This is the text of a paper written in 1999 and currently under
review by a journal.
HEAVY DRINKING: RISK, RESTRAINT AND OTHER DETERMINANTS.
I.INTRODUCTION.
The literature on drinking has tended to focus on a number
of distinct areas: determinants of the demand for drinking across
the population at large, episodes of binge drinking [Humara &
Sherman(1998), health effects, recourse to treatment by problem
drinkers/alcoholics [Hajema et al.(1999)], the general area of
personality traits [Stacy & Newcomb(1998),Skuttle(1998)]. General
demand analysis has naturally tended to be the province of
economists [see e.g. Godfrey(1989) Goel & Morey(1995),Averett &
Hochman(1994)] whilst the other areas have been mostly studied by
specialist addiction studies researchers and psychologists. This
paper adopts an eclectic approach, using a large national sample
survey, to estimate the significance of risk factors associated
with falling into the category of 'heavy drinker' as defined by
European Union health guidelines. For comparative purposes, the
same factors are used to predict the likelihood of falling into
the category of daily drinker which, whilst it does not preclude
health damaging activity, may be a contained lifestyle absorption
of alcohol usage. It is shown, using a large national survey,
that risk taking attitude influences both heavy and daily
drinking. There are also significant effects of age, religion,
education and situational factors.
II. BACKGROUND LITERATURE.
In conventional thought there are two basic features which
distinguish alcohol from most other consumption activities:
health damage and possible addictiveness. Recent economics
literature, inspired by Becker [see Grossman(1995), downplays the
existence of addictive goods per se in favour of the notion that
all goods are potentially addictive so long as there is a
connection between the current rate of satisfaction and the
amount of consumption in the past. This incorporates rather than
negates the traditional 'tolerance and reinforcment' notions of
physiologically based approaches to alcohol usage. In the
specialist addiction literature the definition of addiction is
"repetitive and compulsive use of a psychoactive substance"
[Goldstein (1994,p.3.]. There is a profound epistemological
problem in distinguishing frequent behaviour from that which is
repetitive and compulsive: in the economic approach there is no
compulsiveness per se as all acts are seen as rational and the
individual takes full account of the risks entailed in their
conduct. Not all economists endorse this account; there is for
example the bounded rationality espoused by Akerlof (1991,p.5.)
who says ".. I do not agree that the model of forward-looking,
rational behavior accurately describes the way in which
individuals decide on drug or alcohol intake. Most drug abusers,
like most chronically overweight individuals, fully intend to cut
down their intake, since they recognize that the long-run cost of
their addiction exceeds its benefits. They intend to
stop-tomorrow. Individuals following the procrastination model
are both maximising and knowledgeable and yet their decisions are
not fully rational."
Although notions of rationality and determinism may
fascinate those on the highground of battle for the validity of
a discipline it is of more practical relevance whether intrinsic
and situational determinants of health damaging behaviour can be
identified in the available statistics. Ultimately (unless it is
of a moral-ethical nature) concern about addictiveness boils down
to worry about health damages: that is, someone who is fully
aware of health damages and is not disposed to take the risk may
find themselves 'over risking' against their will due to failures
of self control. On an observational level, without experimental
controls such as in Vucinich & Simpson(1998), it is somewhat
difficult to distinguish the rational individual with a high
taste for risk (or a high rate of discounting the future) from an
individual with addictive weaknesses in self control.
Psychological studies of the discounting behaviour of different
types of drinker indicate that the 'problem' drinker has a higher
discount rate than the 'social' drinker [Vuchinich &
Simpson(1998). Whilst this may be true, it is possible that the
difference in discount rates is itself a reflection of other
factors such as age,religion, education and so forth.
Health production consists of prevention,promotion and
protection [Rothman & Salovey(1997)]. A recent spate of
literature on labour markets has emphasized the possible health
promotion effects of drinking that may lead to higher earnings
[see e.g. for the USA,Zarkin et al,1998, Canada,Hamilton &
Hamilton(1997) and Australia [Barrett(1999)]. These papers are
somewhat loose on the definition of what constitutes light,
moderate and heavy drinking. Nevertheless, they generally find in
favour of a beneficial impact of up to moderate drinking which is
taken as far as 'daily' drinking albeit that daily in some papers
in the literature may mean 20 days in the month. As this adds
further weight to a 'health gains of drinking' argument, then
moderate drinking may be seen, in contrast to heavy drinking, as
prevention and promotion of health. It follows that moderate to
light drinking may be associated with factors that are correlated
with positive health attitudes.
III. MODEL.
To look at the daily/heavy drinking distinction we need a
functional model which accounts for the psychopharmacological and
socio-structural determinants of consumption levels. This can be
found in Cameron (1999). There follows a version of this applied
to alcohol. Let there be two goods; one (X) which is 'normal' in
the sense that it does not deliver any chemical alterations to
the brain state nor is it imbued with any learned components and
another (Z) which delivers chemical 'hits' to the brain, through
the bloodstream, and also provides utility from being a
constituent of 'scripts' (M) [Van Raaij(1990) p.168/9]. A script
is a scenario of interlocking consumption elements which have a
reinforcing quality. We can distinguish between private and
social scripts. In the private script one might be drinking alone
as a reward for completing a solo work task or simply because one
finds an alcoholic drink relieving on a hot day or to alleviate
feelings of loneliness. In the social script, the tension and
excitement of the presence of other people are factors in
generating the desired level of alcohol consumption. The social
script element is enhanced by the impact of alcohol on lowering
inhibition. The basic idea of the script is that the image of
drinking becomes intertwined with the experiences which are
temporally joint with the consequences of the bloodstream
absorption of alchohol units. Over a process of time, these
mental associations [or imaging] may be a non-separable element
of the utility from the psychopharmacological processes. The
mental process of association may engender brand loyalty as the
style of bottle, name of the product etc. can potentially become
part of the overall pleasure.
The individual utility function is :
(1) U = U(H,S,X,W)
where H= 'hits' , S is the volume of 'scripts' in which Z
features, W is the perceived stock of health. There is assumed
to be an exogenously given menu of scripts which the individual can
work her way through. The delivery of scripts,hits and health is
through the following production functions:
(2) H = H(Z)
(3) W=W(Z,X,k) = &(k)W(Z,X) + [1-&(k)] W(X)
where &(k) = {1,k >= k*
{0,k <= k*
k is a filter variable to represent cognitive dissonance [Gilad
et al.(1987)]. If k* is exceeded by the flow of messages,
indicating that drinking is a serious health risk or anti-drink
driving messages, then it will be curtailed as the switchover in
(4) works through (1). This model is similar to the 'flip-flop'
utility function formulation of Winston(1980) who emphasizes the
problem of self control, but does so without incorporating hits
or scripts.
The 'k*' filter will be exceeded by events which make the
risk more 'salient' such as witnessing at first hand the details
of drink related deaths,illnesses and accidents If there is a
switch in perceptions, the smoker could simply reduce alcohol
unit intake by drinking less volume per se or for example moving
to weaker lagers/beers. The market has an incentive, as
Winston(1980) points out, to provide devices to assist self
control. Hence the emergence of lower ethanol content and even
alcohol free lagers. Clearly such devices can not change the
underlying preference structure viz. the latent demand function
for hits and scripts. Thus there is the problem of backsliding.
The drinker who is apprehensive about personal health could
backslide to an old higher danger brand or stick with the 'safe'
brand and consume more to keep up the steady state dosage level
of psychopharmacological effects. There is a possible
distinction here in terms of scripts, that is volume of drinks
taken is mediated by the social situation. For someone who
manages a switch to a safer brand [or perhaps type of alcohol],
whilst staying within a social context where the relevance of
chemical stimulation is low, it may be more likely that pure
alcohol intake can be reduced compared with a private drinker or
one whose social scripts are heavily dependent on the
psychopharmacological element e.g. to overcome, shyness, stress,
lose inhibition and so forth.
A major determinant of k is general attitudes to risk
management and health. Although there are are a plethora of
extremely sophisticated models of decision-making under risk
[Viscusi(1992), Machina(1989)] they are not greatly relevant to
this paper. Indeed, it is hard to implement these models fully
even in the ideal environment of the controlled experiment which
generally has the disadvantage of not shedding light on the
broader social and cultural factors. In this paper, drinking
style is related to some measures of risk derived from stated
attitudes and behaviour along with controls for religion,
education and other factors which might predispose a change of
style due to alterations in the scripts engaged in.
IV.DATA AND VARIABLES.
Most studies of alcohol usage are derived from small scale
experimental studies or specific surveys for alcohol. There are
some possible problems with this such as bias due to selectivity
in participation along with the lack of a broad range of control
variables which are needed for estimating models with a wider
focus on drinking style than is normally found in the specialist
literature. This paper deploys a U.K. survey which was primarily
concerned with the health risk of sexual activity but was, in
fact, turned into a much more wide ranging health survey. The
National Survey of Sexual Attitudes and Lifestyles has a large
sample and a wide range of variables. It also affords the luxury
of constructing some control variables for the possibility that
people might not be telling the whole truth in the interview
situation. The NSSAL survey was carried out, by professional
market researchers, in the UK in 1990/1. This was a rigorously
conducted survey of 18,876 adults of ages from 16 to 59. The
questions were extensively piloted and a lengthy in person
interview was combined with separate self completion schedules in
an attempt to ensure greater accuracy in response to the more
'sensitive' questions. Although the primary purpose of the study
was to examine whether there had been discernible behavioural
responses to the threat of HIV/AIDS, it was conducted in the
context of a thorough examination of the overall lifestyles and
socio-economic circumstances of the individuals. The sexual
content was revised carefully in order to ensure the maximum
participation, across all demographic groups, by avoiding fall
out through embarrassment or distaste. Two samples are available.
The 'long' version comprises 4,548 individuals and contained a
detailed investigation of religious beliefs and sexual attitudes.
The 'short' version is used here giving us a usable sample of
7,726 women and 11,150 men. The major reason for using the short
version was to get a larger sample of data. The data were
provided, on a CD-rom, by the ESRC Data Archive at the University
of Essex.
The alchohol questions were part of a battery of questions
on general health and lifestyle. The survey file provided comes
with additional variables, constructed from these, in which women
and men are classified as 'heavy drinkers' according to the EU
definitions of exceeding 35 units per week, for women and 50
units per week, for men. This is obviously a medically (health
damage assessment) derived measure rather than a behaviourally
developed one. However, as we shall see it produces so few female
respondents in the heavy category that we were forced to move to
daily drinking as an alternate measure of concentration in this
tail.
The definitions and names of the other variables are given
in Table I. One of our main intentions is to perform more
systematic assessment of the role of risk taking attitude than
has hitherto been possible. Direct risk attitude is measured by
two variables: one for perception of the danger of HIV-Aids
[AIDSHI] and one for the usage of a condom during one's first
sexual intercourse. The AIDSHI item is constructed from 'lay'
assessment of a hypothetical situation rather than introspection
from own behaviour. An indirect measure of risk is available in
the LAXWELL variable which indexes a somewhat carefree attitude
to health. As the other major legal health damaging activity,
smoking is often linked in the social script of alchohol usage we
include a measure of heavy smoking as a covariate of heavy
drinking. Script altering measures are included which derive from
geographical mobility; that is moving areas (INMOVE) and being
taken away from home regularly for work reasons (AWAYLOT) may
induce an increase in drinking. Education is a standard variable
in health promotion so, in so far as moderate drinking is a
health promoting activity, it might be expected to have a
positive effect on daily drinking but a potential negative effect
on heavy drinking. Education may also be a measure of income
level which is more directly represented by long term
unemployment and the AFFLUENT, MIDDLE and POOR variables derived
from the wealthiness of the housing areas resided in. There are
no direct measures of earnings or wealth in this survey. Social
factors my exert different pressures on women than on men, and
the impact of risk attitudes may also differ by gender, so we
have estimated all equations separately for men and women. To
control for the possibility of biased response, two variables are
included to represent the environment of the interview. The
LISTEN variable is coded to 1 if any one appears to be present
and listening in the household at the time of the interview. The
EMBARRASS variable is coded to 1 if the interviewer judges the
subject to be embarrassed during the interview.
A set of controls is also included for residence in the
standard regions of the U.K. These may simply be portmanteau
variables for a miscellany of influences. There are sterotypical
notions in the U.K. of strong regional variations in drinking.
The historical evidence (OHE 1981) of heavy drinking in Scotland
was based on proxies for heavy drinking in the form of higher
level of arrests for drunknenness and deaths from cirrhosis of
the liver. However, this may indicate a celtic tendency to binge
drinking rather than higher mean levels of alchohol intake.
V. THE EMPIRICAL WORK.
This is a 'tail' analysis of the odds of being at the
heavier end of the spectrum of drinking behaviour over being in a
moderate or nil region. It was decided to use the European Union
definition of weekly intake of all alcohols converted to units,
via a content based formula, as this is a suitable focus of
policies towards alcohol usage. As indicated, a comparison is
made with daily drinking. Logit analysis is used to estimate the
impact of the selected variables on the probability of being a
heavy or daily drinker.
Let us look first at the male results. Of the 11150
sampled, 301 self reported as exceeding the EU 'safe' limit of 50
units of alchohol per week. The embarrassment check variable is
not significant but there is a strong impact of the LISTEN
variable suggesting that those who suspect eavesdropping on their
interview are more likely to report as heavy drinkers. Having a
university degree is an inhbiting factor. The age pattern is
suggestive of an inverted U-shape; probability of heavy drinking
rises from age 30 but is less in the 50-9 group than the 40-9
group. Measured aversion to risks [in the AIDSHI and CONDO1ST
variables] and moral restraints, in the form of church
attendance, both reduce the predicted incidence of heavy
drinking. Situational factors [possibly stress or loneliness]
show some tendency to dispose men to heavy drinking as the long
hours of work and frequently working away from home variables are
positive and significant. These factors may also be reflected in
the positive significant effect of the heavy smoking variable.
There is no strong evidence of any sterotypical regional
variations as the celtic regions do not have significant positive
differentials with Scotland even having weak signs of a negative
differential.
Comparing the heavy drinking logit with that for daily
drinking we find a lot of similar results. A major contrast
occurs with education. Rising levels of education are associated
with rising likelihood of daily drinking whereas the possession
of a university degree had a significant negative impact on the
incidence of heavy drinking. Recent arrival in the area is a
significant precipitator of daily drinking but not of heavy
drinking. Those signalling negligent attitudes to health are
significantly more likely to be heavy drinkers. Again there are
no particularly marked regional tendencies of a stereotypical
nature.
For women, only 27 of the 7726 report as heavy drinkers
which is not really sufficient for us to put too much weight on
the logit analysis. Only 3 variables are significant: aidshi
(-), smoking (+) and Roman Catholic(+).
Turning to the daily drinking, 350 women self report in
this category. Comparing the female daily drinking with the
male, we find that the significant precipitating factors for the
former are also found in the latter: that is, age education and
long hours of work. Church attendance is a significant deterrent
in both. However there some very notable differences. Whilst
being away from home a lot is a precipitating factor for men it
is an inhibiting factor for women. This is not very surprising as
it is doubtless indicative of the differently gendered social
context of drinking alone in a strange place. There is some
contrast in the measures of risk attitude [AIDSHI,CONDO1ST] and
laxity in health attitudes [LAXWELL] as only the first of these
is significant in the female equation. The female equation also
shows a contrasting lack of adjustment to the interview situation
as the embarrassment and potential eavesdropping variables are
not significant.
VI. DISCUSSION.
This paper has analyzed a large sample survey for the
determinants of daily and heavy drinking. A large number of
socio-economic and demographic variables were considered along
with measures for risk taking attitude, moral restraint and
situational disposition to drinking. For both males and females,
measures of risk taking disposition are positively correlated
with regular engagement in drinking. The distinct inhibitors of
heavy, as opposed to daily, drinking can only be assessed
adequately for men and seem to be of a social class nature:
living in a middle class housing area and possessing a university
degree both significantly INCREASE the likelihood of daily
drinking but DECREASE the likelihood of heavy drinking.
REFERENCES.
Akerlof,G.1991. Procrastination and obedience. American Economic
Review. 81(2) 1-19
Averett,S. & Hochman,H.1994. Addictive behavior and public
policy. Public Finance. 49. Supplement. 244-258
Barrett,G.F.1999. The effect of alcohol consumption on earnings.
Unpublished Ms. Department of Economics. University of New South
Wales. Sydney.
Cameron,S.1999. Nicotine addiction and cigarette consumption:
a psycho-economic model. Journal of Economic Behavior and
Organization 20 (FORTHCOMING)
Festinger,L. 1957. A theory of cognitive dissonance. Stanford,
CA. Stanford University Press.
Gilad,B.,Kaish,S. & Loeb,P.D. 1987. Cognitive dissonance and
utility maximisation. Journal of Economic Behavior and
Organization, 8 61-73
Godfrey,C.1989. Factors influencing the consumption of alcohol
and tobacco: the use and abuse of economic models. British
Journal of Addiction 84 1123-1138
Goel,R.K. & Morey,M.J.1995. The interdependence of cigarette and
liquor demand. Southern Economic Journal. 62(2) 451-459
Goldstein,A.1994. Addiction: from biology to drug policy. New
York.W.H.Freeman.
Grossman,M. 1995. The economic approach to addictive behaviour.
pp.157-171. in Tomassi,M. & Ierulli,K. The new economics of human
behaviour. Cambridge University Press.
Hajema,K.,Knibbe,R.A. & Drop,M.J. 1999. Social resources and
alcohol-related losses as predictors of help seeking among male
problem drinkers. Journal of Studies on Alcohol. 60(1) 120-129
Hamilton,V & Hamilton,B.H.1997. Alcohol and earnings, does
drinking yield a wage premium? Canadian Journal of Economics.
30(1) 135-151
Humara,H.J. & Sherman,M.F.1998. Situational determinants of
alcohol abuse amongst Caucasian and African-American College
Students. Addictive Behaviors. 24(1) 135-138
Machina,M.1989. Choice under uncertainty: problems solved and
unsolved. pp.12-46 in Hey,J.(ed) Current Issues in
Microeconomics. Macmillan 1989
Office of Health Economics. 1981. Alcohol: Reducing the Harm.
Rothman,A.J. & Salovey,P.1997. Shaping perceptions to motivate
healthy behavior: the role of message framing. Psyhcological
Bulletin. 121. 3-19
Skuttle,A. The relationship among self-efficacy expectancy and
severity of alcohol abuser and psychological benefits from
drinking. Addictive Behaviors.24(1) 135-138
Stacy,A.W. & Newcomb,M.D.1998. Memory association and personality
as predictors of alcohol use: mediation and moderator effects.
Viscusi,K. 1992. Smoking: the Risky Decision.
Vucinich,R.E. & Simpson,C.A.1998. Hyperbolic temporal discounting
in social drinkers and problem drinkers. Experimental and
Clinical Psychopharmacology. 6(3) 292-305
Wellings,K. et al. 1994. Sexual behaviour in Britain. Penguin.
Harmondsworth.
Winston,G.C.1980.Addiction and backsliding: a theory of
compulsive consumption. Journal of Economic Behavior and
Organization 1 295-324
Zarkin, G.A French, M.T Mroz, T and Bray, J.W 1998. Alcohol use
and wages : new results from the National Household Survey on
Drug Abuse. Journal of Health Economics, 17, pp 53-68.
TABLE I: VARIABLE NAMES AND DEFINITIONS
NAME DEFINITION
GOCHURCH frequency of atendance at church converted to
annual equivalents from interval answers
FAITH DUMMIES
JEW =1 if Jewish
RCATH =1 if Roman Catholic
all remaining faiths set =0
BLACK =1 if gives ethnic origin as black
ASIAN =1 if gives ethnic origin as Asian
OTHRACE =1 if non white,black or Asian
AGEL20 =1 if aged 16-19
AGEL40 =1 if aged 30-39
AGEL50 =1 if aged 40-49
AGEL60 =1 if aged 50-59
OLEVEL =1 if highest educational qualification is 'O'levels
ALEVEL =1 if highest educational qualification is 'A'levels
DEG =1 if highest educational qualification is degree
OTHQL =1 if highest educational qualification is a other
AFFLUENT =1 if lives in affluent housing area
MIDDLE =1 if lives in middle income housing area
POOR =1 if lives in poor housing area
MILBASE =1 if lives on a military base
COHAB =1 if cohabitatiing
NATKIDS =1 if has any natural children
LONGUNEM =1 if unemployed for more than six months
AWAYLOT =1 if away from home a lot
AWAYOCC =1 if away from home occasionally
LONGHOUR =1 if works >50 hours
INMOVE =1 if have been in the area < 5 years
BIGFAG =1 if heavy smoker
BIGDRINK =1 if heavy drinker based on EU definitons in terms
of alcohol 'units'.
EMBARRASS=1 if the respondent appeared to be embarassed
LISTEN =1 if there was someone else possibly listening
during the interview
CONDO1ST =1 if used a condom in first sexual experience
LAXWELL =1 if expressed lack of concern about personal health
AIDSHI =1 if thought that multiple partners caused a high
risk of catching HIV/AIDS
The following dummies are =1 for the name of the region stated
EANGL East Anglia
EMIDS East Midlands
GLOND Greater London
NWEST North West
SCOTL Scotland
SOUTHE South East
SOUTHW South West
WALES Wales
WMIDS West Midlands
YANDH Yorkshire and Humberside
[omitted categroy: NWEST=North West]
TABLE II: LOGIT EQUATION FOR HEAVY DRINKING. MALE SAMPLE.
Model Chi-Square (45 D.F.) = 270.362
Variable B S.E. Wald Sig Exp(B)
AFFLUENT .1072 .1731 .3840 .5355 1.1132
AGEL20 -.4782 .3896 1.5062 .2197 .6199
AGEL40 .5159 .1868 7.6295 .0057 1.6751
AGEL50 1.0996 .1912 33.0602 .0000 3.0028
AGEL60 .7574 .2251 11.3184 .0008 2.1328
AIDSHI -.4877 .1222 15.9188 .0001 .6140
ALEVEL -.0926 .1664 .3096 .5779 .9116
ASIAN -3.8568 4.2738 .8144 .3668 .0211
AWAYLOT .5288 .2165 5.9675 .0146 1.6968
BIGFAG .9107 .1252 52.8919 .0000 2.4862
BLACK .0680 .4307 .0249 .8745 1.0704
COHAB -.3024 .2568 1.3867 .2390 .7390
CONDO1ST -.3471 .1418 5.9925 .0144 .7067
CONTROL .1922 .1232 2.4325 .1188 1.2119
DEG -.9035 .2829 10.2002 .0014 .4052
DIVORCED -.2660 .2139 1.5467 .2136 .7664
EANGL -.7596 .4281 3.1487 .0760 .4678
EMBARASS -.4893 .3180 2.3671 .1239 .6131
EMIDS -.4609 .3363 1.8779 .1706 .6307
GLOND -.0050 .2823 .0003 .9858 .9950
GOCHURCH -.0261 .0072 12.9486 .0003 .9743
INMOVE .0076 .1528 .0025 .9605 1.0076
JEW .2488 .7464 .1112 .7388 1.2825
LAXWELL .8350 .2110 15.6547 .0001 2.3048
LISTEN .3972 .1287 9.5244 .0020 1.4877
LONGHOUR .4936 .1427 11.9716 .0005 1.6382
MARRIED -.7551 .2004 14.1959 .0002 .4699
MIDDLE -.4416 .1733 6.4909 .0108 .6430
MILBASE -4.2178 8.8365 .2278 .6331 .0147
NATKIDS .0234 .1637 .0205 .8862 1.0237
NWEST -.1686 .2826 .3561 .5507 .8448
OLEVEL -.0321 .1602 .0401 .8412 .9684
OTHQL .0936 .3904 .0576 .8104 1.0982
OTHRACE -.1712 .6008 .0812 .7757 .8426
POOR -.2708 .1823 2.2063 .1375 .7628
RCATH .1225 .2171 .3184 .5725 1.1304
SCOTL -.4933 .3102 2.5294 .1117 .6106
SHIFT -.1533 .1600 .9170 .3383 .8579
SOUTHE -.6152 .2788 4.8682 .0274 .5406
SOUTHW -.0580 .3021 .0369 .8477 .9436
WALES -.5271 .3697 2.0336 .1539 .5903
WIDOWED .3324 .3327 .9980 .3178 1.3942
WMIDS .1641 .2804 .3427 .5583 1.1784
YANDH -.1226 .2969 .1705 .6797 .8846
LONGUNEM -.3545 .1877 3.5682 .0589 .7015
Constant -3.5214 .3348 110.6118 .0000
TABLE III: LOGIT FOR DAILY DRINKING. MALE SAMPLE
Model Chi-Square (45 D.F.) 544.476
Variable B S.E. Wald Sig Exp(B)
AFFLUENT .2827 .0923 9.3884 .0022 1.3267
AGEL20 -.3121 .2244 1.9334 .1644 .7319
AGEL40 .4715 .1053 20.0661 .0000 1.6024
AGEL50 1.0108 .1098 84.6995 .0000 2.7479
AGEL60 1.1245 .1185 90.0173 .0000 3.0788
AIDSHI -.3441 .0664 26.8647 .0000 .7089
ALEVEL .2073 .0941 4.8552 .0276 1.2304
ASIAN -.3336 .3415 .9541 .3287 .7163
AWAYLOT .5857 .1171 25.0277 .0000 1.7963
BIGFAG .4912 .0736 44.5372 .0000 1.6342
BLACK -.2061 .2446 .7100 .3995 .8138
COHAB .2321 .1466 2.5067 .1134 1.2613
CONDO1ST -.1854 .0715 6.7316 .0095 .8308
CONTROL .2121 .0672 9.9554 .0016 1.2363
DEG .2446 .1186 4.2502 .0392 1.2771
DIVORCED .0519 .1248 .1726 .6778 1.0532
EANGL -.1674 .2190 .5843 .4446 .8459
EMBARASS -.2593 .1523 2.8994 .0886 .7716
EMIDS .0751 .1798 .1745 .6761 1.0780
GLOND .3276 .1641 3.9840 .0459 1.3876
GOCHURCH -.0119 .0026 20.7854 .0000 .9882
INMOVE .1643 .0789 4.3402 .0372 1.1786
JEW .0719 .3644 .0389 .8436 1.0746
LAXWELL .4805 .1280 14.0855 .0002 1.6168
LISTEN -.0882 .0708 1.5493 .2132 .9156
LONGHOUR .2746 .0778 12.4555 .0004 1.3160
MARRIED .1105 .1087 1.0330 .3095 1.1168
MIDDLE -.2550 .0928 7.5512 .0060 .7749
MILBASE -.5510 .7360 .5605 .4541 .5764
NATKIDS -.1949 .0872 4.9906 .0255 .8229
NWEST -.0939 .1701 .3047 .5809 .9104
OLEVEL .0617 .0951 .4207 .5166 1.0636
OTHQL .1963 .2234 .7725 .3795 1.2169
OTHRACE -.5518 .3394 2.6428 .1040 .5759
POOR -.2751 .1094 6.3249 .0119 .7595
RCATH .2731 .1135 5.7883 .0161 1.3141
SCOTL -.4777 .1873 6.5045 .0108 .6202
SHIFT -.1732 .0898 3.7212 .0537 .8410
SOUTHE .0907 .1550 .3424 .5584 1.0949
SOUTHW -.0538 .1801 .0893 .7650 .9476
WALES -.5782 .2268 6.4963 .0108 .5609
WIDOWED .2257 .2058 1.2023 .2729 1.2532
WMIDS .2714 .1681 2.6055 .1065 1.3118
YANDH -.1290 .1796 .5153 .4728 .8790
LONGUNEM -.1838 .1017 3.2664 .0707 .8321
Constant -2.8531 .1984 206.7379 .0000
DAILY ALCOHOL : FEMALE
TABLE IV: LOGIT EQUATION FOR HEAVY DRINKING. FEMALE SAMPLE.
Model Chi-Square (45 D.F.) 69.863
Variable B S.E. Wald Sig Exp(B)
AFFLUENT .8600 .6446 1.7797 .1822 2.3631
AGEL20 .8427 1.0322 .6666 .4143 2.3227
AGEL40 .5852 .7155 .6688 .4135 1.7953
AGEL50 .9963 .7594 1.7215 .1895 2.7083
AGEL60 1.2413 .8228 2.2763 .1314 3.4603
AIDSHI -1.0208 .4060 6.3210 .0119 .3603
ALEVEL -.7012 .7122 .9694 .3248 .4960
ASIAN 1.0620 1.1524 .8493 .3568 2.8922
AWAYLOT -6.7326 41.2604 .0266 .8704 .0012
BIGFAG 1.9489 .4561 18.2595 .0000 7.0207
BLACK -5.8743 37.3987 .0247 .8752 .0028
COHAB -1.0162 1.0841 .8787 .3486 .3620
CONDO1ST .0658 .4322 .0232 .8789 1.0681
CONTROL -.3503 .4135 .7176 .3969 .7045
DEG .6922 .7346 .8880 .3460 1.9982
DIVORCED -1.1822 1.2217 .9365 .3332 .3066
EANGL -5.4421 23.6038 .0532 .8177 .0043
EMBARASS .5614 .7653 .5381 .4632 1.7531
EMIDS 1.0557 1.2626 .6990 .4031 2.8739
GLOND 1.0531 1.1687 .8119 .3675 2.8666
GOCHURCH -.0184 .0181 1.0253 .3113 .9818
INMOVE .4561 .4910 .8626 .3530 1.5778
JEW -6.4090 60.3629 .0113 .9154 .0016
LAXWELL -.1142 1.0770 .0112 .9156 .8921
LISTEN -.3133 .4263 .5403 .4623 .7310
LONGHOUR 1.0611 .6338 2.8033 .0941 2.8897
MARRIED -.9424 .9480 .9883 .3202 .3897
MIDDLE .0037 .6638 .0000 .9956 1.0037
MILBASE -5.4212 58.9886 .0084 .9268 .0044
NATKIDS .7956 .7634 1.0862 .2973 2.2158
NWEST .5840 1.1790 .2454 .6204 1.7932
OLEVEL -.2308 .5122 .2030 .6523 .7939
OTHQL -6.8487 38.0564 .0324 .8572 .0011
OTHRACE -6.6974 41.6742 .0258 .8723 .0012
POOR .6878 .6559 1.0994 .2944 1.9893
RCATH 1.1574 .5271 4.8209 .0281 3.1817
SCOTL -.5490 1.4635 .1407 .7076 .5775
SHIFT .6291 .5376 1.3694 .2419 1.8758
SOUTHE 1.3419 1.1186 1.4391 .2303 3.8265
SOUTHW .4901 1.4547 .1135 .7362 1.6324
WALES 1.1078 1.2599 .7731 .3793 3.0277
WIDOWED -7.9104 45.2777 .0305 .8613 .0004
WMIDS .5857 1.2555 .2176 .6409 1.7962
YANDH 1.3332 1.1488 1.3466 .2459 3.7930
LONGUNEM -.2126 .5116 .1727 .6778 .8085
Constant -7.4856 1.5220 24.1888 .0000
TABLE V. LOGIT FOR DAILY DRINKING. FEMALE SAMPLE.
Model Chi-Square (45 D.F.) 332.969
Variable B S.E. Wald Sig Exp(B)
AFFLUENT .3767 .1579 5.6902 .0171 1.4575
AGEL20 .0385 .4104 .0088 .9253 1.0392
AGEL40 1.2395 .2257 30.1690 .0000 3.4538
AGEL50 2.0205 .2339 74.6114 .0000 7.5421
AGEL60 2.0601 .2548 65.3752 .0000 7.8466
AIDSHI -.2396 .1232 3.7831 .0518 .7869
ALEVEL .6491 .1744 13.8586 .0002 1.9138
ASIAN -.8222 .6149 1.7880 .1812 .4394
AWAYLOT -1.9018 1.0267 3.4311 .0640 .1493
BIGFAG .4950 .1407 12.3708 .0004 1.6405
BLACK -.9350 .7261 1.6580 .1979 .3926
COHAB .0078 .3271 .0006 .9811 1.0078
CONDO1ST -.0286 .1173 .0594 .8075 .9718
CONTROL .0197 .1170 .0284 .8663 1.0199
DEG 1.1797 .2093 31.7747 .0000 3.2533
DIVORCED -.1015 .3950 .0661 .7972 .9035
EANGL 1.5370 .4782 10.3314 .0013 4.6507
EMBARASS .0827 .2519 .1079 .7426 1.0862
EMIDS 1.2338 .4650 7.0394 .0080 3.4342
GLOND 1.5545 .4474 12.0706 .0005 4.7329
GOCHURCH -.0148 .0044 11.5750 .0007 .9853
INMOVE .1818 .1406 1.6710 .1961 1.1994
JEW -.7091 .7381 .9231 .3367 .4921
LAXWELL .0217 .3308 .0043 .9478 1.0219
LISTEN -.0458 .1190 .1484 .7000 .9552
LONGHOUR .7432 .2178 11.6439 .0006 2.1027
MARRIED -.1551 .3037 .2608 .6095 .8563
MIDDLE -.2012 .1626 1.5305 .2160 .8177
MILBASE 1.0735 .6288 2.9143 .0878 2.9255
NATKIDS -.2499 .1742 2.0568 .1515 .7789
NWEST 1.2550 .4450 7.9532 .0048 3.5079
OLEVEL .3313 .1605 4.2575 .0391 1.3927
OTHQL -.2468 .6101 .1637 .6858 .7813
OTHRACE -1.6933 1.0197 2.7572 .0968 .1839
POOR -.1672 .2086 .6421 .4230 .8460
RCATH .1770 .1970 .8071 .3690 1.1936
SCOTL .2039 .5033 .1642 .6853 1.2262
SHIFT -.3637 .2047 3.1568 .0756 .6951
SOUTHE 1.3482 .4335 9.6697 .0019 3.8503
SOUTHW 1.0772 .4617 5.4440 .0196 2.9364
WALES .8590 .4986 2.9678 .0849 2.3608
WIDOWED -1.7831 1.0602 2.8287 .0926 .1681
WMIDS 1.4152 .4501 9.8864 .0017 4.1172
YANDH .6878 .4747 2.0989 .1474 1.9893
LONGUNEM -.1867 .1396 1.7869 .1813 .8297
Constant -5.4665 .5402 102.3989 .0000
ABSTRACT
HEAVY DRINKING: RISK, RESTRAINT AND OTHER DETERMINANTS.
It is shown, using a large national survey, that risk
taking attitude influences both heavy and daily drinking. There
are also significant effects of age, religion, education and
situational factors. Some factors also differ by gender notably
the effect of working away from home.
Samuel Cameron
Economics
Development and Project Planning Centre
University of Bradford
Richmond Road
BRADFORD BD7 1DP
England
Tel: 01274 234772
E-mail: samcameron@lineone.net
Website: http://website.lineone.net/~samcameron