Sei sulla pagina 1di 28

JOURNAL

ARTICLE
10.1177/0022427803256074
Tittle et al. /OF
A CHALLENGE
RESEARCH IN
TOCRIME
SELF-CONTROL
AND DELINQUENCY
THEORY

GENDER, AGE, AND CRIME/


DEVIANCE: A CHALLENGE
TO SELF-CONTROL THEORY
CHARLES R. TITTLE
DAVID A. WARD
HAROLD G. GRASMICK

Focusing on gender and age variations and using various measures of self-control
and of crime/deviance, the authors’provide additional evidence concerning the stron-
gest implications of self-control theory—that self-control interprets the main demo-
graphic facts about crime/deviance and is of approximately equal import for all sub-
categories of individuals. On one hand, the results are strongly supportive of the
theory, showing that some measures of self-control not only predict misbehavior but
they interpret the associations between gender and age and measures of crime/
deviance. On the other hand, self-control does not appear to predict misbehavior
equally well among various subcategories of individuals, particularly not for age
groups, even failing to predict misbehavior at all for some groupings. Moreover, sup-
port for the strongest claims of the theory are not robust, varying depending on how
self-control and crime/deviance are measured.

Keywords: self-control; age; gender

Currently one of the most widely cited theories is Gottfredson and


Hirschi’s (1990) theory of self-control (see Pratt and Cullen 2000; Vazsyoni
et al. 2001). It claims to have identified the major cause of criminal/deviant
behavior and to have universal application, which if true implies that the ef-
fects of its main variable change little from situation to situation or subgroup
to subgroup. Hypotheses from the theory have been widely tested but much
remains problematic. In this article we report additional research that tries to
determine (1) the extent to which self-control accounts for gender and age
differences in crime, and (2) whether the association between self-control
and crime differs among gender and age categories such that they serve as
contingencies for the operation of self-control in affecting crime or deviance.

JOURNAL OF RESEARCH IN CRIME AND DELINQUENCY, Vol. 40 No. 4, November 2003 426-453
DOI: 10.1177/0022427803256074
© 2003 Sage Publications
426

Downloaded from jrc.sagepub.com at UCSF LIBRARY & CKM on April 4, 2015


Tittle et al. / A CHALLENGE TO SELF-CONTROL THEORY 427

HYPOTHESES

The theory contends that people differ in their ability to control impulses
for immediate gratification because of differences in socialization, mainly
before puberty. Those with low self-control are theorized to have a much
greater probability of criminal or analogous behaviors than do those with
high self-control. The less controlled presumably have more difficulty taking
long range consequences of their behavior into account. Because opportuni-
ties for misbehavior are theorized to be ubiquitous and misbehavior to be
inherently gratifying but costly to the individual in the long run, those with
low self-control will succumb more often.
The theory seems to embody two major claims about self-control. First,
low self-control is said to be “for all intents and purposes, the individual-level
cause of crime” (Gottfredson and Hirschi 1990:232). Although the authors
later soften that stance (Hirschi and Gottfredson 1995:140), there is a clear
implication that self-control should account for most of the differences
among individuals in criminal/deviant behavior that is not due to random
variation or errors of measurement. The authors of the theory grant that the
predicted negative association between self-control and criminal/deviant
behavior is nondeterministic (low self-control does not inevitably lead to
crime and numerous conditions presumably may affect whether it does or
does not produce crime) (see, for example, Hirschi and Gottfredson
2000:65). But in fact, they theorize that the relationship is little affected by
variables that sociologists and criminologists have long favored, including
morality, strain, peer influences, social bonds, cultural influences, social dis-
advantages, and others (Gottfredson and Hirschi 1990:chaps. 7 and 8), all of
which are either overshadowed by self-control or are themselves products of
self-control.
Opportunity for crime, however, is one condition about which the theo-
rists have expressed ambivalence. They at first seem to have appreciated it as
having an important influence on the self-control/crime relationship
(Gottfredson and Hirschi 1990:127, 147-48; see Grasmick et al., 1993).
Later they discounted its importance, contending that opportunities are “lim-
itless” and that in any case, opportunity and misbehavior are independent of
each other (Hirschi and Gottfredson 1993:50). More recently (Hirschi and
Gottfredson 1995:140), they seem to have granted a larger part to criminal
opportunities. Nevertheless, low self-control theoretically stands as the pri-
mary cause of misbehavior.
Second, the theory implies that the effects of self-control are similar in all
circumstances (Gottfredson and Hirschi 1990:87, 117). If postchildhood
experiences or conditions have minimal influence on the operation of self-
control, and low self-control is the main cause of crime/deviance, then the

Downloaded from jrc.sagepub.com at UCSF LIBRARY & CKM on April 4, 2015


428 JOURNAL OF RESEARCH IN CRIME AND DELINQUENCY

effects of self-control should be more or less equal across all conditions.


Thus, even though some groups or categories of individuals have higher or
lower self-control than other groups or categories, variation in self-control
among individuals within any group or category should predict the probabil-
ity of misbehavior about the same as in any other group or category, provided
self-control varies sufficiently within the groups or categories and individu-
als within those subgroups are exposed to sufficient opportunities to permit
low self-control to manifest itself in criminal/deviant behavior.
The claims of self-control theory, therefore, give rise to two main hypoth-
eses. First, barring statistical irregularities and measurement deficiencies, the
effect of self-control on criminal/deviant behavior should be universal and
strong, accounting for much if not most of the variance in measures of crime/
deviance among all categories of individuals. If self-control does not account
for most of the variance in criminal/deviant behavior, it should at least show
effects that are approximately equally strong for all subcategories of individ-
uals, and its effects should not be conditional on particular subgroups or cate-
gories. Second, statistically controlling for self-control should, to a large
extent, interpret or explain observed associations between any social or
demographic characteristic and criminal/deviant behavior. Following the
standard logic of analysis, whether it helps interpret or explain the associa-
tion between self-control and some characteristic depends on whether the
characteristic precedes or follows the establishment of self-control during
childhood. In either case, however, such associations should be substantially
reduced when self-control is introduced into a predictive equation.
Much research bearing on these hypotheses has already been reported
(Pratt and Cullen 2000; Vazsonyi et al. 2001). Typical studies have examined
the association between various measures of self-control and various mea-
sures of crime or deviance under different conditions and with different con-
trol variables. In one sense, those tests have provided strong support. Mea-
sures of self-control have been shown to predict, to some extent, criminal and
other misbehavior among established criminals (Longshore 1998; Long-
shore and Turner 1998) as well as general samples of individuals (Evans et al.
1997; Grasmick et al. 1993), college students (Cochran et al. 1998; Gibbs and
Giever 1995; Nagin and Paternoster 1993; Piquero and Tibbetts 1996; Sellers
1999), and youth (Brownfield and Sorenson 1993; Junger and Tremblay
1999; LaGrange and Silverman 1999; Vazsonyi et al. 2001; Wood,
Pfefferbaum, and Arneklev 1993); among people in different cities (Winfree
and Bernat 1998); among those of some different age categories (Burton
et al. 1999); among males as well as females (Burton et al. 1998; Keane,
Maxim, and Teevan 1993; LaGrange and Silverman 1999); and among
research subjects in various countries (Forde and Kennedy 19971; Keane
et al. 1993; LaGrange and Silverman 1999; Nakhaie, Silverman, and

Downloaded from jrc.sagepub.com at UCSF LIBRARY & CKM on April 4, 2015


Tittle et al. / A CHALLENGE TO SELF-CONTROL THEORY 429

LaGrange 2000; Polakowski 1994; Wright et al. 1999; Vazsonyi et al. 2000).
Moreover, statistically significant associations between measures of self-
control and crime or other misbehavior have been found for cross-sectional
(e.g., Evans et al. 1997; Grasmick et al. 1993; Nagin and Paternoster 1993)
and longitudinal samples (Avakeme 1998; Junger and Tremblay 1999;
Lynam et al. 2000; Polakowski 1994; Paternoster and Brame 1998; Wright
et al. 1999), as well as among experimental subjects (Finkel and Campbell
2001; White et al. 1994).
However, significant associations between self-control and measures of
crime or deviance have not been found among all categories of individuals,
and even when associations have proven to be better than chance, they have
not been large (usually in the range of .20 to .30). This calls into question the
hypothesis of strong, similar associations. In addition, evidence relative to
the claim that self-control is the key variable accounting for criminal/deviant
behavior is especially meager. In particular, few tests have employed the
most convincing strategy for assessing the centrality of self-control—that of
determining if it accounts for known demographic variations in misbehavior
(LaGrange and Silverman 1999). Because the best-established patterns of
misbehavior are by gender and age, accounting for those variations is the
most demanding test for theoretical claims. If self-control is the master vari-
able, it should, to a large extent, interpret associations between gender and
age and crime, and if the effects of self-control are universally and equally
strong, there should be no substantial differences in the associations of self-
control with misbehavior among different categories of gender and age.

Self-Control and Gender

With the exception of a few relatively minor offenses, males show a higher
probability and frequency of committing criminal/deviant acts than do
females (Elliott 1994; Nagel and Hagan 1983; Smith and Visher 1980;
Steffensmeier and Allan 2000; Steffensmeier, Allan, and Streifel 1989a;
Sutherland, Cressey, and Luckenbill 1992; Tittle and Paternoster 2000:316-
24; Warren 1991; Wilson and Herrnstein 1985:104-15). The theory logically
accounts for many of these gender differences in criminal/deviant behavior
because self-control is theorized to be a product of child rearing. Children are
said to develop high self-control, and therefore to commit little misbehavior
throughout life, when childhood care givers (1) love a child enough to moni-
tor its behavior and react actively to misconduct, (2) actually practice surveil-
lance, (3) recognize misbehavior when it occurs, and (4) punish or disap-
prove deviant behavior (Gottfredson and Hirschi 1990:97-8). Because most
people regard females as in more danger from misbehavior than males, par-
ents may tend to monitor the behavior of girls more carefully than that of

Downloaded from jrc.sagepub.com at UCSF LIBRARY & CKM on April 4, 2015


430 JOURNAL OF RESEARCH IN CRIME AND DELINQUENCY

boys. This excess monitoring may also extend beyond childhood as commu-
nity institutions try to reduce opportunities for misconduct among adult
females. In addition, because the female role is more dependent on social
approval, care givers are typically more concerned with effective socializa-
tion for girls (Heimer and DeCoster 1999). As a result, they are usually more
attuned to, and recognize, misbehavior among female children. Finally,
being more concerned about the consequences of misbehavior and the
greater demand for social acceptability of female behavior, parents or other
care givers are more likely to impose costly consequences on childhood mis-
conduct by females. As a result, if Gottfredson and Hirschi are correct,
females should usually develop stronger self-control than males and there-
fore commit less crime/deviance. Although past research does not demon-
strate that males and females are necessarily treated differently, some of it
does show differences in self-control (see Gibbs, Giever, and Martin 1998).
The theorists acknowledge (Gottfredson and Hirschi 1990:147-49) that
male/female differences in misbehavior may be influenced by other things,
chief among which are differences in opportunity and social control (super-
vision and enhanced penalties for misbehavior). Nevertheless, they contend
that self-control is among the main reasons that females offend less than
males. This implies a model in which gender has indirect effects on crime/
deviance through several intervening variables, primary among which is
self control. A compelling test of self-control theory, therefore, would show
that statistically controlling for self-control significantly reduces the asso-
ciation between gender and crime/deviance, even though such controls
would not necessarily be expected to completely eliminate the gender-crime
relationship.
Though many studies testing hypotheses from self-control theory have
used gender as a control variable (e.g., Cochran et al. 1998; Longshore 1998;
Nakhaie et al. 2000), have tested effects of self-control on misbehavior for
subgroups of males and females separately (e.g., Burton et al. 1998; Caspi
et al. 1994; LaGrange and Silverman 1999), or have examined male-female
differences in self-control scores (Keane et al. 1993), only three have actually
tested to see if self-control accounts for gender differences in crime/
deviance. Burton et al. (1998) found that holding self-control (measured by a
12-item index) constant eliminated the effect of gender in predicting scores
on an additive index of 20 criminal acts among a random sample of residents
of one city in the United States. LaGrange and Silverman (1999) report that
various psychological measures of self-control (such as risk seeking,
impulsivity, and temper) reduced an association between gender and scores
on an additive, 20-item delinquency scale, among a sample of Canadian sec-
ondary school students. Gibbs et al. (1998), using a convenience sample of
college criminology students, found that gender had a direct effect on a

Downloaded from jrc.sagepub.com at UCSF LIBRARY & CKM on April 4, 2015


Tittle et al. / A CHALLENGE TO SELF-CONTROL THEORY 431

measure of noncriminal deviance (including alcohol use, cheating, and cut-


ting classes) as well as an indirect effect through self-control.
While these three studies are very useful and all support self-control the-
ory, they do not collectively establish that self-control interprets sex/devi-
ance associations. Given that gender differences represent the strongest and
most reliable variation in criminal/deviant behavior, it is crucial for a good
theory to account for it, and it is important to establish that self-control has
the effect that the theorists suggest. Hence, the issue should be examined in a
variety of different contexts. In addition, given that there is no fully agreed
upon measure of self-control, it is essential for the assertions of the theory to
be tested using a variety of measurement instruments.
Not only has the research literature failed to fully establish that self-
control is a potent factor accounting for gender differences in criminal/
deviant behavior, but it has called into question the hypothesis of universal,
equally strong effects. At least three studies have found substantial differ-
ences between males and females in the effects of self-control on criminal/
deviant behavior (Burton et al. 1998; Keane et al. 1993; LaGrange and
Silverman 1999), suggesting that self-control effects may be conditional on
gender.

Self-Control and Age

Though not as clear cut or consistent as differences by gender, patterns of


criminal and deviant behavior also vary markedly by age (Steffensmeier and
Allan 1995; Steffensmeier et al. 1989b). The exact nature of this association
seems to depend somewhat on the type of crime or deviance in question (Tit-
tle and Paternoster 2000), but involvement in most misbehavior appears to be
distributed by age in an inverted j pattern. The trend is for crime and deviance
to increase throughout the adolescent years, reaching a maximum in late ado-
lescence or early adulthood. Thereafter, it appears to decline steadily to the
latest ages (Hirschi and Gottfredson 1983; Tittle and Grasmick 1997; Tittle
and Ward 1993).
The self-control theorists do not claim their theory accounts for age varia-
tions. According to them, self-control cannot explain the age/crime relation-
ship because self-control does not vary enough over the life course, being
basically set before adolescence. In addition, Gottfredson and Hirschi con-
tend that crime varies inherently, naturally, and in invariant ways by age, so
that no theory is obligated to account for that relationship (Gottfredson and
Hirschi 1990; Hirschi and Gottfredson 1983: see Tittle and Grasmick 1997).
This may appear like a maneuver to escape potential criticism for an inability
to explain all major correlates of crime and deviance (although their conten-
tions about age and crime preceded the theory of self-control). If it is a

Downloaded from jrc.sagepub.com at UCSF LIBRARY & CKM on April 4, 2015


432 JOURNAL OF RESEARCH IN CRIME AND DELINQUENCY

defensive tactic, it appears to be unnecessary, at least on theoretical grounds,


because there is good reason within the logic of the theory to expect its own
variables to account for age variations in crime.
First, despite inconsistent research results (Burton et al. 1998, 1999;
Cochran et al. 1998; Grasmick et al. 1993; Longshore 1998; Longshore and
Turner 1998; Sellers 1999), probably due to weak and oblique measurement,
and the skepticism of the authors of self-control theory (Hirschi and
Gottfredson 1993:50), “opportunities” for crime/deviance may be important
preconditions for low self-control to result in deviant behavior (Baumeister,
Heatherton, and Tice 1994:69). With the exception of white collar crime,
opportunities for criminal behavior seem to vary with age in ways consistent
with their having an effect on the age/crime relationship, being more abun-
dant among youth and less available among the aged.
Second, though Gottfredson and Hirschi (1990:106) maintain that self-
control is largely stable throughout the life course, they do acknowledge
some possible improvement in self-control as people grow older. According
to their general perspective, people are basically rational. If so, then one
would expect those who suffer the consequences of low self-control to actu-
ally develop greater self-control over time to avoid further bad consequences.
Therefore, as people grow older, having adapted continuously to the conse-
quences of earlier bad decisions, their self-control should improve, leading to
the reductions in crime revealed in the typical age/crime curve (Tittle and
Grasmick 1997:314). Indeed, there is experimental evidence suggesting that
self-control acts like a muscle, improving in strength over time as it is exer-
cised more (Baumeister and Exline 1999, 2000; Muraven, Tice, and
Baumeister 1998). And Wright et al. (1999:502) found that self-control in
childhood is far from perfectly correlated with self-control in adolescence
(.43), suggesting that self-control is not necessarily stable over the life
course.
Hence, though Gottfredson and Hirschi disagree, one would expect self-
control to account for much of the age variations in crime/deviance just as it is
expected to help account for gender and all other variations. It would cer-
tainly add credibility if this turns out to be the case. So far, various studies
have used age as a control variable (Cochran et al. 1998; Longshore 1998;
Longshore and Turner 1998; Nakhaie et al. 2000) or have reported self-
control effects separately for different age categories (Burton et al. 1999).
Only one study has attempted to ascertain whether controlling for self-
control interprets the age/crime association (Tittle and Grasmick 1997), and
it found that self-control had little effect. However, one study is a weak basis
for a conclusion. Moreover, the only study that has examined associations
between self-control and crime within age categories (Burton et al. 1999) has
reported substantial differences, with self-control failing completely to

Downloaded from jrc.sagepub.com at UCSF LIBRARY & CKM on April 4, 2015


Tittle et al. / A CHALLENGE TO SELF-CONTROL THEORY 433

predict misbehavior within one of three age categories, thereby challenging


the hypothesis of universal, similar effects.

THE STUDY

We use data from a survey of the residents of a southwestern city, concen-


trating specifically on gender and age variation in self-control. Although
many issues about self-control bear further attention, such as how self-con-
trol fares relative to variables from other leading theories, the focus on gender
and age variations is particularly pertinent. It permits us to evaluate two key
hypotheses derivable from self-control theory, that self-control is a major
cause of differences among the categories of people in criminal/deviant
behavior and that the effects of self-control are strong and similar in all cir-
cumstances. In this respect our study adds to a growing body of evidence.
However, it should be especially pertinent to the issues addressed because it
uses four alternative measures of self-control, including the popular
Grasmick et al. (1993) scale, an additive index of the Grasmick et al. items,
and two behavioral measures. In addition, we use several different measures
of crime/deviance.

METHOD

Sample

Data are taken from the 16th annual Oklahoma City Survey, which was
conducted in the spring of 1994. A total of 350 respondents, age 18 and over,
were selected by simple random methods from the R.L. Polk Directory of
households. People were initially contacted by letter and told that they would
be approached by a member of the research team to schedule an interview.
Trained interviewers negotiated the appointment schedule. Any in the target
sample who could not be scheduled were replaced randomly until the total
sample size of 350 was reached. Of the initial target sample, 40 percent com-
pleted the interview, leaving 60 percent of the final sample as random
replacements. At the time of the face-to-face interview, each respondent
recorded on a separate sheet, unseen by the interviewer, information about
criminal/deviant behavior.
Census data for 1990 suggests the sample may contain a slightly dispro-
portionate number of females (56 percent vs 53 percent for those 18+ in the
census), Whites (80 percent vs 78 percent for those 18+ in the census), and
people of older ages (mean of 46.4 for this sample of people over age 17 vs

Downloaded from jrc.sagepub.com at UCSF LIBRARY & CKM on April 4, 2015


434 JOURNAL OF RESEARCH IN CRIME AND DELINQUENCY

32.3 for the census population that includes all ages). Some of the differences
may be because the sample contains only adults or because of demographic
changes from 1990 to 1994. However, the sample probably also reflects
inherent biases in household surveying. As Hirschi and Gottfredson
(1993:48) note, surveys are likely to miss people with extremely low self-
control whose lifestyles do not lend themselves to the stability necessary for
inclusion, and it is likely to miss people who commit a lot of criminal behav-
ior. The theorists contend that effective surveys must reflect differences
among individuals that are relevant to their theory. Consistent with that
requirement, this survey included drinkers (l7% report drinking more than
two or three alcoholic beverages a week), smokers (28%), those who fail to
use seat belts (38%), and people who admit acts of force or fraud (means
range from 1.98 to 1.48 on various dichotomous self-reports). Because these
behaviors presumably reflect low self-control, there should be sufficient
variations to permit meaningful tests of the issues at hand.
These 1994 data have not previously been used to examine hypotheses
from self-control theory, although data from an earlier Oklahoma City Sur-
vey were employed in three widely cited studies concerning self-control
(Arneklev, Grasmick, and Bursik 1999; Arneklev et al. 1993; Grasmick et al.
1993). The data we are analyzing here are from a completely independent
sample drawn 3 years after the previous survey.

Measures of Independent Variables

SELF-CONTROL: A 23-ITEM FACTOR-BASED COGNITIVE SCALE

As conceptualized by Gottfredson and Hirschi (1990), self-control refers


to tendencies to behave in certain ways or to have certain kinds of prefer-
ences. Our cognitive scale of self-control uses the Grasmick et al. (1993)
items, which have been widely employed by others (see Pratt and Cullen
2000; Vaszonyi et al. 2001). Their 23-item composite scale was derived from
a set of items with a four response format. There were four items for each of
six “elements” of self-control, as specified by Gottfredson and Hirschi
(1990:89-90)—impulsivity, preference for simple tasks, risk seeking, prefer-
ence for physical activities, being self-centered, and having trouble control-
ling one’s temper. However, one of the items was eliminated by Grasmick
et al. (1993) because it did not contribute well to the overall measure.
We factored the same items for the 1994 sample with the principal compo-
nents method and, like others who have worked with these or similar items,
found ambiguous evidence about unidimensionality (see Arneklev et al.
1993, 1999; Grasmick et al. 1993; Longshore, Turner, and Stein 1996; Long-
shore, Stein, and Turner 1998; Miller and Lynam 2001; Piquero and Rosay

Downloaded from jrc.sagepub.com at UCSF LIBRARY & CKM on April 4, 2015


Tittle et al. / A CHALLENGE TO SELF-CONTROL THEORY 435

1998; Piquero, MacIntosh, and Hickman 2000; Vaszonyi et al. 2001). As


in the 1991 data (Table 1, Column 1), six factors with eigenvalues greater
than one are produced (Table 1, Column 2) from the 23 items, and there is
discontinuity in the decreasing differences between eigenvalues for adjacent
factors (between the fifth and sixth factors), suggesting multidimensionality
(Nunnally and Bernstein 1994:482-84). On the other hand, the break in dif-
ferences in adjacent eigenvalues between the first and second factors (2.48),
compared to the break between the second and third factors (.076), suggests
that a one factor model is appropriate. Hence, like others have, we assume
there is enough evidence of unidimensionality to justify combining the 23
items into one general scale.
Our scale was derived by multiplying the regression factor score for each
item by the z score for each individual and summing across all items, a proce-
dure we use for all subsequent factor-based scales. In constructing this scale
and others to be described later, cases with missing values were deleted
listwise. The final scale has a mean of zero and a standard deviation of 1, with
scores ranging from –.31 to 2.12, and an alpha of .82.2 Our factor-based scale
is a departure from previous practice in that Grasmick et al. (1993) simply
summed z scores. We also score the items so that a high score indicates strong
self-control whereas Grasmick et al. let higher score reflect lower self-
control. It seems to make interpretation easier when all of the measures run in
the same direction with higher scores reflecting greater amounts of whatever
is being measured.

A FACTOR-BASED BEHAVIORAL SELF-CONTROL MEASURE

Hirschi and Gottfredson (1993:48) contend that the best measures of self-
control are behaviorally based. To accommodate that perspective, we employ
two behaviorally based measures of self-control. One was derived from self-
reports about 10 forms of noncriminal, problem behavior that according to
the logic of the theory, reasonably ought to reflect self-control. The items
concern marital arrangements, smoking, drinking, taking medicine with
minor illnesses, overeating, using seat belts, having accidents, financial plan-
ning or lack thereof, and education. Although factor analysis suggests a num-
ber of different factors instead of one underlying dimension assumed by the
theorists, we tried to follow their directives by combining into a single scale
the items that produced the most reliable measure. We began with a one-
factor forced solution to derive regression factor scores for the 10 best load-
ing items. They are listed in Table 2. Seven of them (2, 4, 5, 6, 8, 9, and 10)
used yes/no formats, which we coded so that higher scores indicate greater
self-control. Item #1 (frequency of drinking) used six response categories
from never to nearly every day, with the never end of the continuum getting

Downloaded from jrc.sagepub.com at UCSF LIBRARY & CKM on April 4, 2015


436 JOURNAL OF RESEARCH IN CRIME AND DELINQUENCY

TABLE 1: Differences in Eigenvalues for 1991 and 1994 Self-Control Scale

Factor 1991, Self-Control 1994, Self-Control

1 4.66 5.24
2 2.34 (2.32) 2.75 (2.48)
3 2.07 (0.27) 2.00 (0.76)
4 1.81 (0.26) 1.77 (0.23)
5 1.78 (0.03) 1.63 (0.14)
6 1.11 (0.67) 1.10 (0.53)

NOTE: Numbers in parentheses are the differences in eigenvalues between the de-
scending adjacent factors.

higher scores to reflect greater control. Item #3 (number of drinks in a typical


day during the last year) used the actual number reported, which we coded so
that fewer drinks (including zero for those who did not drink at all) indicate
greater self-control. Item #7, marital status, was coded into “presently mar-
ried,” “single, never married,” and “separated or divorced,” with being mar-
ried taken as more indicative of high self-control and being separated or
divorced as indicative of the least self-control.
In three instances, the high self-control response for those behavioral
items is not obvious. First, married people presumably have more self-
control than divorced or separated people, but they may not necessarily have
more self-control than single people. Second, it is not clear whether taking
medicine for minor illness is more indicative of self-control than “bucking
up” and taking none. We experimented with the scoring of the these two
items, settling on the scoring listed in Table 2, which permitted the most reli-
able scale (.62). The items were combined using the procedure described
before (multiplying the z score for each individual by the regression factor
score and summing across items), which produces a scale ranging from 2.34
to 1.18 with a mean of zero and a standard deviation of one.

A VARIETY TYPE BEHAVIORAL SELF-CONTROL INDEX

In one place, Hirschi and Gottfredson (1995:134) contend that the best
measures of a propensity to offend (presumably meaning low self-control)
are indexes of the number of different kinds of problem behaviors in which a
person engages. To implement that directive and to provide the most theoreti-
cally grounded test of the hypotheses at issue here, we created an additive
(variety) behavioral measure of self-control using the 10 forms of noncrimi-
nal, problem behavior noted above. Eight of the “problem behaviors” were
dichotomized into yes (0) and no (1) while two of them (frequency and vol-
ume of drinking) were dichotomized into some (0) and (1) none, with higher

Downloaded from jrc.sagepub.com at UCSF LIBRARY & CKM on April 4, 2015


TABLE 2: Items Used in Factor Scale of Behaviorally Based Self-Control
Downloaded from jrc.sagepub.com at UCSF LIBRARY & CKM on April 4, 2015

Factor Percentage with Lower


Item Loading Mean SD Self-Control Response

1. How often during the past year did you drink beer, wine, or hard liquor? .785 4.51 1.51 66% (any drinking)
2. Do you usually drink more than two or three alcoholic beverages during
a span of a week? .721 1.83 .38 17% (yes)
3. On a typical day when you drank in the past year, how many drinks did
you have? That is, how many beers, glasses or wine, mixed drinks and
shots of liquor did you have? .707 4.53 1.54 63% (any drinking)
4. Have you ever had a blackout while drinking, that is, where you drank
enough so that you couldn’t remember the next day what you had said or done? .638 1.80 .40 20% (yes)
5. Have you ever had difficulty stopping drinking before you became intoxicated? .425 1.79 .41 21% (yes)
6. Do you smoke tobacco products? .369 1.72 .45 28% (yes)
7. Current marital status .348 2.58 .72 15% (separated or divorced)
8. When you are in an automobile, do you always use the seat belt? .239 1.62 .49 38% (no)
9. Do you sometimes get so far in debt that it’s hard to see how you will get
out of it? .260 1.78 .42 22% (yes)
10. When you have a cold or some other minor ailment, do you usually take
some kind of medication? .217 1.74 .44 25% (no)
Alpha = .62
437
438 JOURNAL OF RESEARCH IN CRIME AND DELINQUENCY

scores on both sets of items indicating more self-control. Scores on the 10


items were then summed to create an index ranging from 0 to 10, with a mean
of 6.67 and a standard deviation of 2.12.

GENDER AND AGE

Our measure of gender is simply a dichotomized categorization, with


females scored one and males scored two. Age is the respondent’s report of
age on last birthday, with recorded ages ranging from 18 to 90. The full range
of ages is used in our regression analyses but in examining subgroups we use
four sets: 18 to 24, 25 to 44, 45 to 64, and 65 and older.

Measures of Dependent Variables

THE GENERAL SCALE OF CRIME/DEVIANCE

Respondents reported their past misbehavior and projected their future


chances of misbehavior, focusing on five offenses—assault (four items), tak-
ing something that did not belong to them worth less than $20 (two items),
cheating on income tax (two items), illegal gambling (two items), and driving
while under the influence (two items)—and using dichotomies, with one
being no or never and two being yes or some. The items are displayed in Table
3, top panel, which also shows their means, standard deviations, and factor
loading. We used our previously described factor-based procedure to con-
struct a scale ranging from –.88 to 3.39, with a mean of zero, a standard devi-
ation of one, and an alpha of .76.

A GOTTFREDSON/HIRSCHI-BASED CRIME INDEX

There is some ambiguity about whether Gottfredson and Hirschi’s (1990)


theory applies strictly to acts of crime as they define them (force or fraud for
self-gratification) or whether it also explains acts that involve neither force
nor fraud but are nevertheless illegal. Given that the theory is said to explain
acts that are “analogous” to crime in the sense of reflecting weak self-control,
the theory should apply to a wide variety of legal and illegal acts. Neverthe-
less, in order to give the theory its best chance to succeed in explaining gender
and age variations, we constructed a crime scale based entirely on acts of
force or fraud, presumably for personal gratification. This subset of misbe-
havior items is listed in the bottom panel of Table 3. Consistent with our ear-
lier described procedure, we conducted a forced single-factor analysis to
derive regression factor scores to use in constructing a composite scale

Downloaded from jrc.sagepub.com at UCSF LIBRARY & CKM on April 4, 2015


TABLE 3: Crime/Deviance Items

Factor
Item Loading Mean SD

General crime index items


Downloaded from jrc.sagepub.com at UCSF LIBRARY & CKM on April 4, 2015

1. As an adult, how often have you threatened someone with physical violence? .412 1.48 .74
2. In the future will you ever gamble illegally on a sporting event or other situation? .670 1.20 .45
3. In the future will you ever drive an automobile while under the influence of a moderate amount of alcohol? .557 1.23 .41
4. In the past 5 years have you ever driven an automobile while under the influence of a moderate amount of alcohol? .510 1.31 .46
5. As an adult, that is, since you were 18, how often have you beaten or punched another adult? .360 1.42 .72
6. In the future will you ever take something from someplace worth less than $20 that does not belong to you? .465 1.05 .23
7. In the future will you ever fail to report a certain income or claim an undeserved deduction on your income tax return? .611 1.17 .38
8. In the past 5 years have you ever taken something worth less than $20 that did not belong to you? .528 1.13 .33
9. In the future will you ever physically hurt another person on purpose? .500 1.08 .27
10. In the past 5 years have you ever physically hurt another person on purpose? .486 1.11 .31
11. In the past 5 years have you ever failed to report a certain income or claimed an undeserved deduction on
your income tax return? .641 1.18 .39
12. In the past 5 years have you ever gambled illegally on a sporting event or other situation? .722 1.28 .45
Gottfredson/Hirschi Crime Index Items
1. As an adult, that is, since you were 18, how often have you beaten or punched another adult? .413 1.42 .72
2. In the future will you ever take something worth less than $20 that does not belong to you? .674 1.05 .23
3. In the past 5 years have you ever failed to report certain income or claimed an undeserved deduction on your
income tax return? .663 1.18 .39
4. In the past 5 years have you ever taken something from someplace worth less than $20 that did not belong to you? .673 1.13 .33
5. In the past 5 years have you ever physically hurt another person on purpose? .480 1.11 .31
6. In the future will you ever physically hurt another person on purpose? .623 1.08 .27
7. In the future will you ever fail to report a certain income or claim an undeserved deduction on your income tax return? .674 1.17 .38
8. As an adult, how often have you threatened someone with physical violence? .410 1.48 .74
439
440 JOURNAL OF RESEARCH IN CRIME AND DELINQUENCY

ranging from –.70 to 4.45, with a mean of zero, a standard deviation of one,
and an alpha of .70.

VARIETY INDEXES

As noted before, Hirschi and Gottfredson (1995: 134) favor variety mea-
sures of misbehavior to reflect self-control. Presumably they also favor vari-
ety measures of crime/deviance. Moreover, some people might think that
analyses should be done separately for past and future offenses. Therefore,
we constructed separate variety indexes of future and past crime by summing
dichotomized indicators of having engaged in or projecting assault, theft, tax
cheating, illegal gambling, and DWI. The two range in scores from zero to
five. The future variety index has a mean of .80 and a standard deviation of
1.09 while the past variety index has a mean of 1.01 and a standard deviation
of 1.23.

Control Variables

We are interested in determining whether self-control, as an intervening


variable, interprets associations between gender and age and crime. How-
ever, long before self-control is fully established, other conditions associated
with gender and age may affect the gender/age-crime relationships. To elimi-
nate those conditions as possible contaminants, we include in all equations
nine control variables that are antecedent to self-control. They include gen-
der (in equations about age), age (in equations about gender), race, size of the
place where the respondent spent most time growing up, father’s occupation,
mother’s occupation, childhood supervision by the mother (supervision by
father contained too many missing cases to use), the person’s attachment to
the father, and attachment to the mother. Race is non-White (1) and White
(2). In the father and mother occupation index, responses are scored to allow
a higher score to indicate a more authoritative position. Unemployed equals
one; working for someone else and not supervising anybody is scored two;
working for someone and supervising people equals three; being self-
employed with no one working for them is scored four; and being self-
employed and an employer has a score of five. Mother’s supervision com-
bines answers to questions about whether she knew where the child was
when away from home and whether she knew who was with the child (four
responses from strongly agree to strongly disagree). Mother’s supervision
index, therefore, ranges from 2 to 8. Finally, attachment to the parent is
tapped by degree of agreement or disagreement with the following statement:
“I would like to be the kind of person my father [mother] was.”

Downloaded from jrc.sagepub.com at UCSF LIBRARY & CKM on April 4, 2015


Tittle et al. / A CHALLENGE TO SELF-CONTROL THEORY 441

Summary

We use four measures of self-control, four measures of crime/deviance,


and nine control variables; however, when results are similar for different
measures we report figures for only one. The self-control measures are (1)
the Grasmick et al. (1993) factor-based scale, (2) an additively scored version
of the Grasmick et al. scale (results for which we do not report), (3) a
behaviorally based factor scale, and (4) a 10-item behaviorally based variety
index. The dependent deviance indexes include (1) a factor-based general
deviance scale, (2) a factor-based G-H crime scale, (3) a variety measure of
past offenses, and (4) a variety measure of projections of future offenses.

Analysis

First, we examine the nature and magnitude of the relationships between


gender and age and the various measures of crime/deviance, as well as the
measures of self-control. Then, using multiple regression we determine the
effect on the relationships between gender and age and crime/deviance of
adding measures of self-control to the equations. In these analyses multi-
collinearity is not a problem because none of the relevant variables is related
to each other beyond .40. Finally, we examine the nature of the self-control/
misbehavior relationship by observing differences in the magnitude of the
regression coefficients for self-control predicting the crime/deviance mea-
sures among categories of gender and age. In testing the significance of dif-
ference in the regression coefficients between categories of individuals we
use the z test, calculated according to the recommendations of Paternoster
et al. (1998).

RESULTS

Consistent with previous research, the data show that for every crime
index in our study, males score significantly higher than females. The top
panel of Table 4 reports the mean scores on each of the four crime/deviance
indexes and each of the three measures of self-control separately for males
and for females, along with the zero-order regression coefficients reflecting
associations between gender and each of the crime measures as well as asso-
ciations between gender and self-control measures. Although all of the gen-
der differences in mean crime/deviance measures are substantial and statisti-
cally significant (row 3 of Table 4 reports the significance of the beta
representing the association of gender with specific crime indexes and mea-
sures of self-control, which we use instead of a test of the significance of

Downloaded from jrc.sagepub.com at UCSF LIBRARY & CKM on April 4, 2015


442

TABLE 4: Gender, Age, Crime/Deviance, and Self-Control


Downloaded from jrc.sagepub.com at UCSF LIBRARY & CKM on April 4, 2015

Crime/Deviance Measures Self-Control Measures


General G-H Past Future Cognitive Behavioral Behavioral
Deviance Index Variety Variety Factor Factor Variety

Gender
Mean scores:
a
Female (n = 197) –.22 –.17 .59 .77 .04 .23 7.1
Male (n = 148) .29 .23 1.08 1.33 –.04 –.29 6.6
b
Association .25* .20* .22* .22* –.04 –.26* –.23*
Age
Mean scores:
18 to 24 (n = 35) .62 .69 1.31 1.71 –.55 –.58 5.51
25 to 44 (n = 142) .12 .05 .89 1.23 –.05 –.19 6.12
45 to 64 (n = 102) –.04 –.04 .84 .94 .28 .16 7.11
65+ (n = 63) –.53 –.40 .30 .27 –.04 .50 7.85
Association –.30* –.26* –.23* –.33* .14* .32* .36*

a. Number of cases varies for each measure, due to listwise deletion of cases.
b. Zero-order beta.
*p < .05.
Tittle et al. / A CHALLENGE TO SELF-CONTROL THEORY 443

differences of means), the largest zero-order association is for the general


crime index (.25) where females have a mean score of –.22 and males have a
mean score of .20, and the smallest zero order association between gender
and a measure of crime/deviance is for the G-H Crime index (.20) where the
females have a mean of –.17 and the males have a mean of .23. Clearly, these
data show a gender difference in crime/deviance, which self-control might
interpret.
Our data also show age to be negatively associated with all four of the
crime/deviance indexes. This is consistent with the literature concerning
“ordinary” crime. Because the youngest age of our respondents is 18, near
the age where ordinary crime/deviance typically peaks, we would expect a
negative relationship between age and the indexes of crime/deviance. As the
bottom portion of Table 4 shows, there is a steady decline in mean scores on
the crime/deviance indexes from the 18 to 24 age group through the two mid-
dle age groups until the lowest levels of misbehavior are recorded for the cat-
egory of 65 and older. These patterns are confirmed by the coefficients in the
last row of Table 4, which show that age has significant negative zero order
associations with each of the crime/deviance measures, ranging from –.23 to
–.33. Hence, there is also a clear age association with misbehavior that self-
control might interpret.
However, whether self-control itself varies between females and males, as
the reasoning set forth by Gottfredson and Hirschi (1990:97-105, 144-49)
suggests, and among the age groupings, as our application of the Gottfredson/
Hirschi logic contends, depends somewhat on how self-control is measured.
The cognitive, factor-based self-control scale (the well known Grasmick
et al. [1993] scale) shows no significant gender difference (upper right hand
panel of Table 4) while the behavioral, factor-based scale and the behavioral
variety index both show significant relationships between gender and self-
control (–.26 and –.23, respectively). All three of the self-control measures
show significant associations with age (.14,.32, and .36, respectively), as our
interpretation of the logic of self-control theory predicts (lower right panel of
Table 4). Because the behavioral measures of self-control are both associated
significantly with gender and age, one might expect them to largely interpret
the associations between gender and age and crime/deviance, thereby pro-
viding support for self-control as the master variable.
Table 5 reports data bearing on whether the various measures of self-
control interpret the gender and age associations with the general deviance
scale (results are similar for all four of the measures of crime/deviance). The
columns in the table show regression coefficients from equations in which
different sets of variables are used to predict the general crime scale. Model 1
(Column 1) includes age and gender and the control variables that are ante-
cedent to self-control; model 2 (Column 2) includes gender and age, all of the

Downloaded from jrc.sagepub.com at UCSF LIBRARY & CKM on April 4, 2015


444 JOURNAL OF RESEARCH IN CRIME AND DELINQUENCY

TABLE 5: Standardized Regression Coefficients Representing Effects of Gender and Age in


Predicting General Deviance Scale, Controlling Different Measures of Self-
Control

Predictors Model 1 Model 2 Model 3 Model 4

Race –.02 .00 –.08 –.01


Size of place .07 .08 –.02 .07
Father occupation .03 .03 –.01 .03
Mother occupation .08 .09 .07 .06
Father attachment .07 .09 .05 .06
Mother supervision –.21* –.18* –.21* –.20*
Mother attachment –.08 –.09 –.09 –.07
Gender .16* .18* .08 .10
Age –.22* –.20* –.10 –.11
Self-control:
Cognitive factor –.24*
Behavioral factor –.40*
Behavioral variety –.38*
2
Adj. R .18 .24 .31 .30

NOTE: The italicized figures highlight the coefficients for age and gender that are ren-
dered insignificant by the control variable.
*p < .05; t test for regression coefficients.

control variables, and the cognitive, factor-based measure of self-control.


Model 3 includes gender and age, the control variables, and the behavioral,
factor-based scale of self-control. And model 4 includes gender and age, the
control variables, and the behavioral variety measure of self-control.
This analysis shows three noteworthy outcomes. First, as the figures in
Column 1 demonstrate, gender and age continue to be significantly associ-
ated with misbehavior even with the control variables included (.16 and –.22,
respectively). Second, consistent with the theory and with other research (see
Pratt and Cullen 2000), all of the measures of self-control show at least mod-
est, statistically significant associations with crime/deviance (–.24, –.40, and
–.38, respectively) even with the other variables controlled.
Third, and most important, the coefficients for gender and age are reduced
below significance when the behaviorally based measures of self-control are
added to the equations. The coefficient for gender is reduced from a signifi-
cant .16 to an insignificant .08 (Model 3, in bold) when the behavioral, factor-
based scale is controlled and from .16 to an insignificant .10 (Model 4) when
the behavioral variety index is controlled. Similarly the coefficient for age is
reduced from a significant –.22 to an insignificant –.10 when the factor-based
scale is included in the equation and from –.22 to an insignificant –.11 when
the variety self-control index is controlled.. Therefore, consistent with prior
research (Burton et al. 1998; Gibbs et al. 1998; LaGrange and Silverman

Downloaded from jrc.sagepub.com at UCSF LIBRARY & CKM on April 4, 2015


Tittle et al. / A CHALLENGE TO SELF-CONTROL THEORY 445

1999), it appears that at least with some measures, self-control predicts


offending and interprets relationships between gender and age and measures
of crime/deviance just as the logic of the theory implies. With the Grasmick
et al. (1993) scale, self-control has an independent and significant associa-
tion with the crime/deviance measure but including it in the equation does not
have much effect on the coefficients for gender and age.
Our analyses provide considerable support for the implications of the the-
ory that self-control is the master variable in explaining crime/deviance.
However, the results are not so supportive of the second major implication of
the theory—that the effects of self-control are general; that is, they are
equally likely for all subcategories of the population. The idea of generality
implies that self-control operates similarly for males and females and for the
various age grouping. The figures reported in Table 6, using the behavioral
measures of self-control that proved so effective in interpreting the gender
and age associations, however, provide mixed evidence about the generality
hypothesis.
Table 6 has four column-panels, one for each of the measures of crime/
deviance, and two row-panels, one for gender differences and one for age dif-
ferences. Within each column panel there are two columns of figures. The
first set of figures in each column panel are produced when the factor-based
measure of self-control is used as the predictor. The second set of figures in
each column panel result from using the variety self-control measure as the
predictor. The upper portion of the table shows how well, with the control
variables in the equations, the self-control measures predict each of the
crime/deviance measures within categories of male and female, with signifi-
cant associations designated by an asterisk. The third row of the upper row-
panel shows the differences in magnitude of the predictive coefficients
between females and males, with a positive difference indicating a higher
coefficient for males, and a superscripted plus indicating a significant differ-
ence between the coefficients for females and males.
Two observations are relevant. First, all of the self-control coefficients for
females and males are negative and statistically significant, confirming that
self-control, behaviorally measured, predicts deviance for both sexes even
with other variables controlled. Second, for three of the measures of crime/
deviance the predictive coefficients do not differ significantly between
females and males, a result consistent with the generality implications of the
theory. However, the coefficients for females and males do differ signifi-
cantly in their ability to predict scores on the future variety index of
misbehavior.
The second row-panel of Table 6 shows similar figures for age categories.
Each coefficient represents the effect of self-control among those of a given
age on a particular measure of crime/deviance, and its predictive significance

Downloaded from jrc.sagepub.com at UCSF LIBRARY & CKM on April 4, 2015


446 JOURNAL OF RESEARCH IN CRIME AND DELINQUENCY

a
TABLE 6: Standardized Regression Coefficients Representing Effects of Factor-Based,
Behavioral Measure of Self-Control and Variety Based Measure of Self-Control on
Various Measures of Crime/Deviance, within Gender and Age Categories.

Self-Control Measures
General
Deviance G-H Index Past Variety Future Variety
Factor Variety Factor Variety Factor Variety Factor Variety

Female –.38* –.41* –.22* –.30* –.32* –.36* .35* –.34*


Male –.41* –.36* –.26* –.26* –.41* –.36* –.27* –.20*
+
Difference .04 .05 .04 .04 .09 .00 .08 .14
18 to 24 –.77* –.68* –.87* –.71* –.49 –.51* –.65* –.48
25 to 44 –.39* –.33* –.20* –.20* –.38* –.33* –.31* –.22*
45 to 64 –.40* –.39* –.17 –.23* –.44* –.39* –.34* –.33*
65+ –.22 –.28 –.14 –.21 –.18 –.26 –.04 –.08
+ + + + + +
Maximum difference .55 .40 .73 .50 .31 .25 .61 .40

a. All equations contain controls for age, gender, race, size of place of childhood resi-
dence, father’s occupation, mother’s occupation, attachment to father, attachment to
mother, and mother’s supervision, except when a subcategory of the variable is being
examined.
+
*p < .05; t test for regression coefficients. p < .05; z test according to formula presented
by Paternoster et al. (1998) for differences in coefficients.

is indicated by an asterisk. In addition, the fifth row of this second panel


shows the maximum difference in magnitude of the predictive coefficients
between the age categories. For example, in Column 1 the highest predictive
coefficient (–.77) is for the young age category and the smallest coefficient is
for the oldest age category (–.22), producing a maximum difference of .55. A
superscripted + in this fifth row indicates that this maximum difference in the
coefficients for the two most extreme age categories is statistically significant
according to the Paternoster et al. (1998) z test. In addition, we test the signifi-
cance of differences for the coefficients in all combinations of age categories
(not shown).
The results show several things bearing on the generality hypothesis.
First, consistent with Burton et al. (1999) the self-control measures do not
significantly predict any of the four measures of deviance among all of the
age categories. For example, both measures significantly predict the general
deviance scale among the first three age categories but not among the aged.
And the factor measure predicts past variety deviance significantly among
the middle two age groups but not among the young and the aged, while the
variety measure of self-control predicts past variety deviance significantly
among the first age groups but not among the aged. This failure to predict sig-
nificantly within all age categories is contrary to what would be expected if
the generality hypothesis were correct. In addition, the differences in pre-

Downloaded from jrc.sagepub.com at UCSF LIBRARY & CKM on April 4, 2015


Tittle et al. / A CHALLENGE TO SELF-CONTROL THEORY 447

dictive coefficients between various age groups often prove to be significant,


as suggested by the “maximum differences” for three of the four measures of
crime/deviance being significant. In most cases, the youngest age category
shows significantly different coefficients from all of the other age categories,
and in many instances, other age categories differ among themselves as well
(not shown). It is only the coefficients among the age categories when the
past variety measure of crime/deviance is used that prove to be no different
from each other. To make sure these results are not merely a function of
restricted variation within some of the age groups, we reconstituted the mea-
sures, standardizing within age groups. Substantive results using the within-
age-group standardized scores (not shown) are similar to those shown. These
findings again provide mostly contradictory evidence for the generality the-
sis as we interpret it.

SUMMARY AND DISCUSSION

Our results add somewhat to the body of confirming evidence for self-
control theory but they do not provide unambiguous support. The claims of
the theory that self-control is the main variable leading to criminal/deviant
behavior are strengthened by some of our results, but the parallel claims that
self-control accounts similarly for misbehavior in all of its variations and dis-
tributions appears to be overly optimistic. Moreover, support for the hypothe-
ses we examine here appears to depend on how self-control is measured. We
regard this lack of robustness as damaging to the stronger claims of the
theory.
Our data show that all of the measures of self-control significantly predict
crime/deviance when other variables antecedent to self-control are con-
trolled. In addition, controlling the behavioral measures of self-control
reduces relationships between gender and age and misbehavior below signif-
icance. These results are consistent with previous studies and they are what
would be expected if self-control has the potency its proponents assume.
However, the evidence for the generality argument is quite ambiguous, again
consistent with past research showing much evidence of contingencies in the
operation of self-control (Burton et al. 1998, 1999; Giner-Sorolla 2001;
Keane et al. 1993; LaGrange and Silverman 1999; Lynam et al. 2000; Nagin
and Paternoster 1993; Nakhaie et al. 2000). Moreover, the fact that sup-
portive results appear only when self-control and crime/deviance are
measured in specific ways and fail to appear when measured in other ways
detracts from a strong conclusion. To be sure, the best support for the the-
ory in these analyses—being able to interpret gender and age associations by
controlling self-control—appears when self-control is measured behaviorally,

Downloaded from jrc.sagepub.com at UCSF LIBRARY & CKM on April 4, 2015


448 JOURNAL OF RESEARCH IN CRIME AND DELINQUENCY

as the theorists recommend. However, even those measures do not generally


support expectations that the effects of self-control will be universal and
approximately equally strong in all subgroups. The collective body of evi-
dence that has already been compiled draws much of its strength from results
showing that a wide range of different measures of self-control are persis-
tently associated with a range of measures of misbehavior. To find that the
strongest claims of the theory are not always supported and, even when sup-
ported, appear to depend on particular forms of measurement, suggests that
self-control per se may not be as powerful a variable as the theory suggests it
is. On the other hand, it is important to note that the strongest implications of
the theory can be empirically supported if researchers selectively pick and
chose outcomes based on particular measures of self-control and particular
measures of crime/deviance.

Caveats

We must be careful, of course, in drawing conclusions from these results.


First, they bear on what we regard as the strongest implications of the theory.
Regardless of the confidence one might place in our findings about self-
control interpreting gender and age effects on crime/deviance or about the
generality of self-control effects, self-control does appear to be an important
factor in criminal/deviant behavior. To the extent that our data challenge the
theory, it is does so with respect to self-control being the primary, or perhaps
only, cause of misbehavior and the implication that its effects are universal
and similar in magnitude in all conditions, that is, that self-control operates
without contingencies. It may be that the theorists’ actual assumptions/
claims are much more modest than our interpretation suggests. After all,
given their tendency to state things in one place but then to qualify them in
another venue, it is often difficult to identify their exact position.
Second, the data are self-reported and come from the residents of only one
city. The theorists are critical of surveys on the grounds that self-control itself
affects the nature of the sample as well as response. Potential errors built into
survey data might render our results suspect. In addition, it is possible,
though hardly likely, that people in Oklahoma City are quite different in rele-
vant ways from people elsewhere. If other research consistently and more
firmly shows the predicted results, we could easily concede that our some-
what ambiguous findings fall into the category of “factoid,” a minor and
unimportant deviation from a general pattern. So far, however, other research
has not shown that self-control accounts for almost all of the variance in
crime/deviance; it has not shown that self-control and misbehavior are
related under all circumstances, certainly not to an approximately equal
degree, and correlatively it has not shown conclusively that self-control is the

Downloaded from jrc.sagepub.com at UCSF LIBRARY & CKM on April 4, 2015


Tittle et al. / A CHALLENGE TO SELF-CONTROL THEORY 449

fundamental element in crime and deviance. Therefore our results and inter-
pretation are in line with the collected body of evidence, though in some ways
more supportive of the stronger claims of the theory.
Finally, as noted before, some of the results could stem from statistical
artifacts. Our behavioral measures of self-control contain indicators of
excessive drinking while some of our measures of crime/deviance include
DWI, so there is the possibility of a degree of tautology. However, the results
are similar using the Gottfredson-Hirschi Crime Index, which contains no
indicator of DWI, and they are similar using variety indexes of both self-
control and crime/deviance, which simply count drinking indicators as one
step additions to the index scores. Therefore, tautology would seem to have
minimal impact on the conclusions. In addition, if variation is too restricted,
which it may be because of natural concentrations of self-control or tenden-
cies toward misbehavior, or because of faulty representativeness for the gen-
der and age categories, then our methods cannot unambiguously show the
results anticipated by self-control theory. Although the alternative standard-
ization procedures we used to counter this possibility do not contradict the
conclusions, one cannot be absolutely sure that restricted variation is not still
at work because people with extremely low self-control may have been
missed in the survey. The only remedy for this potential problem is to investi-
gate the hypotheses in a wide variety of contexts, using different methods. In
the meantime, we advance the tentative conclusion that self-control theory is
much better than its critics recognize but is not as good as its proponents
think. Before it can become the theory it wants to be, the authors, or others,
will probably have to expand the notion of self-control to include more than
ability to exercise it and theorists will probably have to systematically bring
other variables into the theory, showing how they intermesh with, and condi-
tion the effects of, self-control. In addition, they will have to provide clearer
guidelines about appropriate measurement.

CONCLUSION

Our results provide supportive but somewhat ambiguous evidence con-


cerning self-control theory. On one hand, the data seem to indicate that self-
control, at least in some forms of measurement, performs largely as its propo-
nents might wish. Some of our measures of self-control show consistent and
at least modest associations with measures of crime/deviance and appear to
interpret gender and age variations. We regard this as quite strong support for
the theory. However, the effects of self-control do not appear to be entirely
general, failing to appear among some subcategories of individuals. And, the
conclusions depend on how self-control and crime/deviance are measured,

Downloaded from jrc.sagepub.com at UCSF LIBRARY & CKM on April 4, 2015


450 JOURNAL OF RESEARCH IN CRIME AND DELINQUENCY

calling into question the robustness of the theory’s hypotheses. This is partic-
ularly true because measures supporting the theory’s implications are not
consistently of the same form.

NOTES

1. There may be some question as to whether this study shows an effect for self-control. The
authors examine the relationship between self-control and criminal behavior, controlling a mea-
sure of “imprudent behaviors.” However, Gottfredson and Hirschi (1990) contend that engaging
in imprudent behaviors is also a product of low self-control and can serve as a measure of low
self-control. Therefore, treating imprudent behavior as an intervening variable would seem to
inappropriately eliminate a potential direct association between low self-control and criminal
behavior. However, there is debate on this point (Britt 2000; Paternoster and Brame 1998, 2000).
2. We also combined the items additively and used the additive index as an alternative mea-
sure of self-control.

REFERENCES

Arneklev, Bruce J., Harold G. Grasmick, and Robert J. Bursik, Jr. 1999. “Evaluating the
Unidimensionality and Invariance of ‘Low Self-Control.’ ” Journal of Quantitative Crimi-
nology 15:307-31.
Arneklev, Bruce J., Harold G. Grasmick, Charles R. Tittle, and Robert J. Bursik, Jr. 1993. “Low
Self-Control and Imprudent Behavior.” Journal of Quantitative Criminology 9:225-47.
Avakame, Edem F. 1998. “Intergenerational Transmission of Violence, Self-Control, and Conju-
gal Violence: A Comparative Analysis of Physical Violence and Psychological Aggression.”
Violence and Victims 13:301-16.
Baumeister, Roy F. and Julie Juola Exline. 1999. “Virtue, Personality, and Social Relations: Self-
Control As the Moral Muscle.” Journal of Personality 67:1165-94.
Baumeister, Roy F. and Julie Juola Exline. 2000. “Self-Control, Morality, and Human Strength.”
Journal of Social and Clinical Psychology 19:29-42.
Baumeister, Roy F., Todd F. Heatherton, and Dianne M. Tice. 1994. Losing Control: How and
Why People Fail at Self-Regulation. New York: Academic Press.
Britt, Chester L. 2000. “Comment on Paternoster and Brame.” Criminology 38:965-70.
Brownfield, David and Ann Marie Sorenson. 1993. “Self-Control and Juvenile Delinquency:
Theoretical Issues and An Empirical Assessment of Selected Elements of a General Theory
of Crime.” Deviant Behavior 4:243-64.
Burton, Velmer S., Jr., Francis T. Cullen, T. David Evans, Leanne Fiftal Alarid, and R. Gregory
Dunaway. 1998. “Gender, Self-Control, and Crime.” Journal of Research in Crime and
Delinquency 35:123-47.
Burton, Velmer S., Jr., T. David Evans, Francis T. Cullen, Kathleen M. Ovares, and R. Gregory
Dunaway. 1999. “Age, Self-Control, and Adults’ Offending Behaviors: A Research Note
Assessing a General Theory of Crime.” Journal of Criminal Justice 27:45-54.
Caspi, Avshalom, Terrie E. Moffitt, Phil A. Silva, Magna Stouthamer-Loeber, Robert F. Krueger,
and Pamela S. Schmutte. 1994. “Are Some People Crime-Prone? Replications of the Person-
ality-Crime Relationship Across Countries, Genders, Races, and Methods.” Criminology
32:163-95.

Downloaded from jrc.sagepub.com at UCSF LIBRARY & CKM on April 4, 2015


Tittle et al. / A CHALLENGE TO SELF-CONTROL THEORY 451

Cochran, John K., Peter B. Wood, Christine S. Sellers, Wendy Wikerson, and Mitchell B.
Chamlin. 1998. “Academic Dishonesty and Low Self-Control: An Empirical Test of a Gen-
eral Theory of Crime.” Deviant Behavior 19:227-55.
Elliott, Delbert S. 1994. “Serious Violent Offenders: Onset, Developmental Course, and Termi-
nation.” Criminology 32:1-21.
Evans, T. David, Francis T. Cullen, Velmer S. Burton, Jr., R. Gregory Dunaway, and Michael L.
Benson. 1997. “The Social Consequences of Self-Control: Testing the General Theory of
Crime.” Criminology 35:475-501.
Finkel, Eli J. and W. Keith Campbell. 2001. “Self-Control and Accommodation in Close Rela-
tionships: An Interdependence Analysis.” Journal of Personality and Social Psychology 81:
263-77.
Forde, David R. and Leslie W. Kennedy. 1997. “Risky Lifestyles, Routine Activities, and the
General Theory of Crime.” Justice Quarterly 14:265-94.
Gibbs, John J. and Dennis Giever. 1995. “Self-Control and Its Manifestations Among University
Students: An Empirical Test of Gottfredson and Hirschi’s General Theory.” Justice Quar-
terly 12:231-35.
Gibbs, John J., Dennis Giever, and Jamie S. Martin. 1998. “Parental Management and Self-
Control: An Empirical Test of Gottfredson and Hirschi’s General Theory.” Journal of
Research in Crime and Delinquency 35:40-70.
Giner-Sorolla, Roger. 2001. “Guilty Pleasure and Grim Necessities: Affective Attitudes in
Dilemmas of Self-Control.” Journal of Personality and Social Psychology 80:206-21.
Gottfredson, Michael R. and Travis Hirschi. 1990. A General Theory of Crime. Palo Alto, CA:
Stanford University Press.
Grasmick, Harold G., Charles R. Tittle, Robert J. Bursik, Jr, and Bruce K. Arneklev. 1993.
“Testing the Core Empirical Implications of Gottfredson and Hirschi’s General Theory of
Crime.” Journal of Research in Crime and Delinquency 30:5-29.
Heimer, Karen and Stacy DeCoster. 1999. “The Gendering of Violent Delinquency.” Criminol-
ogy 37:277-317.
Hirschi, Travis and Michael R. Gottfredson. 1983. “Age and the Explanation of Crime.” Ameri-
can Journal of Sociology 89:552-84.
———. 1993. “Commentary: Testing the General Theory of Crime.” Journal of Research in
Crime and Delinquency 30:47-54.
———. 1995. “Control Theory and the Life-Course Perspective.” Studies on Crime and Crime
Prevention 4:131-42.
———. 2000. “In Defense of Self-Control.” Theoretical Criminology 4:55-69.
Junger, Marianne and Richard E. Tremblay. 1999. “Self-Control, Accidents, and Crime.” Crimi-
nal Justice and Behavior 26:485-501.
Keane, Carl, Paul S. Maxim, and James J. Teevan. 1993. “Drinking and Driving, Self-Control,
and Gender: Testing a General Theory of Crime.” Journal of Research in Crime and Delin-
quency 30:30-46.
LaGrange, Teresa C. and Robert A. Silverman. 1999. “Low Self-Control and Opportunity:
Testing the General Theory of Crime As an Explanation for Gender Differences in Delin-
quency.” Criminology 37:41-72.
Longshore, Douglas. 1998. “Self-Control and Criminal Opportunity: A Prospective Test of the
General Theory of Crime.” Social Problems 45:102-13.
Longshore, Douglas and Susan Turner. 1998. “Self-Control and Criminal Opportunity: Cross-
Sectional Test of the General Theory of Crime.” Criminal Justice and Behavior 25: 81-98.
Longshore, Douglas, Susan Turner, and Judith A. Stein. 1996. “Self-Control in a Criminal Sam-
ple: An Examination of Construct Validity.” Criminology 34:209-28.

Downloaded from jrc.sagepub.com at UCSF LIBRARY & CKM on April 4, 2015


452 JOURNAL OF RESEARCH IN CRIME AND DELINQUENCY

Longshore, Douglas, Judith A. Stein, and Susan Turner. 1998. “Reliability and Validity of a Self-
Control Measure: Rejoinder.” Criminology 36:175-82.
Lynam, Donald R., Avshalom Caspi, Terrie E. Moffitt, Per-Olof Wikstrom, Rolf Loeber, and
Scott Novak. 2000. “The Interaction Between Impulsivity and Neighborhood Context in
Offending: The Effects of Impulsivity Are Stronger in Poorer Neighborhoods.” Journal of
Abnormal Psychology 109:563-74.
Miller, Joshua D. and Donald Lynam. 2001. “Structural Models of Personality and Their Rela-
tion to Antisocial Behavior: A Meta-Analytic Review.” Criminology 39:765-98.
Muraven, Mark, Dianne M. Tice, and Roy Baumeister. 1998. “Self-Control As a Limited
Resource: Regulatory Depletion Patterns.” Journal of Personality and Social Psychology
74:774-89.
Nagel, Ilene H. and John Hagan. 1983. “Gender and Crime: Offense Patterns and Criminal Court
Sanctions.” Pp. 91-124 in Crime and Justice: An Annual Review of Research, Vol. 4, edited by
Michael Tonry and Norval Morris. Chicago: University of Chicago Press.
Nagin, Daniel S. and Raymond Paternoster. 1993. “Enduring Individual Differences and Ratio-
nal Choice Theories of Crime.” Law and Society Review 27:467-96.
Nakhaie, M. Reza, Robert A. Silverman, and Teresa C. LaGrange. 2000. “Self-Control and
Social Control: An Examination of Gender, Ethnicity, Class and Delinquency.” Canadian
Journal of Sociology 25:35-9.
Nunnally, Jum C. and Ira H. Bernstein. 1994. Psychometric Theory, 3rd ed. New York: McGraw-
Hill.
Paternoster, Raymond and Robert Brame. 1998. “The Structural Similarity of Processes Gener-
ating Criminal and Analogous Behaviors.” Criminology 36:633-70.
———. 2000. “On the Association Among Self-Control, Crime, and Analogous Behaviors.”
Criminology 38:971-82.
Paternoster, Raymond, Robert Brame, Paul Mazerole, and Alex Piquero. 1998. “Using the Cor-
rect Statistical Test for the Equality of Regression Coefficients.” Criminology 36:859-66.
Piquero, Alex R., Randall MacIntosh, and Matthew Hickman. 2000. “Does Self-Control Affect
Survey Response? Applying Exploratory, Confirmatory, and Item Response Theory Analy-
sis to Grasmick et al.’s Self-Control Scale.” Criminology 38:897-929.
Piquero, Alex R. and Andre B. Rosay. 1998. “The Reliability and Validity of Grasmick et al.’s
Self-Control Scale: A Comment on Longshore et al.” Criminology 36:157-74.
Piquero, Alex R. and Stephen Tibbetts. 1996. “Specifying the Direct and Indirect Effects of Low
Self-Control and Situational Factors in Offender’s Decision Making: Toward a More Com-
plete Model of Rational Offending.” Justice Quarterly 13:481-510.
Polakowski, Michael. 1994. “Linking Self- and Social-Control with Deviance: Illuminating the
Structure Underlying a General Theory of Crime and Its Relation to Deviant Activity.” Jour-
nal of Quantitative Criminology 10:41-78.
Pratt, Travis C. and Francis T. Cullen. 2000. “The Empirical Status of Gottfredson and Hirschi’s
General Theory of Crime: A Meta-Analysis.” Criminology 38:931-64.
Sellers, Christine S. 1999. “Self-Control and Intimate Violence: An Examination of the Scope
and Specification of the General Theory of Crime.” Criminology 37:375-404.
Smith, Douglas A. and Christy Visher. 1980. “Sex and Involvement in Deviance/Crime: A Quan-
titative Review of the Empirical Literature.” American Sociological Review 45:691-701.
Steffensmeier, Darrell and Emilie Andersen Allan. 1995. “Age-Inequality and Property Crime:
The Effects of Age-Linked Stratification and Status-Attainment Processes on Patterns of
Criminality Across the Life Course.” Pp. 95-115 in Crime and Inequality, edited by John
Hagan and Ruth D. Peterson. Stanford, CA: Stanford University Press.
———. 2000. “Looking for Patterns: Gender, Age, and Crime.” Pp. 83-113 in Criminology,
edited by Joseph F. Sheley. Belmont, CA: Wadsworth.

Downloaded from jrc.sagepub.com at UCSF LIBRARY & CKM on April 4, 2015


Tittle et al. / A CHALLENGE TO SELF-CONTROL THEORY 453

Steffensmeier, Darrell, Emilie Anderson Allan, and Cathy Streifel. 1989a. “Development and
Female Crime: A Cross-National Test of Alternative Explanations.” Social Forces 68:262-
83.
Steffensmeier, Darrell , Emilie Andersen Allan, Miles D. Harer, and Cathy Streifel. 1989b. “Age
and the Distribution of Crime.” American Journal of Sociology 94:803-31.
Sutherland, Edwin H., Donald R. Cressey, and David F. Luckenbill. 1992. Principles of Crimi-
nology, 11th ed. Dix Hills, NY: General Hall.
Tittle, Charles R. and Harold G. Grasmick. 1997. “Criminal Behavior and Age: A Test of Three
Provocative Hypotheses.” Journal of Criminal Law and Criminology 88:309-42.
Tittle, Charles R. and Raymond Paternoster. 2000. Social Deviance and Crime: An Organiza-
tional and Theoretical Approach. Los Angeles: Roxbury Publishing.
Tittle, Charles R. and David A. Ward. 1993. “The Interaction of Age with the Correlates and
Causes of Crime.” Journal of Quantitative Criminology 9:3-53.
Vazsonyi, Alexander T., Lloyd E. Pickering, Marianne Junger, and Dick Hessing. 2001. “An
Empirical Test of a General Theory of Crime: A Four-Nation Comparative Study of Self-
Control and the Prediction of Deviance.” Journal of Research in Crime and Delinquency
38:91-131.
Warren, Marquerite. 1991. Comparing Male and Female Offenders. Newbury Park, CA: Sage.
White, Jennifer, Terrie E. Moffitt, Avshalom Caspi, Dawn J. Bartusch, Douglas S. Needles, and
Magna Stouthamer-Loeber. 1994. “Measuring Impulsivity and Examining Its Relationship
to Delinquency.” Journal of Abnormal Psychology 103:192-205.
Wilson, James Q. and Richard J. Herrnstein. 1985. Crime and Human Nature. New York: Simon
and Schuster.
Winfree, L. Thomas and Frances P. Bernat. 1998. “Social Learning, Self-Control, and Substance
Abuse by Eighth Grade Students: A Tale of Two Cities.” Journal of Drug Issues 28:539-38.
Wood, Peter B., Betty Pfefferbaum, and Bruce J. Arneklev. 1993. “Risk Taking and Self-Control:
Social Psychological Correlates of Delinquency.” Journal of Crime and Justice 16:111-30.
Wright, Bradley R. Entner, Avshalom Caspi, Terrie E. Moffitt, and Phil A. Silva. 1999. “Low
Self-Control , Social Bonds, and Crime: Social Causation, Social Selection, or Both?” Crim-
inology 37:479-514.

Charles R. Tittle earned his PhD at the University of Texas, Austin, in 1965. He is cur-
rently professor of sociology and Goodnight/Glaxo-Wellcome Endowed Chair of Social
Science at North Carolina State University, after having previously served on the facul-
ties of Washington State University, Florida Atlantic University, and Indiana University.
His scholarly interests focus on theory development and testing.

David A. Ward received his masters and doctoral degrees from the University of Florida
at Gainesville. Upon completing his graduate studies, he accepted a faculty position at
Washington State University. After having been at Washington State for over a decade,
he moved to the Department of Sociology at Clemson University. He has published in a
wide range of professional journals, including The Journal of Health and Social Behav-
ior, Social Psychology Quarterly, and Deviant Behavior.

Harold G. Grasmick is professor of sociology at the University of Oklahoma. His re-


search over the years has concentrated on various issues in control theory and on cul-
tural and social factors that shape attitudes toward punishment.

Downloaded from jrc.sagepub.com at UCSF LIBRARY & CKM on April 4, 2015

Potrebbero piacerti anche