Riskante gewoonten en zorg voor eigen welzijn

[p. 426]

Summary

Research design and methodology

The goal and the scope of the present study

In a country with a highly developed web of statistical services, prior to our research, no data was available pertaining to the intensity and occurrence of the drinking and smoking habits of the population. In order to meet the existing need, the sociological co-workers of the Department of Mental Health of the Netherlands' Institute of Preventive Medicine (Head: Dr. J. Koekebakker) in Leyden decided to launch a cross-national survey of these habits and of the current opinions about smoking and drinking. Once a national survey of public opinion had been decided upon, additional items of general interest to the research-workers in the field of mental health were included in the interview schedules. The main problem-areas of the study were defined as follows:

1.	In what sociological categories can the smoking and drinking habits be described?
2.	What is the intensity and the occurrence of these habits in the group of the adult population in the Netherlands?
3.	What place do these habits take in the social structure in this country; what is the social ecology of these habits? Which social factors do account for the distribution of these habits over the population?
4.	To what extent can the social norms be applied to the field of drinking and smoking behaviour? Which aspects of this behaviour are approved of and which are censored by the population?
5.	Is there a conscious care of one's well-being? Are people inclined to mutual aid and assistance?
6.	Is one conscious of the shady side of the drinking and the smoking? In what way has this consciousness been affected by the recent publicity campaigns in these fields? Are there social and psychological factors that adversely affect the efficacy of health instruction campaigns?
7.	What topics make people worry? What factors affect the satisfaction and the psychic well-being of the population?

These issues suggest the scope and the limits of the present study: it is concerned with the habits, attitudes and opinions of the adult-people in

[p. 427]

the Netherlands in 1958; any generalisation beyond this group surpasses the limits of the statistical (though not empirical) inference¹.

1 Finally, a note on the personnel aspects of the project: Mr.L.T.v.d.Laar helped with the coordination of the teams of interviewers and coders; Drs.Ch.A.G.Nass, the head of the Statistical Department, designed the sample. After the present writer's appointment at the University of Groningen, two other institutes co-operated. Mr.J.Oorburg, from the Sociological Institute, helped with the construction of scales and with the computation of some statistical tests. Professor Dr.I.A.van de Vooren and Mr.H.J.Burema from the Mathematical Institute greatly eased the evaluation of data by designing the programmes for the electronic computor under their administrative control.

Research-design

In order to answer the above-raised questions we decided for an opinion survey of a representative sample of the Dutch adult population. By means of individual interviews in the homes of the respondents we hoped to collect the data from some 1300 persons. This would be processed by means of Hollerith sorting and tabulating machines and evaluated statistically. After the construction of scales and after the preliminary statistical analysis, a factor-analysis and a systematic analysis by means of partial correlations of all significant relationships would be applied.

Interview-design

The data were gathered by a team of 25 interviewers recruited from the ranks of the advanced students in sociology and in psychology. They all were given the addresses of persons living in their own region together with the interview-sheets. The exact meaning of every question has been discussed with them in the presence of the research-staff; they received instructions to interview a few persons outside the sample, the first two interview-sheets were checked by the staff. The interview-sheets were previously tested out by the staff in a pilot study for their comprehensibility and for their psychological aspects: since a large number of questions was included, one had to build a certain dynamics and pauses into the process of interviewing; interviewers were asked to reverse the rôles during the pauses and to invite questioning by the interviewees.

The interview-schedule (see Appendix i, p. 469) comprised some multiple-choice questions; these were administered in the forms of cards comprising various items. The interviewees received instructions to ‘pick one’ item they adorsed, or to ‘pick two’, or to order the items.

As may be seen in Appendix I (e.g. questions 51 and 57) check-questions have been included in the schedule. On the title-page the interview-situation was described and the personal impressions of the interviewer were presented.

[p. 428]

The interviewer was introduced by a personal letter of the research-director containing the name and the address of the Institute sponsoring the research and describing the goals of the interview in vague terms: ‘a talk about the habits and the way of living’. The Institute was described as ‘an independent scientific institution that tries to foster health and well-being of the population by sponsoring research’ (see Bijlage 2). No mention was made of drinking or smoking in the letter of introduction.

Sample

The size of the sample depended largely from the available means. Originally, only some fl. 6,000.-were allocated to the process of collection of data. In order to bestow the sum as effectively as possible, we decided for a two-phase heterogeneous sample-plan that would allow us to draw a probability sample from clusters: the Dutch communities. Drs. Ch.A.G. Nass, the Head of the Statistical Department of the Institute designed the sample and the method that would enable us to take the effect of clustering into account by means of a modified analysis of variance.

All big cities (with 100,000 inhabitants or more) were included in the sample proportionally to the number of inhabitants. For every 5000 inhabitants one card was lifted from the Community Registrars (Bevolkings-registers); if it concerned a person younger than 20 years it was dropped. The selection was done blindly in order to avoid the bias owing to the different colour of the cards designating women and men. As to the minor communities, the procedure was as follows: in an alphabetical list of all communities with less than 100,000 inhabitants, the numbers of inhabitants were progressively added. A figure from 0 to 9 was chosen at random as a starting point (7); consequently, this number was multiplied by 10,000 and progressively added to 100,000; thus: 70,000, 170,000, 270,000, 370,000, etc. Communities corresponding to the intervals in which these progressive totals fell were included in the sample. Each of them would contribute 20 persons to our original sample so that the likelihood of being included in the sample was the same as in the big cities, namely 1 : 5,000.

The actual sample to be interviewed differed from this original sample in that we dropped at random the persons younger than 20 years. This accounts for the different numbers of persons actually to be interviewed in various communities.

Table 2.4.1 presents the basic data referring to the two sub-samples, both for the 14 cities and the 71 smaller communities. In addition, it refers to the sample that has been actually reached and registers the number of refusals. We notice that there were 2,182 inhabitants and 1,382

[p. 429]

adults to be interviewed and 1,297 persons who adequately answered our interview-questions; 85 either refused or could not be reached. Those who refused or were not reached represent 3.9% of the original sample and 6.15% of the actual sample to be interviewed. Table 2.4.2 enables us to compare the two groups, the one refusing and the one cooperating with regard to six characteristics: sex, age, marital status, religious denomination, occupation, residence-place. Only with regard to age, a significant difference was found, the responding group being slightly of lower age than the mean of the sample to be interviewed.

It is interesting to note that the direction of the deviation in the composition of respondents is consistently in agreement with the composition of the group of interviewers (see Table 2.3.1). There are more men than women, more younger persons, more Roman Catholics and more persons of higher occupations in either groups. The bias due to the principle of homiphily (R.K.Merton) is the more striking since the interviewers received a list of names (not adresses of households but of particular persons in these households, for whom they were not allowed a substitute). It seems that the co-operation in the interview situation is adversely affected by the different composition of the group of interviewers.

It is noteworthy to say that the above mentioned tendency did not interfere with the representative nature of our sample. With regard to several characteristics we were able to compare the final sample of respondents with the population-parameters as ascertained by the Central Bureau for Statistics (C.B.S.) in the Hague. None of the five characteristics examined showed differences larger than four per cent. The probability of the sampling error of the properties whose parameters are unknown is roughly described by the following facts: (a) the low percentage of those refusing the co-operation (6.15%); (b) fairly equal compositions of the refusing and the co-operating groups (Table 2.4.1); (c) the standard error of 1.8% for one of the key-variables (smoking - nonsmoking) at a 60%-40% split (see further). In view of these facts the group of 1,297 respondents may be considered as fairly representative for the adult population of the Netherlands in 1958.

Coding

The bulk of material that had been gathered by means of more than 150 questions from almost 1,300 persons had to be evaluated by mechanical means. Four Hollerith-cards were designed for the tabulation and the sorting-process. Nine coders working in the same room and using identical code-books were employed to translate the answers to interview-questions into numerical symbols. No less than 194 interview-sheets, i.e. 14%

[p. 430]

of the total number, have been check-coded by the research-staff who supervised the work of the student-coders. As Table 2.5.1 demonstrates, the results of the coding have not been very satisfactory. The number of mistakes varied from 3.35 per cent to 6.45 per cent per individual coder (one coder with 9.88 per cent of mistakes was dismissed and his work redone). There have been 26 questions or items that accounted for more than five per cent error level (see Table 2.5.2). The analysis of mistakes shows that most mistakes could be ascribed to the lack of concentration and attention and not to the difficulty of interpretation (most questions were of the pre-coded type; see Appendix i).

For the purpose of qualitative analysis (case-studies), the material was once more coded on hand-sorting punching cards that made the quick selection of various categories of drinkers, smokers, etc. possible.

Scales and indices

In order to pool the information that was gained by the large number of stimuli in an interview-situation, various devices have been used to contract the items into the scales and more reliable variables. They varied from simple cumulative indices (e.g. ‘worries’, variable no.22 in Table 2.7.5) to unidimensional scales of Guttman-type (e.g. ‘culture-involvement’, variable no. 10 in Table 2.7.5) or scales that were based on one or another factor-analytical design (e.g. attitude to smoking, general satisfaction, variables no. 14 and 21).

Table 2.6.1 refers to the scale of Guttman-type to measure the attitude towards drinking. The questions A, B, D, E, and F satisfied reasonably Guttman's criteria: the percentage of errors was 6.7%; the errors in negative or positive categories did not surpass the limit of 50% (the highest percentage was 39% with this scale); the errors were distributed at random over the single scores (though we found 6.2 per cent of errors, which is 1.2 per cent more than the usual 5-per cent level). Green's coefficient of consistency, being the function of the actual reproducibility and the reproducibility by chance, was



This is just below the critical value of .50 that Green gives for scalability¹. By lack of a better scale, the interviewees have been scored according to

1 We drew on Robert N. Ford, ‘A rapid scoring technique for scaling attitude questions’, Public Opinion Quarterly, Fall 1950, pp. 507-532; and B.F.Green, ‘A method for scalogram analysis using summary statistics’, Psychometrika, Vol. 21, i, March 1956, pp. 79-88.

[p. 431]

the scheme as given in Table 2.6.2 and the scale has been included in our Basic Correlation Matrix (Table 2.7.5, Variable 17).

The already mentioned scale of the attitude to smokers and smoking fulfilled all criteria of Guttman, but the index of consistency was still lower, namely .41. Mr.Nass applied to the same questions (see Table 2.6.3) a method of correlational analysis (the modified principle component analysis) and found that the longest correlations-ax did not correspond to the attitude-continuum; the second longest ax could easily be identified with the variable sought. Again owing to the lack of a better scale the questions have been scored accordingly and contracted into the variable no. 14 of our Basic Correlation Matrix (Table 2.7.5). Both methods, the principle component analysis and the Guttman-scaling, confirm each other, unfortunately through their negative results.

Another scale, that of ‘culture-involvement’ consisted of six questions that met all four criteria of scalability as mentioned by Ford; Green's index of consistency amounted to I = .47¹.

In order to make comparisons with the American studies of social participation possible, F.Stuart Chapin's scale has been translated and validated for the Netherlands in our earlier study². Since the results did not deviate from the Stuart Chapin's results in the States, the scale was used in the present study and included in the Basic Matrix (variable no.9, Table 2.7.5).

In several cases we abstained from the construction of the objective scales and chose either subjective semantic scales (the case of social, altruistic attitude, no.20 in the Basic Correlation Matrix) or cumulative indices. In the semantic scales, the individual items have been classified according to their meaning and scored accordingly; in the cumulative indices, we made the unwarranted presumption that all items were equivalent. The nature of data (order two or pick two techniques) made an objective scaling either impossible or a dreary enterprise (‘unfolding technique’ of Coombs).

The following Table 2.7.4 sums up the main variables and scales that have been included in the final statistical analysis together with the distributions of answers and the questions on which they are based.

1 For more details see: I. Gadourek, J. Oorburg, L.T.van de Laar, ‘Involvement in cultural system in the Netherlands: its measurement and social correlates’ in Social Forces, Vol.40, No.4, May, 1962, pp.302-308.

2 I. Gadourek, A Dutch Community. Social and Cultural Structure and Process in a Bulb-Growing Region in the Netherlands, J.B. Wolters, Groningen, 1961.

[p. 432]

Table 2.7.4 Specification of the 34 variables of the basic correlation matrix
No.	Variable Sub-category	Source: Question no.	Frequency	Sub-categories used
1.	Sex:	1		i, ii
	i. man		612
	ii. woman		685
2.	Age:	2		i to vii
	i. 21 to 25 years		138
	ii. 26 to 30 years		154
	iii. 31 to 40 years		309
	iv. 41 to 50 years		249
	v. 51 to 60 years		219
	vi. 61 to 70 years		150
	vii. 71 years and older		76
	viii. not filled in adequately		1
3.	Incomes:	18		i to viii
	i. none - to 40 guilders per week		136
	ii. fl. 41.- to fl. 50.- per week		48
	iii. fl. 51.- to fl. 60.- per week		93
	iv. fl. 61.- to fl. 70.- per week		182
	v. fl. 71.- to fl. 80.- per week		150
	vi. fl. 81.- to fl. 100.- per week		229
	vii. fl. 101.- to fl. 150.- per week		194
	viii. fl. 150.- per week or more		130
	ix. no adequate answer		135
4.	Educational level:	5		i to vi
	i. none (no response)		11
	ii. only Elementary School		765
	iii. Secondary School		395
	iv. College		99
v. Post-College, non-University level		7
	vi. University level		20
5.	Kind of work:	9		i, ii
	i. outdoors		228
	ii. indoors		1069
6.	Size of community:	Community		i to viii
	i. - 3,000 inhabitants	Registers	69
	ii. - 6,000 inhabitants		91
	iii. - 10,000 inhabitants		104
	iv. - 15,000 inhabitants		121
	v. - 20,000 inhabitants		62
	vi. - 50,000 inhabitants		216
	vii. - 100,000 inhabitants		169
	viii. 100,000 inhabitants or more		465

[p. 433]

No.	Variable Sub-category	Source: Question no.	Frequency	Sub-categories used
7.	Migration-index:	4		i to ix
	i. did not move		626
	ii. moved once		407
	iii. moved twice		122
	iv. moved three times		62
	v. moved four times		33
	vi. moved five times		20
	vii. moved six times		10
	viii. moved seven times		2
	ix. moved eight times or more		15
8.	Family-integration:	6 to 8		i to v
	i. single, without children		199
	ii. engaged		41
	iii. married without children		133
	iv. single with children		72
	v. married with children		852
9.	Social Participation Index:	13, 146-150		i to viii
	(low)
	0 to 2 points		171
	3 to 4 points		216
	5 to 6 points		301
	7 to 8 points		273
	9 to 10 points		185
	11 to 12 points		72
	13 to 14 points		39
	15 points or more		39
	(high)		39
10.	Culture- involvement:	125b, c, e, 120, 122, 123		i to vii
	(low)
	i. score 0		234
	ii. score 1		108
	iii. score 2		295
	iv. score 3		262
	v. score 4		161
	vi. score 5		173
	vii. score 6		64
	(high)
11.	Contacts with mass- communication media:	151, 153, 156		i to iv
	i. no contacts		7
	ii. contacts with one medium		52
	iii. contacts with two media		218
	iv. contacts with three media		553
	v. all three media and lectures		467

[p. 434]

No.	Variable Sub-category	Source: Question no.	Frequency	Sub-categories used
12.	Church-affiliation A:	146		i to v
	(by collective pressure)
	i. without denomination		301
	ii. Mennonite, Remonstrant, etc.		41
	iii. Dutch Reformed		306
	iv. Christian Calvinist Churches		105
	v. Roman Catholic		544
13.	Intensity of smoking habits:	51		i to vii
	i. 0- 1 cigarettes (or g tobacco) a day		549
	ii. 2- 5 cigarettes (or g tobacco) a day		126
	iii. 6- 9 cigarettes (or g tobacco) a day		87
	iv. 10-13 cigarettes (or g tobacco) a day		129
	v. 14-22 cigarettes (or g tobacco) a day		243
	vi. 23-29 cigarettes (or g tobacco) a day		65
	vii. 30- cigarettes (or g tobacco) a day		82
	no (adequate) answer		16
14.	Attitude towards smoking:	81 to 82		i to v
	(tolerant)
	i. 0- 3 points		81
	ii. 4- 8 points		388
	iii. 9-10 points		435
	iv. 11-12 points		363
	v. 13-16 points		30
	(intolerant)
15.	Belief in smoking causing lung-cancer:	77		i, ii, iii
	i. does not believe		263
	ii. does not know, no answer		463
	iii. believes so		571
16.	Intensity of drinking habits:	89		i to vii
	i. does not drink		752
	ii. 1- 3 glasses		301
	iii. 4- 6 glasses		101
	iv. 7- 9 glasses		52
	v. 10-15 glasses		42
	vi. 16-21 glasses		18
	vii. 22 glasses or more		13
	viii. no (adequate) answer		18
17.	Attitude towards drinking:	101,103		i to vi
	(tolerant)
	i. score 0		155
	ii. score 1		121
	iii. score 2		336

[p. 435]

No.	Variable Sub-category	Source: Question no.	Frequency	Sub-categories used
	iv. score 3		200
	v. score 4		278
	vi. score 5		207
	(intolerant)
18.	Intensity coffee-drinking:	46		i to viii
	i. drinks no coffee		59
	ii. drinks 1 cup a day		103
	iii. drinks 2 cups a day		239
	iv. drinks 3 cups a day		249
	v. drinks 4 cups a day		255
	vi. drinks 5 cups a day		131
	vii. drinks 6 cups a day		124
	viii. drinks 7 cups or more a day		137
19.	Eating sweets:	47		i to v
	i. eats no sweets		132
	ii. sometimes, not every day		563
	iii. 1-2 cookies a day		357
	iv. 3 cookies or sweets a day		203
	v. every day a lot		42
20.	Social attitude:	126		i to iv
	(asocial)
	i. score o		106
	ii. score 1		198
	iii. score 2		545
	iv. score 3		426
	(altruistic, helpful)
	v. no (adequate) answer		22
21.	General satisfaction:	12, 17, 23, 33		i to v
	i. satisfied with regard to 4 stimuli		618
	ii. satisfied with regard to 3 stimuli		405
	iii. satisfied with regard to 2 stimuli		198
	iv. satisfied with regard to 1 stimulus		60
	v. satisfied with regard to none of the 4 stimuli		16
22.	Worries:	116		i to viii
	i. no worries		448
	ii. worried about 1 problem-area		332
	iii. worries about 2 problem-areas		254
	iv. worried about 3 problem-areas		130
	v. worried about 4 problem-areas		73
	vi. worried about 5 problem-areas		36
	vii. worried about 6 problem-areas		17
	viii. worried about 7 problem-areas		7

[p. 436]

No.	Variable Sub-category	Source: Question no.	Frequency	Sub-categories used
23.	Normative pattern:	108		i to vii
	(sexual, collectivistic)
	i. rejects ‘j’		438
	ii. rejects ‘f’, ‘c’		316
	iii. rejects ‘g’		202
	iv. rejects ‘h’		103
	v. rejects ‘e’		38
	vi. rejects ‘b’, ‘d’		108
	vii. rejects ‘a’, ‘i’		26
	(individualistic)
	viii. no (adequate) answer		66
24.	Symptoms of lack of well-being:	109, 115, 138, 139		i to vii
	i. no negative symptom		490
	ii. negative symptom		367
	iii. 2 negative symptoms		218
	iv. 3 negative symptoms		136
	v. 4 negative symptoms		50
	vi. 5 negative symptoms		26
	vii. no (adequate) answer		10
25.	Traumatic youth experiences:	135		i, ii, iii, iv
	i. no such experience		635
	ii. 1 experience		365
	iii. more than 1 experience		277
	iv. no answer		20
26.	Evaluation of one's youth	134		i to iv
	i. pretty		995
	ii. does not know, so, so		55
	iii. less pretty		148
	iv. unpleasant		89
	v. no (adequate) answer		10
27.	Optimism- pessimism:	117-118		i to iv
	i. the future will be better		293
	ii. the same; does not know		466
	iii. the future will be worse; is not worried hereabout		341
	iv. the future will be worse than the present; is worried hereabout		181
	v. no (adequate) answer		14
28.	Change in incomes:	19		i, ii, iii
	i. earns now more than before		548
	ii. earns just as much		619
	iii. earns now less than before		111
	iv. no (adequate) answer		19

[p. 437]

No.	Variable Sub-category	Source: Question no.	Frequency	Sub-categories used
29.	Change in smoking pattern:	62		i, ii, iii
	i. smokes more now than before		79
	ii. just as much, does not know		838
	iii. smokes now less than before		143
	iv. no (adequate) answer		237
30.	Change in drinking pattern:	97		i, ii, iii
	i. drinks more now		204
	ii. drinks just as much, does not know		790
	iii. drinks less now		303
31.	Region where one lives:			i, ii
	i. in the North		928
	ii. in the South		369
32.	Church affiliation B:	146		i to v
	(Roman Catholic vs. Protestant)
	i. Christian Calvinist		105
	ii. Dutch Reformed		306
	iii. Other Protestant Church		41
	iv. Without denomination		301
	v. Roman Catholic		544
33.	Contacts with family-doctor:	27		i to vii
	i. this week		123
	ii. 7 to 14 days ago		51
	iii. 3 to 4 weeks ago		123
	iv. 1 to 3 months ago		168
	v. 4 to 6 months ago		119
	vi. 7 to 12 months ago		193
	vii. longer than 1 year ago		495
	viii. no (adequate) answer		27
34.	Working-hours:	11		i to vii
	i. none		64
	ii. - 5 hours per day		105
	iii. - 6 hours per day		67
	iv. - 7 hours per day		49
	v. - 8 hours per day		206
	vi. - 10 hours per day		428
	vii. 11 hours per day or more		345
	viii. no (adequate) answer		33

[p. 438]

Statistical Evaluation

In choosing the proper technique of statistical analysis we paid attention to the following objectives: (1) to estimate-as precisely as possible-the population parameters on the basis of the information that had been gathered through our survey; (2) to test the hypotheses about the contingency between the variables; (3) to estimate relationships between a large number of variables that would make a systematic analysis of correlated bias possible.

ad 1: In estimating the standard error due to the sampling, a major obstacle appeared in the evaluation of the so called ‘clustering effect’, i.e. the bias caused by the deviation of our actual sample from the strict random sample design. Instead of interviewing persons from some 1000 communities in the Netherlands, only the people clustered in 85 communities were selected by the researchers. What is the effect of this residential clustering upon the population-estimates? A general method for the estimation of the variance of proportions was used:
(1)



For the sub-sample comprising the big cities, we can substitute as follows:

Mi = the number of respondents per city that have been selected from the Registrars;

fix = the number of persons with the property ‘x’ among the M persons from the community ‘i’;

Pix = fix/Mi;M = the total number to be examined in the cities, i.e. 742 persons;

s^2ix = fix(Mi - fix)/Mi(Mi - 1), is the variance of the property ‘x’ in the community ‘i’;

fixy = the number of persons in the city ‘i’ possessing both the property ‘x’ and the property ‘y’.

[p. 439]

For the sub-sample comprising the smaller communities we substitute in formula (1) as follows:

n = 71 smaller communities in our sample;

m = 20 persons selected at random from each community;

i = 1, 2, 3 ... 71;

fix = the number of persons with a property ‘x’ among the m-persons from the community i;

The information from both sub-samples can be pooled by applying the formula:
(2)



i.e. the proportion in the two sub-samples together; thus, e.q., the number of smokers divided by the number of persons above 20 years of age in the total sample;
W = the total number of respondents in the two sub-samples;
W_a, W_b = the number of respondents in the corresponding sub-samples;
Var q_a, Var q_b = the variance of the proportion in each separate subsample.

An illustration of the application of this method is given in tables 2.7.1 and 2.7.2 where the data for the proportion of smokers (non-smokers) is given. The standard-error for the combined sample amounts to 1,77 per cent; this means that the percentage of smokers in the adult Dutch population lies somewhere between 53,1 and 60,3 per cent (the statistical probability of this statement being P = .95).

As can be read from Table 2.7.3, the dreary technique of estimating the variance of proportion by taking the effect of clustering into con-

[p. 440]

sideration has been applied several times. It shows significant deviations from the standard error as computed for the strict random samples; in one case (the proportion of non-smokers among the male inhabitants), the standard error has been more than halved by the clustering effect. In spite of the soundness of this technique, we abstained from generalizing it to the technique of testing associations. Though it was, in principle, possible to test the difference between two proportions by taking the clustering effect into account, the procedure turned out to be too laborious to be applied generally. Moreover, the extension of the technique to more than two proportions would be necessary to control the correlated bias of the assumed associations. A short-cut method for it was not available at the time we needed it¹.

For this reason we contented ourselves with estimating the standard error of the main variables under study (distributions of smoking, drinking, and culture-involvement). As we planned to examine systematically the influence of various aspects of the social structure (sex, age, income, religious denomination, etc.), and as the effect of clustering can be interpreted in terms of these various social structure aspects, the analysis of variance has not been pushed further than the seven proportions that are mentioned in Table 2.7.3.

ad 2: In examining the distribution of the main variables, we were struck by the differences and variations in various social groups. Smoking habits turned out to be much more intensive with men than with women; the age, the religious denomination, the income-level, and several other factors brought about a differentiation in the subject under study. In order to show that these differences could not be ascribed to the sampling error but were likely to be found in the general population, simple non-parametric tests were applied. The chi-square test was used in most cases. The correction for continuity was usually computed whenever the frequency in a cell of the table was lower than 50. A simple measure of the degree of association (usually Cramér's V; we drew upon Hubert M. Blalock, Social Statistics, MacGraw Hill, 1960), has usually been derived from the chi-square values. All these measures can be found directly in the text, under the corresponding tables.

ad 3: The independent variables (sex, age, education, income-level, religious denomination, etc.) were intercorrelated to the extent that we de-

1 We learnt so from the personal communication of Professor L. Kish, one of the most advanced workers in this field, whose ‘Confidence intervals for clustered samples’ in American Sociological Review, Vol. 22 (1957), pp. 154-165, drew the attention of sociologists to the problems of clustering.

[p. 441]

cided to look for a systematic analysis of the correlated bias. In order to confirm the hypotheses that we derived from the theories about alcoholism, smoking habits and other problems we studied, one should try to control the simultaneous influence of as many other variables as possible. To achieve this, we needed a common measure that would reduce the associations between variables to one common denominator and that would enable us to examine the relationships of at least three interrelated variables. Not without hesitation, we decided for the product moment correlations that we computed from the grouped data and for the partial correlations derived from them. The correlation-technique has been applied to all 34 variables of the Table 2.7.4 (see page 432ff.) regardless of the nature of scale, (interval or merely ordinal; in some cases even the nominal scales have been transformed into the ordinal ones in an artificial way; see e.g. variables no. 12 and 32). Care was taken to utilize the information of the entire sample and only the evasive answers or the groups for whom the questions had not been designed were omited. The correlations are presented in Table 2.7.5. Those printed in italics are considered significant at the probability level of at least P = .02. Since the N (total number of individuals) was in no correlation table lower than 1000, we were able to estimate that the correlation of r = .065 corresponded roughly to the .05-level of probability. In Table 2.7.5 all correlations amounting to r = .075 or an higher value were denoted as statistically significant and were chosen for a further analysis. (See page 384A).

This analysis proceeded along two lines:

1. all clusters of three significantly correlated variables have been identified in the basic matrix of relationships (Table 2.7.5) by the simple device of the matrix-inspection and multiplication as devised for the analysis of the sociograms¹.

There are exactly 426 clusters of three variables that are inter-correlated in a significant way. They all are presented in Table 2.8.6 together with three partial correlations for each identified cluster. All this systematic work has been done by means of a single programme by an electronic computor. Following the reasoning of Herbert A. Simon (Models of Man, Social and Rational, New York, 1957, chapters 1 and 2) we are able to eliminate the spurious correlations by means of the vanishing partials. A suitable example is the correlation between the variables 5 and 13 (‘kind of work’ and ‘intensity of smoking habits’) being r = - .299 (see the entry in the cell at the cross-section of column 5 and row 13 in Table

1 A description of this systematic procedure is presented in our A Dutch Community, pp. 330-342; also ‘A substitute for a randomization design in sociological research’, in Indian Journal of Social Research, III, no. 1, 12-24 (1962).

[p. 442]

2.7.5). From Table 2.8.6, cluster no. 15, we learn that the partial correlation is only r_1.5-13 = - .020 i.e. approaching zero. We therefore conclude that not the kind of the work but the variable no. 1 (sex) influences the smoking habits, the original correlation r_5-13 = - .299 being of spurious nature. The interpretation of this finding is quite obvious: housewives work at home, and this is the only reason why people working indoors smoke less than people working outdoors, there being, of course, fewer, and less intensive, smokers among housewives and women in general.

There are more vanishing correlations in the Table 2.8.6 that can be interpreted in an equally meaningful way.

2. One could object against the method as described sub 1 that only correlations visible in the basic correlation matrix are taken account of. It is possible that due to an intervening variable, due to the fact that a certain variable is correlated with two other variables in opposite directions, a genuinely significant association does not show up in the basic matrix at all. To meet this and similar objections, a factor-analytical design has been applied. We decided for Kaiser's varimax procedure¹ that had been programmed for the electronic computor of the Mathematical Institute of the University of Groningue on request of the psychologists at this University (Professor Dr. B.J. Kouwer).

Tables 2.8.1 and 2.8.2 present the results of this analysis as applied to the 34 variables of our Basic Correlation Matrix (Tables 2.7.5 and 2.7.4). No less than ten orthogonal factors have been extracted; the rotation has been applied to nine factors. Owing, probably, to the heterogeneous nature of our data, only some 48 p.c. of the total variance has been ‘explained’ by these nine factors. On the other hand the factors could easily by identified and interpreted thanks to the meaningful combinations of variables with high loadings on each of the identified factors:

Factor I
		Loading
Variable 31	Region where one lives	-.741
Variable 32	Church-affiliation B	-.717
Variable 12	Church-affiliation A	-.653
Variable 17	Attitude towards drinking	.462
Variable 23	Normative pattern	.380
Variable 19	Eating sweets	.343
Variable 6	Size of residential community	.265
Variable 27	Optimism	.212

1 H.F. Kaiser, The Varimax Method of Factor Analysis, University of California, Microfilm; also his ‘The varimax criterion for analytical rotation in factor analysis’, in Psychometrika, 23 (1958), pp. 187-200.

[p. 443]

There is little doubt that Factor I denotes the differentiation owing to the church-affiliation and to the regional division (the Roman Catholic South and the Protestant and the less church-bound North). All variables with higher loadings stand in meaningful relation to this division into the main sub-cultures in the Netherlands. Some of them (the more optimistic attitudes of the Roman Catholics and the more tolerant attitude towards alcohol of the latter) have been signalled out by previous studies.

Factor II
		Loading
Variable 1	Sex	-.838
Variable 13	Smoking habits	.809
Variable 5	Kind of work	-.553
Variable 16	Drinking habits	.464
Variable 30	Change in drinking pattern	.290
Variable 23	Normative pattern	.288
Variable 19	Eating sweets	-.279
Variable 18	Coffee drinking	.274

This Factor II can be identified with the different way-of-life of men and women in the Netherlands. Women smoke less, work indoors, drink less alcoholic beverages though their drinking habits do not show the same decrease with age as stated for men; they eat more sweets but drink less often coffee than men.

Factor III
		Loading
Variable 4	Educational level	.669
Variable 2	Age	-.553
Variable 7	Migration-index	.497
Variable 8	Family-integration index	-.346
Variable 16	Drinking habits	.313
Variable 17	Attitude to drinking	-.302
Variable 3	Income-level	.296
Variable 28	Change in income level	-.261
Variable 19	Eating sweets	.245
Variable 6	Size of the residential community	.231

This factor was somewhat less evident. It denotes the groups of young people with more school-education, with higher migration-index, i.e. those who migrated at least once and probably left their parental homes and community of birth, still unmarried. We found more intensive drinking habits, more tolerant attitudes towards drinking, higher incomes and increase of incomes in these groups. We would suggest that this Factor III denotes the general emancipation: the younger generation with more school education than parents received.

[p. 444]

Factor IV
		Loading
Variable 10	Culture-involvement	.722
Variable 9	Social participation	.638
Variable 20	Social attitude	.570
Variable 12	Church-affiliation A	.453
Variable 6	Size of residential community	-.237
Variable 4	Educational level	.218

We can easily identify this factor as denoting the general socio-cultural integration of the individual. The high loading with ‘social’ (better: ‘altruistic’) attitude is interesting in this context.

Factor V
		Loading
Variable 24	Psychical well-being	.708
Variable 22	Worries	.625
Variable 21	General satisfaction	.592
Variable 25	Traumatic youth experience	.473
Variable 26	Evaluation of one's childhood	.438
Variable 33	Contacts with family-doctor	-.404

This factor ties subjective variables that we do not hesitate to denote as indicators of general satisfaction and psychical well-being, together with some of their more objective correlates (especially the two ‘childhood’ variables).

The remaining factors that we extracted can be interpreted in a similar way. Factor VI ties together the higher income-level and family-integration. It may be termed ‘socio-economic status’, though without the usual differentiating meaning in terms of various jobs and occupations. Social participation and contacts with mass-communication media (the latter probably through the possession of them) show higher loadings on it.

Factor VII ties together higher age, shorter working-hours and small residential community. It denotes probably the group of retired persons; the low intensity of coffee-drinking is related to it.

Factor VIII suggests the common basis of the change of the individuals status and habits while Factor IX ties the attitude towards smoking with the acceptance of the disseminated knowledge about lung-cancer in a meaningful way; optimism-pessimism is loaded on it in the congruent direction: optimists accept the knowledge less frequently and are more tolerant towards smoking.

The factor-analysis of the 34 variables of Table 2.7.5 was not the only one which we carried out in the present research. The questions pertaining to the individual's perception of his (her) own mental states and feelings

[p. 445]

have been singled out for a special correlation-analysis. The following questions have been used:

Question no.:	Specification:	Dichotomization:	Variable no.:
12	Work-satisfaction	‘yes’ - the rest	1
16	Fear of unemployment	‘no’ - the rest	2
17	Satisfaction with one's life-career	‘satisfied’ - the rest	3
21	Chances for the future	‘yes’ - the rest	4
23	Health-satisfaction	‘satisfied’ - the rest	5
33	Satisfaction with one's lodging	‘satisfied’ - the rest	6
109-115	Tension-symptoms	none - the rest	7
116	Worries about death	none - the rest	8
129	Broken confidence	‘no’ - the rest	9
135	Traumatic experiences	none - the rest	10
134	Evaluation of one's childhood	‘nice’ - the rest	11
136-139	Loneliness, lack of purpose, etc.	none - the rest	12

Instead of computing the product-moment correlations, estimates were made of the tetrachoric correlations that are presented in Table 2.8.3. The corresponding values have been found by means of the tables of Davidoff and Goheen in Psychometrika Vol. 18 and 19; the Jenkin's tables (Psychometrika, Vol.20, September, 1955) made it possible to estimate the standard errors of the tetrachoric correlations. The values of r's that are significant at least at the .02-level of probability have been printed in italics in the tables. Tables 2.8.4 and 2.8.5 present the three factors that we extracted from this correlation-matrix by means of the Varimaxtechnique.

Factor I' ties the four satisfaction indices together in a meaningful way; Factor II' shows high loadings on symptoms of tension, worries about death and the two childhood-variables. Each of the two factors has been, therefore, included as a seperate variable in the scheme of 34 variables of our final design.

Before closing this description of our methodological design, we wish to draw the attention of the reader to the fact that some variables of our Basic Correlation Matrix cannot be considered as apriori independent. ‘Church-Affiliation A’ and ‘Church-Affiliation B’ are both based on the same factual material; it is therefore no surprise to find these two tied together in Factor I. Similarly the variable 9 (social participation) is tied to the church-affiliation owing to its operational definition. With regard to other variables the principle of independence holds; the correlations as well as the loadings add to our knowledge of their mutual inter-relationships.

[p. 446]

Concepts and findings

1. Smoking habits

In choosing the smoking and drinking habits as the object of our study, we realized that we were using a concept that enjoyed a short popularity among behavioral scientists at the beginning of the century, but passed into oblivion with the intensified research into attitudes and perceptions.

Habits are distinguished from attitudes by their nature of actual, overt behaviour, behaviour that can be directly observed. We differentiated social habits from both the ‘individual habits’ (being chiefly a psychological concept) and ‘customs’ (being rather a concept of social anthropology). We conceived of social habits as being elementary forms of collective conduct that is learnt by the individual from his social environment in an almost unconscious way, is embedded in his personality-structure, and is manifested in his actions that repeat themselves in not too long intervals. Social habits are at the same time habits of the individual, but not the other way round: not all habits become social, only those taken over by a group or a community. Taking this as a starting point, we may apply some psychological theories of habit-formation to social habits: they are ready-made forms of conduct that present themselves to the individual in certain situations; they can be seen as mechanisms of releasing psychological (perhaps even: physiological) tensions. As to the way they are acquired by the individual, the psychological learning theory can be generalized to a lesser extent: only the habit of acquiring social habits can be ‘learned’ in this narrow sense of the learning theory¹. Other than purely psychological ‘mechanisms’ are needed to describe the process of socialisation and acculturation through which the individual acquires the social habits. Reference group theory may offer a cue: social habits are acquired through imitation of persons with whom one identifies oneself, who belong to one's reference groups.

It can be pointed out, in this context, that social scientists have paid more attention to the function of the social habits than to their diffusion throughout the social universe. W.James mentioned the conservative function of habits, calling them ‘a fly-wheel of society’. Another function would be that of social differentiation: social classes distinguish themselves from each other by different social habits. Sumner combined this conception with the theory of social selection: social habits would help the more able individuals to maintain themselves. With the decline of social Darwinism, social habits began to be interpreted in terms of social control and just ready-made action-patterns.

1 N.E. Miller, J. Dollard, Social Learning and Imitation, Yale University Press, 1941.

[p. 447]

More prolific than the thinking about the diffusion of social habits in general has been the thinking about the factors that make an individual to acquire specific social habits. As far as smoking habits are concerned, the following theories have been found in the bulk of available literature:

1.	Theory of physiological predisposition to smoking. Some persons with predilection for bitter food would have a larger tolerance and greater predilection for nicotine. Both would depend on the physiological constitution of the persons in question.
2.	Psycho-analytical theory of oral fixation of smokers. Persons who did not successfully pass through the oral phase of development of the satisfaction of their drives would find in smoking an often preferred way of satisfying their inner needs.
3.	Theory of neurotic predisposition to smoking. Smoking is considered as an indicator of general ‘nervousness’ in a similar way as e.g. nailbiting, thumb-sucking, etc. Several studies comparing the groups of smokers and non-smokers on tests such as the Cornell Medical Index or Neuropsychiatric Screening Adjunct suggest higher frequency of neurotic traits in smokers than in non-smokers.
4.	The compensation-theory of smoking maintains that heavy smokers can frequently be found among the men with feminine physical traits; males with a robust physic would be less included to develop intensive smoking habits.
5.	The social rôle-theory of smoking is related to the theory sub (4): smoking helps to distinguish some subtler rôles, such as: that of an adult, partly also that of a man as compared with that of a woman.
6.	The normative theory of smoking. This is usually stated in negative terms, referring to norms that prevent smoking from spreading among the large masses of population. In one version, it assumes less intensive smoking habits with persons whose parents have been opposed to smoking. In another version, it assumes a relationship between the ethics of protestantism (‘fundamentalists’) and abstinence from smoking or mild smoking habits.

Several operational hypotheses have been derived from these theories. Though the survey-research is not the most appropriate instrument for testing hypotheses, the data collected from a sample of the general population is not without relevance for the question under study.

In presenting the findings of our survey, we first report the distribution of the variables, since the knowledge even of the basic facts regarding the smoking and the drinking habits was missing at the time of the publication of this report. From Table 1.2.6 we learn that smoking is almost

[p. 448]

universal in the group of adult men in the Netherlands; nine out of ten male adults smoke as compared with four out of ten adult women. Intensity of smoking is much higher in the group of men: we found an average (median) of 17 cigarettes or g of tabacco per day for smoking men and only some 3 cigarettes (or equivalent) for smoking women. According to our estimates (taking into consideration the clustering effect owing to the selection of our sample) there are 58.5% to 65.7% smokers in the Netherlands.

According to the information in Table 1.2.2, the majority is formed by the cigarettes smokers. Nine out of every ten men-smokers of the cigarettes inhale the smoke into their lungs. Only about a half of the women cigarette smokers do so. Though there is some indication that one smokes more on Sundays and holidays, about a quarter of the population indicates that this does not make any difference; the rate of smoking is quite regular throughout the week. According to Table 1.2.3, about three quarters of all smokers begin to smoke before ten o'clock in the morning. Table 1.2.4 sums up the answers to the question inquiring after the motives to smoke more than usually; 519 who do not smoke or do so only occasionally are not included; neither are 415 persons who underlined more than one answer-categories, since the limited space of the Hollerithcard did not allow for combinations. These 415 persons were analysed separately; the results were in agreement with those of the Table 1.2.4. A great majority of smokers mention ‘contact with people’ as the main reason why they occasionally smoke more. Tables 1.2.5 and 1.2.6 show the association between the age and the age at which one starts smoking. This association is positive for men (P < .001; ϕ = .17) and negative for women (P < .001; ϕ = .19); women start smoking at an older age, in general, but there is a trend to start earlier. Men show a tendency to start at an older age than the older generations. This suggest that smoking reached (or passed) already its zenith with men while it is still spreading among women. Another interesting conclusion is that the median age at which men start smoking is 16 years and three months; smoking sets in at the age of adolescence.

Our conclusion about the differential smoking trends of men and women is further confirmed by tables 1.2.8 and 1.2.9: older men appear to smoke more, older women less than their corresponding younger groups.

As Table 1.2.10 shows, especially women with more education develop more intensive smoking habits (the association is significant; Chi-square test as applied to the table amounted to the value of 19,18 at 4 degrees of freedom, P < .001). Another factor is the urban milieu: women in cities smoke more than women in rural communities. All this supports

[p. 449]

the assumption that smoking habits are still likely to spread among women-groups, chiefly owing to the ongoing emancipation process.

Other interesting aspects of the social ecology of smoking habits are the following: higher income-groups smoke more and more frequently than lower income groups, though there is no correlation between the occurence of smoking and the social prestige of occupation. The religious factor also plays a rôle: Roman Catholics smoke somewhat more than Protestants do, give at least significantly higher quotas of tobacco that they consume.

These findings bring us back to our starting point-the hypotheses about the development of intensive smoking habits by the individual.

By inspecting the Basic Correlation Matrix (Table 2.7.5), the table of partial correlations (Table 2.8.6) and by taking the results of our factor-analysis (Table 2.8.2) into account, we were able to examine the operational hypotheses that were derived from the theories about the smoking habits. On psycho-analytical grounds we expected correlations of smoking with other forms of oral satisfaction: drinking habits (alcohol), the habit of drinking coffee, the habit of eating sweets. Both coffee and alcohol drinking was positively correlated with smoking; eating sweets and cookies was, on the other hand, correlated in the opposite direction: smokers munch less sweets than the non-smokers. This finding seems to support our theory (2) about the physiological predisposition. We indeed found less predilection for sweets with the smokers; the question, however, arises whether this is not an effect rather than a cause of intensive smoking habits in man. A social survey, taken at one moment of time, was not a suitable instrument to determine this.

The theory of neurotic predisposition of smokers received no support in our data: smokers did not more frequently than non-smokers admit to the interviewers of having anxieties and fears, having worries, or experiencing inner tensions, feeling lonely, having nothing to live for and being bored - symptoms that formed a basis of our variable no. 24 (the lack of psychical well-being). In so far as our questions referring to these symptoms adequately measure the neurotic predisposition (which the clinicians would probably not easily admit) the theory has to be rejected. We hesitate to go so far and content ourselves with pointing out that persons inclined to more complaints in this field are not smokers.

In order to shed some light on the compensation-theory (see above sub 4), we compared smoking habits of those holding jobs requiring physical strength (farmers, agricultural labourers, unskilled workers) and those holding jobs of a more subtle nature (teachers, preachers, clerks). No difference of any significance was found.

By far the strongest evidence received the hypotheses that were derived

[p. 450]

from the social rôle theory of smoking. We mentioned already that the greatest part of smokers acquire this social habit in early adolescence: there is good reason to believe that smoking helps to make the rôle of a young adult visible. Smoking also can be ascribed to the rôle of a man in our society; the sex-factor accounted for no less than 45 p.c. of the total variance in the smoking habits over the general population. The fact that women with more education, living in cities also show a tendency to smoke, suggests that education and urbanization smooth the traditional differences between the sex-rôles. Women take over occupational rôles of men and the rôles of action-initiators; smoking helps to make their emancipation and independence visible.

As Table 1.2.7 clearly shows, no association was found between the intensity of smoking habits and the parental norms with regard to smoking. Those who were not allowed to smoke in their youth smoke rather more at present, than those who enjoyed freedom in this respect. The other hypothesis that was derived from the normative theory has been confirmed: persons of Protestant denomination report less intensive smoking habits than the Roman Catholics. Because of the complex nature of the influence of social norms, we shall treat it once more, after having briefly reviewed the main aspects of the drinking habits in the Netherlands.

2. Drinking habits

Drinking of alcoholic beverages has puzzled many behavioural scientists. Several of them launched interesting theories to explain drinking, especially in its more excessive form. To a certain extent, these theories overlap with the theories referring to smoking: the psycho-analytical theory of oral fixation, the social rôle theory of drinking and smoking, the normative theory. There are, however, theories that stand on their own, from which hypotheses have been deduced to be tested in this research:

1.	Escapist theory of drinking: alcohol should help to cope with worries, dearth and want of any kind. It forms a social mechanism to escape from cruel, rude social reality into a better world of dreams and harmony.
2.	Anxiety theory of drinking, in Horton's formulation perhaps only a variation, though an important variation, of the theory sub 1. We might examine whether the persons who report more anxiety report more frequent use of alcoholic beverages.
3.	Social isolation theory of drinking. Several writers pointed out that drinking may be seen as a substitute for the lost social contacts. Both formal and informal social participation of the drinkers would be lower than that of the corresponding non-drinking groups.

[p. 451]

4.	The loss of frame of reference. Somewhat related to the above mentioned theory (sub 3) is the attempt to explain drinking out of a sudden disintegration of one's frame of reference. Migration, social revolution and other sudden upheavals that affect a person's Weltanschauung would result in more frequent and intensive drinking patterns.
5.	The social anomie theory of drinking is sometimes interpreted in terms of lacking normative scales; in that case it approaches the theory mentioned sub (4). There is, however, a narrower theory of anomie of Durkheim that relates both suicidal tendency and alcoholism to a sudden upward social mobility: the change of social position places the fortunate person in a normless vacuum while the material conditions enable him to satisfy his sensual needs. But in order to keep his new position, he has to strain energy and to double his efforts. A general fatigue of the organism ensues, often accompanied with crises when the individual sees through the senseless nature of his effort. Suicide and alcoholism are then often an issue.
6.	The availability-theory of drinking: People drink simply because alcohol is available to them. This availability of alcohol can be translated and interpreted in social and geographic, but also in economic terms: those with higher incomes are according to this theory more likely to imbibe.

The data that is presented in the tables of Chapter 3 of our book clearly show the relative low intensity of alcohol drinking habits in the Netherlands-this in contradistinction to the smoking habits. The answers tot he questions about the consumption of alcohol in the week preceding the interview display both quantitative and qualitative differences between the drinking patterns of men and women. The number of persons who never use alcoholic beverages lies between 15 and 21 p.c. (18% ± 3.5%).

Only 113 respondents (out of 1297) acknowledge to drink regularly; 37 of them drink in the evening, 26 before dinner, 16 in the afternoon, 19 in the morning. These 19 morning drinkers could not be identified with ‘alcoholics’; by means of a detail analysis (of the intensity of their consumption and the total life-history as far as this could be reconstructed), we found that various other reasons accounted for the morning drinking in the general population (occupation, doctor's advise, chance-factor).

Another factor accounting for the general lower consumption of alcoholic beverages is the fact that alcohol, as a rule, is not used at table. Only some two per cent of the total sample drink during their warm meals: 22 mention beer, 5 wine. Yet another interesting aspect is that drinking usually takes place in the homes of the participants. As Table

[p. 452]

1.3.3 shows, the habit of offering a drink is quite general; only one out of eight adult Dutchmen does never offer a drink. One offers drinks, as a rule, at home. About a half of the population has got a stock of alcoholic beverages at home; Table 1.3.4 gives the necessary information in this respect. Socio-economic status brings about a differentiation about this: out of 73 professionals, managers, and business-men only 16 had no stock of alcohol at home; out of 244 working-class persons, 159 had no alcohol at home at the time of the interview.

The question whether one drinks more on certain days of the week was answered by 58 p.c. of our sample in a negative way. Out of those who gave a positive answer (256 in total), a large majority mentioned the week-end (only 18 spoke of other days than Sundays and Saturdays). The week-end drinkers turned out to be prevalently beer-drinkers. As in the case of smoking, congenial and sociable motives were mostly given for the use of alcohol by the respondents who adequately answered our corresponding question (no. 101 of our Questionnaire, see Appendix on p. 477).

As Table 1.3.5 shows, for certain persons drinking is probably equivalent to eating sweets; 71 drink because it tastes well. (For the negative correlation between drinking and eating sweets, see further).

In spite of the generally low figures that we presented so far, one should not conclude that intoxicative effects of alcohol are unknown to the respondents. From Table 1.3.6 we learn that men drink on average as much as 5.6 glasses and women 3.2 glasses per evening (owing to the fact that the size of glasses varies with the strength of the beverage, we simply registered the number; a cross-tabulation with the kind of beverage used made, whenever needed, a more precise picture possible). No less than 821 persons, i.e. 64 p.c. of the population according to our estimate, answered our question no. 94 (whether one drinks until the intoxicative effect of alcohol is felt, now and then; ‘just a glass too much’) in a negative way. To this group some 18 p.c. who never drink should be added. This leaves us with 221 persons (17.3 p.c. of the sample) who openly acknowledged to the interviewers to drink in an excessive way. Most of them experienced an elation, merry and relaxed state of mind; about one quarter complained of sadness.

Men report to need on average 8.5 glasses, women 5.5. glasses to experience the effects of alcohol. By closer analysis we found that the previous question, of normative nature, influenced the responses. We hesitate to draw the conclusion that women need a lower quantum of alcohol to enjoy or experience intoxication.

About ten per cent drink, according to the evidence from our sample, in connection with their jobs or work. These are mostly the salesmen, innkeepers, etc.

[p. 453]

We also were interested in the development of the individual drinking habits in the general population; 731 persons (56.4 p.c. of the sample) reported no change in their drinking pattern; 303 persons (23.4 p.c.) reported to drink less at present than at an earlier period of their life; 204 persons (15.7 p.c.) noticed an increase of their alcohol consumption (3 persons, i.e. 0.2 p.c. of the sample) mentioned rather a qualitative instead of quantitative change.

By cross-tabulating our data by age and sex (see Tables 1.3.7 and 1.3.8), we found that with the ongoing age men drink less, while the drinking patterns of women remain about the same. It seems that especially young men (and especially the young week-end drinkers in the pubs and café's in the South of the country) acquire more intensive drinking habits. When married and settled down, they limit their drinking to the social drinking with occasional guests.

Age and sex appear important factors in other aspects of drinking-habits, too. Table 1.3.9 shows that intoxication is significantly concentrated in the younger groups (P < .01 for a dichotomic table of persons above and below 50 years of age). The intoxication is more frequently acknowledged in the groups of workers than in the groups of persons working for their own (difference of 17.5 p.c. in the positive categories between the groups of respectively 231 and 172 persons). A small group of clerks and shop-assistants (20 persons in total) showed a still higher percentage (50% as compared with 37.5% of the workers and 20% of the professionals and those working on their own admitting intoxication, now and then). This finding is the more interesting since we know that actual drinking is more frequent with the higher income groups who also keep more alcohol beverages in stock than do the lower income groups to which the workers belong.

Another factor that brings about a differentiation in the drinking habits in the Netherlands is the religious denomination. The Calvinist groups (members of the Gereformeerde Church and of the Dutch Reformed Church) report less alcohol consumption than the Roman Catholics and the persons without religious affiliation. As Table 1.3.10 shows, this difference could not be ascribed to the regional differences (the Roman Catholic South vs. the Protestant and secular North); we found still a significant difference between Protestants and the other denominations in the South of the country (P < .05 for a dichotomized table). This does not imply that there are no differences between the two important cultural subregions: one drinks more outdoors, experiences (or reports) more intoxication in the South.

Returning to the theories and hypotheses that we derived from them, we have to state that many hypotheses were rejected: the hypothesis of

[p. 454]

a concomittant variation of all forms of oral satisfaction (no relation was found between eating sweets and drinking; the correlation between the drinking of alcohol and the drinking of coffee vanished in the sub-categories of smoking habits: r_16-18.13 = .008; the remaining correlation between smoking and drinking could easily be explained by means of other theories, e.g. the rôle-theory of smoking and drinking); the hypothesis that lower income groups would drink more, as derived from the escapist-theory of drinking: in fact income is correlated with the drinking habits in a positive way. Thus an alternative hypothesis, that was deduced from the ‘availability-theory’ of alcohol, was confirmed. More educated persons drink more and this association partly accounts for the correlation between incomes and drinking. Social drinking seems to be embedded in the way of life of the educated classes in the Netherlands, who earn higher wages than the average

r_3-16 = .089; r_4-16 = .119; r_3-16.4 = .057.

Contrary to our expectation, no significant relationship was found between the drinking habits and social isolation. Participants in clubs and associations drink even somewhat more than the non-participants.

The anxiety-theory made us expect higher occurrence of drinking with the persons who mentioned fears. The findings, again, were discouraging, since no association in this direction has been found.

The theory of the loss of frame of reference gave rise to a hypothesis about the more intensive drinking habits of persons without involvement in a cultural system or sub-system. This hedonistic theory of drinking has hardly been supported by our data. The scale of Guttman-type, that was construed for this purpose (the culture involvement scale), did not correlate with the drinking habits in a significant way, r_10-16 = - 0.46. When, however, the social participation as a test-factor was introduced, the value of the partial correlation rose to reach the significance-level, r_10-16.9 = - .065 (all these and the previously mentioned correlations can easily be identified in Tables 2.7.5 and 2.8.6). Though significant, the association is of extremely low intensity.

According to the anomie-theory, we expected more frequent drinking-habits with the persons who mentioned to earn more at present than before. Even this expectation has been contradicted by facts. An originally weak tendency (r_16-28 = - .060) vanished completely when age, educational level, and other variables were introduced as test-factors.

Thus the rôle-theory of drinking received the strongest support of our data. Women drink considerably less than men (r_1-16 = - .273); age also is a factor of importance (r_2-16 = - .097): especially men start

[p. 455]

drinking rather heavily with the onset of adolescence and abandon their habits with ongoing age. Once more: better educated groups, and groups of men, drink more. Social drinking seems an expected pattern of behaviour in the groups of young adult men in the Netherlands who gain their own wages and wish to manifest their maturity and independence.

The normative theory also got support: the correlation between the attitude towars drinking and the actual consumption of alcohol is r_16-17 = - .250. As we shall show presently, the variables that are correlated with this normative attitude, such as the religious denomination, also correlate with the drinking habits.

An explanation has been sought for these startingly negative results of our research. Instead of ascribing them to the defect of our methodological tools and to the improper research-design, we wish to propose the following interpretation. Social drinking habits are basically different from alcoholism and from excessive drinking as described, up to now, by the alcoholism research. As most hypotheses have been deduced from the literature on alcoholism, it is not surprising that they have not been borne out by the data collected about the social drinking habits. Differences are, indead, striking. By the conclusive evidence of the social scientists in the United States and elsewhere, we became used to picture an alcoholic as an elderly, lonely man, without organizational or family-ties, vagrant or migrant, periodically unemployed and unable to sustain himself, of lower educational and social economic status. Instead, we notice that in a general population alcohol is being consumed by young, vital men, by persons of higher educational an socio-economic status, by sociable people participating in clubs ans free associations. In other terms, we see in our data a confirmation of Selden D. Bacon's conception of alcoholism as a dysfunctional aspect of social drinking. Learning to drink in a group, in a congenial sphere, at the onset of adolescence, the peak of virility, at the time when one assumes emotional and financial independence, some people continue to associate drinking with these circumstances and characteristics even when becoming a prey of excessive habits. For them drinking becomes the very impediment of the goals they are, consciously and unconsiously, striving for.

Very little can be learnt from the present survey of the social drinking habits about the situations and factors generating this excessive drinking pattern. Only a question enquiring after the need of craving for alcohol in the past, gave a cue. Table 1.3.11 shows that persons with disagreable recollections of their childhood and their parental homes display somewhat more inclination for alcoholism. The association, however, is very weak and we possess no certainty that a response-set does not account for it. The rôle of the agreeing response set should not be exagger-

[p. 456]

ated. The high reliability of information tapped from two sources (i.e. the question about smoking habits in general and the question about the consumption of tobacco on the day preceding the interview resulted in a correlation r = .91) and the meaningful pattern of the correlations strengthen our confidence in the validity of our data.

3. Normative aspects of smoking and drinking

For obvious reasons, we were interested not only in the factual behaviour but also in the social norms regarding smoking and drinking in the Netherlands. The latter were defined in their non-medical and non-statistical terms as the collective ideas of how an individual should behave in a given situation, usually (but not necessarily) strengthened by means of social sanctions and resulting in a certain uniformity of conduct. Obviously, these collective ideas could be tapped by means of an interview-schedule, the degree of their uniformity (and of their collective nature) could be evaluated once the answers had been summed up and compared.

Social normativity in this sense has repeatedly been brought in to account for the habits of smoking and drinking.

Selden D. Bacon attributed the existence of alcoholism, among others, to the fact that drinking is customary in our societies and that our mores very seldom forbid drinking itself; only the effects of excessive drinking are ostracized by society. In this basic fact that social drinking is an allowed and even prescribed form of conduct in certain situations, may be seen the main cause of failure of certain prohibition-movements.

Against the opinion that social norms are causally linked with actual behaviour stand the results of studies (e.g. Kinsey c.s.) showing a large discrepancy between the social norm and the actual conduct. Charles K. Warriner signalled a similar difference between the official morality and the drinking patterns; E. Allardt, on the other hand, found in Finland a more tolerant normative attitude towards drinking with the drinkers. It was interesting, in the light of these studies, to examine the relationship between the norms and the actual drinking behaviour in the Netherlands.

As far as smoking is concerned, its normative aspects came to the fore through the research on lung-cancer and the publications about it. Here was the opportunity to study the social norms in forming, since irresponsible neglect of one's health is censored by certain groups. We decided to study not only the correlation of the norm and the actual conduct, but to examine the patterns of relationships of the normative judgment and of the actual smoking or drinking behaviour, separately. In this way, we hoped to obtain a deeper insight into this controversial issue.

In Chapter 1.4.3 the main operational aspects of the social normativ-

[p. 457]

ity are presented in several tables. It is striking to see that in the year of our survey (1958), smoking is, generally speaking, not censored by the population. Only 8.6% of the population would object against the smoking of the sons, though 37.8% had objections against the smoking of the daughters. Women and the young ones are being protected by the normative judgment: only 3.5 p.c. of our sample would impose no age limit upon the sons' smoking habits, some 15 p.c. would allow smoking of sons below 14 years of age. The majority had no objections when the boys older than 14 years (for many: the wage-earning age!) smoked.

By means of a battery of questions we tried to probe the attitudes of the respondents towards the smokers and towards the non-smokers. The negative statements about each of these groups have not been adorsed by the respondents, as a rule. Neither smokers nor non-smokers are socially ostracized, we cannot speak about the double morality, in this respect. The normative judgment showed a striking consistency in the groups of men and women. A slight tendency has been noted of each sex-group to impose a more stringent norm upon their sex-partners.

The drinking norms showed a pattern fairly similar to the smoking norms. Only 1.6 per cent are of opinion that an adult man should not drink alcoholic beverages at all; the corresponding percentage was 7.9 p.c. (103 persons of our sample of 1297) with regard to the drinking of women; 23 p.c. of the sample are of opinion that a man may drink six or more glasses per evening, while only 16 p.c. display the same degree of tolerance with regard to the drinking of women.

As far as attitudes are concerned: 97.6 p.c. of the total sample would resent it if a family member was tipsy every day towards the evening-hours; the same percentage adorsed the objection against a family-member being drunk every week; 85.4 p.c. objected against ‘now and then drunk’; 68.2 p.c. against ‘now and then tipsy (aangeschoten)’; 30 per cent against regular drinking of a glass before dinner; 10 per cent against a family member who would abstain completely. These data suggest that intoxication is ostracized, is morally censored by the population in the Netherlands. As Table 1.4.4, however, shows, the transgression of drinking norms is judged much less severely than other forms of a-moral conduct. Drinking and neglect of one's health are less censored than e.g. theft, contacts with prostitutes, reckless driving, treason, or blasphemy (‘goddeloosheid’).

As expected, some aspects of social structure influence the norms of the inhabitants. Religious denomination is one of them. If we take three glasses per evening that one allows an adult man to drink as a limiting case, we find that the percentage of persons who would limit the use of the alcohol to less than this number varies as follows:

[p. 458]

Roman Catholics	7.5 per cent (out of 544)
Dutch Reformed Church members	15.6 per cent (out of 306)
Without denomination	16.0 per cent (out of 301)
Calvinists (Gereformeerd)	30.6 per cent (total: 105 persons)
Other denominations	37 per cent (total: 41 persons)

The percentage of those who impose stronger norms on drinking is four times as high in the groups of Calvinists and the members of smaller churches than in the circles of Roman Catholics. With regard to smoking, no differences in social norms were found between the persons of different denomination. This is the more striking since other norms, too, are associated with the church-affiliation, as Table 1.4.6 shows: Protestants especially censor the transgression of norms guiding the economic conduct, Roman Catholic are more severe with regard to the sexual norms.

As Tables 1.4.7 and 1.4.8 shows, similar differential finding was stated with regard to the association between the normative judgment and the actual conduct: drinking habits are significantly correlated with the normative judgment, smoking habits are not. However when sex is introduced as a test-factor, two associations of opposite directions emerge (Table 1.4.9): smoking women (mothers) object more against the smoking of sons, smoking men are more tolerant of the smoking of sons.

No less interesting are the differential correlations of the attitude towards drinking and the actual drinking behaviour. While sex (man- and woman-rôles) is directly associated with actual drinking behaviour, other variables such as age, religious denomination, residential region, involvement in a cultural system, traumatic childhood experiences and worries influence the actual drinking pattern only indirectly, through the social norms on drinking. This is the interpretation which we give to the parallel rows of correlations in Table 2.7.5 (rows 16 and 17) and to the vanishing partial correlations in Table 2.8.6. In general, we cannot say that the normative judgment is the cause of actual behaviour. While smoking was not correlated at all, only some 6 p.c. of the total variance in drinking can be accounted for by the normative attittude. Moreover, the question of cause and effect cannot be decided with any certainty. A longitudinal study would be better suited to shed light on this problem. Finally, most of the correlated variables are of existential kind; they cannot be manipulated with. Thus the practical meaning of our findings is still questionable. There is, with regard to smoking, one exception: the knowledge of effects of smoking is correlated with the attitude towards smoking. It is this variable that will be treated separately in the closing chapter.

[p. 459]

4. Material welfare and the satisfaction of the population

As a counterweight to the basically spiritual factor of ethical and religious norms, the material factor and the satisfaction with the material circumstances received a closer attention in our study. The functional justification of both drinking and smoking could be, that they contribute to the well-being of the population. It was for this reason that we decided to study the well-being as such, if possible in connection with the objects to which it can be related.

The question whether the psychical well-being and satisfaction are primarily determined by the objective situation in which a person lives, or by his (previously formed) personality structure was among the first ones to be solved. Questions concerning the material circumstances of the respondents have been included in our survey. Answers to them are tabulated in Chapter 1.5.4 of our book: the reader is kindly requested to consult the tables concerned. Several differential findings are reported in the text: younger people, thanks to more education, earn higher wages, so do persons participating in more clubs and voluntary associations, persons without religious denomination, and, more or less obvious, the persons with longer working-hours. Other correlations from Table 2.7.5, besides those with drinking and smoking, turned out to be of spurious nature.

The longer working-hours were concentrated in the groups of farmers, farm-hands and construction-workers. Men more than women complain of pressing work, the employed women complain more than the housewives (Tables 1.5.3 and 1.5.4). With these data in mind, we shall find that the work-satisfaction, as surveyed in Table 1.5.5, is extremely high in the general population in the Netherlands. Men show slightly more dissatisfaction than women, yet the percentage of those who are dissatisfied amounts to 5.6 p.c. only. Compared with the national survey results of the N.I.P.O. (Netherlands' Institute for Public Opinion) of some ten years ago, this means a shift to greater satisfaction of no less than 25 per cent. As the following Tables 1.5.7 and 1.5.8 show, the work-satisfaction is relatively unrelated to income-level and to income-increase. Though fear of unemployment is positively correlated with work-satisfaction (r_tetr = .35 in Table 2.8.3), no significant shift towards the feeling of greater security could be stated when comparing the results of our own survey with those of the N.I.P.O.-survey some 10 years ago. It is suggested that perhaps the work-circumstances themselves have changed, more attention being paid to good human relations at a period of a general lack of manpower.

Somewhat less satisfaction has been registered with regard to one's individual life-career (question no. 17, see Appendix). As Table 1.5.6 shows, the independent persons, working on their own, display more satis-

[p. 460]

faction in this respect than employees. This is the more striking, since farmers and independent fishermen think that in financial respect they are excluded from the general boom; they report financial progress in relatively fewer cases than other occupational categories.

An even less favourable picture is that of future chances for the improvement of one's material conditions. According to Table 1.5.7, most persons think to have no chances for improvement in the future. Women are somewhat more pessimistic, in this respect. In order to estimate the frustration resulting from the limited possibilities for upward social mobility, question no. 22 was presented to those who gave no positive answer to the foregoing question about the chances for betterment. A larger group, namely 386 persons (i.e. 51.4 p.c. of those who were asked) resented not being able to improve their living conditions, since they had hoped to do so; 303 persons (41 p.c.) gave a negative and 55 persons (7.6 p.c.) an hesitant answer.

The other satisfaction-questions met with a similarly positive response as the question probing the work-satisfaction; 76.7 p.c. were satisfied with their house or lodging; 12.3 p.c. were ‘fairly satisfied’, 10.8 p.c. were dissatisfied; 3 persons (0.2 per cent) gave no adequate answer. This finding is striking in the light of other information that we present in Chapter 5: we found a median of 1.9 inhabitants per room which suggests a relative overcrowdedness; moreover, about a half of the population is doomed to live under very primitive sanitary conditions, having only a pump or one water-tap (no basin) at their disposal; only about a third of the population can take a bath or a shower at home. No less than 38.7 p.c. of the population thought of the sanitary equipment in their own home as quite insufficient, an additional 15.3 p.c. found it rather insufficient. The reason why the satisfaction with lodging is, nevertheless, so high, can be seen, according to our opinion, in the low rent one pays for the houses. One realizes that the changes in housing conditions can only be achieved at the costs of other expenditures (motors, cars, television-sets, etc.) that one prefers.

It sounds repetitious, but the distribution of answers to the question enquiring after the satisfaction with one's health and physical condition was quite similar; about 80 p.c. were satisfied, 12 p.c. rather satisfied and 8 p.c. dissatisfied. A larger part of the respondents reported never to have suffered from a serious illness (64 per cent); those reporting illness were, of course, less satisfied. Frequency of the contacts with the house-doctors was also correlated with the health-satisfaction in a negative way.

Several indicators of satisfaction were thus distinctly correlated with the objective factors; the objective situation in which the respondent finds himself is of importance for the prediction of this satisfaction score. In

[p. 461]

scrutinizing the patterns of relationships of these indices, we were struck by the fact that the intercorrelations of the subjective variables were higher than the correlations with the external, objective factors.

For this reason we decided to look for a general satisfaction concept, based rather on personality than on situationnal characteristics.

The factor-analysis of 12 subjective variables which were described on p. 445 resulted in a successive extraction of the general satisfaction factor. The four satisfaction-indices with high loadings on it have been included as variable no. 21 in the Basis Correlation Matrix (Table 2.7.5) comprising 34 variables. If we examine the correlations of this variable, we are struck by the absence of associations with the basic social variables. General satisfaction is a property that is not influenced by the income-level, sex, age or any other social circumstance. Of greater importance are the two childhood-variables: people with unhappy childhood are more frequently to be found among those whom we should denote as ‘dissatisfied’ in the terms of our operational concept. Social isolation, as expressed in terms of the low index of social participation (after F. Stuart Chapin) and the lack of church-affiliation, shows also low but significant correlations.

Our conclusion is that the reference-group theory as well as the older theory of ego-involvement in reference-groups offers a suitable explanation for the genesis of dissatisfaction, being a generalized attitude or even a personality-trait. The disruption of parental family-group, lack of ties and emotional security in parental family as well as the lack of ties with associations in present context, account for the distribution of this trait.

One warning should be mentioned. General Satisfaction should not be considered as a unitary concept; we were not able to build up a unidimensional scale of Guttman's type out of the four satisfaction-indices. This corroborates the correlations between the separate satisfaction indices and the social texture. Both subjective and objective (situational) factors seem to influence the satisfaction of man.

5. Worries and other symptoms of the lack of well-being

According to one definition, the social problem is what is experienced and considered as such by the population. Starting from this standpoint, we found that very little is known about the problem-areas of the general population. This was the reason why a complex question referring to ‘worries’ was included in the questionnaire (no. 116, Appendix p.478).

Considering the general tension-releasing function of alcohol and smoking, questions about tensions, fear and other states of uneasiness or discomfort have been added.

The literature about the social pathology and the recently developing

[p. 462]

sociology of mental illness (as well as social psychiatry) offered some theories from which the working hypotheses could be deduced.

If our questions measured to some degree the lack of mental equilibrium or the neurotic tendency, then we would expect concentration of negative symptoms in the groups of persons with unhappy youth, in the groups of migrants or those who are either socially isolated or without cultural ties. Social status would perhaps also be of importance though the literature survey presented alternative hypotheses in this respect.

As to the worries, we expected a concentration in the Calvinist groups and in the group with weakened religious ties. Situational factors were expected to account for the occurrence of worries about specific subjects.

The data is presented in the usual way. Table 1.6.0 shows the distribution of answers to the complex question no. 116. As expected, different responses were obtained from the specific social groups. Women mention more frequently worries about family or children; less obvious: persons with more school-education and those living in the cities also mention more worries in this field. Worries about money and affairs are, indeed, concentrated in the groups earning less than fl. 3,000.- per annum. Protestants more than Roman Catholics are inclined to worries about their relation to God. The worries about death and a life hereafter are also concentrated in the groups of Dutch Reformed and Calvinist people.

Though our assumption about the association with the objective situation has partly been corroborated, the intercorrelations of the separate worry-questions are higher than their correlations with the basic aspects of the social structure (see Table 1.6.1).

Roughly the same can be stated about the other questions probing the psychic well-being, the distribution of which is given in Table 1.6.2. The feelings of anxiety, pressure and despondency are correlated with each other at the level of significance of P < .001; the mean correlation of contingency was Q = .41. Feelings of loneliness, boredom, lack of purpose of life gave an even higher correlation, Q = .70. The high mean coefficient of contingency gave us the justification to contract each of the two clusters and to include the variables thus obtained into our matrix of twelve subjective variables on p. 445. As can be read from Table 2.8.3, both variables (no. 7 and 12) themselves are intercorrelated, r_tetr = .32. The corresponding loadings on Factor II' are .726 (variable 7: tensions, fears and despondency) and .436. We decided to contract these two clusters in a simple cumulative index of lack of well-being and place it as variable no.24 in the Basic Correlation Matrix. ‘Worries’ and the two childhood-variables were included too, as respectively variables no. 22, no. 25, and no. 26. This done, the further analysis proceeded according to the already familiar pattern.

[p. 463]

As the factor-analysis and the analysis of partial correlations show, the two childhood-variables were significantly correlated with the symptoms of lack of well-being. The ‘traumatic experiences’ in childhood consisted in most cases of the sudden disappearance of one or both of the parents or of a sibling through bereavement or divorce; only 4. p.c. mentioned also drunkenness in the parental home. It is interesting to note that the variable ‘broken confidence’ also shows a high loading on Factor II'. In both variables we see an element of the disintegration of the reference-group. This fits into the disintegration-theory of mental disorder. Other findings corroborated this theoretical frame-work: those with more symptoms of the lack of well-being score lower on the social participation scale, they are more often persons without religious ties or belong to the more individualistic churches.

It is no surprise to note that they more often consult their doctors; r_24-33 = -.171.

The correlations of the other variable (‘worries’) confirmed some of our hypotheses: Protestants are more ‘worried’ than Roman Catholics; persons with stronger cultural ties (the higher scores on the culture-involvement scale) underline more questions about worries. Sex is a causal factor, too: women in the Netherlands are more willing to admit worries than men. Finally, an association was found with ‘optimism’ as measured by question no. 117.

Since the childhood-experiences and the evaluation of one's youth appeared to be of importance in predicting the lack of well-being, we mention the main correlates of these variables. We found that the desintegration of parental families was concentrated significantly more in the cities than in the country, significantly more with the migrants than with persons who still live in their place of brith, finally, more in the North of the country.

As far as the more subjective variable is concerned, we were glad to note that the younger generation significantly less frequently mention unhappy childhood. In the North of the country (this means also the western urban region), we found less positive evaluation of one's own youth; this, however, could be ascribed to the already mentioned larger frequency of family disintegration, there; when the latter variable was controlled, no significant partial correlation was found: r_25.26-31 = - .026.

6. Integration of the individual in the society

The weak but significant correlation between the need for alcohol and the disruption of a confidential relation that one experienced (φ = .082, P = .01), together with the correlations that were mentioned in the previous

[p. 464]

section, justifies a closer examination of the integration-problems. Instead of confining our analyses to the formal interaction patterns as measured by Stuart Chapin's scale, we decided to add some new dimensions: (1) the involvement in and interaction with a cultural system; (2) the general altruistic attitude or the feeling of solidarity with one's neighbours.

As to the first concept, we can be short since the theoretical backgrounds have been published elsewhere¹ We used a scale of Guttman's type to measure the degree of involvement in a cultural (sub)system.

‘Solidarity’ was measured by a simple semantical scale that was based on question no. 126.

As we can read from the first tables of Chapter 7, interesting associations have been signalled in our preliminary analysis. Formal participation (as measured by S. Chapin's index), though correlated with the informal participation (visits of friends, acquaintances and family-members), showed a distinct pattern of relationships. Men score less on visits, women on formal participation; moreover, in small (rural) communities formal participation is high while the visits are less frequent; in the cities the situation is just reversed. Both factors (sex, and size of residential community) reinforce each other.

In spite of these differential patterns of relationships with various aspects of social structure, the indicators of social integration are all intercorrelated with each other in the expected direction: social participants in associations score higher on culture-involvement scales, on solidarity scales, and show, upon the whole, a higher frequency of visits. Thus the reinforcement theory instead of the compensation theory receives a support through our data.

If we consult the variables with a high loading on Factor IV, we find here all three ‘integration’-variables together with some variables that fit the picture: church-affiliation continuum (persons without church-ties, the members of less integrated church-communities, the members of the Calvinist Church, the members of the Roman Catholic church), the size of the residential community, educational level, etc. It is interesting to note that stronger rejection of alcohol also is associated with this factor together with higher use of coffee.

In spite of the high loadings on one common factor, more differential associations can be identified on our Basic Correlation Matrix and the table of the partials. Formal participation correlated with incomes, the rôle of the man, happy childhood, more smoking, more eating of sweets. None of these correlations could be found with regard to culture-in-

1 I. Gadourek, J. Oorburg, L.T. van de Laan, ‘Involvement in cultural system in the Netherlands: its measurement and social correlates’, in Social Forces, 40, no.4, pp.302-307 (1962).

[p. 465]

volvement. The latter, on its turn, correlates with more worries, more contacts with mass-communication media, less use of coffee, more rejection of alcohol and more credence given to the reports about the harmful aspects of smoking. Solidary attitude was correlated with culture-involvement, social participation, church-affiliation and the size of residence (there are more altuistic, solidary attitudes in smaller communities). None of the remaining 30 variables was correlated with it; thus we may accept an earlier surmise by Sorokin that altruistic feelings and a helpful attitude are not directly socially conditioned: neither income-level, nor sex, age, or some other basic factor determine this attitude. The general integration of the community is, on the other hand, of importance.

Contacts with the mass-communication media depend on the income-level, educational level, social participation, residential region (in the North more intensive contacts) and culture-involvement.

These latter two variables bring us to the final theme of our book. Before switching to it we wish to point out that, though not a unitary concept, social integration as such, as well as its components, are each of importance for the study of well-being and mental hygiene in the broad sense as launched in our project. While the social participation helps the individual to maintain his healthy mental balance, culture-involvement helps to prevent him from sinking into the pit of hedonism with its risk-implying habits. Both foster a more co-operative solidary attitude and keep him abreast of the current development through the better developed radio-listening and news-reading habits.

7. Public instruction and the care of one's welfare

Welfare is an ambiguous term. Partly, we understand by it material circumstances, prosperity, good luck. Partly we think of good health. It may be viewed as an objective counterpart to the more subjective ‘well-being’. We chose the concept on purpose, since we thought that individual healthcare cannot be treated without its context: the individual's general pattern of satisfaction of his needs. In other terms, consumptive behaviour as far as determined by the level of incomes and stimulation of needs, will without doubts influence the care bestowed upon one's hygienic or preventive health-measures. For health-care, as any other need-satisfaction, has its economic basis; it costs money to choose an hygienic lodging, to pay a doctor, to buy elementary first-aid requisites or medicines, to afford sufficient night-rest. In contradistinction to other needs, the needs in the health-field are only felt when it is usually too late to take the preventive measures. Thus a new factor enters the field: that of the knowledge of ‘the good way’ and the degree to which this knowledge is accepted and

[p. 466]

taken account of in the dislocation of one's spendings. Any attempt to measure the degree of responsibulity for one's health will thus have to take the main variables of our three-dimensional model into account: 1. the material circumstances; 2. the availability of knowledge; 3. the acceptance of this knowledge by the individual¹.

The workable nature of the model becomes obvious when looking for the suitable indicators of individual health-care, the degree of responsibility for one's health. The material (economic) factor usually played the rôle of an intervening variable; from Table 1.8.1 we learn that possession of first-aid requisites (as a possible indicator) is almost linearly dependent on the income-level. From the following three tables we learn that the frequency of contacts with the house-doctor (as a possible indicator of preventive care owing to doctor's advise) shows an inverse relationship: the higher one's socio-economic status, the less frequent contacts with the doctor. To understand this, one has to know that in the Netherlands up to a certain wage-limit all employees are bound to be insured and, consequently, receive free medical assistance. As the correlations of this variable (no. 33 in Basic Correlation Matrix, Table 2.7.5) show, the material factor is, of course, not the only social factor influencing the frequency of contacts with doctors. Women see doctors more often, so do, obviously, older persons; somewhat more interesting: inhabitants of cities consult doctors more frequently; even higher were the correlations with the ‘subjective’ variables: dissatisfaction, the lack of well-being, and worries. Probably both the more neurotic and the more responsible people consult frequently their doctors.

The time reserved for one's night-rest was considered another indicator for the responsible attitude to health. Once more, the material factor intervened: we found a correlation with the average number of working-hours; those working longer, take, obviously, shorter night-rest. The next table (1.8.7) shows that the more subjective aspect of jobs plays a rôle too: those complaining of busy work, sleep less. Again, the material factor is not the only one: women complaining of inner tensions sleep less, as Table 1.8.9 convincingly shows. Our surmise that the anxiety-question somehow taps the neurotic tendency, is strengthened by this result.

The basic assumption of our explanatory model seems to be confirmed: individual health-care is materially conditioned, though the economic factor is not the only one: neurotic tendency and responsible attitude seem other factors of importance.

1 It was especially Leon Festinger who drew our attention to the two aspects of cognitive process, the confrontation with new knowledge and its acceptance or rejection. His theoretical conclusions formed a guiding line for the present study, at least in one specific field: that of dissemination of knowledge about the health-aspects of smoking.

[p. 467]

In order to ascertain the influence of knowledge and of acceptance of knowledge, we singled out the campaign against smoking as a case. In the Netherlands, news about a relation between cancer and smoking started to flow to the public ever since 1953. The public-instruction campaign reached its peak after the public declaration of the government of March, 1957, warning the public and advising the youth not to learn this habit. Our survey took thus place about one year after the press, the broadcast and the television had all publicized the results of the cancer-research in this field. The distribution of answers to our questions referring to smoking should be interpreted in this light. Tables 1.8.10 to 12 show that a majority consider smoking as unhealthy, though the heavy smokers are more hesitant to acknowledge this. The next tables show that smokers are better informed about the harmful effects of their habit than the non-smokers.

Tables 1.8.15 and 1.8.16, on the other hand, clearly show that the smokers are less willing to give credence to the information they received. There is an almost linear correlation between the number of cigarettes (or g of tobacco) one smokes and the percentage of persons accepting the public instruction as true. The analysis of partials made it possible to test this correlation with regard to possible intervening variables and to look for other factors that influence the acceptance of knowledge. Education, culture-involvement and the lack of well-being (‘neurotic tendency’) were the factors found.

The main hypothesis of Festinger's theory of cognitive dissonance was thus confirmed: smokers reject disagreeable, threatening knowledge. How about another mechanism of resolving the tension: giving up smoking?

Our findings were striking, in this respect: as tables 17 and 18 in Chapter 8 show, those accepting the knowledge do intensify rather than abandon their smoking habits. Table 19 shows that this is certainly not because of the irrelevant nature of the subject: out of eight ailments or forms of illness (including mental illness, blindness, epilepsy), cancer is indicated by the largest part of the population as the illness one fears the most. We also learn from Table 20 that almost a half of all smokers in our sample considered abandoning smoking when the campaign reached them. Yet the following tables (21 and 22) show that a good many failed and consider themselves not strong enough to give up smoking. If we interpret this self-classification of smokers in terms of probable development of the smoking habits, the hypothesis predicting the change of habits with those accepting the knowledge is somewhat more tenable. Perhaps we reached the smokers in a too early period after the knowledge had reached them, to draw a final conclusion.

The fact, that smokers in a period of confrontation with threatening knowledge accept this knowledge and intensify their habits, does not fit

[p. 468]

the theory and asks for interpretation. We tentatively draw an analogy with the war situation where intensification of eating habits followed the first threatening news about food-rationing: eat while you can! Possibly, the attention of smokers is unduely drawn to their own habit by the preoccupation with its possible harmful effects and more intensive smoking results from the ensuing tension.