The Scottish Government

« Previous | Contents | Next »

Listen

ANNEX A RESULTS OF MULTIVARIATE ANALYSIS

Chapter Nine

The following tables show the results of some of the key logistic regression analyses (binary or ordinal, as appropriate) we undertook in preparing Chapter Nine. In each case we first of all entered into our models, where significant, a respondent's individual characteristics. The characteristics typically considered for inclusion at this stage were age, gender, education, social class, religious attendance and ethnic origin. Then we allowed to enter into our model whichever of a number of measures of the characteristics of the respondent's area that proved to be significantly associated with the dependent variable. These are level variables comprised the following:

From 2001 Census (output area level data)

• % White, divided into two categories, 97% and above, and less than 97%.
• % Muslim, divided into two categories, less than 2%, and 2% or more.
• % of those aged 16-74 who are graduates, divided into four categories, 8% or less, 8-16%, 16-28%, more than 28%
• % aged 18-24, divided into four categories, 5% or less, 5-7%, 7-10%, more than 10%
• % aged 65 and above, divided into four categories, 7% or less, 7-14%, 14-22%, more than 22%.
• % in good health, divided into four categories, 60% or less, 60-69%, 69-77%, more than 77%.
• % aged 16-74 in professional or managerial occupations, divided into four categories, 13% or less, 13-21%, 21-35%, more than 35%.

From other sources:

2006 Scottish Index of Multiple Deprivation.

Index Score, divided into five categories:

• 7.7446 or less;
• 7.7472 to 13.5627,
• 13.5640 to 21.0436,
• 21.0521 to 33.6982,
• 33.7252 and above.

Scottish Government Urban/Rural Classification.

This has six categories defined as follows:

• Large urban areas: settlements of over 125,000 people
• Other urban areas: settlements of 10,000-125,000 people
• Accessible small towns: settlements of between 3,000 and 10,000 that are within 30 minutes drive of a settlement of 10,000 or more.
• Remote small towns: settlements of between 3,000 and 10,000 that are more than 30 minutes drive from a settlement of 10,000 or more.
• Accessible rural areas: settlements of less than 3,000 people that are within 30 minutes drive of a settlement of 10,000 or more.
• Remote rural areas: settlements of less than 3,000 people that are more than 30 minutes drive from a settlement of 10,000 or more.

Table A.1 Logistic Regression of General Measure of Prejudice

Model: Binary Logistic
* significant at 5% level
Main cell entries are parameter coefficients; figures in brackets are associated standard errors

Table A.2 Logistic Regression of where prefer to live

Model: Binary Logistic
* significant at 5% level
Main cell entries are parameter coefficients; entries in brackets are associated standard errors

Table A.3 Logistic Regression of Unhappiness at Close Relative marrying / forming long-term relationship with a Muslim

Model: Ordinal Logistic
* significant at 5% level
Main cell entries are parameter coefficients; entries in brackets are associated standard errors

Table A.4 Logistic Regression of whether older people should have to retire at a certain age

Model: Binary Logistic
* significant at 5% level
Main cell entries are parameter coefficients; entries in brackets are associated standard errors

Table A.5 Logistic Regression of Attitudes to Refusal of Bed and Breakfast Booking from same sex Couple

Model: Ordinal Logistic
* significant at 5% level
Main cell entries are parameter coefficients; entries in brackets are associated standard errors

Chapter Ten

The following table gives full details of the factor analysis summarised in Table 10.1 of Chapter Ten.

Table B.1 Factor Analysis of Psychological Orientation Questions

Note: Only factor loadings above .3 are shown.

Table A.2 Multivariate analysis of attitudes towards prejudice

* statistically significant at the 5% level.

Note to table:
This table shows the results of a logistic regression of thinking there is sometimes good reason to be prejudiced as opposed to stating that Scotland should do everything to get rid of prejudice. The main cell entries are parameter coefficients while the entries in brackets are the associated standard error. The cultural threat and cultural diversity scales were entered into the regression as interval level variables, using the original scores. The keep customs vs. adapt and blend scale was entered as a categorical variable, with those placing themselves in the adapt and blend box as the reference category. Those who placed themselves in one of the three middle boxes have been combined into a single category. The coefficients therefore indicate the 'effect' being in a particular category as opposed to being in the adapt and blend category.

Chapter Eleven

The following tables give further details of the multivariate analyses to which reference is made in Chapter Eleven.

Table C.1 Models of Attitudes to Prospective Relationship Partners

Note to table:
The entries in the first four columns are the parameter coefficients from an ordinal logistic regression of reported unhappiness with a close relative marrying someone from that group. The latter two of these four columns show the effect of being in that category as opposed to favouring the 'adapt and blend' option. The final column shows the Nagelkerke R-squared for that model. In each model the standard error for both the cultural threat and the cultural diversity score was .06. The standard error or 'keep customs' was either .25 or .26, while that for those who ticked one of the middle boxes was .11.
* significant at the 5% level

Table C.2 Models of attitudes towards sexual orientation

See note to Table C.1. Dependent variables have been scored so that a low score indicates a less discriminatory attitude.

Table C. 3 Models of Attitudes to Possible Teachers

See notes to Tables C.1 and C.2

Table C.4 Models of Attitudes towards Bed and Breakfast Bookings

See notes to tables C.1 and C.2

Table C.5 Models of Covert Discrimination and Attitudes to Positive Action

See notes to tables C.1 and C.2

« Previous | Contents | Next »