- Email: [email protected]
Self-esteem and smoking in youth—muddying the waters?
Journal of Adolescence 2002, 25, 415–425 doi:10.1006/jado.2002.0485, available online at http://www.idealibrary.com on
Self-esteem and smoking in youthFmuddying the waters? ANTHONY GLENDINNING Longitudinal analysis by McGee and Williams (2000, Journal of Adolescence, 23, 569– 582, doi: 10.1006/jado.2000.0344) indicates that global self-esteem is not related to substance use in early youth. In the case of tobacco use Glendinning and Inglis (1999, Journal of Adolescence, 22, 673–682, doi: 10.1006/jado.1999.0262) have looked at the ‘‘problem’’ of self-esteem in youth and its relevance for smoking, and they also note that the evidence from the survey literature has been inconclusive. However, rather than suggest that the survey methods and data have been inadequate, Glendinning and Inglis argue that an explanation is provided by looking at the relationship between global self-esteem and smoking in much greater detail, specifically within the peer context, and at peer culture, and the meanings that different groupings of young people attach to smoking or not smoking. That is, rather than a direct link between global self-esteem and smoking behaviour in youth, both are bound up with peer status and differentiation in early youth, although it must be said, not necessarily in the ‘‘expected’’ way. In the present study longitudinal data from the British Household Panel Study (BHPS, 2001, Economic & Social Science Research Council, Research Centre on Micro-Social Change. British Household Panel Survey Colchester, Essex: The Data Archive, 28 February 2001.SN: 4340.) confirm this more complex picture, and qualify the conclusions of McGee and Williams, in that global self-esteem year-on-year at around age 12–14Fwhen young people take up smoking in increasing numbersFis clearly linked to experimentation and to smoking in subsequent years, in the shorter term. However, a longer-term linkage between self-esteem in early youth and smoking in later youth is less clear cut, and less compelling; but then, pursuing the longitudinal analysis still further, the findings lend force to the argument that putative links between self-esteem and smoking must be understood in context, specifically the peer context. r 2002 The Association for Professionals in Services for Adolescents. Published by Elsevier Science Ltd. All rights reserved
Introduction McGee and Williams (2000) have recently re-assessed the relationship between self-esteem and a range of ‘‘health-compromising’’ behaviours in youth by examining data collected over time in pre-, early- and mid-adolescence from a large sample of young New Zealanders. At the outset they remark that: It is often believed that low self-esteem is associated with such health-compromising behaviours in adolescence as substance use, early sexual activity, eating problems and suicidal ideation. Surprisingly, there is little longitudinal research addressing this issue.
(McGee and Williams, 2000, p. 569) Addressing this gap in the literatureFand the assumption that low self-esteem is a ‘‘problem’’ in youth (Kahne, 1996), resulting in the uptake of smoking, or drug use, or early Reprint requests and correspondence should be addressed to A. Glendinning, Department of Sociology & Anthropology, University of Aberdeen, Edward Wright Building, Dunbar Street, Old Aberdeen AB24 3QY, U.K. (E-mail: [email protected]) 0140-1971/02/$3500+000
# 2002 The Association for Professionals in Services for Adolescents. Published by Elsevier Science Ltd. All rights reserved
416
A. Glendinning
sexual activityFMcGee and Williams found that prior self-esteem was predictive of subsequent reports of eating problem and suicidal ideation in mid-adolescence, but that prior self-esteem was not related to substance use, for example, smoking.1 The British Household Panel Study (BHPS, 2001) provides a further excellent opportunity to look in detail over time at this more specific issue of young people’s smoking and its relationship to self-esteem, on a national scale in Britain. That is the aim of the present article. Previously, when reviewing the literature on smoking in youth, Glendinning and Inglis (1999) had also observed that survey results were inconclusive, and that there was no clear-cut relationship between self-esteem and smoking; at least, that was, they argued, until the peer context was taken into account, along with the meanings and ‘‘practical logic’’ that underlie young people’s smoking predispositions and practices. In particular, two separate surveys of young Scots by Glendinning and colleagues Fone national and the other ruralFhad found that self-esteem was, at first sight, apparently unrelated to smoking. West and Sweeting (1997) also looked at low selfesteem and lifestyles at this age, including smoking, this time in two urban surveys of young Scots in one city, and reported inconclusive results. Once again, they also argued that youth culture and the peer context must be taken into account in order to make sense of these findings. The basis for an explanation is perhaps provided by the in-depth work of Michell and colleagues (Michell, 1997a; Michell and Amos, 1997b) with two small groups of young urban Scots in one location, over an extended time period, using a range of methods, which emphasized the importance of peer group differentiation and peer status for self-image and smoking in early-to mid-adolescence. Most importantly, the findings from this qualitative study agreed with and allowed for interpretation and extension of the large-scale survey work of Glendinning and colleagues. Put simply, smoking or not smoking was seen to act as one marker of status and differentiation amongst peers, and of youth lifestyles more generally, but not necessarily in a straightforward way. For example, at this earlier age, from the perspective of peers, attempts at smoking by socially isolated youth may be viewed as another ‘‘sad’’ act serving to confirm low status among peers, and serving to delay the uptake of smoking among those with low self-esteem, and to nullify or transform its significance as a marker of status. Returning to the aim of the present article and to the British Household Panel Study this provides an opportunity to extend research in the particular case of cigarette smoking, working with new longitudinal data. Bergman and Scott (2001, p. 191, 195) have already begun this process by using data from the BHPS to consider ‘‘gender and socio-economic differences in well-being and health-risk behaviours,’’ and as one aspect of this, they examine the relationship between psychological well-being, including self-esteem and negative selfefficacy, and smoking among 11–15-year olds. In line with much previous research, their conclusion is that ‘‘among young adolescents, smoking is not consistently or strongly related to self-esteem, self-efficacy, happiness, or past worries,’’ although there are some significant correlations in the data. However, Bergman and Scott do not utilize the full potential of the BHPS, in that: they look at gender and socio-economic differences, but aggregate data for all 11–15-year olds, across a period of considerable change in young people’s lives, specifically in the uptake of smoking; they use only one of the self-report measures available to them; and 1 In addition, another article by McGee et al. (1998) using data from the same study found little evidence to support the idea that earlier mental health disorders in childhood put young people ‘‘at risk’’ of smoking.
Self-esteem and smoking in youth
417
also, they conduct cross-sectional analysis for each year, separately, rather than exploit the longitudinal nature of the data, including the ability to keep track of young people in subsequent years, over time, and into later youth. So, moving on from Bergman and Scott’s analysis, what does a more detailed examination of longitudinal data from the BHPS reveal, first, about the year-on-year relationship between self-esteem and smoking in early youth, and second, about the longer-term relationship between self-esteem in early youth and smoking in later youth? Data from the BHPS makes it possible to address the first question by looking at one cohort of young people over 4 years, starting from age 11–12, to 12–13, to 13–14, and then through to age 14–15. In this, comparisons are made not only between self-esteem at an earlier age and smoking at a later age but also between smoking at an earlier age and self-esteem at a later age, with no prior assumptions about ‘‘what leads to what.’’ Additionally, it is possible to address the second question about the longer term by looking at a further group of BHPS respondents, who were initially 14–15 years of age, and who were followed up some 6 years later at age 19–20. Here, the relationship between earlier self-esteem and later smoking behaviour is re-examined, whilst taking into account peer context and peer status.
Methods Sample Most of the data are taken from the youth component of the British Household Panel Study (BHPS). The main BHPS is a national survey, launched in 1991, which takes place every year (Institute for Social & Economic Research, 2001). It involves interviews with members of some 5500 households throughout Britain, and it tracks households and household members over time. The main survey collects information from every adult in the household (age 16+) and includes data on income, employment, health, and housing. The main survey was extended in 1994, and in subsequent years, to include the British Household Panel Study Youth Survey (BHPS-YS) of 11, 12, 13, 14 and 15 year olds who were also resident in the main panel study households. Over a 5-year period, from 1994 onwards, successive waves of the BHPS-YS were sponsored by the Health Education Authority (HEA). Each young person was interviewed separately from other members of the household by providing the survey questions on tape, allowing respondents to record their answers using a personal stereo (Scott et al., 1995). A supplement to the main BHPS also asked parents about their children. Typically, around 750 young people from some 600 households were interviewed in each wave of the BHPS-YS. (At each wave respondents moved into the main survey on reaching the of age 16, to be replaced by a new cohort of 11-year olds from the panel study households.) Response rates in 1994 to the initial survey of 11–15-year olds were reassuringly high (Brynin and Scott, 1996). At follow-up in 1997 almost 90 per cent of the original sample of respondents was re-contacted, either as a respondent to the youth survey (age 11–15), or else as a respondent to the main survey (age 16+) in that year. By 1999 around 85 per cent of the original sample of respondents from 1994 were re-contacted as part of the main survey.
418
A. Glendinning
MEASURES
&
SAMPLE PROFILES
Self-esteem: Five items were used in the BHPS-YS as a measure of general self-esteem (deriving from a 10-item scale originally developed by Rosenberg, 1965): Please say whether you strongly agree, agree, disagree or strongly disagree, that the following statements apply to yourself: 1. 2. 3. 4. 5.
I feel that I have a number of good qualities I certainly feel useless at time I am a likeable person All in all, I am inclined to feel I am a failure At times, I feel I am no good at all
(Rosenberg: (Rosenberg: (Rosenberg: (Rosenberg: (Rosenberg:
Item 3) Item 6) Correlate) Item 9) Item 2)
A scale was constructed from these five items (affi 0?7).2 Each item was coded as 0, 1, 2, or 3 (the second and fourth items were re-coded so higher scores corresponded to positive selfevaluations). The mean was calculated and multiplied by 10. Scores therefore lay in the range 0–30. As there were no absolute cut-points indicative of ‘‘low’’ or ‘‘high’’ self-esteem the sample was divided into thirds, on the basis of 1994 responses, into those reporting relatively ‘‘low’’ [0-17], ‘‘moderate’’ [18, 19, and 20], or ‘‘high’’ self-esteem [21-30]. The same cut-points were used for self-esteem scores in subsequent years. (Although profiles of emotional well-being were available in the main BHPS, a measure of self-esteem was not included in the main BHPS, that is, beyond age 15.) Prevalence of low self-esteem: To provide a reference point for the measure of self-esteem, and in terms of negative self-evaluations made by 11–15-year olds in 1994: around one-inten of the sample disagreed with the statement, ‘‘I am a likeable person;’’ and around one-inten agreed that, ‘‘All in all, I am inclined to feel I am a failure;’’ whilst around one-in-three agreed that, ‘‘At times, I feel I am no good at all.’’ Thus, on the whole, young people’s selfimage and self-evaluations appeared to be relatively positive.
Socially isolated youth. In order to identify distinctive social groupings of young people a range of items from the baseline BHPS-YS in 1994 were subjected to cluster analysis, including friendships, activities and time spent with friends and peers, and getting into ‘‘fights’’ and ‘‘trouble’’ (an approach used in previous work, Glendinning and Inglis, 1999). From this analysis four separate groupings were identified: (1) ‘‘peer-oriented’’ youth, with extended friendship networks; (2) more ‘‘conventional’’ family- and school-oriented youth; (3) more ‘‘socially isolated’’ youth; and, (4) peer-oriented, but ‘‘disaffected’’ youth. (A more detailed account is provided in Glendinning, 2000.) On the basis of this characterization, and in relative terms, according to young people’s own accounts about one-in-five of the baseline sample in 1994 could be said to be more isolated, whilst amongst 14–15-year olds around 15 per cent (N=50) could be described in this way.
Smoking behaviour. The BHPS-YS included a fairly standard set of items used by other U.K. researchers, relating to smoking, including experimentation, age of onset, current status, and amount smoked in the last 7 days, alongside orientations and beliefs. Responses to these items in 1994, 1995, 1996 and 1997 were re-categorized to produce three variables: (1) (2)
whether or not the young person had ever smoked, even a single puff; the number of cigarettes the young person had smoked in the last 7 days, if at all;
2 Even although further analysis of the first year of youth survey data in 1994 suggested that it was possible to identify two components, rather than just one (also, see Bergman and Scott, 2001) but not in subsequent years.
Self-esteem and smoking in youth
(3)
419
and, the young person’s current smoking statusFas having never smoked, as having tried once or twice, as having given-up, as smoking occasionally, or as smoking regularly, i.e., at least one cigarette per week.
In the BHPS-YS in 1997, for example, only a minority of 15-year olds had never tried a cigarette, but more than half said that they had never smoked a whole cigarette, and threequarters had not smoked a cigarette in the last 7 days. Nonetheless, one-in-three 15-year olds described themselves as current occasional or regular smokers, and one-in-five 15-year olds currently smoked an average of one or more cigarettes a day. Unlike the BHPS-YS, profiles of adult smoking for most years of the main BHPS have been partial, offering only a limited pictureFput simply, the questions were not asked. However, by the time that the first wave of 11–15 year olds from the initial BHPS-YS in 1994 had entered the main BHPS in 1999, some 6 years later, a much fuller set of survey items relating to smoking had been included within the main BHPS, including ever smoked, current status, amount currently smoked per day, and age when first smoked regularly. This made it possible to look at longer-term links between earlier self-esteem and smoking in later youth within the main BHPS.
Comparisons with other Health Education Authority (HEA) sponsored studies. It is worth noting that the size of the sample available for analysis in any 1 year of the BHPS-YS was relatively small, for example, at around only 150 15-year olds, and so estimates of smoking prevalence therefore involve a significant degree of uncertainty. However, the BHPS-YS sample of households can be viewed as representative, and can be weighted, so it was certainly worthwhile making comparisons with other large-scale surveys. For example, a series of major school-based surveys sponsored by the Health Education Authority in 1989, 1995 and 1997 make this possible (HEA, 1992a, b, 1997, 1999). The fit between estimates of smoking from these larger scale cross-sectional surveys and the BHPSYS data was generally good (whether unweighted or weighted data were used).3 (Table 1)
Statistical Analyses The data were analysed using SPSS for Windows. After a preliminary examination of selfesteem and smoking for all 11–15-year olds at baseline in 1994, analyses were restricted to 11–12-year olds from around 250 households (N=316) where these same young people were followed up at age 12–13 years in 1995 (N=303), at age 13–14 years in 1996 (N=294), and at age 14–15 years in 1997 (N=279). In addition, analyses were restricted to basic crosstabulations along with chi-square tests for the examination of self-esteem and ‘‘ever smoked’’ and self-esteem and ‘‘current smoking’’ and to non-parametric median tests for the examination of self-esteem and ‘‘amount smoked’’. The aim was to produce a simple picture by simple means, and to examine the extent to which self-esteem anticipated smoking, or conversely, smoking anticipated self-esteem. Next, the relationship between smoking and peer groupings at age 14–15 years in 1994 was examined (N=308) and also that between self-esteem and peer groupings at the same age in the same year (N=304). Finally, to develop the picture, possible links between self-esteem and peer groupings at age 14–15 years in 1994 and subsequent smoking at age 19–20 in 1999 was also examined using basic crosstabulations (N=242) and data relating to the age at which the respondent first smoked 3 Smoking prevalence did appear to be a little lower in the first year of the BHPS-YS, in 1994, and also, those stating that they had never smoked was consistently higher in the BHPS-YS, in 1994, 95, 96 and 97).
420
Table 1
A. Glendinning
The smoking behaviour of 15-year olds. HEA-sponsored surveys, 1989 to 1997
Self-reported current smoking
Never smoked (%) Once or twice (%) Used to smoke (%) Smoke occasionally (%) Smoke regularly (%) N (=100%)
BHPS-YS
WHO-HBSC
HEA
1994
1995
1996
1997
1995
1997
1989
48 28
39 30
40 26
43 22
34 32
32 32
39 31
24
31
34
33
34
35
30
152
146
138
146
1891
2135
1639
Sources: BHPS-YS (1994, 95, 96 and 97), HEA (1992), HEA (1997) and HEA (1999).
regularly was also examined using analysis of variance and an associated F-test (N=96). This second stage to the analyses was done in order to test whether or not low self-esteem, social isolation, and low status among peers were predictive of smoking in the longer term in later youth. Fortunately, as mentioned above, 1999 was the first year for which more detailed selfreport smoking data were available within the main BHPS.
Findings Smoking in early youth The findings paint a clear picture of the relationship between self-esteem and smoking in early youth. Overall, young people who reported lower self-esteem were more likely to smoke (see, Table 2 for an aggregate picture using concurrent data at one time point). Looking at the much more detailed picture over time that is represented in Figure 1 (from age 11–12, to age 12–13, to age 13–14, and through to age 14–15) there was a compelling and emergent pattern of significant associations between earlier self-esteem and smoking behaviour in subsequent years, particularly by age 14–15, as is illustrated in Table 3, where this reports a typical set of significant longitudinal results. Thus, earlier self-esteem was related to present and future smoking behaviour, whilst current self-esteem was associated with experimentation. The results in Figure 1 clearly show that the converse was not trueFsmoking ‘‘now’’ was not related to self-esteem in future years. Table 2
Self-esteem and current smoking, 11–15-year olds. BHPS-YS, 1994 (N=765)
Smoking status
Never smoked (%) Experimented (%) Ex-smoker (%) Occasional smoker (%) Regular smoker (%) Note: w2(8)=27?24, po0?001.
General self-esteem
Total
Low
Middle
High
65 7 10 9 9 100
75 10 8 4 3 100
77 11 5 5 2 100
72 9 9 6 4 100
Self-esteem and smoking in youth
421
Figure 1 The relationship between self-esteem and smoking over time, starting at age 11–12. BHPSYS, 1994, 95, 96 and 97 (N=264). General self-esteem score Smoking behaviour at age 11–12 (low, middle, high) Never Current Number smoked status last week
Smoking behaviour at age 12–13 Never smoked
Current status
Number last week
At 11–12 years old
NS
NS
NS
NS
NS
NS
At 12–13 years old
NS
NS
NS
*
NS
NS
At 13–14 years old
NS
NS
NS
NS
NS
NS
At 14–15 years old
NS
NS
NS
NS
NS
NS
General self-esteem score Smoking behaviour at age 13–14 (low, middle, high) Never Current Number smoked status last week
Smoking behaviour at age 14–15 Never smoked
Current status
Number last week
At 11–12 years old
NS
NS
NS
NS
NS
NS
At 12–13 years old
**
NS
NS
**
*
NS
At 13–14 years old
*
NS
NS
**
*
*
At 14–15 years old
NS
NS
NS
*
NS
NS
In terms of the general self-esteem scale: lower=[0-17]; middle=[18,19,20]; and, higher=[21-30] p-values based on cross-tabulation w2 or non-parametric median tests, as appropriate Key: NS=not significant; *=po0?05 and **=po0?01.
Table 3
Self-esteem at age 13–14 and smoking at age 14–15. BHPS-YS, 1996 & 97 (N=264)
Smoking status at age 14–15
Never smoked (%) Experimented (%) Ex-smoker (%) Occasional smoker (%) Regular smoker (%)
General self-esteem at age 13–14
Total
Low
Middle
High
33 12 12 20 23 100
46 20 12 6 16 100
52 19 7 11 11 100
46 17 9 12 16 100
Note: w2(8)=16?56, po0?05.
Hence, by following one cohort of young people from age 11–12, a very clear pattern of associations emerged from the BHPS-YS data over time. Associations were not really evident until around age 13–14. At this age, self-esteem in the previous year was related to current smoking status and uptake of ‘‘regular’’ smoking, and by age 14–15 self-esteem in the previous 2 years was related to current smoking status and to amount smoked.
422
A. Glendinning
Table 4 Prevalence of low self-esteem and smoking at age 14–15 by peer groupings at age 14–15. BHPS-YS, 1994 (N=304 and 308) Peer social groupings at age 14–15
Total
Conventional Peer oriented Socially isolated Disaffected a
Low self-esteem at 14–15 (%) Current smoker at 14–15b (%)
33 9
19 14
47 17
36 39
34 20
a 2
w (3)=7?91, po0?05. w (3)=30?24, po0?001.
b 2
At ages 12–13, 13–14 and 14–15 concurrent feeling of self-esteem were associated with experimentation at the same age. It would appear that for 12, 13 and 14 year olds feelings of low self-esteem were associated with increased likelihood of smoking in subsequent years, and raised self-esteem with reduced prevalence. Next, peer social groupings were introduced into the picture to suggest that, as well as selfesteem directly, these were also important for smoking in early youth, where from this perspective self-image and self-esteem were seen as bound up with peer status, along with smoking. For example, at age 14–15 under 20 per cent of ‘‘popular, peer-oriented’’ youth reported relatively low self-esteem, whilst approaching 50 per cent of more ‘‘socially isolated’’ youth reported low self-esteem. Yet the picture was not straightforward, and as can be seen from Table 4, the likelihood of smoking at age 14–15 was little different between ‘‘isolated’’ youth and ‘‘popular’’ youth4 although it was very clear that ‘‘conventional’’ youth reported lower levels of smoking whilst ‘‘disaffected’’ youth reported higher levels of smoking.
SMOKING
IN LATER YOUTH
It was also possible to extend the analysis by tracking a group of 14–15-year olds from the BHPS-YS through to the main BHPS at age 19–20. Unlike early youth, the results were inconclusive for direct links between earlier self-esteem and later smoking. Thus, results were not significant for respondents’ current smoking behaviour at age 19–20, although in respect of smoking history there was evidence to suggest that those who had earlier reported lower self-esteem were more likely to have been regular smokers at some point, whilst those who had reported higher self-esteem were less likely to have ever smoked, beyond trying a cigarette once or twice. Turning to a consideration of peer social groupings at age 14–15 and smoking some six years later at age 19–20, as can been seen from Table 5, differences in smoking prevalence among peer groupings were very evident. Twenty-four per cent of those who had reported extended friendship networks at an earlier age currently smoked at age 19–20, and unlike the picture in early youth, forty-four percent of those who had reported feeling socially isolated at an earlier age currently smoked by age 19–20. Interestingly, as shown in Table 5, this was the case even though ‘‘popular’’ smokers appeared to have started regular smoking earlier (twothirds at age under 16) whilst ‘‘isolated’’ smokers had started late (two-thirds at age 16 or older). To sum up this last set of findings, social isolation had been associated with low self-esteem in mid-adolescence, but not with increased smoking, and yet an earlier sense of isolation was 4
However, amongst ‘popular’ youth, girls were more likely to be smokers when compared to boys.
Self-esteem and smoking in youth
423
Table 5 Prevalence of smoking at age 19–20 by peer groupings at age 14–15. BHPS-YS, 1994 & BHPS, 1999. (N=242 and 96) Peer social groupings at age 14–15 Conventional 18 Current smoker at 19–20a (%) Mean age first smoked regularly 15?2 (95% CI)b (14?5, 15?9)
Peer oriented
Socially isolated
Total
Disaffected
24 44 52 33 13?9 16?1 14?9 15?1 (12?6, 15?1) (15?3, 16?8) (14?3, 15?5)
a 2
w (3)=23?52, po0?001. F(3,92)=3?67, po0?05.
b
subsequently linked to the uptake of smoking beyond this ageFreinforcing the idea that low self-esteem in youth when understood in context does matter in the longer term for smoking. Notably, over time smoking was consistently high among ‘‘disaffected’’ youth and consistently low among ‘‘conventional’’ youth, where neither group had reported significantly higher or significantly lower self-esteemFfor one group smoking was the ‘‘norm’’ and for the other group non-smoking was the ‘‘norm’’, irrespective of overall feelings of self-esteem.
Conclusion The present findings for self-esteem and smoking do not concur with what has emerged as a largely consensus view within the recent research literature, if not necessarily within health promotion and health education circles, which says that self-esteem and smoking in early youth are largely unrelated, so do they muddy the waters? Indeed, the analysis of British Household Panel Study data provide an unambiguous longitudinal picture of self-esteem year-on-year at around the age 12–14 yearsFwhen young people take up smoking in increasing numbersFas linked to experimentation with smoking and also to subsequent smoking behaviour, at least in the shorter term. However, further analysis of British Household Panel Study data, suggest that a longer-term linkage between low self-esteem in early youth and smoking in later youth is less clear cut, and less compelling; but also, that putative links between self-esteem and smoking must be understood in context, specifically within the peer context, and by reference to peer status and differentiation, and the meanings that smoking or not smoking have for different peer social groupings. As has been proposed by Glendinning and Inglis (1999), following the lead provided by Michell (1997), and borrowing a theoretical framework from Bourdieu (1992) as applied to the ‘‘youth field’’, what matters is youth culture, and the ‘‘practical logic’’ as to why young people do smoke or do not smoke (cf. ‘‘habitus’’). Self-image and general feelings of selfesteem are bound up with this, rather than self-esteem directly causing young people to smoke or not to smoke. From that perspective the findings from the present study do not muddy the waters, but instead simply serve further to refine the picture offered by recent studies. The findings also suggest that there is really no need to look at what has been termed ‘‘domain-specific self-esteem’’ unless the intention is to point to underlying processes within
424
A. Glendinning
the peer and other contexts.5 When considered as a whole, what is striking is that research on self-esteem and smoking in youth now offers a cumulative picture, where results from different studies using different methods can be said to build upon each other, including large-scale longitudinal quantitative work and small-scale in-depth qualitative work. Finally, the study has not looked at gender differences, and once again following the lead of Lynn Michell, this may well be a further important factor within the picture that has been presented here.
Postscript At the time of completion of the present article in late 2001 an excellent review, SelfesteemFthe cost and causes of low self-worth, by Emler (2001) has just been published. As in the previous article by Glendinning and Inglis (1999) the general point is made by Emler that: There is a widespread view that self-esteem is a risk factor for a broad range of psychological and behavioural problems [and yet] neither public discussion nor decisions to invest in prevention and treatment have been strongly informed or guided by hard evidence, either about the effects or causes of low self-esteem.
In the case of smoking in youth, however, Emler (2001, 22, 87) largely leaves the story with McGee and WilliamsFthe relationship between low self-esteem and smoking is at most weakFwhich is the point of departure for the present article. He also only considers the peer context from the perspective of ‘‘peer group pressure,’’ with self-esteem acting there as a buffer or moderating factor, rather than from the perspective of social interaction, status, and position; although he does point out elsewhere that social standing within the wider peer group may be important, even if these are only young people’s own self-perceptions or selfevaluations.
Acknowledgements The views expressed are solely those of the author. However, I would like to acknowledge the following people. Pat West for the encouragement that he has given me over the years, and for pointing out to me that there was an issue here in the first place. Lynn Michell for the insights that her work gave me. David Inglis for his sociological inspiration. Hilary Graham for suggesting that I analyse the BHPS-YS, and for making the introductions. Julie Bull at HEA (and Helen Ryan) for making the analysis possible through financial support. Nick Buck and colleagues for the welcome at ISER, and the assistance of the ESRC Research Centre for Micro-Social Change, and the ESRC Data Archive, in supplying BHPS data and documentation. Jackie Scott and Malcolm Brynin for their work on the initial design and reporting of the BHPS-YS. Lynda Clarke, Julie Williams, and Jonathan Bradshaw for access 5 Glendinning et al. (1995) had earlier looked at distinct social groupings in youth to include elements of the family and school contexts, as well as the peer context, and they then argued that self-image and smoking, along with aspects of youth culture, such as, consumption patterns, were bound up with these, where social groupings were seen as representative of distinctive youth lifestyles.
Self-esteem and smoking in youth
425
to work in progress. And to all of the young people who participated in the BHPSYSFThank you.
References Bergman, M. and Scott, J. (2001). Young adolescents’ wellbeing and health-risk behaviours: gender and socio-economic difference. Journal of Adolescence, 24, 183–197, doi:10.1006/jado.2001.0378. British Household Panel Study (BHPS) (2001). Economic & Social Science Research Council, Research Centre on Micro-Social Change. British Household Panel Survey. Colchester, Essex: The Data Archive, 28 February 2001. SN: 4340. Bourdieu, P. (1992). The logic of practice. Cambridge: Polity Press. Brynin, M. and Scott, J. (1996). Young people and the familyFfamily health research report. London: Health Education Authority. Emler, N. (2001). Self-esteem: the costs and causes of low self-worth. York: Joseph Rowntree Foundation, York Publishing Services. Glendinning, A., Shucksmith, J. and Hendry, L. (1995). Lifestyle, health & social class in adolescence. Social Science & Medicine, 41, 235–248. Glendinning, A. and Inglis, D. (1999). Smoking behaviour in youth: the problem of low self-esteem? Journal of Adolescence, 22, 673–682, doi:10.1006/jado.1999.0262. Glendinning, A. (2000). Health variations in youth: a picture of health, family & peer culture from the BHPS youth survey. London: Health Education Authority. Health Education Authority (HEA) Rudat, K., Speed, M. and Ryan, H. (1992a). Tomorrow’s young adults. London: Health Education Authority. Health Education Authority (HEA) Rudat, K., Ryan, H. and Speed, M. (1992b). Today’s young adults. London: Health Education Authority. Health Education Authority (HEA) Turtle, J., Jones, A. and Hickman, M. (1997). Young people & health. London: Health Education Authority. Health Education Authority (HEA) Haselden, L., Angle, H. and Hickman, M. (1999). Young people & health. London: Health Education Authority. Institute for Social & Economic Research (ISER) (2001). http://www.iser.essex.ac.uk/bhps 19 April, 2001. Kahne, J. (1996). The politics of self-esteem. American Educational Research Journal, 33, 3–22. McGee, R., Williams, S. and Stanton, W. (1998). Is mental health in childhood a major predictor of smoking in adolescence? Addiction, 93, 1869–1874. McGee, R. and Williams, S. (2000). Does low self-esteem predict health-compromising behaviours among adolescents? Journal of Adolescence, 23, 569–582, doi:10.1006/jado.2000.0344. Michell, L. (1997a). Loud, sad or bad: young people’s perceptions of peer groups and smoking. Health Education Research, 12, 1–14. Michell, L. and Amos, A. (1997b). Girls, pecking order and smoking. Social Science & Medicine, 44, 1861–1869. Rosenberg, M. (1965). Society and the adolescent self-image. : Princeton, NJ.: Princeton University Press. Scott, J. Brynin, M. and Smith R. (1995). Interviewing children in the British household panel study. In Advances in Family Research, Hoy, J.J. et al. (Eds). Amsterdam: Thesis Publishers. West, P. and Sweeting, H. (1997). ‘Lost souls’ and ‘rebels’: a challenge to the assumption that low selfesteem and unhealthy lifestyles are related. Health Education, 97(5), 161–167.
Self-esteem and smoking in youthFmuddying the waters? ANTHONY GLENDINNING Longitudinal analysis by McGee and Williams (2000, Journal of Adolescence, 23, 569– 582, doi: 10.1006/jado.2000.0344) indicates that global self-esteem is not related to substance use in early youth. In the case of tobacco use Glendinning and Inglis (1999, Journal of Adolescence, 22, 673–682, doi: 10.1006/jado.1999.0262) have looked at the ‘‘problem’’ of self-esteem in youth and its relevance for smoking, and they also note that the evidence from the survey literature has been inconclusive. However, rather than suggest that the survey methods and data have been inadequate, Glendinning and Inglis argue that an explanation is provided by looking at the relationship between global self-esteem and smoking in much greater detail, specifically within the peer context, and at peer culture, and the meanings that different groupings of young people attach to smoking or not smoking. That is, rather than a direct link between global self-esteem and smoking behaviour in youth, both are bound up with peer status and differentiation in early youth, although it must be said, not necessarily in the ‘‘expected’’ way. In the present study longitudinal data from the British Household Panel Study (BHPS, 2001, Economic & Social Science Research Council, Research Centre on Micro-Social Change. British Household Panel Survey Colchester, Essex: The Data Archive, 28 February 2001.SN: 4340.) confirm this more complex picture, and qualify the conclusions of McGee and Williams, in that global self-esteem year-on-year at around age 12–14Fwhen young people take up smoking in increasing numbersFis clearly linked to experimentation and to smoking in subsequent years, in the shorter term. However, a longer-term linkage between self-esteem in early youth and smoking in later youth is less clear cut, and less compelling; but then, pursuing the longitudinal analysis still further, the findings lend force to the argument that putative links between self-esteem and smoking must be understood in context, specifically the peer context. r 2002 The Association for Professionals in Services for Adolescents. Published by Elsevier Science Ltd. All rights reserved
Introduction McGee and Williams (2000) have recently re-assessed the relationship between self-esteem and a range of ‘‘health-compromising’’ behaviours in youth by examining data collected over time in pre-, early- and mid-adolescence from a large sample of young New Zealanders. At the outset they remark that: It is often believed that low self-esteem is associated with such health-compromising behaviours in adolescence as substance use, early sexual activity, eating problems and suicidal ideation. Surprisingly, there is little longitudinal research addressing this issue.
(McGee and Williams, 2000, p. 569) Addressing this gap in the literatureFand the assumption that low self-esteem is a ‘‘problem’’ in youth (Kahne, 1996), resulting in the uptake of smoking, or drug use, or early Reprint requests and correspondence should be addressed to A. Glendinning, Department of Sociology & Anthropology, University of Aberdeen, Edward Wright Building, Dunbar Street, Old Aberdeen AB24 3QY, U.K. (E-mail: [email protected]) 0140-1971/02/$3500+000
# 2002 The Association for Professionals in Services for Adolescents. Published by Elsevier Science Ltd. All rights reserved
416
A. Glendinning
sexual activityFMcGee and Williams found that prior self-esteem was predictive of subsequent reports of eating problem and suicidal ideation in mid-adolescence, but that prior self-esteem was not related to substance use, for example, smoking.1 The British Household Panel Study (BHPS, 2001) provides a further excellent opportunity to look in detail over time at this more specific issue of young people’s smoking and its relationship to self-esteem, on a national scale in Britain. That is the aim of the present article. Previously, when reviewing the literature on smoking in youth, Glendinning and Inglis (1999) had also observed that survey results were inconclusive, and that there was no clear-cut relationship between self-esteem and smoking; at least, that was, they argued, until the peer context was taken into account, along with the meanings and ‘‘practical logic’’ that underlie young people’s smoking predispositions and practices. In particular, two separate surveys of young Scots by Glendinning and colleagues Fone national and the other ruralFhad found that self-esteem was, at first sight, apparently unrelated to smoking. West and Sweeting (1997) also looked at low selfesteem and lifestyles at this age, including smoking, this time in two urban surveys of young Scots in one city, and reported inconclusive results. Once again, they also argued that youth culture and the peer context must be taken into account in order to make sense of these findings. The basis for an explanation is perhaps provided by the in-depth work of Michell and colleagues (Michell, 1997a; Michell and Amos, 1997b) with two small groups of young urban Scots in one location, over an extended time period, using a range of methods, which emphasized the importance of peer group differentiation and peer status for self-image and smoking in early-to mid-adolescence. Most importantly, the findings from this qualitative study agreed with and allowed for interpretation and extension of the large-scale survey work of Glendinning and colleagues. Put simply, smoking or not smoking was seen to act as one marker of status and differentiation amongst peers, and of youth lifestyles more generally, but not necessarily in a straightforward way. For example, at this earlier age, from the perspective of peers, attempts at smoking by socially isolated youth may be viewed as another ‘‘sad’’ act serving to confirm low status among peers, and serving to delay the uptake of smoking among those with low self-esteem, and to nullify or transform its significance as a marker of status. Returning to the aim of the present article and to the British Household Panel Study this provides an opportunity to extend research in the particular case of cigarette smoking, working with new longitudinal data. Bergman and Scott (2001, p. 191, 195) have already begun this process by using data from the BHPS to consider ‘‘gender and socio-economic differences in well-being and health-risk behaviours,’’ and as one aspect of this, they examine the relationship between psychological well-being, including self-esteem and negative selfefficacy, and smoking among 11–15-year olds. In line with much previous research, their conclusion is that ‘‘among young adolescents, smoking is not consistently or strongly related to self-esteem, self-efficacy, happiness, or past worries,’’ although there are some significant correlations in the data. However, Bergman and Scott do not utilize the full potential of the BHPS, in that: they look at gender and socio-economic differences, but aggregate data for all 11–15-year olds, across a period of considerable change in young people’s lives, specifically in the uptake of smoking; they use only one of the self-report measures available to them; and 1 In addition, another article by McGee et al. (1998) using data from the same study found little evidence to support the idea that earlier mental health disorders in childhood put young people ‘‘at risk’’ of smoking.
Self-esteem and smoking in youth
417
also, they conduct cross-sectional analysis for each year, separately, rather than exploit the longitudinal nature of the data, including the ability to keep track of young people in subsequent years, over time, and into later youth. So, moving on from Bergman and Scott’s analysis, what does a more detailed examination of longitudinal data from the BHPS reveal, first, about the year-on-year relationship between self-esteem and smoking in early youth, and second, about the longer-term relationship between self-esteem in early youth and smoking in later youth? Data from the BHPS makes it possible to address the first question by looking at one cohort of young people over 4 years, starting from age 11–12, to 12–13, to 13–14, and then through to age 14–15. In this, comparisons are made not only between self-esteem at an earlier age and smoking at a later age but also between smoking at an earlier age and self-esteem at a later age, with no prior assumptions about ‘‘what leads to what.’’ Additionally, it is possible to address the second question about the longer term by looking at a further group of BHPS respondents, who were initially 14–15 years of age, and who were followed up some 6 years later at age 19–20. Here, the relationship between earlier self-esteem and later smoking behaviour is re-examined, whilst taking into account peer context and peer status.
Methods Sample Most of the data are taken from the youth component of the British Household Panel Study (BHPS). The main BHPS is a national survey, launched in 1991, which takes place every year (Institute for Social & Economic Research, 2001). It involves interviews with members of some 5500 households throughout Britain, and it tracks households and household members over time. The main survey collects information from every adult in the household (age 16+) and includes data on income, employment, health, and housing. The main survey was extended in 1994, and in subsequent years, to include the British Household Panel Study Youth Survey (BHPS-YS) of 11, 12, 13, 14 and 15 year olds who were also resident in the main panel study households. Over a 5-year period, from 1994 onwards, successive waves of the BHPS-YS were sponsored by the Health Education Authority (HEA). Each young person was interviewed separately from other members of the household by providing the survey questions on tape, allowing respondents to record their answers using a personal stereo (Scott et al., 1995). A supplement to the main BHPS also asked parents about their children. Typically, around 750 young people from some 600 households were interviewed in each wave of the BHPS-YS. (At each wave respondents moved into the main survey on reaching the of age 16, to be replaced by a new cohort of 11-year olds from the panel study households.) Response rates in 1994 to the initial survey of 11–15-year olds were reassuringly high (Brynin and Scott, 1996). At follow-up in 1997 almost 90 per cent of the original sample of respondents was re-contacted, either as a respondent to the youth survey (age 11–15), or else as a respondent to the main survey (age 16+) in that year. By 1999 around 85 per cent of the original sample of respondents from 1994 were re-contacted as part of the main survey.
418
A. Glendinning
MEASURES
&
SAMPLE PROFILES
Self-esteem: Five items were used in the BHPS-YS as a measure of general self-esteem (deriving from a 10-item scale originally developed by Rosenberg, 1965): Please say whether you strongly agree, agree, disagree or strongly disagree, that the following statements apply to yourself: 1. 2. 3. 4. 5.
I feel that I have a number of good qualities I certainly feel useless at time I am a likeable person All in all, I am inclined to feel I am a failure At times, I feel I am no good at all
(Rosenberg: (Rosenberg: (Rosenberg: (Rosenberg: (Rosenberg:
Item 3) Item 6) Correlate) Item 9) Item 2)
A scale was constructed from these five items (affi 0?7).2 Each item was coded as 0, 1, 2, or 3 (the second and fourth items were re-coded so higher scores corresponded to positive selfevaluations). The mean was calculated and multiplied by 10. Scores therefore lay in the range 0–30. As there were no absolute cut-points indicative of ‘‘low’’ or ‘‘high’’ self-esteem the sample was divided into thirds, on the basis of 1994 responses, into those reporting relatively ‘‘low’’ [0-17], ‘‘moderate’’ [18, 19, and 20], or ‘‘high’’ self-esteem [21-30]. The same cut-points were used for self-esteem scores in subsequent years. (Although profiles of emotional well-being were available in the main BHPS, a measure of self-esteem was not included in the main BHPS, that is, beyond age 15.) Prevalence of low self-esteem: To provide a reference point for the measure of self-esteem, and in terms of negative self-evaluations made by 11–15-year olds in 1994: around one-inten of the sample disagreed with the statement, ‘‘I am a likeable person;’’ and around one-inten agreed that, ‘‘All in all, I am inclined to feel I am a failure;’’ whilst around one-in-three agreed that, ‘‘At times, I feel I am no good at all.’’ Thus, on the whole, young people’s selfimage and self-evaluations appeared to be relatively positive.
Socially isolated youth. In order to identify distinctive social groupings of young people a range of items from the baseline BHPS-YS in 1994 were subjected to cluster analysis, including friendships, activities and time spent with friends and peers, and getting into ‘‘fights’’ and ‘‘trouble’’ (an approach used in previous work, Glendinning and Inglis, 1999). From this analysis four separate groupings were identified: (1) ‘‘peer-oriented’’ youth, with extended friendship networks; (2) more ‘‘conventional’’ family- and school-oriented youth; (3) more ‘‘socially isolated’’ youth; and, (4) peer-oriented, but ‘‘disaffected’’ youth. (A more detailed account is provided in Glendinning, 2000.) On the basis of this characterization, and in relative terms, according to young people’s own accounts about one-in-five of the baseline sample in 1994 could be said to be more isolated, whilst amongst 14–15-year olds around 15 per cent (N=50) could be described in this way.
Smoking behaviour. The BHPS-YS included a fairly standard set of items used by other U.K. researchers, relating to smoking, including experimentation, age of onset, current status, and amount smoked in the last 7 days, alongside orientations and beliefs. Responses to these items in 1994, 1995, 1996 and 1997 were re-categorized to produce three variables: (1) (2)
whether or not the young person had ever smoked, even a single puff; the number of cigarettes the young person had smoked in the last 7 days, if at all;
2 Even although further analysis of the first year of youth survey data in 1994 suggested that it was possible to identify two components, rather than just one (also, see Bergman and Scott, 2001) but not in subsequent years.
Self-esteem and smoking in youth
(3)
419
and, the young person’s current smoking statusFas having never smoked, as having tried once or twice, as having given-up, as smoking occasionally, or as smoking regularly, i.e., at least one cigarette per week.
In the BHPS-YS in 1997, for example, only a minority of 15-year olds had never tried a cigarette, but more than half said that they had never smoked a whole cigarette, and threequarters had not smoked a cigarette in the last 7 days. Nonetheless, one-in-three 15-year olds described themselves as current occasional or regular smokers, and one-in-five 15-year olds currently smoked an average of one or more cigarettes a day. Unlike the BHPS-YS, profiles of adult smoking for most years of the main BHPS have been partial, offering only a limited pictureFput simply, the questions were not asked. However, by the time that the first wave of 11–15 year olds from the initial BHPS-YS in 1994 had entered the main BHPS in 1999, some 6 years later, a much fuller set of survey items relating to smoking had been included within the main BHPS, including ever smoked, current status, amount currently smoked per day, and age when first smoked regularly. This made it possible to look at longer-term links between earlier self-esteem and smoking in later youth within the main BHPS.
Comparisons with other Health Education Authority (HEA) sponsored studies. It is worth noting that the size of the sample available for analysis in any 1 year of the BHPS-YS was relatively small, for example, at around only 150 15-year olds, and so estimates of smoking prevalence therefore involve a significant degree of uncertainty. However, the BHPS-YS sample of households can be viewed as representative, and can be weighted, so it was certainly worthwhile making comparisons with other large-scale surveys. For example, a series of major school-based surveys sponsored by the Health Education Authority in 1989, 1995 and 1997 make this possible (HEA, 1992a, b, 1997, 1999). The fit between estimates of smoking from these larger scale cross-sectional surveys and the BHPSYS data was generally good (whether unweighted or weighted data were used).3 (Table 1)
Statistical Analyses The data were analysed using SPSS for Windows. After a preliminary examination of selfesteem and smoking for all 11–15-year olds at baseline in 1994, analyses were restricted to 11–12-year olds from around 250 households (N=316) where these same young people were followed up at age 12–13 years in 1995 (N=303), at age 13–14 years in 1996 (N=294), and at age 14–15 years in 1997 (N=279). In addition, analyses were restricted to basic crosstabulations along with chi-square tests for the examination of self-esteem and ‘‘ever smoked’’ and self-esteem and ‘‘current smoking’’ and to non-parametric median tests for the examination of self-esteem and ‘‘amount smoked’’. The aim was to produce a simple picture by simple means, and to examine the extent to which self-esteem anticipated smoking, or conversely, smoking anticipated self-esteem. Next, the relationship between smoking and peer groupings at age 14–15 years in 1994 was examined (N=308) and also that between self-esteem and peer groupings at the same age in the same year (N=304). Finally, to develop the picture, possible links between self-esteem and peer groupings at age 14–15 years in 1994 and subsequent smoking at age 19–20 in 1999 was also examined using basic crosstabulations (N=242) and data relating to the age at which the respondent first smoked 3 Smoking prevalence did appear to be a little lower in the first year of the BHPS-YS, in 1994, and also, those stating that they had never smoked was consistently higher in the BHPS-YS, in 1994, 95, 96 and 97).
420
Table 1
A. Glendinning
The smoking behaviour of 15-year olds. HEA-sponsored surveys, 1989 to 1997
Self-reported current smoking
Never smoked (%) Once or twice (%) Used to smoke (%) Smoke occasionally (%) Smoke regularly (%) N (=100%)
BHPS-YS
WHO-HBSC
HEA
1994
1995
1996
1997
1995
1997
1989
48 28
39 30
40 26
43 22
34 32
32 32
39 31
24
31
34
33
34
35
30
152
146
138
146
1891
2135
1639
Sources: BHPS-YS (1994, 95, 96 and 97), HEA (1992), HEA (1997) and HEA (1999).
regularly was also examined using analysis of variance and an associated F-test (N=96). This second stage to the analyses was done in order to test whether or not low self-esteem, social isolation, and low status among peers were predictive of smoking in the longer term in later youth. Fortunately, as mentioned above, 1999 was the first year for which more detailed selfreport smoking data were available within the main BHPS.
Findings Smoking in early youth The findings paint a clear picture of the relationship between self-esteem and smoking in early youth. Overall, young people who reported lower self-esteem were more likely to smoke (see, Table 2 for an aggregate picture using concurrent data at one time point). Looking at the much more detailed picture over time that is represented in Figure 1 (from age 11–12, to age 12–13, to age 13–14, and through to age 14–15) there was a compelling and emergent pattern of significant associations between earlier self-esteem and smoking behaviour in subsequent years, particularly by age 14–15, as is illustrated in Table 3, where this reports a typical set of significant longitudinal results. Thus, earlier self-esteem was related to present and future smoking behaviour, whilst current self-esteem was associated with experimentation. The results in Figure 1 clearly show that the converse was not trueFsmoking ‘‘now’’ was not related to self-esteem in future years. Table 2
Self-esteem and current smoking, 11–15-year olds. BHPS-YS, 1994 (N=765)
Smoking status
Never smoked (%) Experimented (%) Ex-smoker (%) Occasional smoker (%) Regular smoker (%) Note: w2(8)=27?24, po0?001.
General self-esteem
Total
Low
Middle
High
65 7 10 9 9 100
75 10 8 4 3 100
77 11 5 5 2 100
72 9 9 6 4 100
Self-esteem and smoking in youth
421
Figure 1 The relationship between self-esteem and smoking over time, starting at age 11–12. BHPSYS, 1994, 95, 96 and 97 (N=264). General self-esteem score Smoking behaviour at age 11–12 (low, middle, high) Never Current Number smoked status last week
Smoking behaviour at age 12–13 Never smoked
Current status
Number last week
At 11–12 years old
NS
NS
NS
NS
NS
NS
At 12–13 years old
NS
NS
NS
*
NS
NS
At 13–14 years old
NS
NS
NS
NS
NS
NS
At 14–15 years old
NS
NS
NS
NS
NS
NS
General self-esteem score Smoking behaviour at age 13–14 (low, middle, high) Never Current Number smoked status last week
Smoking behaviour at age 14–15 Never smoked
Current status
Number last week
At 11–12 years old
NS
NS
NS
NS
NS
NS
At 12–13 years old
**
NS
NS
**
*
NS
At 13–14 years old
*
NS
NS
**
*
*
At 14–15 years old
NS
NS
NS
*
NS
NS
In terms of the general self-esteem scale: lower=[0-17]; middle=[18,19,20]; and, higher=[21-30] p-values based on cross-tabulation w2 or non-parametric median tests, as appropriate Key: NS=not significant; *=po0?05 and **=po0?01.
Table 3
Self-esteem at age 13–14 and smoking at age 14–15. BHPS-YS, 1996 & 97 (N=264)
Smoking status at age 14–15
Never smoked (%) Experimented (%) Ex-smoker (%) Occasional smoker (%) Regular smoker (%)
General self-esteem at age 13–14
Total
Low
Middle
High
33 12 12 20 23 100
46 20 12 6 16 100
52 19 7 11 11 100
46 17 9 12 16 100
Note: w2(8)=16?56, po0?05.
Hence, by following one cohort of young people from age 11–12, a very clear pattern of associations emerged from the BHPS-YS data over time. Associations were not really evident until around age 13–14. At this age, self-esteem in the previous year was related to current smoking status and uptake of ‘‘regular’’ smoking, and by age 14–15 self-esteem in the previous 2 years was related to current smoking status and to amount smoked.
422
A. Glendinning
Table 4 Prevalence of low self-esteem and smoking at age 14–15 by peer groupings at age 14–15. BHPS-YS, 1994 (N=304 and 308) Peer social groupings at age 14–15
Total
Conventional Peer oriented Socially isolated Disaffected a
Low self-esteem at 14–15 (%) Current smoker at 14–15b (%)
33 9
19 14
47 17
36 39
34 20
a 2
w (3)=7?91, po0?05. w (3)=30?24, po0?001.
b 2
At ages 12–13, 13–14 and 14–15 concurrent feeling of self-esteem were associated with experimentation at the same age. It would appear that for 12, 13 and 14 year olds feelings of low self-esteem were associated with increased likelihood of smoking in subsequent years, and raised self-esteem with reduced prevalence. Next, peer social groupings were introduced into the picture to suggest that, as well as selfesteem directly, these were also important for smoking in early youth, where from this perspective self-image and self-esteem were seen as bound up with peer status, along with smoking. For example, at age 14–15 under 20 per cent of ‘‘popular, peer-oriented’’ youth reported relatively low self-esteem, whilst approaching 50 per cent of more ‘‘socially isolated’’ youth reported low self-esteem. Yet the picture was not straightforward, and as can be seen from Table 4, the likelihood of smoking at age 14–15 was little different between ‘‘isolated’’ youth and ‘‘popular’’ youth4 although it was very clear that ‘‘conventional’’ youth reported lower levels of smoking whilst ‘‘disaffected’’ youth reported higher levels of smoking.
SMOKING
IN LATER YOUTH
It was also possible to extend the analysis by tracking a group of 14–15-year olds from the BHPS-YS through to the main BHPS at age 19–20. Unlike early youth, the results were inconclusive for direct links between earlier self-esteem and later smoking. Thus, results were not significant for respondents’ current smoking behaviour at age 19–20, although in respect of smoking history there was evidence to suggest that those who had earlier reported lower self-esteem were more likely to have been regular smokers at some point, whilst those who had reported higher self-esteem were less likely to have ever smoked, beyond trying a cigarette once or twice. Turning to a consideration of peer social groupings at age 14–15 and smoking some six years later at age 19–20, as can been seen from Table 5, differences in smoking prevalence among peer groupings were very evident. Twenty-four per cent of those who had reported extended friendship networks at an earlier age currently smoked at age 19–20, and unlike the picture in early youth, forty-four percent of those who had reported feeling socially isolated at an earlier age currently smoked by age 19–20. Interestingly, as shown in Table 5, this was the case even though ‘‘popular’’ smokers appeared to have started regular smoking earlier (twothirds at age under 16) whilst ‘‘isolated’’ smokers had started late (two-thirds at age 16 or older). To sum up this last set of findings, social isolation had been associated with low self-esteem in mid-adolescence, but not with increased smoking, and yet an earlier sense of isolation was 4
However, amongst ‘popular’ youth, girls were more likely to be smokers when compared to boys.
Self-esteem and smoking in youth
423
Table 5 Prevalence of smoking at age 19–20 by peer groupings at age 14–15. BHPS-YS, 1994 & BHPS, 1999. (N=242 and 96) Peer social groupings at age 14–15 Conventional 18 Current smoker at 19–20a (%) Mean age first smoked regularly 15?2 (95% CI)b (14?5, 15?9)
Peer oriented
Socially isolated
Total
Disaffected
24 44 52 33 13?9 16?1 14?9 15?1 (12?6, 15?1) (15?3, 16?8) (14?3, 15?5)
a 2
w (3)=23?52, po0?001. F(3,92)=3?67, po0?05.
b
subsequently linked to the uptake of smoking beyond this ageFreinforcing the idea that low self-esteem in youth when understood in context does matter in the longer term for smoking. Notably, over time smoking was consistently high among ‘‘disaffected’’ youth and consistently low among ‘‘conventional’’ youth, where neither group had reported significantly higher or significantly lower self-esteemFfor one group smoking was the ‘‘norm’’ and for the other group non-smoking was the ‘‘norm’’, irrespective of overall feelings of self-esteem.
Conclusion The present findings for self-esteem and smoking do not concur with what has emerged as a largely consensus view within the recent research literature, if not necessarily within health promotion and health education circles, which says that self-esteem and smoking in early youth are largely unrelated, so do they muddy the waters? Indeed, the analysis of British Household Panel Study data provide an unambiguous longitudinal picture of self-esteem year-on-year at around the age 12–14 yearsFwhen young people take up smoking in increasing numbersFas linked to experimentation with smoking and also to subsequent smoking behaviour, at least in the shorter term. However, further analysis of British Household Panel Study data, suggest that a longer-term linkage between low self-esteem in early youth and smoking in later youth is less clear cut, and less compelling; but also, that putative links between self-esteem and smoking must be understood in context, specifically within the peer context, and by reference to peer status and differentiation, and the meanings that smoking or not smoking have for different peer social groupings. As has been proposed by Glendinning and Inglis (1999), following the lead provided by Michell (1997), and borrowing a theoretical framework from Bourdieu (1992) as applied to the ‘‘youth field’’, what matters is youth culture, and the ‘‘practical logic’’ as to why young people do smoke or do not smoke (cf. ‘‘habitus’’). Self-image and general feelings of selfesteem are bound up with this, rather than self-esteem directly causing young people to smoke or not to smoke. From that perspective the findings from the present study do not muddy the waters, but instead simply serve further to refine the picture offered by recent studies. The findings also suggest that there is really no need to look at what has been termed ‘‘domain-specific self-esteem’’ unless the intention is to point to underlying processes within
424
A. Glendinning
the peer and other contexts.5 When considered as a whole, what is striking is that research on self-esteem and smoking in youth now offers a cumulative picture, where results from different studies using different methods can be said to build upon each other, including large-scale longitudinal quantitative work and small-scale in-depth qualitative work. Finally, the study has not looked at gender differences, and once again following the lead of Lynn Michell, this may well be a further important factor within the picture that has been presented here.
Postscript At the time of completion of the present article in late 2001 an excellent review, SelfesteemFthe cost and causes of low self-worth, by Emler (2001) has just been published. As in the previous article by Glendinning and Inglis (1999) the general point is made by Emler that: There is a widespread view that self-esteem is a risk factor for a broad range of psychological and behavioural problems [and yet] neither public discussion nor decisions to invest in prevention and treatment have been strongly informed or guided by hard evidence, either about the effects or causes of low self-esteem.
In the case of smoking in youth, however, Emler (2001, 22, 87) largely leaves the story with McGee and WilliamsFthe relationship between low self-esteem and smoking is at most weakFwhich is the point of departure for the present article. He also only considers the peer context from the perspective of ‘‘peer group pressure,’’ with self-esteem acting there as a buffer or moderating factor, rather than from the perspective of social interaction, status, and position; although he does point out elsewhere that social standing within the wider peer group may be important, even if these are only young people’s own self-perceptions or selfevaluations.
Acknowledgements The views expressed are solely those of the author. However, I would like to acknowledge the following people. Pat West for the encouragement that he has given me over the years, and for pointing out to me that there was an issue here in the first place. Lynn Michell for the insights that her work gave me. David Inglis for his sociological inspiration. Hilary Graham for suggesting that I analyse the BHPS-YS, and for making the introductions. Julie Bull at HEA (and Helen Ryan) for making the analysis possible through financial support. Nick Buck and colleagues for the welcome at ISER, and the assistance of the ESRC Research Centre for Micro-Social Change, and the ESRC Data Archive, in supplying BHPS data and documentation. Jackie Scott and Malcolm Brynin for their work on the initial design and reporting of the BHPS-YS. Lynda Clarke, Julie Williams, and Jonathan Bradshaw for access 5 Glendinning et al. (1995) had earlier looked at distinct social groupings in youth to include elements of the family and school contexts, as well as the peer context, and they then argued that self-image and smoking, along with aspects of youth culture, such as, consumption patterns, were bound up with these, where social groupings were seen as representative of distinctive youth lifestyles.
Self-esteem and smoking in youth
425
to work in progress. And to all of the young people who participated in the BHPSYSFThank you.
References Bergman, M. and Scott, J. (2001). Young adolescents’ wellbeing and health-risk behaviours: gender and socio-economic difference. Journal of Adolescence, 24, 183–197, doi:10.1006/jado.2001.0378. British Household Panel Study (BHPS) (2001). Economic & Social Science Research Council, Research Centre on Micro-Social Change. British Household Panel Survey. Colchester, Essex: The Data Archive, 28 February 2001. SN: 4340. Bourdieu, P. (1992). The logic of practice. Cambridge: Polity Press. Brynin, M. and Scott, J. (1996). Young people and the familyFfamily health research report. London: Health Education Authority. Emler, N. (2001). Self-esteem: the costs and causes of low self-worth. York: Joseph Rowntree Foundation, York Publishing Services. Glendinning, A., Shucksmith, J. and Hendry, L. (1995). Lifestyle, health & social class in adolescence. Social Science & Medicine, 41, 235–248. Glendinning, A. and Inglis, D. (1999). Smoking behaviour in youth: the problem of low self-esteem? Journal of Adolescence, 22, 673–682, doi:10.1006/jado.1999.0262. Glendinning, A. (2000). Health variations in youth: a picture of health, family & peer culture from the BHPS youth survey. London: Health Education Authority. Health Education Authority (HEA) Rudat, K., Speed, M. and Ryan, H. (1992a). Tomorrow’s young adults. London: Health Education Authority. Health Education Authority (HEA) Rudat, K., Ryan, H. and Speed, M. (1992b). Today’s young adults. London: Health Education Authority. Health Education Authority (HEA) Turtle, J., Jones, A. and Hickman, M. (1997). Young people & health. London: Health Education Authority. Health Education Authority (HEA) Haselden, L., Angle, H. and Hickman, M. (1999). Young people & health. London: Health Education Authority. Institute for Social & Economic Research (ISER) (2001). http://www.iser.essex.ac.uk/bhps 19 April, 2001. Kahne, J. (1996). The politics of self-esteem. American Educational Research Journal, 33, 3–22. McGee, R., Williams, S. and Stanton, W. (1998). Is mental health in childhood a major predictor of smoking in adolescence? Addiction, 93, 1869–1874. McGee, R. and Williams, S. (2000). Does low self-esteem predict health-compromising behaviours among adolescents? Journal of Adolescence, 23, 569–582, doi:10.1006/jado.2000.0344. Michell, L. (1997a). Loud, sad or bad: young people’s perceptions of peer groups and smoking. Health Education Research, 12, 1–14. Michell, L. and Amos, A. (1997b). Girls, pecking order and smoking. Social Science & Medicine, 44, 1861–1869. Rosenberg, M. (1965). Society and the adolescent self-image. : Princeton, NJ.: Princeton University Press. Scott, J. Brynin, M. and Smith R. (1995). Interviewing children in the British household panel study. In Advances in Family Research, Hoy, J.J. et al. (Eds). Amsterdam: Thesis Publishers. West, P. and Sweeting, H. (1997). ‘Lost souls’ and ‘rebels’: a challenge to the assumption that low selfesteem and unhealthy lifestyles are related. Health Education, 97(5), 161–167.