of Medicine, 1966, 59,
Lester, D. Relation between attempted suicide and completed suicide.
Psychological Reports, 1970, 27, 719-722.
Lennard-Jones, J. E. and Asher, R. Why do they do it? A study of
pseudocide. Lancet, 1959, 1, 1138-1140.
Litman, R. E., Farberow, N. L., Wold, C. I., and Brown, T. R. Predic-
tion models of suicidal behaviors. Unpublished paper, 1971.
Lukianowicz, N. Suicidal behavior: an attempt to modify the environ-
ment: II. Psychiatria Clinica, 1973, 6, 171-190.
MacMahon, B., Johnson, S., and Pugh, T. F. Relation of suicide rates
to social conditions. Public Health Reports, 1963, 78, 285-293.
MacMahon, B. and Pugh, T. F. Suicide in the widowed. American Journal
of Epidemiology, 1965, 81, 23-31.
Maris, R. W. Suicide, status, and mobility in Chicago. Social Forces,
1967, 46, 246-256.
Maris, R. W. The sociology of suicide prevention: policy implications
of differences between suicidal patients and completed suicides.
Social Problems, 1969, 17, 132-149.
Massey, J. T. Suicide in the United States. Vital Health Statistics,
1967, 20, 1-34.
McCarthy, P. D. and Walsh, D. Suicide in Dublin. British Medical
Journal, 1966, 1, 1393-1396.
McGee, R. K. and Hegert, T. A detailed analysis of suicide in Orange
County, Florida, 1963-1965. Unpublished manuscript. Gainesville,
Florida: University of Florida, 1966.
Menninger, K. A. Man Against Himself. New York: Harcourt Brace, 1938.
Mintz, R. S. Prevalence of persons in Los Angeles who have attempted
suicide: a pilot study. Bulletin of Suicidology, 1970, 7, 9-16.
Motto, J. A. Suicide attempts: a longitudinal view. Archives of
General Psychiatry, 1965, 13, 516-520.
Motto, J. A. and Greene, C. Suicide and the medical community. Archives
of Neurology and Psychiatry, 1958, 80, 776-781.
Murphy, G. E. Recognition of suicide risk: the physician's responsi-
bility. Southern Medical Journal, 1969, 62, 723-728.
Murphy, G. E. The physician's failure in suicide prevention. In:
Litman, R. E. (Ed.), Proceedings: 6th International Conference for
Suicide Prevention. Ann Arbor: Edwards Brothers, 1972.
Murphy, G. E., Clendenin, W. W., Darvish, H. S., and Robins, E. The
role of the police in suicide prevention. Life Threatening Be-
havior, 1971, 1, 96-105.
Murphy, G. E. and Robins, E. Social factors in suicide. Journal of
the American Medical Association, 1967, 199, 303-308.
Murphy, G. E. and Robins, E. The communication of suicidal ideas. In:
Resnick, H. L. P. (Ed.), Suicidal Behaviors: Diagnosis and Manage-
ment. Boston: Little, Brown, 1968, 163-170.
Neuringer, C. Methodological problems in suicide research. Journal of
Consulting Psychology, 1962, 26, 273-278.
Neuringer, C. Rigid thinking in suicidal individuals. Journal of
Consulting Psychology, 1964, 28, 54-58.
Neuringer, C. and Lettieri, D. J. Cognition, attitude, and affect in
suicidal individuals. Life Threatening Behavior, 1971, 1, 106-124.
Parkin, D. and Stengel, E. Incidence of suicide attempts in an urban
community. British Medical Journal, 1965, 2, 133-138.
Paykel, E. S., Myers, J. K., and Lindenthal, J. J. Thoughts of
suicide: a general population survey. V World Congress of
Psychiatry, Abstracts. La Prensa Medica Mexicana, 1971.
Pokorny, A. D. Characteristics of 44 patients who subsequently committed
suicide. Archives of General Psychiatry, 1960, 2, 314-323.
Robins, E., Gassner, S., Kayes, J., Wilkinson, R. H. and Murphy, G. E.
The communication of suicidal intent: a study of 134 consecutive
cases of successful (completed) suicide. American Journal of
Psychiatry, 1959a, 115, 724-733.
Robins, E., Murphy, G. E., Wilkinson, R. H., Gassner, S., and Kayes, J.
Some clinical considerations in the prevention of suicide based on
a study of 134 successful suicides. American Journal of Public
Health, 1959b, 49, 888-899.
Rosen, A. Detection of suicidal patients: an example of some limita-
tions in the prediction of infrequent events. Journal of Con-
sulting and Clinical Psychology, 1954, 18, 397-403.
Ruegsegger, P. Selbstmordversuche. Psychiatria et Neurologia, 1963,
Schwab, M. D., Warheit, G. J., and Holzer, C. E. Suicidal ideation
and behavior in a general population. Diseases of the Nervous
System, 1972, 33, 745-748.
Sclare, A. B. and Hamilton, C. B. Attempted suicide in Glasgow.
British Journal of Psychiatry, 1963, 109, 609-615.
Segal, B. E. and Humphry, J. A comparison of suicide victims and
suicide attempters in New Hampshire. Diseases of the Nervous
System, 1970, 31, 830-838.
Shneidman, E. S. Orientations toward death: a vital aspect of the
study of lives. In: White, R. W. (Ed.), The Study of Lives.
New York: Atherton, 1963, 200-227.
Shneidman, E. S. Classifications of suicidal phenomena. Bulletin of
Suicidology, 1968, 1, 1-9.
Shneidman, E. S. Suicide, lethality, and the psychological autopsy.
International Psychiatry Clinics, 1969, 6, 225-250.
Shneidman, E. S. and Farberow, N. L. Some comparisons between genuine
and simulated suicide notes. Journal of General Psychology, 1957a,
Shneidman, E. S. and Farberow, N. L. The logic of suicide. In: Shneid-
man, E. S. and Farberow, N. L. (Eds.), Clues to Suicide. New York:
McGraw-Hill, 1957b, 31-40.
Shneidman, E. S. and Farberow, N. L. Statistical comparisons between
attempted and committed suicides. In: Farberow, N. L. and Shneid-
man, E. S. (Eds.), The Cry for Help. New York: McGraw-Hill, 1961,
Stengel, E. Enquiries into attempted suicide. Proceedings of the
Royal Society of Medicine, 1952, 45, 613-620.
Stengel, E. Suicide and Attempted Suicide. Baltimore: Penguin Books,
Stengel, E. Attempted suicides. In: Resnick, H. L. P. (Ed.), Suicidal
Behaviors: Diagnosis and Management. Boston: Little, Brown, 1968,
Stengel, E. and Cook, N. G. Attempted Suicide: Its Social Significance
and Effects. London: Chapman & Hall, 1958.
Tuckman, J. and Lavell, M. A study of suicide in Philadelphia. Public
Health Reports, 1958, 73, 547-553.
Tuckman, J. and Youngman, W. F. Suicide risk among persons attempting
suicide. Public Health Reports, 1963a, 78, 585-587.
Tuckman, J. and Youngman, W. F. Identifying suicide risk groups among
attempted suicides. Public Health Reports, 1963b, 78, 763-766.
Tuckman, J. and Youngman, W. F. Assessment of suicide risk in attempted
suicide. In: Resnick, H. L. P. (Ed.), Suicidal Behaviors: Diagnosis
and Management. Boston: Little, Brown, 1968a, 190-197.
Tuckman, J. and Youngman, W. F. A scale for assessing suicidal risk in
attempted suicide. Journal of Clinical Psychology, 1968b, 24,
Tuckman, J., Youngman, W. F., and Bleiberg, B. M. Attempted suicide
by adults. Public Health Reports, 1962, 77, 605-614.
United Nations. Demographic Yearbook. New York: United Nations, 1966,
U.S. Bureau of Census. U.S. Census of Population and Housing: 1960.
Census Tracts, St. Louis, Mo.-Ill. Washington, D.C., Department
of Commerce, 1962.
U.S. Bureau of Census. U.S. Census of Population and Housing: 1970.
Census Tracts, St. Louis, Mo.-Ill. Washington, D.C., Department
of Conmerce, 1972.
Weissman, A. D. Suicide, death, and life-threatening behavior. Pre-
sented at meeting of the Task Force on "Suicide Prevention in the
Seventies," Phoenix, Arizona, January 30 February 1, 1970.
Welssman, A. D. and Wordcn, J. Risk-rescue rating in suicide assess-
ment. Archives of General Psychiatry, 1972, 26, 553-560.
Weissman, M. M. The epidemiology of suicide attempts, 1960 to 1971.
Archives of General Psychiatry, 1974, 30, 737-746.
Weissman, M. M., Fox, K., and Kerman, G. L. Hostility and depression
associated with suicide attempts. American Journal of Psychiatry,
1973, 130, 450-455.
Whitlock, F. A. and Schapira, K. Attempted suicide in Newcastle upon
Tyne. British Journal of Psychiatry, 1967, 113, 423-434.
Wilkins, J. Suicidal behavior. American Sociological Review, 1967,
Wilson, K. E. Implications of reconceptualizing suicide risk and self-
injury behavior. Presented at Seventh Annual Meeting of the
American Association of Suicidolcgy, Atlantic Beach, Florida,
World Health Organization. Prevention of suicide. Public Health Papers,
Karl Eugene Wilson was born June 14, 1946, at Wayne, Michigan.
He was graduated from a Detroit Public High School in August of 1964.
He immediately went to work as a laborer, for an automobile manufacturing
corporation, and completed several college courses before enlisting as
a private in the United States Marine Corps in February, 1966. He
completed two years of active duty, including a fourteen month tour of
duty in Viet Nam, and was able to travel around the world.
Mr. Wilson re-enrolled in Monteith College, Wayne State University
in April, 1968 and graduated with the Bachelor of Philosophy with high
distinction in June of 1971. During this period he worked at such
diverse jobs and positions as bartender, nursery school teacher, and
research assistant at the Merrill-Palmer Institute of Human Development.
He was married in December, 1969, to Rose Janet Goodman Wilson. Ms.
Janet Wilson is a psychiatric social worker.
Mr. Wilson has been enrolled in the graduate program in clinical
psychology at the University of Florida since September, 1971. He
received his Master of Arts degree in December, 1972. During his
graduate career he was associated with the Suicide and Crisis Inter-
vention Service of Gainesville, Florida and the Center for Crisis
Intervention Research. He completed his one year internship in
clinical psychology, with emphasis in community psychology, in
September, 1975, at Malcolm Bliss Mental Health Center in St. Louis,
Missouri. Mr. Wilson began his current position as Visiting Assistant
Professor, Department of Psychology, Washington University, in
January, 1976. Mr. Wilson has published articles in the areas of
child development, suicidology, and the utilization of paraprofessionals
in crisis intervention.