Willingness to pay for rehabilitation versus punishment to reduce adult and juvenile crime.
Jones, Craig G.A. ; Weatherburn, Don J.
Introduction
One of the most common findings from research on public attitudes
to sentencing is that, at face value, members of the public are very
punitive. When asked broad questions such as "do you think that
sentences handed down by the courts are too lenient, about right or too
harsh", somewhere between two-thirds and three-quarters of people
respond that sentences are too lenient. These attitudes are pervasive
across all western countries where surveys have been conducted and do
not vary greatly over time (Roberts et al. 2003).
While surveys such as these show that the public express widespread
dissatisfaction about sentencing, it is equally clear that the public
are largely misinformed about trends in crime, conviction rates and
sentencing. Studies spanning several decades have shown that people
think that crime is increasing when it is declining. They overestimate
the proportion of crimes that involve violence, while underestimating
conviction and imprisonment rates. There is a widespread lack of
knowledge about statutory maximum and minimum penalties, and people have
little knowledge about sentencing alternatives (Doob & Roberts 1988;
Hough & Roberts 1998; Indermaur 1987; Jones & Weatherburn 2010;
Mattinson & Mirrlees-Black 2000; Weatherburn & Indermaur 2004).
Public ignorance about crime and criminal justice is hardly surprising
given that most people learn about both through the news media. In one
recent study conducted in New South Wales (NSW), Australia, Jones,
Weatherburn and McFarlane (2008) surveyed a random sample of residents
and asked them to rate the most influential sources of information about
the criminal justice system. The top three responses were 'TV/radio
news', 'broadsheet newspaper' and 'local
newspaper'. Because of their inherent newsworthiness, crimes that
are unusual or violent and sentences that appear to be out of line with
community expectations tend to be over-represented in media coverage of
crime (for example, Roberts & Grossman 1990). Some scholars have
argued that this media portrayal of crime provides a platform that
favours punishment as the popular policy response to crime (Roberts et
al. 2003, p.76).
Roberts and colleagues (2003) argue that this public
misunderstanding about crime and criminal justice outcomes has both
driven and been exploited by politicians to implement more punitive
penal policies. There is certainly evidence of an increase in
punitiveness in many countries. For example, use of imprisonment as a
sanction has grown across much of the Western world --nowhere more so
than in the United States (Roberts et al. 2003). In NSW, where the
current research was conducted, adult and juvenile prison populations
have both substantially increased in recent years. The sentenced adult
prison population has increased by about 20 per cent since the mid-1990s
(to 10,368 inmates in 2009; Corben 2010) and the number of young people
admitted to juvenile justice centres on full-time custodial orders has
increased by 73 per cent over the last five years (to 711 admissions in
the 2008-09 financial year; NSW Department of Juvenile Justice 2009).
Much of the growth in the size of the Australian prison population is
due to harsher sentencing and penal policies (Gorta & Eyland 1990;
Matka 1991). This is reflected by increases in both the number of people
being sentenced to prison and the length of prison sentences handed out
(Lulham & Fitzgerald 2008). These policies appear to have been
introduced on the assumption that the general public holds fairly
punitive attitudes to offenders (Weatherburn & Indermaur 2004).
Recent evidence from the United States calls this assumption into
question. These studies have adopted a methodology from environmental
economics research known as contingent valuation to study how people
value various policy responses to crime. Contingent valuation surveys
ask people to express how much they would be willing to pay for a good
described in the proffered scenario. In environmental economics, this
might reflect willingness to pay for a certain reduction in
environmental pollution. In a health scenario this might reflect
willingness to pay for some core improvement in one's own health
status. In criminology, the surveys are framed in terms of
respondents' willingness to pay for reductions in crime.
Nagin and colleagues (2006) asked members of the Pennsylvanian
public how much additional tax they would be willing to pay to bring
about a measurable reduction in juvenile crime. Half of the sample was
informed that this reduction would be produced by way of a
rehabilitation program, while the other half were told that it would
come about by incarcerating young offenders for longer. The study found
that respondents offered the rehabilitation scenario were, on average,
willing to pay more to reduce crime than those offered the punishment
scenario. Piquero and Steinberg (2010) replicated this study in
Pennsylvania and three other US states (Illinois, Louisiana and
Washington) two years later. Like Nagin and colleagues (2006), Piquero
and Steinberg (2010) found that, when averaged across the four states,
survey respondents were willing to pay more for rehabilitation programs
than for additional incarceration. However, when broken down by state,
only three of the four states showed this preference for rehabilitation.
In Louisiana, the amount participants were willing to pay was equivalent
for rehabilitation and additional prison time.
These findings have important implications for criminal justice
policy decisions. While increases in imprisonment almost certainly have
some effect on the crime rate by incapacitating offenders (Weatherburn,
Hua & Moffatt, 2006), imprisonment is also a very expensive means of
controlling crime (Chan 1995). In 2009-10, the total net recurrent and
capital cost of keeping an offender in prison in NSW is estimated to
have cost taxpayers $271 per inmate per day (Productivity Commission
2011). At the same time as prison rates have been rising, governments in
NSW and elsewhere have shown a willingness to invest in programs that
divert people out of the criminal justice system and into rehabilitation
programs. The growth in problem solving courts such as mental health and
drug courts in many jurisdictions is a good example of this investment.
These programs aim to treat the issues that underlie offending (for
example, mental health, drug use) in an effort to reduce rates of
recidivism. There are currently more than 2,000 drug courts operating in
the United States and more than 200 in planning (U.S. Justice Programs
Office, 2010). Drug treatment courts have also been established in a
number of other countries, including Australia, Bermuda, Brazil, Canada,
the Cayman Islands, England, Ireland, Jamaica, Mauritius, New Zealand,
Scotland and Wales (United Nations Office on Drugs and Crime, 2010). If
public sentiment is just as disposed to efforts to reform offenders as
it is to punish, it could be argued that public expenditure on
imprisonment could be re-invested in more effective--and arguably more
cost-effective--rehabilitation programs such as these.
The current study aimed to shed light on the degree to which the
NSW public are willing to pursue crime control policies that
rehabilitate offenders versus those that punish. To do this, we adapted
the approach used by Nagin and colleagues (2006) to gauge the relative
extent to which members of the public would be willing to pay for a
reduction in crime via rehabilitation or imprisonment. We add to this
debate by also seeking to determine whether willingness to pay varies
according to whether the frame of reference concerned adult or juvenile
offenders. This differentiation between adult and juvenile offending is
particularly important in the NSW context given the very different
legislative framework governing the criminal justice response to adult
and juvenile offenders. In NSW, one of the fundamental guiding
principles of the Young Offenders Act 1997 is to apply the least
restrictive form of sanction against juvenile offender where possible.
This principle does not explicitly guide approaches to adult offending.
Insofar as legislation might reflect community attitudes toward
punishment, we therefore hypothesised that members of the NSW public may
be more disposed to rehabilitating juvenile offenders than adult
offenders.
A subsidiary aim of the current study was to determine whether
willingness to pay for crime reduction differed according to the
socio-demographic characteristics of respondents, their experience as
crime victims and their views about crime in their local area. While we
did not have any a priori hypotheses about the direction these
relationships might take, we were interested in whether some groups
within the community are more or less disposed to paying for crime
reduction. We were also interested in whether experience of crime might
have an impact on willingness to pay for crime reduction. It is hoped
that this exploratory analysis might inform future studies of
willingness to pay within specific subgroups in the population.
Method
Data Collection
The data were collected in mid-2009 via Computer Assisted Telephone
Interviewing (CATI). Only English-speaking people aged 18 years or
older, who were eligible to vote in NSW at the time of interview and who
were required to lodge a tax return in NSW during the previous tax year
were eligible to take part in the study. These restrictions were put in
place because the questionnaire asked respondents how much additional
tax they would pay under a given scenario (see below). This question
would be unintelligible for people who do not pay tax in NSW.
Sample quotas were established according Australian Bureau of
Statistics (ABS) estimates of the age and sex distribution within each
NSW Statistical Division and each Sydney Statistical Subdivision. This
ensured that the resulting sample was representative of the NSW
voting-age population in terms of their age, sex and residential
location. The overall response rate information is shown in Table 1. The
nominal response rate averaged across all respondents was 20 per cent.
Design
The study employed a two (rehabilitation versus punishment) by two
(adult versus juvenile), randomised factorial design to investigate the
primary research aim. The critical part of the questionnaire involved
reading aloud a scenario describing the amount each NSW taxpayer
currently pays to keep offenders in custody and then asking how much
extra tax they would be willing to pay to achieve a 10 per cent
reduction in serious crime. A 10 per cent reduction was selected on the
basis that the best rehabilitation programs tend to have effect sizes in
this range (Aos et al. 2006). Increasing the length of a prison sentence
for burglary from one to two years has also been estimated to produce an
8 per cent reduction in crime (Weatherburn et al. 2006).
Half the participants were presented with a scenario where the 10
per cent crime reduction was to be achieved by a rehabilitation program
while the other half received a scenario where the crime reduction was
to be achieved by imprisoning offenders for longer. Half of each of
these sub-samples received a scenario where the crime reduction was to
be achieved by reducing juvenile offending, while adult offenders were
the proposed focus of intervention for the other half of the sub-sample.
This resulted in four groups, each of whom only received one crime
reduction scenario: adult rehabilitation JAR], adult punishment [AP],
juvenile rehabilitation [JR] and juvenile punishment [JP]. The age, sex
and residential location quota groupings were applied within each of
these four groups to ensure that a representative selection of the
population was exposed to each scenario condition. The nominal response
rates were similar across the four scenario conditions, ranging from
18.3 (JP) to 21.8 per cent (JR).
Questionnaire and Fieldwork
The critical part of the questionnaire involved reading aloud the
following scenario to each respondent:
At the moment each NSW taxpayer spends about $6.50 a week in taxes
keeping [adult/juvenile] offenders in custody and supervising them
in the community. Suppose the Government was thinking of adding [a
rehabilitation program/an extra year] to the sentences of all
[adult/juvenile] offenders [serving one year or more in prison/sent
to prison for one year or more]. Similar programs overseas have cut
serious crime by 10 per cent. How much extra would you be willing
to pay in tax each week to get this 10 per cent reduction in
serious crime? Please state an amount in dollars and cents.
The current weekly amount was estimated from the 2008 Report on
Government Services (Productivity Commission 2009). By anchoring the
amount willing to pay against the approximate amount currently paid to
keep offenders in gaol, we were able to minimise some of the response
problems encountered by past research using open-ended responses in this
field, such as protest answers or inability to respond (Pearce et al.
2006).
In addition to age, sex and residential location, several other
respondent characteristics were measured that might have co-varied with
willingness to pay for these crime reduction alternatives. These
measures were: highest level of formal education completed (year 10 of
high school or less, year 11 or 12 of high school, a qualification
received through a vocational educational service [TAFE], or
university); primary source of income (no income, self-employed, full
time employment, part time employment, government benefit, student
allowance, retirement fund, other); level of income (measured in $10,000
brackets to a high of $130,000 or more); whether the respondent or a
member of their family had ever been the victim of a crime and, if so,
whether the crime involved violence; and perceived frequency with which
crimes occur in their neighbourhood. Three measures of financial stress
were also collected and aggregated to create a single measure of
financial stress. The three individual items, scored on a scale of one
to five (strongly disagree to strongly agree), were: "I generally
consider myself to be financially well off" (reverse scored);
"I usually have very little money left after I pay all of my
bills"; and "If I needed to raise $2,000 in a week for
something important, I would probably experience financial
hardship"
A pilot sample of 200 respondents was collected to ensure that the
questions were interpretable, that the responses were meaningful and
that the randomisation was being implemented correctly. The only anomaly
to arise from the pilot study was that there was a greater volume of
unsure responses on the dependent variable (amount willing to pay) in
the JR condition. In response, staff who administered the survey
received additional training in how to deal with unsure responses and a
field was added to identify reasons why people were unsure in their
responses. This imbalance in unsure responses across scenario conditions
was not apparent in the main study.
Analysis
Analysis of whether there were any differences in amount willing to
pay across the four groups was complicated by the fact that the
dependent variable contained a significant number of zero responses
(that is, respondents who were not willing to pay any extra tax),
rendering standard analyses of variance inappropriate. As a result, the
data were analysed in two steps by first modelling the likelihood of
spending any additional tax dollars on the crime reduction alternative
and then analysing the amount willing to pay among those who were
willing to pay anything.
In the first stage of this analysis, the response to each scenario
condition was cross-tabulated against a binary indicator that took the
value one if the respondent was willing to pay anything, and zero
otherwise. Because there was some evidence of an imbalance in prior
victimisation rates across the scenario conditions (ranging from 49.5
percent in the AR condition, compared to 58.4 per cent in the JR
condition; see Table 2), a binary logistic regression model was also
fitted to ensure that this imbalance was not accounting for the results.
This regression model included terms for method of crime reduction
(rehabilitation vs. punishment), population (adult vs. juvenile), the
interaction between crime reduction method and population, and prior
victimisation.
The second stage was assessed by taking the natural log of all
non-zero responses to normalise the distribution of scores and fitting
an ordinary least squares regression model with terms for crime
reduction method, population and the interaction between the two. Again,
prior victimisation was included in the model to ensure that it was not
influencing the results. (1)
Results
Participants
The survey participants were 1,885 people who resided in, were
eligible to vote in and who were required to lodge a tax return in NSW
during the 2007-2008 financial year. The characteristics of the
participants are displayed in Table 2 by the scenario condition to which
they were assigned. The first row of Table 2 shows that 73.6 per cent of
respondents in the AP condition resided in Sydney, compared with 73.1
per cent in the AR condition, 71.3 per cent in the JP condition and 72.3
per cent in the JR condition. This difference was not statistically
significant. The remainder of the table can be interpreted in a similar
manner. Table 2 reveals that the respondent characteristics were well
balanced across scenario conditions. The only statistically significant
difference between groups was in the proportion reporting that they or
their family members had been victims of a crime. Respondents allocated
to the AR group were slightly less likely than those in the other three
groups to report having been victimised (p=0.041).
Descriptive Overview of Willingness to Pay for Crime Reduction
Approximately one in eight respondents (n=227, 12 per cent) could
not estimate how much extra they would be willing to pay for a 10 per
cent reduction in crime. There was a slight tendency for those in the JP
condition to provide unsure responses (14.6 per cent compared with a low
of 9.6 per cent in the AP condition) but the difference was not
statistically significant (p = 0.097). The most common reason cited
among those not willing to make a response was that they did not have
enough information to make an informed decision (n = 77, 39.5 per cent),
followed by a preference to reduce crime by other means (n = 44, 22.6
per cent), having issues with spending money on crime reduction
initiatives (n = 39, 20.0 per cent), and various other reasons (n = 19,
9.7 per cent). A small proportion refused to nominate why they could not
respond (n = 16, 8.2 per cent).
Figure 1 shows the mean amount respondents indicated that they were
willing to pay to achieve a 10 per cent reduction in serious crime, by
the scenario condition to which they were assigned. Figure 1 suggests
that respondents were slightly more willing to spend more on programs
that punish offenders, although the overlapping error bars indicate that
these differences were not statistically significant. Table 3 shows the
summary descriptive statistics by the scenario condition to which they
were assigned. Those who nominated zero values (that is, they were not
prepared to pay any additional taxes to reduce crime) are included in
all of these calculations. It is clear from the higher standard
deviations and higher maximum values shown in Table 3 that the tendency
toward punishment is due to a small number of outliers in the punishment
conditions pushing the mean values upwards.
Willingness to Pay Anything to Reduce Crime
Figure 2 shows the proportion of respondents within each scenario
condition who indicated a willingness to spend any additional tax
dollars on the crime reduction alternative presented to them. The
proportion willing to pay anything ranged from 67.8 per cent in the AP
condition to 73.5 per cent in the JR condition. This difference was not
statistically significant (p=0.351). Fitting a logistic regression model
to control for prior victimisation did not alter this outcome (see
results of model 1 in Table 6).
[FIGURE 1 OMITTED]
Amount Willing to Pay to Reduce Crime
Table 4 shows the results of the ordinary least squares regression
of log willingness to pay (among those willing to pay anything) on crime
reduction alternative (rehabilitation, punishment), population (adult,
juvenile) and the interaction of these two variables. As with
willingness to pay anything, there was no significant difference in the
amount willing to pay across either of the crime reduction alternative
categories or when framed in terms of adult or juvenile crime reduction.
The interaction between these two variables was also not statistically
significant. These results remained after adjusting for prior
victimisation (data not presented).
Other Respondent Characteristics that Relate to Willingness to Pay
Table 5 shows how each of the other measured characteristics of
respondents related to willingness to pay for crime reduction. The
following groups were more willing to spend at least some additional
taxes on one of the crime reduction alternatives: younger respondents,
those who are more highly educated (particularly at TAFE level), those
who were in paid employment (particularly part-time or self-employed),
those under less financial stress, those who had been the victim of a
crime or had a family member who had been the victim of a crime, and
those who reported that crimes occur in their area. Among those willing
to pay anything to reduce crime, level of education was the only
significant correlate of the amount they were willing to pay to reduce
crime. Those who had lower levels of education were willing to pay
significantly more to reduce crime. However, this finding should be
interpreted cautiously because the second column in Table 5 suggests
that this group was also significantly less willing to pay any
additional taxes to reduce crime.
Table 6 shows the results of two logistic regression models
predicting willingness to pay anything to reduce crime. Model 1 gives
the relationship between scenario condition and willingness to pay
anything after adjusting for prior victimisation (the only variable that
was unbalanced across the scenario conditions). Those who reported
personal or familial crime victimisation were more willing to pay to
reduce crime while, as reported above, the scenario to which
participants were assigned was not predictive of willingness to pay.
Model 2 gives a full model adjusting for all respondent characteristics
that were significantly associated with willingness to pay at a
bivariate level. Income source was not considered for this model because
it would result in unacceptable data loss. Using a backward elimination
modelling approach, the variables that remained in the model were age,
education, financial stress and experience with crime in their local
area. Crime victimisation was not predictive of willingness to pay for
crime reduction after adjusting for these other respondent
characteristics. A model was not fitted to predict the amount
respondents were willing to pay because only one variable was found to
be related to this outcome (that is, education).
Discussion
A clear majority of respondents in our survey were willing to pay
some additional taxes to produce a reduction in crime. Younger people,
those educated at TAFE level, those under less financial stress and
those who report that there is crime in their local area
'sometimes' were more likely to indicate a willingness to pay
additional taxes to reduce crime. There were no differences across any
of the four scenario conditions in either willingness to pay any
additional taxes to reduce crime or in the amount respondents were
willing to pay.
At face value, this suggests that members of the NSW public are
just as disposed to reducing crime through programs that seek to
rehabilitate offenders, as they are to pay for longer prison sentences.
It also suggests that they are just as disposed to punishing or
rehabilitating young offenders as they are to punishing or
rehabilitating adult offenders. If this is indeed the case, there would
seem every reason to pursue rehabilitation with greater vigour. A crude
assessment of the relative costs of incarceration and rehabilitation
programs would suggest clear cost-benefit advantages in favour of
rehabilitation over imprisonment. The annual cost of imprisoning an
adult offender is about $271 per day, or more than $100,000 over the
course of a year (Productivity Commission 2011). The cost of
incarcerating a young person is much higher, at approximately $589 per
day (personal communication, Research and Information, Juvenile Justice
NSW). Despite the high cost of imprisonment, there is very little
evidence of its effectiveness in deterring further offending. Indeed,
two recent literature reviews suggest that, instead of deterring future
offending, prison may have a criminogenic effect on recidivism (Nagin et
al. 2009; Villettaz et al. 2006). In their review, Nagin and colleagues
(2009) were able to identify a number of studies that employed the
highest standard of evidence (random assignment to prison and
alternative sanctions) and a number of others that rigorously accounted
for the selection biases that complicate comparisons of recidivism among
custodial and non-custodial populations. At best, prison was estimated
to have no impact on rates of re-offending. A number of studies found
that incarcerated offenders were more likely to re-offend after
receiving a custodial sentence than were offenders who received
non-custodial sentences.
In comparison, there is strong evidence that well-conducted
rehabilitation programs can reduce offending at relatively low cost
(MacKenzie 2002). In the United States, the Washington State Institute
for Public Policy estimated that the best adult-based rehabilitation
programs, such as intensive treatment-based supervision programs, can be
expected to reduce offending by approximately 16-17 per cent at a cost
of slightly more than $US7,000 per offender. The best juvenile programs,
such as family-based therapies while on probation, can be expected to
reduce recidivism by a similar proportion at a cost of around $US2,000
per person (Aos et al. 2006). Some rehabilitation programs are more
expensive than others but, on the basis of these crude estimates, the
relative cost-benefit advantages of rehabilitation programs over
incarceration appear quite stark.
It is not clear, however, that the current findings should be taken
at face value. As noted earlier, using the same general methodology
among residents in the US states of Pennsylvania, Illinois and
Washington, Nagin and colleagues (2006) and Piquero and Steinberg (2010)
found a clear preference for programs that rehabilitate offenders over
those that punish. Our findings are more in line with the results of
Piquero and Steinberg's (2010) Louisiana sample, where no
preference for rehabilitation or punishment was evident. There are two
possible explanations for these findings. The first is that people in
NSW and Louisiana are more punitive than their counterparts in Illinois,
Pennsylvania and Washington. The second is that differences between the
current study and those conducted by Nagin and colleagues (2006) and
Piquero and Steinberg (2010) in the approach taken to measuring
willingness to pay account for the differences in results.
If voting preferences are reflective of punitive attitudes--and
research suggests that they are (Unnever et al. 2008)--residents of
Louisiana may well be more punitive than residents of the other three
states polled by Piquero and Steinberg (2010). In the 2008 Presidential
election, Louisiana voted (centre-right) Republican while the other
three states voted (centre-left) Democrat. On the other hand, it would
seem unlikely that the NSW public would hold more punitive views than
their American counterparts. NSW residents have voted (centre-left)
Labour in the four of the last five state elections. Americans are also
generally more supportive of capital punishment than Australians tend to
be (Roberts et al. 2003; Roberts & Indermaur 2009). Indeed, each of
the states polled by Piquero and Steinberg (2010) retain capital
punishment for some offences (Snell 2010). Nevertheless, in the absence
of more targeted cross-cultural research, we cannot dismiss the
possibility that differences in punitive attitudes explain the
differential findings across studies.
In our view, however, a more likely explanation for the difference
between the current study and two US studies is methodological. There
were slight differences in both the scenario wording and in the outcome
measures employed by the respective studies. In the current study,
participants were informed that the crime reduction program would reduce
serious crime by 10 per cent, whereas Nagin and colleagues (2006) and
Piquero and Steinberg (2010) nominated a 30 per cent reduction. It is
possible that personal preferences for rehabilitation versus punishment
vary with the size of the crime reduction on offer. We selected 10 per
cent for the current study because it is more in line with the effect
sizes that can reasonably be expected by the best rehabilitation
programs (Aos et al. 2006) or, indeed, by increasing the length of
prison sentences (Weatherburn et al. 2006).
A second difference between studies was in the nature of the
dependent variable. The current study gave respondents freedom to
nominate how much they would be willing to pay including an
'unsure' response option, which 12 per cent of respondents
chose. These 12 per cent of respondents were not included in the
analyses. Nagin and colleagues (2006) and Piquero and Steinberg (2010)
used a constrained choice paradigm with no 'unsure' response
option. In the constrained choice approach, participants were presented
with a scenario and told that it would cost each household an additional
$100 per annum to bring about the nominated reduction in crime. Those
who indicated that they would be willing to pay this additional tax were
asked if they would be willing to pay $200 per annum. Those who were
unwilling to pay an additional $100 were asked if they would be willing
to pay $50 per annum. It is not clear why constrained choice
methodologies might result in greater willingness to pay for those
exposed to the rehabilitation condition. However, forcing those who are
unsure whether they would be willing to pay would boost support for
rehabilitation if those who are unsure how to respond generally favour
rehabilitation. The extent to which people who provide unsure responses
also favour rehabilitation would need to be examined in future studies.
There is clearly a need for further research on this issue. It
would be very useful to conduct comparative research using the method of
measuring willingness to pay used by Nagin and colleagues (2006) and
Piquero and Steinberg (2010). This would provide interesting
cross-cultural information about the relative extent to which people are
willing to invest in punishment versus rehabilitation. It would be
useful to include a condition in which the size of the crime reduction
on offer is varied. It would also be interesting to explore whether
types or levels of media exposure have an impact on preferences for
rehabilitation versus punishment. Given that members of the public who
receive information about crime from tabloid newspapers or talkback
radio tend to hold more punitive attitudes towards crime and justice
(for example, Jones et al. 2008), we might expect that sources of media
exposure might also influence willingness to invest in rehabilitation.
Any research of this nature would have to carefully tease out the
inherent selection biases associated with cross-sectional research
through the use of experimental or longitudinal research designs.
In light of the fact that willingness to pay varied across
different sub-groups within this study, it might be interesting to
identify whether there are differences in willingness to pay for
rehabilitation versus punishment within specific populations of voters.
Older people in the current study were the least willing to pay for
crime reduction so it might be interesting to examine whether young
people were more favourably disposed to rehabilitation over punishment.
Unfortunately, the current study had relatively low statistical power
with which to examine these interactions. Perhaps the most interesting
of the covariates of willingness to pay in the current study was the
strong relationship between levels of financial stress and willingness
to pay for crime reduction. Those who were under high levels of
financial stress--as indicated by their general sense of wealth, their
after-expenses income and their ability to raise emergency funds--were
less willing to pay additional taxes to reduce crime. This suggests that
willingness to pay is strongly linked to capacity to pay. This finding
has potential methodological implications for contingent valuation
studies. While some respondents may have been very supportive of
rehabilitation programs or greater use of imprisonment, their
preferences may not have been reflected in this experimental approach
because they are unable to contribute additional taxation to implement
such policies. Future research may examine methodologies that explore
relative investments of existing levels of taxation in programs that
rehabilitate versus those that punish.
One important limitation of this study must be mentioned, as it
must be for all research using CATI sampling. While we made every effort
to ensure that the resulting sample was representative of the voting-age
population in NSW, this study was not representative of the entire
voting population. For example, people who do not own home telephones
were excluded from the sample, as were those for whom English is a
second language. While it is not clear whether or how this impacts on
the representativeness of the sample on factors such as education, we
can at least be confident that the sample was representative in terms of
the age, sex and residential location profile of the community. In
addition, the critical issue to consider in terms of our principal
hypothesis is whether there were any imbalances across the scenario
conditions. As can be seen in Table 2, the four groups were very well
balanced on factors that were measured in this survey. The only observed
imbalance was on prior victimisation. This was not predictive of
willingness to pay after adjusting for other respondent characteristics
(that is, age, education, financial stress and perceived levels of crime
in the local area). Furthermore, adjusting for prior victimisation did
not affect the observed relationship between scenario condition and
willingness to pay.
Conclusion
The results of the current study clearly indicate that surveyed
members of the NSW public are just as inclined to support crime
reduction efforts through programs that seek to rehabilitate offenders
as they are to pay for longer prison sentences. This does not differ
according to whether the crime reduction is framed in terms of adult or
juvenile offending. Those who are more inclined to pay additional taxes
to reduce crime tend to be: younger; educated via vocational training;
under less financial stress; and report that crime occurs in their local
area. The public policy implications are clear. While, on face value,
the public appear to show highly punitive attitudes towards offenders,
the results of the current study suggest that reducing crime might be
their primary motivation and the method by which this is achieved is
secondary. At face value, then, there would seem every reason to pursue
rehabilitation with greater vigour, especially given the relative
cost-effectiveness of rehabilitation programs over incarceration.
Acknowledgements
This study was funded by the NSW Bureau of Crime Statistics and
Research. We thank our many friends and colleagues for their helpful
suggestions on the methodology employed for this study.
References
Aos, S., Miller, M. & Drake, E. (2006) Evidence-based Public
Policy Options to Reduce Future Prison Construction, Criminal Justice
Costs, and Crime Rates, Washington, D.C., Washington State Institute for
Public Policy.
Chan, J. (1995) The Limits of Incapacitation as a Crime Control
Strategy, Crime and Justice Bulletin no. 25, Sydney, NSW Bureau of Crime
Statistics and Research.
Corben, S. (2010) NSW Inmate Census 2009: Summary of
Characteristics, Statistical Publication no. 34, Sydney, NSW Department
of Corrective Services.
Doob, A. & Roberts, J.V. (1988) 'Public punitiveness and
public knowledge of the facts: Some Canadian surveys'. In N. Walker
& M. Hough (eds) Public Attitudes to Sentencing: Surveys from Five
Countries, Aldershot, Gower.
Gorta, A. & Eyland, S. (1990) Truth in Sentencing. Impact of
the Sentencing Act, 1989: Report 1, Research Publication no. 22, Sydney,
NSW Department of Corrective Services.
Hough, M. & Roberts, J. (1998) Attitudes Toward Punishment:
Findings from the British Crime Survey, Home Office Research Study no.
179, London, Home Office.
Indermaur, D. (1987) 'Public perceptions of sentencing in
Perth, Western Australia'. Australian and New Zealand Journal of
Criminology, 20, 163-183.
Jones, C. & Weatherburn, D. (2010) 'Public confidence in
the NSW criminal justice system: A survey of the NSW public',
Australian and New Zealand Journal of Criminology, 43(3), 506-525.
Jones, C., Weatherburn, D. & McFarlane, K. (2008) Public
Confidence in the New South Wales Criminal Justice System, Crime and
Justice Bulletin no. 118, Sydney, NSW Bureau of Crime Statistics and
Research.
Lulham, R. & Fitzgerald, J. (2008) Trends in Bail and
Sentencing Outcomes in New South Wales Criminal Courts: 1993-2007, Crime
and Justice Bulletin no. 124, Sydney, NSW Bureau of Crime Statistics and
Research.
MacKenzie, D.L. (2002) 'Reducing the criminal activities of
known offenders and delinquents'. In L.W. Sherman, D.P. Farrington,
B.C. Welsh & D.L.
MacKenzie (eds), Evidence-based Crime Prevention, London,
Routledge. Matka, E. (1991) NSW Sentencing Act 1989, Legislative
Evaluation Series, Sydney, NSW Bureau of Crime Statistics and Research.
Mattinson, J. & Mirrlees-Black, C. (2000) Attitudes to Crime
and Criminal Justice: Findings from the 1998 British Crime Survey, Home
Office Research Study no. 200, London, Home Office.
Nagin, D.S., Cullen, F. & Jonson, C. (2009) 'Imprisonment
and re-offending'. In M. Tonry (ed.), Crime and Justice: A Review
of Research, 38, 115-200, Chicago, University of Chicago Press.
Nagin, D.S., Piquero, A.R., Scott, E.S. & Steinberg, L. (2006)
'Public preferences for rehabilitation versus incarceration of
juvenile offenders: Evidence from a contingent valuation survey',
Criminology & Public Policy, 5(4), 627-652.
NSW Department of Juvenile Justice (2009) Annual Report 2008-2009,
Sydney, NSW Department of Juvenile Justice.
Pearce, D., Atkinson, G. & Mourato, S. (2006) Cost-benefit
Analysis and the Environment: Recent Developments, Paris, Organisation
for Economic Co-operation and Development.
Piquero, A.R. & Steinberg, L. (2010) 'Public preferences
for rehabilitation versus incarceration of juvenile offenders',
Journal of Criminal Justice, 38, 1-6.
Productivity Commission (2009) Report on Government Services 2008,
Canberra, Commonwealth Government of Australia.
Productivity Commission (2011) Report on Government Services 2011,
Canberra, Commonwealth Government of Australia.
Roberts, J. & Grossman, M. (1990) 'Crime prevention and
public opinion', Canadian Journal of Criminology, 32(1), 75-90.
Roberts, J., Stalans, L.J., Indermaur, D. & Hough, M. (2003)
Penal Populism and Public Opinion, New York, Oxford University Press.
Roberts, L. & Indermaur, D. (2009) What Australians Think About
Crime and Justice: Results from the 2007 Survey of Social Attitudes,
Research and Public Policy Series no. 101, Canberra, Australian
Institute of Criminology.
Snell, T.L. (2010) Capital Punishment, 2009--Statistical Tables,
U.S. Department of Justice, retrieved 30 March 2011, from
http://bjs.ojp.usdoj.gov/content/pub/pdf/cp09st.pdf.
United Nations Office on Drugs and Crime (2010) UNODC and drug
treatment courts, retrieved 15 September 2010, from
http://www.unodc.org/unodc/en/legal-tools/Drug-Treatment-Courts.html.
Unnever, J.D., Benson, M.L. & Cullen, F.T. (2008) 'Public
support for getting tough on corporate crime: Racial and political
divides', Journal of Research in Crime and Delinquency, 45(2),
163-190.
U.S. Justice Programs Office (2010) Summary of drug court activity
by state and county, 7th July, 2010, retrieved 15 September 2010, from
http://wwwl.spa.american.edu/justice/documents/2812.pdf.
Villettaz, P., Killias, M. & Zoder, I. (2006) The Effects of
Custodial vs Non-custodial Sentences on Re-offending, Report to the
Campbell Collaboration Crime and Justice Group, Switzerland, Institute
of Criminology and Criminal Law, University of Lausanne.
Weatherburn, D., Hua, J. & Moffatt, S. (2006) How Much Crime
does Prison Stop? The Incapacitation Effect of Prison on Burglary, Crime
and Justice Bulletin no. 93, Sydney, NSW Bureau of Crime Statistics and
Research.
Weatherburn, D. & Indermaur, D. (2004) Public Perceptions of
Crime Trends in New South Wales and Western Australia, Crime and Justice
Bulletin no. 80, Sydney, NSW Bureau of Crime Statistics and Research.
Endnotes
(1.) To ensure that this two-stage modelling process was not
providing misleading results, the analyses were repeated by rounding all
nominated responses to the nearest integer and fitting a zero-inflated
poisson regression model. While the results of this model are not
presented, the results were not substantively different than those
reported here.
Craig G.A. (1) and Don J. Weatherburn
(1.) Correspondence concerning this article should be addressed to
Craig G.A. Jones, Research Manager, NSW Bureau of Crime Statistics and
Research, GPO Box 6, SYDNEY NSW 2001, Australia. Email:
craig_iones@agd.nsw.gov.au
Table 1: Call outcomes for all respondents
Outcome N %
Refused 7526 68.9
Ineligible 1351 12.4
Unexpended appointments 169 1.5
Completed surveys 1885 17.2
Total contacted 10931 100.0
Nominal response rate-- 20.0%
(complete / complete +
refused) (a)
Total response rate-- 17.2%
(complete / all
contacted)
(a) AP = 20.9 per cent (95% CI: 19.3% to 22.60; AR = 19.4 per cent
(95% CI: 17.9% to 21.1 %); JP = 18.3 per cent (95% CI: 16.9% to
19.9%) and JR = 21.8 per cent (95% CI: 20.1% to 23.6%)
Source: Bureau of Crime Statistics and Research (BOCSAR),
original unpublished survey data (2009)
Table 2: Respondent characteristics, by scenario condition to
which they were allocated
Scenario
AP AR JP JR
Characteristic N (%) N (%) N (%) N (%)
Resided in Sydney 352 (73.6) 340 (73.1) 342 (71.3) 334 (72.3)
(n=1885)
Male (n=1885) 254 (53.1) 245 (52.7) 228 (47.5) 223 (48.3)
Mean age (n=1838) 50.1 49.8 49.7 50.2
Education (n=1871)
Year 10 or less 105 (22.2) 104 (22.6) 102 (21.3) 98 (21.4)
Year 11 or 12 98 (20.7) 82 (17.8) 94 (19.6) 96 (21.0)
TAFE 87 (18.4) 110 (23.9) 90 (18.8) 107 (23.4)
University 184 (38.8) 165 (35.8) 193 (40.3) 156 (34.1)
Income source
(n=1782)
No income 18 (4.0) 12 (2.7) 12 (2.6) 10 (2.3)
Self-employed 75 (16.5) 71 (16.1) 64 (14.0) 58 (13.6)
FT employed 191 (42.0) 195 (44.1) 219 (47.9) 183 (42.8)
PT employed 83 (18.2) 84 (19.0) 88 (19.3) 93 (21.7)
Benefit 25 (5.5) 19 (4.3) 18 (3.9) 19 (4.4)
Student allowance 1 (0.2) 1 (0.2) 1 (0.2) 3 (0.7)
Other 62 (13.6) 60 (13.6) 55 (12.0) 62 (14.5)
Income (n=1338)
Less than $39,999 113 (36.8) 117 (37.9) 125 (39.2) 111 (36.3)
$40,000 - $69,999 99 (32.2) 109 (35.3) 113 (35.4) 113 (36.9)
$70,000 - $99,999 64 (20.8) 59 (19.1) 62 (19.4) 56 (18.3)
$100,000+ 31 (10.1) 24 (7.8) 19 (6.0) 26 (8.5)
Financially well
off? (n=1885)
Disagree 146 (30.5) 113 (24.3) 139 (29.0) 114 (24.7)
Neither agree nor 121 (25.3) 125 (26.9) 104 (21.7) 124 (26.8)
disagree
Agree 211 (44.1) 227 (48.8) 237 (49.4) 224 (48.5)
No money after
bills? (n=1885)
Disagree 158 (33.1) 164 (35.3) 164 (34.2) 154 (33.3)
Neither agree nor 79 (16.5) 83 (17.8) 66 (13.8) 82 (17.7)
disagree
Agree 241 (504) 218 (46.9) 250 (52.1) 226 (48.9)
Unable raise
$2000? (n=1885)
Disagree 213 (44.6 231 (49.7) 213 (44.4) 213 (46.1)
Neither agree nor 55 (11.5) 40 (8.6) 51 (10.6) 50 (10.8)
disagree
Agree 210 (43.9) 194 (41.7) 216 (45.0) 199 (43.1)
Crime victim 269 (56.3) 230 (49.5) 264 (55.0) 270 (58.4)
(n=1885)
Violent crime 108 (40.1) 78 (33.9) 89 (33.7) 94 (34.8)
victim (n=1033)
Crime in area?
(n=1820)
Never 15 (3.2) 10 (2.3) 19 (4.0) 15 (3.4)
Rarely 154 (33.2) 155 (35.1) 171 (36.2) 126 (28.4)
Sometimes 183 (39.4) 160 (36.3) 183 (38.8) 183 (41.3)
Frequently 112 (241) 116 (26.3) 99 (21.0) 119 (26.9)
Scenario
Characteristic p-value
Resided in Sydney 0.854
(n=1885)
Male (n=1885) 0.182
Mean age (n=1838) 0.950
Education (n=1871) 0.361
Year 10 or less
Year 11 or 12
TAFE
University
Income source 0.818
(n=1782)
No income
Self-employed
FT employed
PT employed
Benefit
Student allowance
Other
Income (n=1338) 0.798
Less than $39,999
$40,000 - $69,999
$70,000 - $99,999
$100,000+
Financially well 0.118
off? (n=1885)
Disagree
Neither agree nor
disagree
Agree
No money after 0.548
bills? (n=1885)
Disagree
Neither agree nor
disagree
Agree
Unable raise 0.599
$2000? (n=1885)
Disagree
Neither agree nor
disagree
Agree
Crime victim 0.041
(n=1885)
Violent crime 0.364
victim (n=1033)
Crime in area? 0.186
(n=1820)
Never
Rarely
Sometimes
Frequently
Source: BOCSAR, original unpublished survey data (2009)
Table 3: Descriptive statistics relating to the amount in
additional tax people would be willing to pay to bring about a
10 per cent reduction in serious crime
Scenario N Mean ($) Std. Dev. Median ($)
Adult punishment 432 3.82 7.36 2.00
Adult rehabilitation 405 3.29 4.88 2.00
Juvenile punishment 410 4.04 7.18 2.00
Juvenile rehabilitation 411 3.49 3.99 2.00
Total 1658 3.66 6.05 2.00
Scenario Min. ($) Max. ($)
Adult punishment 0.00 100.00
Adult rehabilitation 0.00 50.00
Juvenile punishment 0.00 100.00
Juvenile rehabilitation 0.00 20.00
Total 0.00 100.00
Source: BOCSAR, original unpublished survey data (2009)
Table 4: Ordinary least squares regression model of (log) amount
willing to pay (zero values excluded) to achieve a 10 per cent
reduction in serious crime, by the scenario to which respondents
were assigned (n=1173)
Population/policy [beta] (se) p-value
Juvenile vs. adult 0.045 (0.078) 0.566
Rehabilitation vs. -0.112 (0.078) 0.151
punishment
Interaction 0.051 (0.110) 0.644
Constant 1.235 (0.055) <0.001
Source: BOCSAR, original unpublished survey data (2009)
Table 5: Respondent characteristics, by scenario condition to
which they were allocated
Mean of
Characteristic % willing [chi square] (log) F-test
to pay p-value $ willing p-value
something to pay
Residential 0.974 0.971
location
(n=1658,1173)
Sydney 70.8 1.214
Other 70.7 1.216
Sex (n=1658, 1173) 0.779 0.405
Male 70.4 1.192
Female 71.1 1.238
Age (n=1618, 1156) <0.001 0.456
18-29 79.8 1.197
30-44 74.7 1.193
45-59 73.4 1.267
60+ 62.2 1.157
Education 0.044 0.029
(n=1644, 1168)
Year 10 or less 66.0 1.362
Year 11 or 12 71.3 1.201
TAFE 75.7 1.113
University 71.2 1.206
Income source <0.001 0.645
(n=1565, 1110)
No income 70.2 1.049
Self-employed 75.8 1.205
FT employed 71.0 1.238
PT employed 75.3 1.237
Benefit / allowance 50.0 1.170
Other 65.9 1.103
Income (n=1123, 837) 0.934 0.556
Less than $39,999 73.8 1.143
$40,000 - $69,999 74.2 1.220
$70,000 - $99,999 75.8 1.261
$100,000+ 76.1 1.207
Fin. Stress 0.004 0.097
(n=1658, 1173)
High 66.3 1.168
Moderate 70.9 1.174
Low 75.3 1.295
Crime victim? 0.031 0.242
(n=1658, 1173)
Yes 72.9 1.242
No 68.1 1.178
Violent crime 0.509 0.941
victim? (n=909, 663)
Yes 71.6 1.238
No 73.6 1.244
Freq. Crime 0.001 0.723
(n=1610, 1173)
Never 54.7 1.311
Rarely 70.4 1.223
Sometimes 76.0 1.198
Frequently 67.6 1.271
Among those willing to pay something.
Source: BOCSAR, original unpublished survey data (2009).
Table 6: Binary logistic regression models predicting willingness
to pay anything to reduce crime (n=1567)
Model 1
[beta]
Variable (se) p-value
AP --
AR 0.15 (0.15) 0.30
JP 0.15 (0.15) 0.31
JR 0.27 (0.15) 0.08
Aged 18-29
Aged 30-34
Aged 45-59
Aged 60+
Year 10 or less
Year 11 or 12
TAFE
University
High financial stress
Moderate financial stress
Low financial stress
Crime victim? 0.23 (0.11) 0.03
Never crime in local area
Rarely crime in local area
Sometimes crime in local area
Frequently crime in local area
Model 2
[beta]
Variable (se) p-value
AP -
AR 0.13 (0.16) 0.43
JP 0.10 (0.16) 0.55
JR 0.22 (0.16) 0.18
Aged 18-29
Aged 30-34 -0.32 (0.26) 0.22
Aged 45-59 -0.43 (0.26) 0.09
Aged 60+ -0.99 (0.26) <0.01
Year 10 or less
Year 11 or 12 0.18 (0.17) 0.30
TAFE 0.37 (0.18) 0.04
University -0.05 (0.16) 0.74
High financial stress -
Moderate financial stress 0.29 (0.14) 0.04
Low financial stress 0.55 (0.15) <0.01
Crime victim? -
Never crime in local area -
Rarely crime in local area 0.57 (0.31) 0.07
Sometimes crime in local area 0.89 (0.31) <0.01
Frequently crime in local area 0.41 (0.31) 0.20
Source: BOCSAR, original unpublished survey data (2009)
Figure 2: Proportion of respondents within each scenario condition who
would be willing to pay something to bring about a 10 per cent
reduction in serious crime (n=1,658)
Scenario Per cent willing to pay something
Adult punishment 67.8
Adult rehabilitation 70.9
Juvenile punishment 71.0
Juvenile rehabilitation 73.5
Note: Table made from bar graph.
