Part 2
I then repeated the sequence of statistical analyses to look at the effects of removing the variables that were common to the three scales. A full list of the variables used in the two NEW attitudinal scales, Attitudes to Religious Values and Attitudes to Social Values. These are the new scales without the variables held in common with the scale, UK Zoroastrian Boundaries.
The variables were recoded so that +1 = support for variable; 0 = missing value or don't know; -1 = no support for variable. Using the program SPSS List Cases I printed out a list of the responses per respondent for each set of variables; religious or social. By hand I added up the total per case and returned the total to the raw data set, creating two NEW variables the scales, Attitudes to Religious Values and Attitudes to Social Values.
A frequency distribution of the variables showed that the scale, Attitudes to Religious Values, had a minimum of -12 and maximum of 19 with a mean of 5.809, standard deviation of 5.679 and mode of 11.000. This compares with the original scale, Attitudes to Religious Values where there was a minimum = -15, maximum = 26, mean = 5.618, standard deviation = 8.361, mode = 0.000.
The scale, Attitude to Social Values, had a minimum of -12 and maximum of 20 with a mean of 5.050, standard deviation of 5.817 and mode of 8.000. This compares with the original scale, Attitudes to Social Values where there was a minimum = - 13, maximum = 20, mean = 4.384, standard deviation = 6.102, mode = 8.000.
Did the symbolic boundaries had any impact on attitudes to religious and social values of Zoroastrianism? The SPSS Plot facility and plotted the scale, Attitudes to Religious Values, against the scale, Attitudes to Social Values. The results are given in Table 14 to Table 16. The exact position of each respondent with regard to the sum total of their attitudes to the religious and social values of Zoroastrianism can be examined.
SPSS Linear Regression procedures were used to look at the relationship between these three variables, UK Zoroastrian Boundaries, Attitudes to Religious Values, Attitudes to Social Values. The results are summarized in tabular form in Tables 17 to 21.
As was expected the scale, Attitudes to Religious Values, explained a very high percentage of the variance in the scale, UK Zoroastrian Boundaries, but it should be noted that the scale, Attitudes to Social Values was dropped from the stepwise regression equation.
For the scale, Attitudes to Religious Values, both scales, UK Zoroastrian Boundaries and Attitudes to Social Values, were entered into the stepwise regression equation with an explained variance of (30%). The effect of the background variables on the scale, Attitudes to Religious Values, is given in Table 19.
The independent action of each scale on the scale, Attitudes to Religious Values, is given in Table 20.
The scale, Attitudes to Social Values, added little to the explanatory of the scale, UK Zoroastrian Boundaries, (6%), and by far the best predictor of Zoroastrian attitudes to the religious values of Zoroastrianism was their attitude to the symbolic boundaries, (beta = .40). A stepwise regression analysis of the scale, Attitudes to Social Values, was conducted and the results are given below in Table 21.
As was expected the scale, UK Zoroastrian Boundaries, was dropped from the regression equation. Religious values explained (11%) of the variance in social values of Zoroastrians.
It is not known in advance whether a model such as linear regression is appropriate therefore it is necessary to look for violation of assumptions. The program SPSS Residuals was utilised to check for violations of assumptions. Only those models have been reported that complied with the assumptions.