The order of scores is the same as that on the social environment measures.
EFFECTS OF LIVING IN AN EGALITARIAN ENVIRONMENT
by David Erdal. 2001.
(summarising part of my 1999 PhD thesis).
INTRODUCTION
Background.
Following work on the likely characteristics of the social environment in which Homo sapiens evolved, Erdal and Whiten identified from a survey of hunter-gatherer ethnographies two universal characteristics as “egalitarian”: meat sharing and counter-dominance (Erdal and Whiten 1994, 1996). In hunter-gatherer groups, these maintained a relatively equal distribution of physical and political resources, of wealth and of influence. To the extent that individual psychological algorithms for behaviours affected by the social environment have been designed by natural selection, it is to this egalitarian social environment that they will be adapted. If counter-dominance was instrumental in maintaining the sharing of meat and broadening access to mating opportunities, with crucial consequences respectively for nutrition and reproduction, then human psychology should measurably reflect that evolutionary history today.
Hypothesis.
A specific hypothesis following from that model is tested in this study. This hypothesis is relatively wide in its scope: it is that since human psychology evolved in and is adapted to an egalitarian social environment, living in such an environment today will have widely beneficial effects, reflected in measurable social psychological variables.
A corollary of this hypothesis is that living in an inegalitarian environment will be to some extent harmful: any organism living in an environment which on relevant variables is different from that to which it is adapted may experience stress at the lack of fit. In extreme form an example was provided by the work of Harlow on young rhesus macaques raised in isolation from their natural mothers (Harlow 1958): even after successful rehabilitation using normally reared macaques as “therapists” the ex-isolates remained more susceptible to stress (Gleitman 1991, p. 585).
DESIGN
The basic design of this study was to test the health of the populations of three towns, as similar as possible in every respect except in their level of egalitarianism.
The independent variable: an egalitarian environment today?
The characteristics of a relatively egalitarian environment in the modern world were assumed to include:
a) The Social Environment: a lack of class differentiation, and a relatively equal distribution of wealth;
b) The Work Environment: co-operatives and businesses owned by their employees, with those employed being given information, involvement in decisions and the right to influence them, and participation in profits made;
c) The Political Environment: a democratic tradition where voters had the experience of influencing policies and events.
It is difficult in practice to determine the extent to which such variables are independent. For example, a high rate of participation in the political process is likely to be more effective, and also to be reinforced by the experience of its efficacy. A high political participation rate will thus be an outcome of such an environment at the same time as it perpetuates it. A feedback process will operate which will sustain the system.
The most independent of these complex interacting variables may be the business ownership system. To the extent that businesses are owned by their employees living in the local community, there is a material factor underlying the egalitarian environment.
Dependent variables
The dependent variables were taken to include a very wide range of social behaviours under the general headings of health, education, crime, social particpation and perception of the social environment.
a) Criminal Behaviour: crime rates should be lower. Crime has been shown to be more severe in areas of greater wealth differentiation, and less severe in areas where there is a strong community tradition (ref??). Clinard (1978) described the main factors behind Switzerland’s low crime rate, emphasising the importance of the cohesive community, where differences in wealth were downplayed and the police were closely linked with their communities.
The literature on one aspect of criminal behaviour, domestic violence, generally focuses on how to intervene constructively (e.g. Martin 1978, Fiorenza and Copeland 1994). Most theoretical discussion concentrates on cultural rather than material explanations: for example, Dobash and Dobash (1979) see inequality in marital relationships but make nothing of wider social inequalities. Nonetheless domestic violence has been linked to unemployment (Komarovsky 1940 discussed in Pahl 1985). According to the present hypothesis domestic violence will be improved by living in a more equal environment.
b) Educational Performance. The large-scale youth cohort studies in the UK have shown that educational performance varies with rates of truancy, with an early cessation of school attendance and with negative attitudes towards school (Bosworth 1992). Given the hypothesis that people living in an egalitarian community will feel less in conflict with the institutions around them, it follows that children in a more egalitarian community should play truant less, stay at school longer and feel more positive towards their schools. Taylor and Spencer showed that negative experiences as a child at school provided a significant disincentive from pursuing education and training later in life (Taylor and Spencer 1994, p. 20). It is likely that an egalitarian community would tend to promote more constructive school experiences, resulting in better performance in adult education and training, while a lack of psychological fit with the social environment would make the problem more intractable.
c) Social Participation: by this hypothesis, social participation should be easier and more natural in an egalitarian environment. Concretely, voluntary societies and charitable activity should be at a higher level, as should rates of participation in the political system. Given the seminal study by Titmus (1970) of blood donation, it was decided to collect data on that activity also. On political involvement, Franklin established that country differences were much larger than individual differences in voting rates, with Italy very highly placed at an average 90% participation in national elections (1996, p.218, Table 8.1). Putnam (1993) identified the whole of Emilia Romagna, the province in which this study was carried out, as having an environment conducive to good local government, due to its long-standing co-operative traditions. The current study was designed to make fine distinctions within this excellent environment. Franklin also showed that low voting rates in some cases were “associated with the widespread use of alternative avenues for participatory activities” (1996, p.216), which might suggest that an egalitarian community with its wider range of opportunities to participate would be associated with lower voting rates. However, it was decided to keep the intuitive prediction that voting rates in an egalitarian society should be high.
d) Social networks should be more developed. By this is meant not the mathematical analysis of networks of relationships (Barnes 1972), but rather the support felt to be available from workmates, friends and family.
e) Levels of trust should be higher than in communities with a lower sense of solidarity, as would be caused by salient inequalities.
f) Health should be better, particularly where stress could affect the outcome. Black et al. (1980) established that material deprivation was associated with poor health outcomes and Whitehead (1988) gave further evidence that this was the case. The bibliographies in those two books provide a full sample of the extensive literature on equality and health.
Two studies seem particularly relevant to this thesis. The first is the large (over 10,000 subjects) longitudinal study, started in 1967 by Marmot and his colleagues, of the health of the people working in the hierarchical Whitehall bureaucracy of the British government. They showed (e.g. Marmot 1991) that mortality and many serious diseases, including heart conditions, were inversely related to status in the hierarchy, after controlling for all other known factors. Given that the egalitarian environment should decrease the sense of hierarchy, it is therefore possible to predict that Imola’s population should have lower morbidity and mortality.
The second relevant work is that of Wilkinson (1996) which surveyed research from many sources showing that once a population has reached a reasonable state of economic development, health improvements appear to follow not from further economic development, but from a more even distribution of wealth. This work tends to be suggestive rather than conclusive, and not to have the rigour of Marmot’s work referred to above, but it produces a range of evidence to support an egalitarian effect.
g) Sex Differences. In addition, an evolutionary perspective suggests a new hypothesis: that there may be a sex difference in the response to such an environment. The first leg of the argument rests upon general evolutionary theory. Intra-species competition is stronger among the members of the sex which invests less in reproduction, which in humans, as in most mammals, is among males (Trivers 1972). The variance in male reproductive success is very much higher than the variance in female reproductive success, so the sexual selection stakes are higher and the readiness to compete greater in males. Among humans, this competition should include efforts to appear competent, and measures of competence would include wealth and social status (Symons 1979, Buss 1994). Miller argues that much current research focuses too simplistically on superficial cues, but does not dissent from the theoretical model (Miller 1997).
The second leg rests on the distinction between modern environments and the environments in which evolution took place. A particular feature of post-hunter-gatherer societies, which modern technology has reinforced strongly, is the ability of some individuals to build up unassailably conspicuous surpluses of resources and status (Frank and Cook 1995). Since this possibility seems never to have arisen before agriculture - that is, during the millennia that passed while Homo sapiens evolved - then to the extent that human psychology is tuned by evolution, it is not tuned to this situation where some individuals can obtain a vastly greater share of wealth and status. So it follows that males in particular, sensitised to evaluations of social competence by sexual selection over vast tracts of evolutionary time, will suffer from a degree of “insult”, living in this social environment and being unable to do anything effective about it.
The third leg of the theory concerns the role of egalitarianism. An egalitarian environment is one in which the display of differences in status and resources is more muted: so egalitarianism represents a moderation of intra-species competition. In view of the greater competition among males than females for these resources, this should be particularly relevant to males rather than females.
A specific prediction following from this perspective would be that moving from a less to a more egalitarian environment should result in a greater improvement for males rather than females, particularly in those measures which are most closely related to sexual selection. This would include sensitivity to the difference between rich and poor, where males in an unequal environment will suffer more severely, since they are measured more than females on demonstrated competence in providing goods for paternal investment (Buss 1994). Another prediction is that there should be a difference in the stress response to an inegalitarian environment, with males suffering more severely: they are constantly reminded, by the obvious disparity in wealth, that they are failing in important ways. This would manifest itself in sex differences in the improvement in health and mortality; it would also predict a smaller gap in age-at-death between the sexes in the more egalitarian environment.
Given the many possible confounds of looking at a wide selection of variables in populations in the real world, it is not possible to be more than tentative in these predictions. For example, it can be argued that if females show less improvement than males in an egalitarian environment, it may be that for females there is less of a difference in the environment. For females a significant level of male dominance in the home – likely to be common across all the towns – could mute the egalitarian effect for the females as opposed to the males, for whom the working environment may be more salient.
Populations
The main goal was to find a population which was living in a relatively egalitarian social environment, and had done so for several decades, and to compare the population of that town with controls. Imola - a town of some 60,000 people, approximately 30 km south east from Bologna in Northern Italy, was chosen because it was believed to have the highest proportion of co-operatives in Italy (Earle 1986, Oakeshott 1990a), and probably in Europe after Mondragon in Spain (Oakeshott 1990b). Mondragon was considered as an alternative site for this study, but it was felt that Mondragon's relative isolation in the mountains of the Basque region, and the fact that it is associated with an unusual set of social-support institutions, would make controls hard to identify. As a result, Imola would provide a clearer and more general test.
The second task was to find a control town or towns for comparison. The target here was to find towns that were approximately the same size, a similar distance from a city, clear of the coast (to avoid the influence of the tourist industry), equally industrialised and as close as possible geographically, politically and culturally.
Two towns were chosen, bracketing Imola in population size and geography: Sassuolo, which is smaller (approx. 40,000) and slightly further away (about the same distance on the other side of Bologna, although closer to its nearest city, Modena, at about 15 km) and Faenza, which is larger (approx. 80,000), very close to Imola, (about 12 km further away from Bologna), and some 25 km from Ravenna, its nearest city, which is on the coast. Both control towns are in the same province as Imola and both appear equally prosperous and industrialised to a broadly similar degree. It was not possible to gather data on the relative prosperity of the towns, as opposed to larger geographical units.
In addition to these three towns, the search for mortality statistics revealed a detailed study conducted by the health authorities in Ferrara. Ferrara is a city a little more than double the size of Imola - with a population of approx. 136,000 - situated some 55 km to the North. Apart from its size, it fits four of the criteria for comparison: it is in the same political province and the same geographical area, it is similarly industrialised and it is not near the coast.
Since Imola (but not the other two towns) also provided data on deaths by age and sex, the specific prediction could be tested that males would benefit more than females from living in an egalitarian environment. Ferrara’s mortality data were only discovered towards the end of the study, too late to include Ferrara in the questionnaire survey.
The hypothesis covered a wide variety of variables in five main areas: health, education, crime, social participation and social environment. In addition, data on demographic variables were gathered to ensure that the populations were not significantly different on these background measures.
Data from Published Statistics
Published statistics were gathered as follows:
Health: each town publishes statistics on mortality. In addition, local health boards were contacted to provide data.
Education: public statistics were checked, and local school boards contacted.
Crime: No data are published at the level of the town. The police and the carabinieri (which are separate organisations but which share responsibility for combating crime) were asked for data in each town.
Social Participation. Statistics on voting rates are publicly available, and the organisation which collects blood, AVIS, was contacted in each town.
Social environment. No data were available from published sources.
Demographics. Statistics are published by each town, covering the population by age and sex, household size, employment, and other details. One national survey included Imola and Faenza on various measures, set out in the Results section below.
Data from a Postal Survey
The second source of data was a postal survey, as follows:
Questionnaire design
Questions were taken from various sources, as follows:
Health: the validated Italian version of the SF36 (see Shiely et al., 1996; Tsai et al., 1997) was obtained from the Medical Outcomes Trust in Boston. There was room to use only the 12 questions of the abbreviated version, the SF12, which were not altered in any way. This questionnaire produces 2 scales, one for physical health and one for emotional health.
The sections on Education, Social Participation and Social Environment were written from scratch. The Crime items were based on questions in the British Crime Survey (Chambers and Tombs, 1984). The Demographics section was designed to allow measures to be compared with the publicly available data.
Sample selection
A representative random sample of addresses was generated from the telephone books in each town, using the most recent possible editions.
Materials
500 questionnaires with stamped return envelopes were posted to each town in November 1998. There were 14 questions on crime, covering victimisation and opinions; 12 questions on health; 6 on educational experience and attitudes; four on membership of voluntary organisations; three on perceptions of the social environment and 12 on general demographic and background information. The questions were multiple choice, with the exception of questions about age.
Appendix 1 gives the essence of the questions in abbreviated form.
Test.
The aim was to construct a table of measures of all these factors so that a sign test could be carried out identifying whether there were consistent differences between the populations.
In deciding when to treat scores as having a difference worthy of inclusion in a sign test two levels of difference were examined:
1. a simple sign difference of any magnitude
2. a sign difference of at least one standard error of the mean. These are referred to as “starred” differences, and marked with an asterisk * in the tables.
RESULTS
Questionnaire Responses
The response rates were low, from 14.9% in Sassuolo to 17.1% in Faenza.
Demographic Comparisons
Demographic comparisons showed that the age distributions of the samples were similar to the age distributions of the towns, and that in all three towns people aged over 60 had responded at a much lower rate. The age distributions of the samples were comparable, except that the Sassuolo sample had a lower proportion of people over 60 and a higher proportion of responders in their thirties.
Comparisons by household size showed that there was an under-representation of one-person households. Members of one-person households were more likely to report being older; having received more training; and having lived in their towns for a shorter time than average.
In view of these disparities between sample and population, and between sample and sample, it was decide to focus on comparisons between sub-samples matched by age and sex.
Matched Samples.
Two sets of matched samples were prepared, matching by age and sex. It was possible to match 44 (23 male, 21 female) of the 49 Sassuolo respondents with same-sex respondents from each of the other two towns, with a maximum gap of four years, but with 94% of all pairs within two years of each other. In addition, 56 same-sex respondents from Imola and Faenza were matched (29 male, 27 female) with an age difference of no more than two years per pair
Relative to the populations, these age band proportions are all characterised by over-representation of the 30s and under-representation of the over 60s, so the exact quanta of the variable scores may not accurately reflect the underlying populations; however, differences between the samples may reasonably be expected to reflect differences between the underlying populations.
Correlations Between Dependent Variables
The total sample was first examined for high correlations indicating two variables measuring a single factor, rather than plausibly cause and effect. In five cases of significant correlations the variables were combined.
Independent variable: Numbers working in co-operatives
The variable treated as independent under the hypothesis being tested was the number of household members in employment who worked in co-operatives. These ranged from 25% in Imola to 0% in Sass, with Faenza at 16%. (In the total samples the figures were 26%, 0% and 13%). This is a significant difference: chi square 2 df = 19.84, p<0.0001)
Thus there is a gradient between the towns. Given these differences in the prevalence of co-operatives, by the hypothesis of the study the wider effects of egalitarianism should be shown most clearly by Imola and least clearly by Sassuolo, with Faenza in between.
Dependent Variables
Perception of the Social Environment
The similarity between rich and poor perceived by respondents
The pattern was parallel to the distrbution of co-operatives: Imola, then Faenza then Sass. This was a highly significant difference (chi square 4 df = 31.81, p<0.0001).
Those who saw a relatively small difference between rich and poor also reported having family members working in co-operatives, and positive perceptions in social and crime factors.
In Imola, the most egalitarian town, the difference between rich and poor was rated as smaller by the men than the women; in the other two towns it was rated as larger by the men than the women, and the difference was significant (z transformation test, Cohen and Cohen 1983, p. 54, equation 2.8.5: Imola vs Sassuolo: z=1.765, p<0.04; vs Faenza: z=2.295, p<0.012). As predicted, the men appeared more sensitive than the women to differences in this variable.
1.2 Social Networks
A) Scores. The question here was: "If you are in difficulty, how many people can you turn to for help?". Again, the pattern of responses was the same as the pattern of employment in co-operatives. People who saw their networks as supportive also reported seeing the authorities as helpful, seeing less difference between rich and poor, seeing domestic violence as less prevalent, and believing education to be important for happiness.
1.3. Helpfulness of Authorities
A). Scores. Authorities were judged to be helpful with the same pattern shown again, again without the differences reaching significance.
Social Environment: Summary
Table 4.13 shows the relative positions in pairs, using the N=44 sample for the Sassuolo comparison and the N=56 sample for the Imola:Faenza comparison. On all four measures Imola > Faenza > Sassuolo.
Table 4.13. Social Environment: Relative Positions
Imola:Sass Imola:Faenza Faenza:Sass
Percent in co-ops 1:0* 1:0* 1:0*
Rich vs. Poor Difference 1:0* 1:0* 1:0*
Social Networks 1:0* 1:0 1:0
Authorities’ Helpfulness 1:0* 1:0* 1:0
Total 4:0 4:0 4:0
* This difference is equal to or greater than one standard error of the mean.
Crime
The questionnaire results relating to crime were not in the same pattern as employment in co-operatives: generally Faenza was best, Imola second and Sassuolo again worst.
Table 4.27 shows the summary results of the crime section, using the N = 56 samples for the Imola comparison with Faenza, and the N = 44 samples for the other two comparisons.
Table 4.27. Summary of Relative Crime Scores
Imola:Sass Imola:Faenza Faenza:Sass
Security from crime 1:0 0:1 1:0*
Police activity 1:0* 1:0 1:0*
Serenity about crime 1:0 0:1 1:0*
Confidence crime controlled 1:0* 0:1 1:0*
Domestic peace 1:0* 1:0 1:0
Total 5:0 2:3 5:0
SE test: 3:0 0:0 4:0
* The “SE test”, i.e. the magnitude of the difference between means
was equal to or greater than one standard error.
One problem with the interpretation of the crime rates was the very high involvement of criminals from outside the towns, as shown by the disproportionate number of arrests of outsiders in Imola (neither of the police forces in either of the other towns would provide any information).
The exception to this was domestic violence, which is clearly a local crime. That again takes the same pattern as employment in co-operatives. The results for that specific variable were as follows:
2.7. Domestic Peace (i.e. absence of domestic violence). The responses to the stimulus: "In the home, a husband acts violently towards his wife...." had the patterns shown in Figure 4.13.
Figure 4.13. Domestic Peace: Mean Scores
The order of scores is the same
as that on the social environment measures.
B). Correlations. Those who felt that domestic violence was relatively low also reported seeing little difference between rich and poor (*), having good social networks (*), being confident that crime was under control and feeling that children were playing truant less.
Health
On the health measures, the results had a different pattern.
The scores on physical health and emotional health were highly correlated, but given that they have been independently verified (ref) they were left separate.
Faenza had the worst self-reported health, both physical and emotional. Sassuolo and Imola were almost identical in both, with Imola marginally ahead in the matched sample but behind in the total sample, where the difference in age spread between the towns had an effect. Independent t tests showed that Imola was significantly better than Faenza in physical (t = 2.13, df = 71, p = 0.037) but not emotional health (t = 1.56, df = 79, p = 0.12 n.s.)
B). Correlations.Those reporting good physical health were more likely also to report being members of voluntary organisations. Those reporting good emotional health were more likely to be male (**),to be living with a spouse or partner (*) and to be confident that crime was under control (*) - also, in the total sample, to be members of voluntary associations (T*), to be in their 50s, in larger households and with good social networks.
Table 4.29 sets out the relative positions on health.
Table 4.29. Relative positions on health
Imola:Sass Imola:Faenza Faenza:Sassuolo
Physical 1:0 1:0* 0:1*
Emotional 1:0 1:0* 0:1*
Total 2:0 2:0 0:2
* This difference is at least one standard error in magnitude.
This gives an order for this section of Imola > Sassuolo > Faenza
Social Participation
The scores in this section were affected by small numbers: in the N=44 sample Imola had 16 members of voluntary associations and each of the other towns had 15. There were no significant differences by any test.
Education and Training
Education: Summary
Table 4.48 summarises the relative positions on education
Table 4.48. Education Summary: Relative Positions
Imola:Sass Imola:Faenza Faenza:Sass
N=44 N=56 N=44
Education: Performance 1:0 0:0 1:0
School Attendance then 1:0* 0:1 1:0*
School Attendance now 0:1 0:1* 1:0*
Training 1:0* 1:0* 0:1
Education and happiness 0:1* 1:0 0:1*
Total 3:2 2:2 3:2
SE test 2:1 1:1 2:1
* Where the SE test applies (higher mean ≥ lower mean+SE) the score
is marked with an asterisk:.
This gives an order for this section of Imola = Faenza > Sassuolo.
Unusually, Imola had the lowest score on two measures: the current expectation that children were attending school rather than playing truant (a very small difference in scores) and the view that education was important for happiness. The truancy result is unexpected for what is essentially a measure of the level of trust that the population feels towards their children, which must count as a point against the hypothesis of this paper. The differences in the feeling that education is important for happiness are discussed below.
Questionnaire Summary Analysis
The null hypothesis is that there should be no significant pattern of differences between these towns in the various measures examined: any differences should be random, once correlated variables have been combined. Table 4.49 draws together the mean scores listed above from the two sets of matched samples, and Table 4.50 sets out the overall relative positions.
Table 4.48. Questionnaire Scores: Summary
N = 44 N = 56
Imola Faenza Sassuolo Imola Faenza
Rich vs. Poor 2.50 2.37 1.82 2.51 2.44
Social Networks 3.36 3.30 3.21 3.36 3.23
Authorities Helpful 2.56 2.29 2.26 2.52 2.37
Security from Crime 6.55 6.73 6.50 6.45 6.63
Police Activity 48 50 32 46 45 R
Serenity re. Crime 1.68 1.89 1.61 1.68 1.90
Crime Confidence 1.97 2.15 1.76 1.96 2.17
Domestic Peace 2.74 2.59 2.46 2.70 2.61
Physical Health 3.09 2.91 3.04 3.09 2.93
Mental Health 3.282 3.06 3.279 3.28 3.10
Member of Vol. Soc 36 34 34 39 29
Education 2.02 2.11 1.88 2.11 1.88 R
Attendance at School 3.27 3.30 3.02 3.27 3.29
Attendance Now 1.77 1.91 1.79 1.87 1.98
Training 56 35 41 53 33
Educ’n/Happiness 3.00 3.02 3.23 3.02 2.91 R
Total N of top scores: 8 7 1 11 5
Top scores are underlined. “R” indicates that the order of scores has reversed between Imola and Faenza in the N=44 and N=56 samples, implying approximate equivalence.
Table 4.50. Relative Positions on each variable
Imola:Sass Imola: Faenza Faenza:Sass
(N=44) (N=56) (N=44)
Rich vs. Poor 1:0* 1:0* 1:0*
Social Networks 1:0* 1:0* 1:0
Authorities Helpful 1:0* 1:0* 1:0
Security fromCrime 1:0 0:1* 1:0*
Police Activity 1:0* 0:0 R 1:0*
Serenity re. Crime 1:0* 0:1* 1:0*
Crime Confidence 1:0* 0:1* 1:0*
Domestic Peace 1:0* 1:0 1:0
Physical Health 1:0 1:0* 0:1*
Mental Health 1:0 1:0* 0:1*
Member of Vol. Soc 1:0 1:0* 0:0
Education 1:0* 0:0 R 1:0*
School Attendance 1:0* 0:1 1:0*
School Atten'ce Now 0:1 0:1* 1:0*
Training 1:0* 1:0* 0:1
Educ'n & Happiness 0:1* 0:0 R 0:1*
Totals All Scores 14:2 8:5 11:4
Totals SE test 10:1 7:4 8:3
* Where the SE test applies (higher mean ≥ lower mean+SE) the score
is marked with an asterisk:. “R” indicates that since there is a reversal
between the N=44 and N=56 samples from Imola and Faenza, their scores
are counted as equal.
Comparing overall scores by the sign test Imola is significantly higher than Sassuolo both overall (p<0.005, 1 tail) and on the SE test (p<0.025, 1 tail) but none of the other differences in totals are significant.
Since the arithmetic means of the scores are arbitrary measures, and in different units, the scores were standardised and the towns compared in pairs using Wilcoxon signed rank tests. Imola's scores on the whole N=44 table are significantly higher than Sassuolo's (N = 16, W = 11, p<0.025, 1 tail) but not Faenza's (W=58), and Faenza's are not quite significantly higher than Sassuolo's (W = 33 vs critical 30). Comparing Imola and Faenza using the N = 56 sample scores also gives no significant difference (Wilcoxon signed ranks test, N = 16, W = 42, n.s).
Finally, to avoid over-weighting any particular category of question, since the different categories have different numbers of questions (Social environment 3; crime 5; health 2; voluntary associations 1; education 5), if the scores are first evaluated by section to give a single point per section, Table 4.51 results.
Table 4.51. Relative positions by section
Imola:Sassuolo Imola:Faenza Faenza:Sassuolo
Health 1:0 1:0 0:1
Education 1:0 0:1 1:0
Crime 1:0 0:1 1:0
Vol. Associations 1:0 1:0 0:1
Social Environment 1:0 1:0 1:0
Total 5:0 3:2 3:2
By this test again the only significant difference is that between Imola and Sassuolo (Sign test p<0.05). These questionnaire results will be discussed further in the Discussion section.
Public Statistics
The second source of data consisted of the sets of statistics published by the authorities and other organisations in the towns and the cities covering their areas. The main statistics targeted were Demographics, Crime, Mortality, Voting and Blood donation.
Crime
Only the Imolese officials provided data on crime. Even with the active help of a local person, it proved impossible to persuade the police or carabinieri in the other two towns to co-operate with this research project.
From both the police and the carabinieri in Imola there was a clear view that outsiders played a significant part in the criminal activity in Imola, and in all these Northern towns.
Taking the police first, a sample of 20 consecutive arrests from the “day-book” for 1997 in the police headquarters in Imola was examined. Of those arrested, only nine were Imolese and of these nine, three came originally from the South of Italy. Analysed by origin, the 20 included 8 from Southern Italy, 6 from Imola, 4 from North Africa, and 2 from other countries. Three of the 20 were local drug addicts: drug addiction was said to be the single biggest cause of crime; that too was said to be caused by incoming drug pushers.
Secondly, the carabinieri in Imola said that from examining their records they believed that most robberies were committed by people from the South of Italy; thefts by foreigners from further afield; and only shoplifting was characteristically committed by locals, mainly elderly people.
Thirdly, it was also clear that the people of Imola believe that a very large proportion of the crime in the whole area is committed by outsiders: in 1997 107 (30%) of the 356 people reported to the police as having committed a crime were foreign residents; on 10th Oct 1996 there were 761 foreigners resident in Imola, only 1.2% of the population.
Demographic Measures
Most of these measures have been reported in the section comparing the survey samples with the populations. In addition, a national survey of medium sized towns was identified, carried out by ISTAT, the Italian national bureau of statistics, in 1991. This compared data on various measures from 67 towns, including Imola and Faenza but not Sassuolo. Imola was rated as having fewer cars per 1000 inhabitants than Faenza (576 vs. 590), and a smaller percentage of those cars over 2000 cc in engine volume (3.0% vs 3.5%: over the whole of Italy the range was 2.8% to 6.2%, so both are towards the lower end of the range). Imola also had a higher average house price (by 4.7%) and more liquid resources per person (by 0.9%) but used 6.8% less electricity per person, had a much larger average school class size (35.2 vs 21.7) and fewer pharmacies (3927 people per pharmacy vs 3859).
One other demographic comparison is worthy of note. Marriage can take place in church or in a civil ceremony. Data were obtained for the 11 year period 1985 to 1995 on the marriages in Imola and Sassuolo. Table 4.52 gives the data showing that Imola with 25% civil marriages is significantly more secular than Sassuolo with 18%. One implication is that Imola’s generally more positive scores than Sassuolo’s are not driven by a religious culture.
Table 4.52. 1985-1995 Total Marriages by Ceremony
Church Civil
Sassuolo 1285 287
Imola 2213 720
This is a significant difference (chi square 1 df = 22.98, p<0.0001).
Mortality
Data were collected on mortality from the three towns and in addition a detailed study on mortality by the city of Ferrara was obtained.
Mortality data sets differed among the three towns. For Imola, mortality rates by 5-year age band (e.g. aged 50-54, aged 55-59 etc.) were provided for six years (1990-1995).
For the same six years Sassuolo provided the age breakdown of the population, and the overall mortality rate, but not the mortality rate by age band.
From Faenza it proved possible only to obtain the overall mortality rates, and only for four of these years: 1991, and 1993-5. Again the population breakdown was provided, but not the mortality by age band.
To make comparisons, the mortality rates by age band from Imola were applied to the population by age band of the other two towns, and the resulting predicted mortality rates were compared with the overall actual mortality rates recorded. The results were as follows.
Imola’s Mortality. The raw Imola mortality rate was increasing over the six-year period measured, as is clear from Figure 4.23.
Figure 4.23. Imola Deaths, 1990-95
At the same time the population remained nearly steady, moving from 62,352 in 1990 to 63,699 in 1995, so that the rate increased from a low of 10.0 per thousand in 1991 to 11.2 per thousand in 1995. The age structure of the population did not alter materially in this period.
Mortality: Imola vs Sassuolo
By contrast, the raw mortality rate in Sassuolo had no trend, and was lower than Imola’s (Figure 4.24).
Figure 4.24. Mortality: Imola vs. Sassuolo
This is the raw mortality rate in the actual populations, before the standardisation of the age profiles.
Adjusting for the difference in age distribution, however, by applying Imola mortality rates per age band to the Sassuolo population, showed that Imola's mortality rates were lower on an age adjusted basis in all six years:
Figure 4.25. Sassuolo Deaths, actual and projected using Imola’s rates
A Wilcoxon signed ranks test shows this difference to be significant (W = 0, N = 6, p < 0.025). The differences in the last three years were smaller, as Imola’s mortality rate rose.
Mortality: Imola vs Faenza
Using the same system on the four years for which data are available for Faenza shows that for 1991 and 1993 the projected deaths at the Imola rates are lower than the actuals, by 48 and 60 respectively, but in the next two years this reverses, and the actual rates are better than the projected by 20 and 24. The actual deaths were 926, 1006, 927 and 955. This implies that the growth in the Imola mortality rate over this period has moved Imola from being better than Faenza to worse than Faenza - not by as much as they were previously better, but nonetheless clearly not better overall.
Thus, to the table of self-reported health differences taken from the survey can be added the mortality differences: Imola:Sassuolo 1:0 (also by the SE test) and Imola:Faenza 1:0 (but not by the SE test). There was no way of testing the difference between Faenza and Sassuolo directly, given that neither town provided mortality by age band.
The data from Ferrara enabled a comparison by sex. The hypothesis here was a) that mortality would be lower in Imola than in Ferrara, as in the other two towns, b) that the improvement would be better among males than among females and c) that the age gap between the sexes at death would be lower in Imola by comparison with Ferrara.
a) Mortality Rates by town.
The analysis was concentrated on deaths in 1994 and 1995 between the ages of 50 (when the death rate starts to rise beyond very low rates) and 74 (above which the age distributions of the populations in the two towns are unknown, which makes comparisons unsafe).
Fig 4.26 shows the mortality rates by sex by age band for the two towns over this age range.
Fig 4.26. Imola and Ferrara: 1994-5 Mortality by Sex by Age Band
For every age band above age 55, Imola’s mortality rates are lower,
for both males and females.
Figure 4.27 shows the differences by age band. Only for those in the age band 50 to 54 - when mortality is low - are the Imola rates higher, for both sexes. The difference is in all cases better for males than females.
Figure 4.27. Imola and Ferrara. 1994-5. Difference in Mortality by Sex by Age Band
The absolute difference in mortality between Imola and Ferrara is greater for males than females. Males appear to benefit more than females from the Imola social environment.
b) Improvement by Sex
Figure 4.28 shows the overall weighted average difference for the whole sample, age 50-74, with the proportions of each age band standardised to the mean across the two towns. The Imola male mortality rate at 13.7 deaths per 1000 is 24% lower than that for the Ferrara males; the Imola female mortality rate at 7.9 deaths per 1000 is 17% lower than that for the Ferrara females.
Figure 4.29. Imola vs. Ferrara. 1994-5 Mortality by Sex. Age 50-74
Imola male mortality rate is better by 24%, female by 17%.
Taking the actual mortality rates in these age groups for 1994 and 1995, and standardising the populations by age band, it is evident that the predicted sex difference exists; Imola males appear to benefit even more than females from living in Imola (Table 4.53).
Table 4.53. Deaths by sex by town, 1994 and 1995. Age 50-74
Adj. Deaths: 94-95 Imola Ferrara
Males 5516 17457
Females 3173 9222
Ratio 1.74 1.89
This sex/town difference is statistically significant
(chi square 1 df = 10.87, p<0.001).
c) Sex difference in age gap. It was not possible to calculate the mean age at death with any confidence, given that such a high proportion of both populations died in the top, open-ended age band (over 75): 63% of all deaths in Imola, 59% in Ferrara. However, it was possible to approach this question by looking at the difference in each town between the mortality rates of the males and females. Fig 4.30 shows this difference, from which it is clear that in Imola age 55-69 male mortality rates are less different from the female rates than they are in Ferrara. So the data are to an extent consistent with the prediction: the age gap at death in Imola is lower than it is in Ferrara, until the age of 70-74, when it is identical. It also appears identical from age 75 upwards, the age band in which most deaths occur, although given the unknown age distributions at this level no definite conclusion can be drawn from this. As predicted, the gap in age-at-death between males and females appears confirmed as lower in the more egalitarian town.
Fig 4.30. Difference between male and female mortality rates by town
Male mortality rate less female mortality rate, deaths per 1000. 1994-5. Age 50+.
Cardiovascular Mortality
Data indicated that the main factor behind Imola’s lower mortality rates was cardiovascular mortality. Imola published data on death by age by cause for 1985, and Italy for 1984. For Italy as a whole, 85.3% of all cardiovascular deaths occurred at age 65 and upwards; in Imola, 85.5% of all cardiovascular deaths ocurred at age 70 and upwards.
Data were obtained on cardiovascular mortality for 1993 for Imola town and for Sassuolo covering a wider area than the town. The data are set out in Table 4.54, showing that Imola’s rate of cardiovascular mortality is significantly lower than that of the Sassuolo area.
Table 4.54. 1993 Cardiovascular and other deaths: Imola and Sassuolo
Cardio Other
Imola 225 468
Sassuolo (area) 399 543
This difference is highly significant (chi square 1 df = 16.13, p<0.001)
3.6. Mortality and Diet.
The control towns were selected to be geographically as close as possible to Imola, to give the maximum probability that there was no difference in diet. However, it was recognised that Sassuolo lies close to the main Italian centre of production of processed salami-type meats known as insaccati. A question was therefore included in the questionnaire to establish how often the respondents had eaten this food in the last week. The inhabitants of Sassuolo had eaten insaccati significantly more often than those of Imola. For the N=44 samples, the mean numbers of times per week were: Imola 1.64, Faenza 2.05 and Sassuolo 2.36 (Imola vs Sassuolo t= 2.18, p=0.032; Imola vs. Faenza t= 1.06, P=0.29 n.s.; Faenza vs Sassuolo t= 0.74, p=0.46 n.s.). A difference in diet cannot therefore be ruled out as a contributory factor to the differences in mortality rate between Imola and Sassuolo.
Social Participation
a) Voting Rates
Comparing the two local elections of 1995, the voting rate for Imola in the election of 23rd April 95 was 87.4% of all those eligible. In Sassuolo, the rate in the election of 31.12.95 was 83% of those eligible. The difference between these rates is highly significant (chi square 1 df = 367.5, p<0.0001).
No data were obtained from Faenza.
b) Blood Donation
Blood collection in Italy is voluntary and unpaid. The largest collecting agency by far is AVIS, who gave us their figures for the three towns.
The members of AVIS represented in 1997-1998 6.6% of the population in Imola, 2.7% in Sassuolo and 3.0% in Faenza. The two lower figures are significantly different from that of Imola (Imola vs Faenza chi square 1 df = 950, p<0.0001).
4.3.9. Overall Summary
Table 4.55 shows the complete picture.
Table 4.55. Relative Positions on each variable, including public statistics
Imola:Sass Imola: Faenza Faenza:Sass
(N=44) (N=56) (N=44)
Rich vs. Poor 1:0* 1:0* 1:0*
Social Networks 1:0* 1:0* 1:0
Authorities Helpful 1:0* 1:0* 1:0
Security fromCrime 1:0 0:1* 1:0*
Police Activity 1:0* 0:0 R 1:0*
Serenity re. Crime 1:0* 0:1* 1:0*
Crime Confidence 1:0* 0:1* 1:0*
Domestic Peace 1:0* 1:0 1:0
Physical Health 1:0 1:0* 0:1*
Mental Health 1:0 1:0* 0:1*
Mortality 1:0* 1:0 1:0
Member of Vol. Society 1:0 1:0* 0:0
Voting 1:0*
Blood donation 1:0* 1:0* 1:0*
Education 1:0* 0:0 R 1:0*
Truant at School 1:0* 0:1 1:0*
Truant Now 0:1 0:1* 1:0*
Training 1:0* 1:0* 0:1
Education & Happiness 0:1* 0:0 R 0:1*
Totals All Scores 17:2 10:5 13:4
Totals SE test 13:1 8:4 9:3
Where the SE test applies (higher mean ≥ lower mean+SE) the score is marked with an asterisk. “R” indicates a reversal between the N=44 and N=56 samples.
By section, with the addition of voting rates and blood donation to voluntary associations to fill out the “social participation” section, Imola’s position ahead of Sassuolo is reinforced (Table 4.56).
Table 4.56. Relative positions by section
Imola:Sassuolo Imola:Faenza Faenza:Sassuolo
Health 1:0 1:0 0:1
Education 1:0 0:1 1:0
Crime 1:0 0:1 1:0
Social Part’n 1:0 1:0 0:1
Social Env't 1:0 1:0 1:0
Total 5:0 3:2 3:2
If Table 4.56 is re-cast using only differences of at least 1SE in magnitude, it remains the same, except for one change: Faenza’s social participation moves ahead of Sassuolo’s, so that the overall score between Faenza and Sassuolo becomes 4:1. This changes nothing for the overall analysis.
Thus, even by this rather blunt instrument, Imola is significantly ahead of Sassuolo, the least egalitarian town (sign test p<0.05, 1 tail). Overall, the scores show that Imola > Faenza > Sassuolo, which is the same pattern as the proportion of people working in co-operatives and the perception of a relatively even distribution of wealth.
CONCLUSION
Given the complexities of comparing the populations of towns, and the impossibility of controlling for the infinite number of potentially relevant variables, this study must be treated as tentative and no more than a pilot. For example, the responses to the questionnaire established there was a significant difference in the consumption of processed salami-type sausage meat, with the inhabitants of Sassuolo eating it more frequently than the others. So the mortality rates might be being affected by different consumption rates of fats. This would need to be controlled in a future study, and there will be other variables that similarly need to be controlled.
With that general caveat this study gives reason to treat seriously the hypothesis that egalitarian environments are better for people than non-egalitarian ones. The predictions were made on purely theoretical grounds, and the results show a strong pattern fitting the predictions. This hypothesis is worth investigating further.
Appendix 1: The Questionnaire Questions (abbreviated).
Crime
A.1 Experience of crime:
Since the beginning of the year, Jan 1st 1998, in your town:
has anyone: • assaulted you? • robbed you? • assaulted you sexually? • broken into your house? • stolen a bicycle belonging to a member of your household? • broken into a car belonging to a member of your household? • stolen a car belonging to a member of your household? • have you been stopped by the police or carabinieri? • have you seen a crime being committed, including vandalism?
A.2 Opinions about crime:
• How much do you worry about crime: a lot/little/not at all? • Alone at night in your town, do you feel: unsafe/fairly safe/completely safe? • Inside the family, do you think that husbands act violently towards their wives: never/rarely/sometimes/frequently? • The police and carabinieri in your area work: well/neither well nor badly/ badly? • In your town crime is: growing/steady/decreasing?
Health
• In general, would you say your health is: excellent/very good/good/fair/poor?
• Daily activities: does your health now limit you in the following activities, and if so, how much? (Categories for each answer: Yes, limits a lot/yes, limits a little/no, does not limit)
a) moderate activities such as moving a table, pushing a vacuum cleaner, bowling or playing golf? b) climbing a flight of stairs?
• Physical health problems: in the past four weeks have you had any of the following problems as a result of your physical health: (yes/no)
a) accomplished less than you would like? b) were limited in the kind of work or other activities?
• Emotional health problems: in the past four weeks, have you had any of the following difficulties as a result of any emotional problems, such as feeling depressed or anxious: (yes/no)
a) accomplished less than you would like?
b) didn’t do work or other activities as carefully as usual?
• During the past four weeks, how much did pain interfere with your normal work (including housework): not at all/ a little bit/ moderately/ quite a bit/ extremely?
• Energy and emotions: in the last four weeks give the answer closest to the way you have been feeling: (all the time/ most of the time/ a good bit of the time/ some of the time/ a little of the time/ none of the time)
a) Have you felt calm and peaceful?
b) Have you had a lot of energy?
c) Have you felt downhearted and low?
• Social activities: in the past four weeks, how much have physical or emotional problems interfered with your social activities: all the time/ most of the time/ a good bit of the time/ some of the time/ a little of the time/ none of the time?
Education
C.1 Experience of Education:
• How old were you when you finished studying: age:.../still studying? • What was the highest educational level you reached: elementary/ middle/ "superiore"/ professional/ degree/ postgraduate/ adult/ other? • When you were at school, did you play truant: never/very rarely/ sometimes/ often? • Have you had training since leaving school?
C.2 Opinions about Education:
• Nowadays, do you think that young people play truant: never/very rarely/ sometimes/ often? • How important is education for happiness: very/fairly /not very/not?
Voluntary Organisations
• Are you a member of a voluntary organisation? If yes: how many? • In the last four weeks, how much time have you spent on activities of voluntary organisations: none/less than one hour/1-3 hours/4-8 hours/ 9 hours or more? • Are you a member of a committee in one of the organisations you belong to? If yes, how many? • What motivates you to belong: duty/to help others/to make friends/ to have fun/ to learn/ other
General Information
• Age. • Sex. • Married/unmarried but living together/separated or divorced/single/widowed or widower/other. • Town of residence. • Time you have lived in your town: less than one year/ 1-5 years/ 6-10 years/ 11-20 years/ over 20 years. • Number of people in your household. • Number in the following age brackets: under 16/ 16-60/ 61 or more. • Do you have a job: yes, full time/yes, part time/no, looking for work/ no, not looking for work/no, pensioner. • How many in your household work? • Do you work in a co-operative? • How many in your household work in a co-operative? • In the last week, how many times have you eaten "insaccati" (high-fat processed meats such as salami)?
Social Environment
Finally, some opinions about life in your town:• If you have personal difficulties, how many people can you turn to for help: many/ a few/ only one/ none? • In your town, is the difference between rich and poor: very big/ big/ not a big difference? • In your town are the authorities helpful: always/ usually/ not usually/ never?