Individual stability of work orientations: experience of longitudinal study

Jüri Saarniit

Department of Sociology

Tartu University

 

1. Introduction

There are two questions that value researchers time and again ask from themselves and argue about. Firstly there is a problem how stable are individual value judgements as such. Are they something that change daily depending on mood and other factors, or are they stable at least to a degree that deserves scientific analysis? The other problem concerns the degree of adequacy to what degree the assessments expressed by the respondents in the research process reflect their actual thoughts and feelings. Such problems apply particularly to mass studies that regardless of standardisation involve rather different survey conditions, respondents’ current mood and other factors that can influence the responses.

Many studies suggest that at the group statistical level value stereotypes of adult population are very stable and both inter- and intra-generational changes in value consciousness in Western as well as East European countries emerge only over a rather long period. (see e.g. Inglehart 1990, Saarniit 1995a). However, non-longitudinal studies fail to give a clear answer to questions discussed here because the statistical stability of value judgements of large groups can hide very big individual fluctuations of value judgements that by counterbalancing one another result in stable group indicators. Such studies also fail to provide a clear answer to the question whether and to what degree respondents express their actual value notions and to what extent their responses are furtuitous or even intentionally false. Longitudinal data enable to consider these problems more thoroughly as this research strategy makes it possible to analyse the individual stability of value judgements over a long period.

 

2. Theoretical comments

The conceptual system of value theory used in our studies has more thoroughly been considered in several articles (Saarniit, 1995a & 1995b). Here it should be mentioned that as an elementary unit of individual’s value consciousness we employ value notion that can be defined as a single notion, opinion, conviction etc determining the intensity of subjective meaning of single value at one definite reference scale. The external equivalent of value notion is value judgement that has three main attributes like value notion: value to which the judgement and respective notion are directed (e.g. affluence, love, religion etc), evaluation scale (e.g. importance, attractiveness, significance, etc) and the intensity of judgement-notion (e.g. very essential, unpleasant, rather important, etc.).

As a rule there are many value notions in individual’s consciousness about all concerning values that differ from one another by estimation criteria and intensity. For example, affluence may be very pleasant, rather essential and quite important.

Value notions are not "placed" irregularly in consciousness but they form certain psychologically connected subgroups or cognitive substructures that in our studies are considered as value orientations. Thus, value orientation can be defined as a rather independent subsystem of certain value notions between which exists certain cognitive relation and that functions as a relatively integral whole in regulating individual’s behaviour. Each value orientation may include the value notions about several values.

The concept of value orientation is very close to that of social attitude and quite often it is defined as an attitude toward (social) values. That is why psychological theories of attitude change, particularly cognitive approaches, can largely be applied to the analyses of value orientations.

One of the most appropriate models, so called assimilation-contrast theory, that also explains the results of factor analysis of value judgements, was worked out by Sherif and Hovland (Insko 1967). On the basis of this theory estimation scales linked to concrete value orientation can be treated as a bundle of vectors. In this case factor can be regarded as a general vector describing original vectors. Value orientation of each respondent or value notions cognitively closely linked between themselves can be depicted as a rubber ring anchored to the bundle of vectors at certain site to each single vector. The rubber ring and its larger or smaller elasticity referres that single value notion belonging to the value orientation can to some extent be influenced, but in case of strong impact to one or several individual notions the whole ring has sooner or later begin to move, i.e. the intensity of all value notions belonging to the given value orientation starts to change.

3. Empirical basis

This article is based on the Estonian data collected in the framework of the study "Paths of a Generation: A Comparative Longitudinal Study of Young Adults in the Former Soviet Union". This is an international comparative longitudinal research project of secondary school leavers of 1983 started under the leadership of professor Mikk Titma in eight that time Soviet republics (see more Titma & Tuma 1995).

In Estonia the first three waves of the research were carried out jointly by Professor Mikk Titma, the department of sociology of Tartu University and the current Institute of International and Social Studies of the Tallinn Pedagogical University. The third stage was supported by the Jacobs Foundation. According to a stratified random sample 3,398 general, specialised and vocational secondary school graduating students completed a questionnaire in late 1982 and early 1983. During the second stage in 1987 data were gathered from 2,178 respondents by the same method. The third survey was conducted in 1993, when 2,128 respondents were interviewed. The data of all three waves have been individually joined together as foreseen by the methodology of longitudinal research.

Keeping in view the goal of this article it is necessary to emphasise that during the third stage when the data were mainly gathered by personal interviews, responses to items concerning values were written by the respondents themselves as during the previous surveys. This enabled to avoid possible auditive and visual side effects on the responses and guaranteed maximum comparability of the results with those obtained during the first and second wave.

Individual changes in value judgements can be studied only on the basis of longitudinal data in which estimations gathered from the same respondents at various points of time are personally combined in the data set. The project as a whole has measured different value orientations. In this article we analyse only individual changes in work orientations. Our approach is mostly methodological, but also includes some substantial aspects. The whole analysis is conducted on the basis of so called fully longitudinal respondents or those from whom data were gathered in all three surveys. There are altogether 1,783 such respondents. The data have been processed by the SPSS for Windows program.

 

4. Methods of measuring work orientations and the structure of data

The methods worked out on the basis of Morris Rosenberg’s work have been employed in all three surveys when respondents were asked to estimate 16 work-related values on a 4-point scale. A fragment of this method is portrayed in Table 1 and the whole list with 16 work values is presented in Table 2.

TABLE 1. A Fragment of Methods of Measuring Work Values.

IN YOUR OPINION, WHICH REQUIREMENTS SHOULD WORK SATISFY?

WORK SHOULD...

Very important

Rather important

Not too important

Unimportant

1.

Enable regular self-improvement

4

3

2

1

2.

Guarantee good earnings

4

3

2

1

...

....................................................

....

....

....

....

16.

Enable to use one’s abilities

4

3

2

1

 

 

 

 

TABLE 2. The indicators of work values and the names of corresponding variables in the data file

Formulation and order of the indicators of work orientations in the questionnaire

Names of variables in the data file

Work should:

stage I variables

stage II variables

stage III variables

Enable regular self-improvement

Guarantee good earnings

Help to win friends’ recognition

Enable a peaceful and secure life

Enable to achieve a high position in society

Enable to socialise

Be useful to society

Enable professional advancement

Be clean and physically easy

Enable sustained growth of professional knowledge and abilities

Provide a possibility for working in a pleasant field

Enable to see the results of one’s work

Enable some freedom in using work time

Be useful to people

Provide a possibility for being creative

Enable to use one’s abilities

T1ARE

T1PAL

T1TUN

T1ELU

T1POS

T1SUH

T1KAS

T1TOU

T1PUH

T1ERI

T1ALA

T1RES

T1VAB

T1INI

T1LOO

T1VOI

T2ARE

T2PAL

T2TUN

T2ELU

T2POS

T2SUH

T2KAS

T2TOU

T2PUH

T2ERI

T2ALA

T2RES

T2VAB

T2INI

T2LOO

T2VOI

T3ARE

T3PAL

T3TUN

T3ELU

T3POS

T3SUH

T3KAS

T3TOU

T3PUH

T3ERI

T3ALA

T3RES

T3VAB

T3INI

T3LOO

T3VOI

 

The results of the measurements of all stages have been saved in SPSS for Windows 6.0. standard format and individually joined together. The names of the work orientation variables have been defined in one system (this should simplify the analysis and the presentation of its results). The names of all variables begin with T followed by one-digit figure marking the research wave (1-3). This number is followed by a 3-letter abbreviation that makes difference between the different work values. As a result there are 48 initial variables in the data file - 16 per each survey.

 

5. Methods of the analysis and the results.

5. 1. The formation of increment variables

One of the simplest ways of analysing individual changes in value judgements is to compute new variables describing interstage changes of estimations that can be called an incremental variable and the values of this variable - individual changes of value judgements. There are several possibilities for this, but in case of a three- wave study it would be most appropriate to form the variables so that they enable to analyse to what degree the second stage estimations differ from those given at the first stage and the third stage estimations from those given at the second stage. If the initial variables are saved under such names as presented in Table 2, a typical command for forming an incremental variable reflecting changes in individual estimations between the second and first stage would look like (presented on the basis of the first indicator in Table 2) as following:

COMPUTE T21ARE = T2ARE - T1ARE.

The given command forms a new variable T21ARE. Figures 21 in variable’s name mark that that we have to do with a variable reflecting the difference between the second and first stage estimations. In the rest the name is analogous with the names of the initial variables. Each individual value of the new variable indicates the difference between the second and first stage estimations given by a concrete respondent. The positive value of the variable indicates to what extent respondent’s orientation toward a given value has risen between two research waves and negative value indicates the decrease of orientation.

As the orientation was measured by a 4-point scale the values of the increment variable vary from -3 (if the given respondent estimated the significance of this value as minimum in the first survey and as maximum in the second survey). to +3 (maximum estimation in the first and minimum estimation in the second survey).

As an example Table 3 portrays the percentage division of the new variable - T21ARE .

TABLE 3. An example of individual inter- stage changes of value judgements.

Variab-

le’s

name

The scale and percentage division of incremental variable

Mean and its standard error at the level p=.05

N

-3.00

-2.00

-1.00

.00

1.00

2.00

3.00

T21ARE

.2

1.8

21.7

50.3

23.3

2.6

.1

.03 ± .02

1715

The division suggests that the work value "enable sustained self-improvement" was estimated by 50.3% of the respondents during the second stage of the research as highly as in the base- year survey. More than two fifths (21.7%+23.3%) estimated it one point lower or higher and only the estimations of a very small part of the respondents (4.7%) differed more than by one point. The arithmetic mean of differences suggests that as average the given value was estimated by 0.03 points higher in the second survey.

Considering that the respondents had to estimate 16 different values it is necessary to form a separate increment variable (T21... type) for all values. As a result we obtain 16 new variables that reflect the differences between the first and second stage surveys that differ from one another by the last three letters in variable’s name.

The same course of action is necessary in computing variables reflecting differences between the third and second stage estimations. The following command was employed for the latter case:

COMPUTE T32ARE = T3ARE - T2ARE .

If we were interested in differences between the third and first stage differences in estimations, then it is possible to form corresponding variables in the same way. The type command would look like as follows:

COMPUTE T31ARE = T3ARE - T1ARE .

It would be wise to form such variables if one is interested only in those summary changes that have occurred in value judgements between the first and third survey.

In this analysis we consider only incremental variables between the first and second and second and third stage. As the number of estimated work values was 16 in all surveys we obtain in total 32 new variables. Half of them (all T21incremental variables) reflect the difference between the arithmetic means of estimations between the second and first stage and the second half (all T32 incremental variables) the difference between the third and second stage estimations. In summary, these variables fix individual changes in estimations over a period between the surveys, that were four and six years respectively between the stages of this longitudinal study.

 

5.2. The analysis of the stability of value judgements on the basis of incremental variables' absolute values.

In analysing the temporal stability of value judgements we are not interested so much in the direction of changes than in their absolute value. For analysing stability it would be wise to form additional variables that reflect changes in absolute values between the surveys. These can be regarded as variables of absolute increment and can be formed either by a command based on the initial variable:

COMPUTE T21AREA = ABS(T2ARE - T1ARE),

or on the basis of formerly formed incremental variable by the command:

COMPUTE T21AREA = ABS(T21ARE).

The result is in both cases the same. Additional letter A in the name of new variable has been used for differentiating them from initial increment variables. The total number of the variables of absolute increment is 32. Half of them (T21...A variables) describe absolute increments between the first and second wave and the second half - between second and third stage (T21...A variables).

Table 4 presents data on absolute increments separately for changes between the first and second stage of study (part A of the table) and between the second and third stage (part B of the table).

In part A it can be seen that on average 48% of the responses coincide on the scale during the first and second survey in the case of all 16 work values. In the case of individual values the coincidence is quite even, fluctuating between 40.9 (see T21PUHA) and 55.4% (see T21RESA). In addition to large absolute coincidence the responses of 40.9% of the subjects differ only by one point. Thus, to express their orientation toward work values, 92.4% of the respondents used at the second stage of study the same or neighbouring scale point than at the first stage. The estimations of only 7.6% of the respondents differ more than by one point.

On average the absolute coincidence of the second and third stage estimations is even higher - 50.9% (see the part B of Table 4). Accordingly, the minimum (see T32PUHA - 44.7%) and maximum (T32ALAA - 58%) values of the absolute coincidence of individual value judgements are also somewhat higher than between the first and second stage of study.

TABLE 4 Percentage division of absolute increments:

A. Absolute increments between the first and second surveys (%)

Variable

0

1

2

3

N

T21AREA

T21PALA

T21TUNA

T21ELUA

T21POSA

T21SUHA

T21KASA

T21TOUA

T21PUHA

T21ERIA

T21ALAA

T21RESA

T21VABA

T21INIA

T21LOOA

T21VOIA

50.3

52.4

41.0

45.1

40.9

52.4

49.6

42.2

40.9

49.8

55.4

51.3

44.1

52.7

47.3

53.3

45.0

44.1

46.3

46.4

47.5

42.7

43.4

47.6

45.1

44.6

40.3

42.8

44.0

42.5

43.9

42.5

4.4

3.3

11.8

7.7

10.2

4.2

6.6

9.4

12.2

5.2

3.3

5.4

10.3

4.4

7.6

3.8

.3

.2

.9

.8

1.4

.6

.3

.7

1.9

.4

1.0

.5

1.6

.3

1.2

.3

1715

1724

1721

1727

1725

1720

1717

1717

1717

1720

1719

1712

1711

1711

1708

1716

Means of the columns

48.0

44.4

7

1

1

B. Absolute increments between the second and third survey (%)

Variable

0

1

2

3

N

T32AREA

T32PALA

T32TUNA

T32ELUA

T32POSA

T32SUHA

T32KASA

T32TOUA

T32PUHA

T32ERIA

T32ALAA

T32RESA

T32VABA

T32INIA

T32LOOA

T32VOIA

54.2

57.7

44.7

45.2

50.4

52.8

46.1

49.4

44.7

52.6

58.3

54.8

46.2

52.7

47.4

57.4

41.7

39.9

44.2

45.6

42.6

42.5

45.7

42.9

43.0

43.2

38.1

40.8

42.7

43.0

43.7

39.7

3.6

2.2

9.9

7.6

6.3

4.1

7.7

6.9

11.4

3.9

3.2

3.7

9.9

3.9

8.0

2.6

.5

.2

1.3

1.7

.7

.6

.6

.7

.9

.3

.5

.7

1.2

.4

.9

.3

1739

1749

1745

1747

1747

1743

1743

1743

1743

1748

1741

1740

1739

1738

1737

1745

Means of the columns

50.9

42.5

6.1

0.9

100%

All these indicators suggest that adults’ individual value judgements are very stable and change very little over the years. Simple calculation shows that if the estimations on the 4-point scale were given totally by accident in the two comparable surveys, the probability of total coincidence of responses would be 0.16, only. This means that the total coincidence of responses at two stages should have been only 6...7% and not nearly 50% as shown by our empirical data. Thus, actual coincidence of responses is nearly seven times higher than it should be in case of accidental answers of respondents. Here should be added those 42 - 43 % of respondents whose responses differ only by one point.

It is necessary to underline that the second and third survey were conducted in conditions where the Soviet social and institutional system was disintegrating and was already being replaced by a quite new - capitalist - social order and respective value system.

 

5.3. Analysis of individual summaries of inter- stage absolute increments.

The next question we can ask in this analysis is how big are the individual inter-stage changes of estimations along the whole list or work values. It may be that some of the respondents have changed their estimations strongly toward all work values but some of them differ very little in different waves of study.

To answer this question we will compute a new more general variable as a sum of individual absolute increments along the whole block of work values. In case of the first and second stage differences, the following command can be employed for this variable:

COMPUTE T21SUM =ABS(t21are) +ABS(t21pal) +ABS(t21tun) +ABS(t21elu) +ABS(t21pos) +ABS(t21suh) +ABS(t21kas) +ABS(t21tou) +ABS(t21puh) +ABS(t21eri) +ABS(t21ala) +ABS(t21res) +ABS(t21vab) +ABS(t21ini) +ABS(t21loo) +ABS(t21voi).

In summarising the changes between the second and third survey increments the command is analogical:

COMPUTE T32SUM =ABS(t32are)+........ +ABS(t32voi).

The divisions of these two new summary variables are depicted in Table 5. It should be noticed that the number of respondents in the case of these variables is considerably smaller than in case of former variables. This is due to the fact that in computing the given variable only those respondents were considered who had estimated all 16 values in both compared surveys.

It should be considered that in evaluating the stability of value judgements within the whole extent of the table of work values the maximum theoretical sum of the increments is 48 (if all values received maximum estimations at one stage and minimum estimations at the other compared stage). But as can be seen in Table 5, actual values of summary variables are between 0 and 28.

TABLE 5 The percentage divisions of absolute summary values of changes in estimations of work values.

Scale of sum of differences

 

T21TOTAL

(division of second and first stage summary variable changes)

T32TOTAL

(division of third and second stage summary variable changes

N

%

Cumulative %

N

%

Cumulative %

0

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

unanswered

1

3

8

24

48

84

132

164

170

184

188

147

112

98

71

50

34

20

10

6

8

3

3

3

0

1

1

0

1

209

.1

.2

.5

1.5

3.0

5.3

8.4

10.4

10.8

11.7

11.9

9.3

7.1

6.2

4.5

3.2

2.2

1.3

.6

.4

.5

.2

.2

.2

.0

.1

.1

.0

.1

---

.1

.3

.8

2.3

5.3

10.7

19.1

29.5

40.3

52.0

63.9

73.3

80.4

86.6

91.1

94.3

96.4

97.7

98.3

98.7

99.2

99.4

99.6

99.8

99.9

99.9

100.0

0

8

15

41

72

117

145

178

190

196

165

150

109

71

51

42

25

12

9

7

5

4

5

4

1

1

1

0

0

159

.0

.5

.9

2.5

4.4

7.2

8.9

11.0

11.7

12.1

10.2

9.2

6.7

4.4

3.1

2.6

1.5

.7

.6

.4

.3

.2

.3

.2

.1

.1

.1

.0

.0

---

 

.5

1.4

3.9

8.4

15.6

24.5

35.5

47.2

59.2

69.4

78.6

85.3

89.7

92.9

95.4

97.0

97.7

98.3

98.7

99.0

99.3

99.6

99.8

99.9

99.9

100.0

TOTAL

1783

100%

 

1783

100%

 

Mean & its standard error

 

9.61 ±.09

 

 

 

 

9.03 ±.09

 

 

In Table 5 zero value means that the respondent evaluated all work values in the same way at both research stages. The divisions reveal that in comparison of the first and second stage there was one such respondent while in the second and third stage there was none.

Value "1" suggests that in the whole table there was one 1 point difference in estimation of one value. Value 2 can express either one- point difference in estimating two values or a 2-point difference in estimating one value, etc. Shortly - the bigger the value of summary variable, the more combinations of individual differences there can be. These combinations can be analysed in detail and undoubtedly they are interesting from the aspect of individual stability of value judgements, but in this article we present only a general picture.

Cumulative percentage divisions in Table 5 suggest that there were very few respondents whose sum of absolute increments exceeds 16 (i.e. as average one point per each value). In comparing the first and second survey there were 3.6 % and the second and third stage 3.0 percent of such respondents. At the same time there were respectively 40.3% and 47.2% those whose sum of differences was 8 or less points (see cumulative divisions in Table 5). The average sum of differences was 9.61±0.09 points between the first and second survey and 9.03 ±0.09 points between the second and third surveys, i.e. 0.60 and 0.56 points as average per each included value. The difference is statistically significant at the level of 0.05.

 

5.4. The differentiation of the stability of value judgements

The next logical step after the clarification of the general picture of individual stability of value judgements would be to ascertain if and to what degree this depends on respondents’ main socio-demographic variables. For this we analyse to what extent the sum of absolute increments (variables T12SUM and T32SUM) is related to four groups of variables:

5.4.1. Educational characteristics and the stability of value judgements

The analysis revealed that most of all the stability of value judgements is related to the educational characteristics of respondents as well as to the educational characteristics of their parents. It means that the sum of absolute increments between the surveys is differentiated significantly by all education-related variables mentioned above.

As an example of interrelations between educational characteristics and stability of value judgements, Figure 1 depicts the connection between academic performance in the last grade or course of the institution of secondary education and the average sum of absolute increments between the first and second surveys. The confidence intervals of the means at the confidence level of 95% are presented in this figure also. We see that the dependence is almost linear: the higher the secondary school academic performance, the smaller the average absolute increment in value judgements. This means that good academic performance is succeeded by stable value judgements in 1983-1987. It is noteworthy that there are big statistically significant differences in the individual stability of value judgements between the students with highest and lowest academic performance. Almost the same correlation can be noted between the 8th grade performance and the absolute increments of value judgements between the first and second surveys. This is quite understandable, as the academic performance tends to be rather stable during the whole educational path of young people.

It should be stressed that the absolute summary increments between the second and third survey also depend significantly on grades in both basic and secondary school. The only difference is in smaller linearity of relationship as can be seen in Figure 1. Thus it can be concluded that during the ten post-secondary school years the stability of value judgements of more successful youths are far more stable than that of their less successful peers.

FIGURE 1. The dependence of the sum of absolute increments between the first and second survey on the average grade at the graduation from the institution of secondary education.

(Arithmetic means of the sum and their confidence range at the confidence level of 95%)

Grades’ codes:

1 less than 3.50

2 3.51-4.00

3 4.01-4.50

4 4.51-4.99

5 5.00

 

Figure 2 portrays data about the stability of value judgements depending on the type of secondary education the respondent attained. It should be marked that the division of the secondary school population between the general, specialised and vocational schools largely depends on students’ academic performance at basic school. As a rule better students continued their studies at general secondary schools and the weakest ones at vocational schools. Therefore, the Table indirectly reflects the impact of academic performance on the stability of value judgements.

We see that the lowest average of absolute increments or the highest stability of value judgements characterises the graduates of the special grades of general secondary and non-technical specialised secondary schools. Average stability is typical of the graduates of ordinary grades of general secondary and agricultural specialised secondary schools. Most extensive changes have undergone the value judgements of those young people who attended vocational or industrial specialised secondary schools. This proves that differences in students' value judgements are not directly related to the type of secondary education (general, specialised and vocational education), but are largely determined by the subtypes of secondary education.

FIGURE 2. The dependence of the sum of absolute increments of value judgements between the first and second stage on the type of secondary education

(arithmetic means are confident at the level of 95%)

  1. Rural vocational school
  2. Urban vocational school
  3. Agricultural specialised secondary school
  4. Technical specialised secondary school
  5. Non-technical (cultural, medical, pedagogical) specialised secondary school
  6. Ordinary grade of general secondary school
  7. Special grade of general secondary school (from grade 9)
  8. Special grade of general secondary school (from grade 1 or 2)

It should be added that the value hierarchies of the graduates of non-technical specialised secondary schools and special grades of ordinary secondary schools are more similar than those of all graduates of either of these basic forms of secondary education. The relationship between the increments of value judgements between the second and third survey and type of education is very much the same. There is only one noteworthy difference. Average increments in Table 5 indicate that the absolute increments between the second and third survey are in most school types more moderate than those between the first and second stage. The only exception concerns the graduates of urban vocational schools whose mean increments have not significantly fallen and their value judgements have changed more than those of any other group of secondary school graduates between the second and third survey.

Educational attainment by the third stage survey in 1993 points even more clearly to the connection between the educational factors and stability of value judgements (see Figure 3).

Figures here apply to the period between the second and third stage. The reason is that at the time of the second survey in 1987 many respondents had not yet completed their educational path and the impact of educational attainment by the third survey becomes more implicit after the completion of educational career. The graph of first and second stage increments differs only somewhat from that depicted in Figure 3. As well as in the graph of school performance (Figure 1), Figure 3 demonstrates an almost linear dependence of value judgements on the educational attainment - the higher the educational attainment, the less have value judgements changed between the considered surveys.

Figure 3. The relationship between the sum of absolute increments between the second and third survey and the educational attainment by the third survey (arithmetic means of sums and their confidence interval at confidence level of 95%)

  1. vocational education after grade 8
  2. vocational education after secondary school
  3. general secondary education
  4. specialised secondary education after grade 8
  5. specialised secondary education after secondary school
  6. incomplete higher education
  7. higher education

5.4.2 Other factors

The relationship of some other variables considered in section 5.3 with the individual stability of value judgements is not so clear as that of educational factors. For instance, there is almost no correlation between the material situation and the stability of value judgements. Of socio-demographic and social background variables some of them correlate with the stability of value judgements, while the others do not. There are only minor differences between the stability of men’s and women’s value judgements. At the same time the indicators of the graduates of Estonian and Russian-language schools differ significantly. The sum of absolute increments of the graduates of Russian schools is 10.46±0.26 between the first and second survey and 11.01±25 between the second and third stage. The same indicators in the case of Estonian schools are 9.45±10 and 8.66±0.09 respectively. More intensive changes of the value judgements of non-Estonian youths are confirmed by several inter- and intragenerational group- statistical comparisons (Saarniit, 1997).

The analysis of relationship between the first and second stage changes and the place of the attainment of secondary education yielded somewhat surprising results. The place of the attainment of secondary education was fixed on a five-point scale (Tallinn, Tartu, other city with 10,000-100,000 inhabitants, small town, rural community). With the exception of Tartu the mean of average absolute increments was 9.7 - 9.8. Only in case of respondents who attained their secondary education in Tartu it was statistically significantly lower - 8.6.

Differences having occurred between the second and third stage were quite the same, only statistically significant difference between those who attained their education in Tartu and in smaller cities had disappeared. On the other hand, differences between those who attended secondary schools in Tartu and Tallinn, had increased. The relationship between respondents’ parental socio-professional background (16-point scale) and the stability of their value judgements is weak. Somewhat more intensive changes characterise those whose parents belong to the military or security workers. But even these differences can be found only in the absolute increments between the second and third survey. Far more significant is the impact of parental education on the stability of respondents’ value judgements. There is a general trend: the more educated the parents are, the more stable are respondents’ value judgements. Differences between the children whose parents have higher or elementary education, are statistically significant. However, this relationship is not as linear as that between respondents’ own educational attainment and the stability of their value judgements. The impact of parental education on youths’ value judgements is probably indirect: the correlation between parents’ education and their children’s academic performance and educational attainment is well-known.

 

6. Some conclusions and hypotheses

In this article we did not delve into the details of the individual changes in value judgements. Our chief goal was more to present research facts than their comprehensive interpretation. However, the considered data enable to draw some important conclusions that might be of interest to scholars studying values.

  1. The value judgements of young people who attained secondary education during the Soviet period have been over the next ten years very stable even in rapidly changing social conditions. In comparison with the base-year measurement, responses obtained four or even ten years later on a 4-point scale coincide to the extent of 50 percent, while 45 percent of the responses fall to the neighbouring scale point.
  2. High individual stability is not typical only of estimating single values, but characterises the assessment of all work values as an integral whole. This is confirmed by the fact that the absolute increment of value judgements of 40-50 percent of the respondents is only 0.5 points as average per each value and in case of 95% this is no more than 1 point.
  3. Thus, this longitudinal study does not confirm a doubt that value judgements given in mass surveys are fortuitous and unstable. Our findings also do not confirm the assumption that the results of Soviet time studies of value judgements cannot be trusted because respondents were afraid to express their real opinions. If it were so we should presume that more than 1,700 of our respondents had to remember how they 10 years ago, in 1983 "lied". A conclusion corresponding to research results is a different one: an overwhelming majority of the respondents expressed their true value ideas during the Soviet period as well as in conditions of re-independent Estonia. The general trends in changes of value judgements reflect actual shifts in the value consciousness of the educational cohort of 1983.
  4. The individual stability of respondents’ value judgements is closely linked with parameters characterising their studies and educational paths. There is a general tendency: the more successful one is at school and the higher the educational attainment, the more stable are his or her value judgements. The relationship of other socio-demographic indicators and material situation with the stability of value judgements is quite weak or absent altogether. The stability of value judgements of Estonians is significantly higher than that of non-Estonians. Another noteworthy finding is that those respondents who attained secondary education in Tartu have most stable value judgements.
  5. Close relationship between the stability of value judgements and educational indicators enables to advance at least three hypotheses justifying further analysis.

The first one of these could be regarded as a hypothesis of school obedience. According to this, educationally more successful young people preserve their value orientations acquired in the Soviet secondary school due to their bigger school obedience. One part of obedience could have been a more profound mastering of official Soviet value system.

The second one can be considered as general erudition according to which good school grades and attained education are the expression of deep and substantial knowledge. Therefore the acquired value picture is connected with wide (incl. social) knowledge and as a result it is more integrated and stable.

There can also be advanced so- called methods’ hypothesis according to which academically less successful respondents may sometimes misunderstand questionnaire items and give fortuitous responses.

References

Inglehart, R. (1990). Culture Shift in Advanced Industrial Society. Princeton University Press.

Insco, C.A. (1967). Theories of Attitude Change. New York

Saarniit, J. (1995a). Changes in the Value Orientations of Youth and Their Social Context. In: L.Tomasi.(ed) Values and post-Soviet youth. Milano. 141-153

Saarniit, J. (1995b). The use of factor analysis in studying the structure of value consciousness (in Estonian). In: Eesti statistika treabevihik 6, Tartu. 160-170

Saarniit, J.(1997) The specifics of non-Estonian youth (in Estonian). P.Järve (ed). Russian youth in Estonia:a sociological mosaic.Tartu University. 69-82

Titma, M. and Tuma, N. (1995). Path of a Generation: a comparative longitudinal study of young adults in the former Soviet Union. Stanford - Tallinn.