SALVATION AND RELIGIOSITY The Predictive Strength and Limitations of a Rokeach Value Survey Item

In this study we explore the possibilities and limitations of using the Salvation item of the Rokeach Value Survey (RVS) as a global indicator of religiosity. Our data come from the data collection connected to the 4th wave of the European Social Survey (2008; N = 1144). First we compare the Salvation item with four global indicators of religiosity. In the second phase of analysis we examine the relationship between the overall indicators of religiosity – including, especially, the RVS Salvation item – and two „classic“ aspects of religiosity, the ideological dimension (beliefs) and the consequential dimension (religious behaviour). In the third step, we analyse the similarities and differences of the behaviour of Salvation and the other indicators as a function of socio-demographic variables (gender, age, level of education, domicile, denominational affiliation) with ANOVA. On the whole, the Salvation item of the RVS is a somewhat weaker indicator of religiosity than the other global indicators. On the one hand, it seems to be a stricter measure than those, and, on the other hand, it is less suitable to identify non-traditional forms of religiosity. With these restrictions, however, it can be considered an adequate indicator of religiosity under the social conditions of Hungary in the twenty-first century.


Introduction
After Durkheim's and Weber's pioneering explorations, religion began being studied as a multidimensional phenomenon from the 1960s on, especially in the wake of Gerhard E. LENSKI's (1961), and Charles Y. GLOCK and Rodney STARK's work (1965, 1966;STARK & GLOCK 1968).This resulted in a breakthrough in the history of sociological research.Glock and Stark developed a model of five different dimensions to assess religious commitment: (1) ideological dimension (i.e.religious beliefs), (2) ritual dimension (including religious practice), (3) experiential dimension (religious sentiments), (4) intellectual dimension (religious knowledge) and (5) consequential dimension (e.g. the impact of religion on lifestyle and moral perception, that is, religious behaviour).This model mirrors the internal complexity of religiosity, but it does not solve the problem of the measurability of various dimensions, nor is it capable of capturing different types of religiosity (FÖLDVÁRI & ROSTA 1998).Following the mid-1960s, the recognition that religiosity cannot be grasped with the help of a single indicator and several independent variables are needed to describe it adequately became a widely held view exerting profound influence on empirical research.
This enrichment of theoretical approaches and analytical tools did not, however, eradicate a desire to interpret religiosity as a unified concept.This can be seen not only in the fact that, already in the 1980s, Mady A. THUNG and Leo LAEYENDECKER (1985) began to reduce the number of religious dimensions to be studied but also in the consensus-building power of the religiosity types developed by Yves LAMBERT (1994;LAMBERT & MICHELET 1992).His threefold system of categories -confessional Christianity, cultural Christianity and secular humanism -is widely accepted today. 1 In his highly influential essay on Hungary's religious landscape (1999), Miklós TOMKA presented a multidimensional approach, but the five-point scale he had developed for the measuring of religiosity (1998) was also an essential part of his work.Based on the self-classification of respondents, the scale differentiates between people (1) who are religious according to the tenets of the church, (2) who are religious in their own way, (3) who cannot decide, (4) who are not religious, and (5) who are resolutely non-religious.In his analysis mobilising a serious mathematical apparatus and examining various dimensions, János SZÁNTÓ (1998) also summarises the results in terms of a few basic types.
Religiosity is thus interpreted as a combination of several irreducible components.Yet at the same time, by identifying a few typical forms of religiosity, the sociology of religion seeks to give a manageably simple description of this extremely complex phenomenon.It is not afraid of developing categories that seem, on the basis of a quantitative criterion, to be linearly arrangeable and thus lead back to a simple one-dimensional scale to measure religiosity.This is, however, by no means to be considered a self-contradiction, for the simplifying step of model building does not disregard but, on the contrary, utilises the lessons learnt from the multi-dimensional analysis.From this hermeneutical stance, with a 'post-critical naivety' we may find a connection to the Salvation item of the Rokeach Value Survey as an indicator of religiosity.
Milton ROKEACH developed his value survey nearly fifty years ago (1969a; 1973), 2 and it was, within a short period of time, adapted to the Hungarian circumstances by Elemér HANKISS (1977;HANKISS et al. 1983).The Rokeach Value Survey (RVS), based on the ranking of eighteen terminal and eighteen instrumental values, has widely remained in use in the twenty-first century.But it has often been criticised, for example, for being 'too general . . .and abstract, since individual values are thought to be attached to specific things and not to overall values' (FÜSTÖS & TIBORI 2011, 46, our trans.). 3 A further deficit associated with the survey is that it does not accurately define individual items (e.g.Self-respect, cf. FÜSTÖS & SZAKOLCZAI 1994, 62).Methodological criticism has centred on difficulties inherent in the ranking process such as reproducibility or the lack of a quantifiable distance of importance between the items.'Although many have criticised the Rokeach test, it has nonetheless proved to be the most usable' (FÜSTÖS & SZAKOLCZAI 1994, 57, our trans.). 4t is a little-known fact today that one of the central issues in Rokeach's early research was the examination of values in connection with religiosity and that his questions and results stirred up a considerable scholarly debate in the US at the turn of the 1960s and 1970s. 5In his study, ROKEACH (1969b) analysed the similarities and differences of the value systems of four groups: Protestants, Catholics, Jews and nonreligious people.He found significant differences between the groups in the case of eighteen of the thirty-six values examined.Of the eighteen terminal values, Salvation EJMH 12:1, June 2017 2 For a review of its early reception history, see VÁRINÉ SZILÁGYI (1987) 150-71. 3 In the original: 'túl általános . . .és elvont, mivel az egyéni értékekről úgy tartják, hogy konkrét dolgokhoz kapcsolódnak nem pedig átfogó értékekhez'.
showed the biggest difference: on average, Protestants ranked it fourth, Catholics thirteenth, and Jews as well as non-religious people put it in the last place.Salvation, at the same time, also moved together with other indicators of religiosity that were independent of the RVS.In terms of the frequency of church attendance, the results were unambiguous.A strong link was displayed between participation at religious events and the ranking of Salvation (weekly: third place; almost weekly: fifth; 2-3 times a month: tenth; once a month: twelfth; a couple of times a year: thirteenth; 1-2 times a year: sixteenth; never: eighteenth place).Among regular church-goers of different Christian denominations, the difference in ranking remained the same.While those belonging to Protestant denominations ranked Salvation first to third place, Catholics and Anglicans placed it tenth and thirteenth, respectively.Another approach to religiosity ('How important is religion for you in everyday life?') led to similar results.Among those who deemed religion important, salvation finished in the first place, while in the case of medium or low importance, it became sixteenth and eighteenth.For all three religiosity criteria, Rokeach's survey showed significant differences between the value systems of religious and non-(or less) religious respondents.The discrepancy was most consistently manifest in the case of the Salvation item.
Other surveys also found correlations between the Salvation item of the RVS and other indicators of religiosity.E.D. TATE and G.R. MILLER (1971) examined the value systems of the members of a Protestant denomination on the basis of ALLPORT and ROSS's (1967) categories of religious orientation.Salvation was ranked most important by those with an intrinsic religious orientation, and least important by those with an anti-religious orientation.In the median rankings, the latter placed the value ninth, and those who belonged to the other three groups, first or second.W.F. RUSHBY and J.C. THRUSH, in their survey examining the social sensitivity of Mennonite students, used the Salvation item of the RVS as a means of measuring religious orthodoxy (1973).Mennonite students ranked this value first, while a control group (students from a state university) ranked it last.The authors found the place of Salvation in the rankings an appropriate indicator of religiosity as long as that was narrowed down to traditional Christian religiosity.
The restriction already points to some limitations of the approach, and there were indeed cautionary results.Bernard SPILKA called attention to the complexity of the concept of salvation as early as 1970, and also to the fact that the ranking of this terminal value may be incidental in the case of persons with different religious dispositions.The item cannot capture the colourfulness of religion (not to mention the shades of the various colours), but, according to Spilka, its application cannot be completely ruled out.Examining Unitarian Universalists, R.L.H. MILLER (1976) found that -unlike members of other denominations -they consistently ranked Salvation last.Salvation, in his opinion, can by no means be considered a measure of religiosity in the case of that denomination.
Both theoretical considerations and empirical data thus give us pause and a warning signal that the Salvation item of the RVS cannot capture the full complexity of religiousness.That does not, however, render it altogether useless, for other data clearly point in that direction.It correlates with other important aspects of religiosity and, in the case of certain populations and certain types of religiosity, it not only functions as an indicator of its own special dimension (salvation orientation) but also has a predictive value to assess the overall religiousness of respondents -at least in the sense of basic types arrangeable along a quantitative scale.Through a secondary analysis of data from a national survey, we explore the possibilities and limitations of using the RVS Salvation item as a global indicator of religiosity in early twentyfirst-century Hungarian society.

Sample and methods
The data we analyse were collected in connection with the 4 th wave (2008) of the European Social Survey (ESS). 6The sample is representative of the population of Hungary.The number of respondents was 3,002, but some subsets of questions were administered to a smaller sub-sample (still meeting the requirements of representativeness).Our results are thus largely based on answers from a sample of 1,144 persons.The questionnaires were administered by interviewers, and answers to the questions included in our analysis were based on the self-classification of respondents.SPSS 21.0 software was used in data analysis.
The original survey reflects developments in the sociology of religion over the past few decades in a limited way.Detailed questions related to different dimensions of religion that may be derived from various theoretical considerations are mostly, although by no means entirely, absent from the questionnaire.Only a few of those appear as coordinated subsets of questions.There are, however, several questions which attempt to capture the religiosity of respondents globally.In the first part of our analysis we compared four of those with the Salvation item of the RVS (R1_18)7 : 1. 'How religious do you consider yourself?' (VAL1) -Respondents classified themselves with the help of cards.Answer categories included 1 = I am markedly anti-religious; an atheist. 2 = I am non-religious. 3 = To a certain degree, perhaps; religion has some role in my life.4 = I consider myself religious, but I am not a regular church-goer.5 = I am a regular church-goer.-We refer to this item as 'categorical self-classification'8 in order to differentiate it from the following question: SALVATION AND RELIGIOSITY

7
EJMH 12:1, June 2017 We thank László Füstös for providing access to the original data for secondary analysis.
2. 'Regardless of whether you belong to any church or denomination, how much of a believer or how religious do you consider yourself?' (VA4) -The response was given on the basis of an eleven-point Likert scale.A score between 0-10 indicates how religious respondents consider themselves: 0 = Not religious at all. . . . 10 = Very religious.3. Interpreting each item of a thirty-question subset as a separate question, respond ents could decide how important the given concept or attitude was for them.One of the variables was 'Religion, faith' (KV1_9).-The response was given on an eleven-point Likert scale: 0 = Not important at all. . . . 10 = Very important.4. Interpreting each item of a six-question subset as a separate question, respondents could decide how much influence a given factor had on their fate.One of the variables was 'I am in the hands of God; God 9 directs, God decides' (SZ2_1).-The response was given on a five-point Likert scale: 1 = Not at all. . . . 5 = To a very great extent.First, then, we examined the RVS Salvation item in the context of these four global indicators of religiosity.In order to standardise the comparison, we divided the values of the variables into three categories: religious; in-between/uncertain; non-religious. 10In cross tabulation analysis, we worked with the recoded variables.In this phase of the analysis, we also included the questions about denominational affiliation: 5. 'Is there a religion or religious denomination to which you belong?' (VA1) 6.If yes, the follow-up question was asked, 'Which one?' (VA2) In the second phase of our analysis we examined the relationship between the overall indicators of religiosity -including, especially, the RVS Salvation item -and certain aspects of religiousness.Available data allowed us to analyse two 'classic' dimensions -beliefs on the one hand and religious behaviour on the other -with the help of the following questions: 7. Interpreting each item of a ten-question subset as a separate question, respondents could decide whether they believed in the given religious concept or not: 'God; eternal life after death; the human soul; Satan; hell; heaven; original sin; resurrection of the dead; reincarnation; angels' (VAL2_1 . . .VAL2_10).-Answer categories included 0 = Item not chosen.1 = Item chosen.8.As the first nine items of a thirteen-question subset, respondents could decide how much they agreed with each of the Commandments (EM1_1 . . .EM1_9).
-The response was given on a five-point Likert scale: 1 = Not at all. . . . 5 = To a very great extent.9. 'Do you live in the spirit of the Ten Commandments?' (EM2) -The response was given by selecting of one of three categories: 1 = Yes, fully. 2 = Partly.3 = No.
Finally, in the third step, we analysed the similarities and differences of the behaviour of Salvation and the other indicators as a function of socio-demographic variables (gender, age, level of education, domicile, denominational affiliation) with ANOVA.

The relationship of the RVS Salvation item with other indicators of religiosity
Following the categorisation of the results of the general indicators of religiosity on a three-point scale, in the case of the four variables -'I am in the hands of God' (SZ2_1); categorical self-classification (VAL1); the importance of religion and belief (KV1_9); 'How religious do you consider yourself?' (VA4) -we got a value between 20% and 50% in all three groups of religiousness (religious, in-between/uncertain, non-religious).The most even distribution we found in the case of the categorical self-classification (approx.⅓-⅓-⅓), while the results of the respondents' evaluation of their own religiosity on a quantitative scale slightly move from the religious pole in the direction of the medial range.By contrast, results on the question about the importance of religion and belief in general without reference to personal faith or the classification of one's own religiosity show just the opposite tendency, shifting from the non-religious pole towards the in-between/uncertain range.The question was asked in a set consisting of thirty items which also included work, family, money and the like.In the frequency table we can see that 38% of respondents rated the importance of religion 7 to 10.At the same time, respondents regarded religion and belief the least important (5.33) after politics.Family received the highest average score (9.74), followed by a few inner qualities such as love (9.18), happiness (9.12), peacefulness (9.03) and well-being (9.02).While on the two previous questions the inbetween/uncertain range is the strongest, on the fourth question -'I am in the hands of God' (SZ2_1) -the non-religious pole stands out.This statement, implying personal involvement, was rejected by more than one in two respondents, who were therefore categorised as non-religious.
Compared to the other global indicators in the survey, the RVS Salvation item acted as an extremely strict measure of religiosity.The in-between/uncertain group here also represents a proportion of about one third, but the proportion of the non-religious is outstanding (nearly two thirds), while only a fraction of respondents marked as religious by the other indicators got into the 'religious' category here, since more than half of the respondents (51.2%) ranked salvation eighteenth (i.e.last) among terminal values, while only 5.5% of them deemed this value very important (1-3) (Table 1).Respondents are almost equally distributed among the other scores (places 4-17), their proportion fluctuating mostly between 2-4%.The average score is 14.66.Respondents' denominational affiliation was also analysed (VA1; VA2).Nearly one in two respondents (46%) do not belong to any denomination; 40% declared themselves Catholic, 13% Protestant, and 1% identified with other, non-Christian denominations.In line with the expectations, the fact that a person belongs to a certain religion/denomination or not shows significant correlation with each indicator of religiosity, including Salvation (Table 2).Between the indicators and specific denominational affiliations, however, there is no statistically verifiable correlation.We generated a religiosity score by performing a principal component analysis of the original, non-categorised variables measuring religiosity without the Salvation item.We experienced a medium-strong correlation between the variables, the KMO value is 0.763, and Bartlett's test is significant.By all relevant criteria (eigenvalue, explained variance, scree plot), all variables load onto one principal component.The communality of each variable exceeds 0.5, that is, the principal component explains more than half of the variance of each variable.The principal component preserved 66.8% of the information carried by the four variables.
Examining the connection between the Salvation item on the one hand and the other variables measuring religiosity and the religiosity principal component on the other, we calculated the Pearson's correlation coefficients (Table 3).We mainly found moderate (min.0.29; max.0.64) correlations between the variables.The RVS Salvation item moves the least with the values of the other variables, but it still proved to be significant at a level of 0.01.A strong correlation (r = 0.643) can be seen between the categorical religious self-classification and 'I am in the hands of God; God directs, God decides'.The religiosity principal component naturally displays a strong correlation with the four variables, for we created it by the reduction of them.The RVS Salvation variable correlates most strongly with the statement 'I am in the hands of God' (r = 0.465), but it shows a strong correlation with the principal component as well (r = 0.447).

The ideological and consequential dimensions of religiosity
The ESS questionnaire allowed for the examination of two classic religiosity dimensions.We were able to analyse the connection between what GLOCK and STARK (1965) called the ideological dimension (beliefs) and the consequential dimension (religious behaviour) on the other hand, and the global indicators of religiosity on the other.The dichotomous variables of the items in the religious beliefs subset (in each case, respondents were asked whether they believed in the given concept or not) significantly correlated with all indicators of religiosity (Table 4).The italicised correlations are the strongest in the matrix.An especially strong correlation appears between the categorical self-classification and the statement 'I am in the hands of God' on the one hand and belief in God and heaven on the other.High values are also assigned to the resurrection of the dead and to eternal life after death.The weakest correlation can be observed with the items related to the soul, particularly with reincarnation (in the case of all indicators of religiosity).In terms of the strength of the connection, angels are in mid-range, and the Satan-hell-original sin concept group is in the lower mid-range of the scale.
Categorical self-classification has the greatest predictive power for each variable.It is followed by the statement 'I am in the hands of God', and then the 'How religious are you?' (1-10) scale.The Salvation item of the RVS also shows significant correlations with the religious concepts, but its predictive power is the weakest of the five indicators, although with half of the items it is stronger than the (impersonal) importance of religion and faith.With medium-strong correlations, Salvation is, on the whole, only slightly behind the strength of the latter indicator.
By analysing adherence to the Ten Commandments, we also examined the correlation between certain behaviours and the indicators of religiosity.The vast majority of respondents (89%) live, according to their self-assessment, at least partly in accordance with the spirit of the Ten Commandments, and only one in ten respondents said they did not follow the Ten Commandments at all.If we compare these data with the different indicators of religiosity, we find that each of the latter is in significant correlation with adherence to the Ten Commandments.The strength of the correlation varies, but each indicator exhibits a medium value (Table 5).
Table 5 The strength of the correlation between the indicators of religiosity and adherence to the Ten Commandments (Cramer's V) Beyond the global assessment of their behavioural conformity, respondents were also asked to express their agreement with the individual commandments on a five-point scale.Answers differed significantly for each item, but their averages clearly demonstrate that the degree of agreement was the lowest for those commandments (1-3) (italicised in Table 6) which do not pertain to the general rules of human coexistence but have specifically religious content.With these commandments, the SALVATION AND RELIGIOSITY

EJMH 12:1, June 2017
The χ 2 test is significant for each variable (p < 0.01).proportion of those who fully agree is strikingly lower than with the other group, where the rate of complete agreement may exceed 80% as in the case of 'You shall not murder'.

Table 6
The degree of agreement with the individual provisions of the Ten Commandments Factor analysis of the commandments clearly reveals a latent structure.The expressly religion-related items belong to one factor, while the others to another factor.The KMO value is 0.873, and Bartlett's test is significant, which suggests that the variables in question are suitable for factor analysis.The MSA values are between 0.760 and 0.923.Each variable is therefore expected to fit well into the factor structure, which is also confirmed by the communalities: each shows a value over 0.45.
In the analysis we used the maximum likelihood method and a varimax rotation.The analysis produced two factors with eigenvalues over 1.The explained variance was 70.8%.We named them 'religion' (Factor 1) and 'ethics' (Factor 2; see Examining the correlation between the factors and religiosity, we find that the religiosity principal component is in significant correlation with both factors, but it correlates only weakly with the ethics factor.If we look separately at the variables that constitute the principal component, we can see similar connections.Salvation moderately correlates (r = 0.352) with the religion factor of the Ten Commandments; with the ethics factor, however, it has no significant correlation (Table 8).

The correlation of the RVS Salvation item with demographic variables
To ensure easier interpretability when comparing the effects of the Salvation item and those of the religiosity principal component, we worked with a standardised version of Salvation (Z-score).With both variables we find that the role of religiosity is increasing with age.While, in the case of the religiosity principal component, the Scheffe test shows that the youngest age group differs significantly from both quintogenarians (50-59) and those over sixty, we see only two homogeneous groups for Salvation (those under/over sixty) (Figure 1).When examining the role of gender, we see the same result for the principal component and Salvation: women tend to be more religious.The correlation is significant in both cases, but the connection is weaker in the case of the Rokeach value.The differences of the averages are much clearer for the principal component (Table 9).The group of those with a low level of education significantly differ from the other groups in the case of both the principal component and Salvation.Those having the lowest level of education are the most religious as measured by the principal component.Skilled workers are the least religious, and then the score of religiosity increases with the level of education.As for Salvation, the Scheffe test shows that college graduates (and higher) do not significantly differ from any other group, and only those with a primary education show a significant difference from skilled workers and high school graduates (Figure 2).We find a weak but significant connection between the principal component and domicile: those living in Budapest significantly differ from those living in the other three types of settlement; their religiosity score is lower.In the case of Salvation, the differences between the groups are very small (can be measured in hundredths).It is not surprising, then, that the correlation is not significant (Figure 3).As we have already mentioned above, there is no statistically significant difference between the different denominations either in the case of Salvation or in the case of the other indicators of religiosity.This is also confirmed by the post-hoc test (Scheffe) applied during ANOVA, which showed that the fact of belonging or not belonging to a denomination determines significantly the scores of the principal component and of Salvation (Figure 4).

Discussion
In the foregoing analysis we have compared the Salvation item of the RVS with other global indicators of religiosity.We have found that Salvation works as a particularly strict measure compared to the other indicators.On those scales, a considerably larger proportion of the sample proved to be religious, and a smaller part expressly non-religious.Just like the other indicators, the RVS Salvation item proved to be inadequate for the prediction of denominational affiliation.We have, however, found significant correlation with the fact of belonging (or not belonging) to a denomination.In this respect, Salvation behaved like the other global indicators.Our principal component analysis has led to a similar result.Salvation displays an at least moderate connection with all the indicators of religiosity, and it is in close correlation with one of the four ('I am in the hands of God') and the principal component.There is, thus, no significant difference between the behaviour of Salvation and the other four indicators included in the analysis.The data available through the ESS-related data collection constituted a limitation for the analysis of the religiosity dimensions.In the case of beliefs, one of the ten items (reincarnation) could not be clearly linked to traditional Christian doctrines, and, compared to the other items, it showed an expressly weaker correlation with Salvation.At the same time it is important to note that, on the one hand, even this connection is statistically significant, and, on the other, Salvation manifests clear similarities to the other global indicators, which also show a weaker than average connection with this item.It is also clear, however, that the predictive power of Salvation for beliefs is, on the whole, smaller than that of the other indicators.In fact, three of the four indicators have a considerably stronger predictive power for beliefs.
We examined the consequential dimension of religion (religious behaviour) through adherence to the Ten Commandments.The available data limited the analysis in several respects.Nearly 90% of respondents claim to live partly in the spirit of the Ten Commandments, but we do not know what that actually means -whether they break only one commandment or they observe only one?Neither do we know what respondents exactly mean when they (11%) say that they do not live according to the Ten Commandments.It may mean that they do not observe all the commandments (strict interpretation), but it may also mean that they do not observe any of them.In view of the low occurrence rate of this answer, however, it seems most probable that the negative answer signals a rejection of deliberateness.The respondent does not seek to observe the Ten Commandments consciously, even if they probably observe some of them (e.g. the prohibition of killing and stealing) 'fortuitously' (that is, without the intention to follow the Ten Commandments playing any role in shaping their behaviour).
The data may be further distorted by the fact that the wording of the first commandment in the questionnaire is not complete.It contains no imperative or prohibitive elements (it lacks the phrase 'you shall have no other gods before me'), and the banning of images has also been left out of the questionnaire.These deficiencies were not compensated by the division, applied by certain denominations, of the last commandment against coveting another person's possessions.Thus only nine commandments were actually included in the questionnaire.Since some versions of the text of the Ten Commandments are aligned with denominational divisions, the wording used in the questionnaire may have contributed to the fact that denominational affiliation did not prove to be a significant factor for any indicators of religiosity.
As with most other aspects of religiosity, the RVS Salvation displayed the weakest correlation with the behavioural dimension as well.Yet it still proved significant and was not substantially weaker than that of the other indicators.The result of the factor analysis is particularly noteworthy because it showed that the indicators of religi osity have a much weaker connection with those commandments that can also be interpreted as general ethical rules than with those of expressly religious content.The difference is obvious in the case of Salvation as well.Moreover, here we found no significant correlation with the ethics factor at all.At this point Salvation deviates from the other indicators of religiosity with the exception of the question 'How religious do you consider yourself?'One possible reason for the difference is that Sal-vation is less suitable to identify non-traditional forms of religiosity, while the ethical commitment synthesised in the Ten Commandments is not restricted to traditional forms of religiosity (cf. KÜNG 1994).Salvation displays a somewhat weaker connection with the religion factor than do the other global indicators.That is in line with our results for other aspects of religiosity.
Our analysis of variance has led to the result that RVS Salvation, considered as a function of the major demographic variables, behaves similarly to the religiosity principal component.Difference between them is only found in relatively minor details.While each background variable correlates significantly with the religiosity principal component, the connection between the Salvation item of the RVS and domicile is not significant.With the religiosity principal component we also find that only the respondents from Budapest differ from the others.With Salvation, however, this difference also disappears.A possible explanation for this may be that, as we have already pointed out, the Salvation item is a strict indicator of religiosity.The results of the F-test show that the religiosity principal component is in a stronger connection with the background variables than Salvation is.The individual steps of the ANOVA thus confirmed our general experience that the scores of the Salvation item fall somewhat below the corresponding scores of the principal component and the other indicators of religiosity.That, however, only qualifies but does not obliterate the fundamental similarities between the RVS Salvation and the other global indicators.

Conclusion
To summarise, we can say that, according to our test criteria, the Salvation item of the RVS is a somewhat weaker indicator of religiosity than the other global indicators, but it is basically comparable to them.It is not suitable in itself to draw a detailed picture of the religiosity of a certain population, but, as our analysis has shown, it is suitable to provide an overall picture of it.We can basically assess the degree of religiosity of a sample with the help of this single question.
Our analysis has identified two limitations of RVS Salvation as a global indicator of religiosity.On the one hand, it seems to be a stricter measure than the other global indicators analysed.The results showed that the RVS item behaves similarly to other indicators of religiosity, but the correlations are weaker in comparison and the variable is stricter in the sense that it indicates a lower level of religiosity in a given sample.Based on the Salvation criterion, we will probably find a smaller part of a population religious and a bigger part expressly non-religious than with other similar measures.On the other hand, Salvation proved to be less suitable than the other indicators to reveal non-traditional forms of religiosity.Thus one important boundary condition of its applicability is a low prevalence of non-traditional forms of religiosity in the sample.The critical edge of our conclusion is not a novelty, for it has been long known in the literature that the RVS Salvation item is not suitable to identify certain types of religiosity (MILLER 1976;SZÁNTÓ 1998).On the contrary, the most important result of our analysis is that under certain conditions, which still seem to apply in Hungary at the beginning of the twenty-first century, the RVS Salvation item can be successfully used as a global indicator of religiosity.
Figure 2The connection of the religiosity principal component and the RVS Salvation with the level of education (average)

Figure 3
Figure 3The correlation of the religiosity principal component and the RVS Salvation with domicile (average)

Figure 4
Figure 4The connection of the religiosity principal component and the RVS Salvation with denominational affiliation (average)

Table 1
Level of religiosity as measured by different indicators (distribution of responses, %)

Table 2
The strength of the correlation between indicators of religiosity and denominational affiliation (Cramer's V)11The χ 2 test is significant for each variable (p < 0.01).

Table 3
Connection between the indicators of religiosity (Pearson's r)

Table 4
The strength of the connection between beliefs and the global indicators of religiosity (Cramer's V)

Table 7
Factor-weights of each of the Ten Commandments

Table 8
The correlation of the indicators of religiosity with factors of the Ten Commandments (Pearson's r)

Table 9
The connection of the religiosity principal component and the RVS Salvation with gender (average)