The Scottish Government

« Previous | Contents | Next »

Listen

Technical annex 3 SHS Analysis

Tukey HSD and mean distance

1. Each table shows how the average commuting distance varies for one particular factor, considered in isolation. Tukey ( HSD) test is a procedure for testing all possible pairs of means from a dataset where there are a relatively large number of groups, with low number groups, considered rather conservative.

2. Taking age and mean distance as an example, Table 1 shows how the average distance varies with age group:

• the first column lists the different age groups
• the second column gives the sample numbers for each age group
• the remaining columns show the mean distance for each age group. Each 'sub-set' contains age-groups whose values do not differ significantly. For example, the difference between the mean distances of 8.1km for those aged 60+ and 9.0 for those aged less than 20 is not significantly significant. Similarly, the difference between the mean distances of 9.0km for those aged less than 20 and 10.4km for those aged 50 to 60 is not statistically significant
• When a 'sub-set' contains the value for only one category, it means that it differs significantly from the values for all other categories. For example, Table C2 (in which the categories are the number of cars available to the household) shows that the mean distances for 'none' (6.6km), one (10.6km) and 'two' (14.5km) all differ significantly
• The Tukey ( HSD) significance measure is given at the bottom row of each table (a value of 1 indicates a strongly significant result)

Table 1: Age and average distance

Table 2: Car ownership and average distance

Table 3: Mode and average distance

Table 4: Employment status and average distance

Table 5: Social class and average distance

Table 6: 6-Fold urban/rural classification and average distance

Table 7: Household income and average distance

Table 8: Household type and average distance

First ANOVA

3. Table 9 lists all variables included in the initial 'full' ANOVA and the sample number within each value label. This first ANOVA includes the following variables: Employment status, mode, car ownership, gender, urban/rural area, household income, age, council area and household type. Table 10 details the findings from the first ANOVA.

Table 9. Variables and sample numbers within the first ANOVA

Table 10: Full ANOVA

R Squared = .197 (Adjusted R Squared = .187)

Second ANOVA

4. The second ANOVA samples only responses where the usual mode of travel to work was car (as driver or passenger). As a consequence, main mode and car ownership were excluded. Employment status was also excluded to shift the focus to those who are commuting by car every day rather than those who may only commute once or twice a week.

5. The second ANOVA includes the following variables: age, gender, urban/rural area, household income and household type. Table 11 lists all variables included in the second ANOVA and the sample number within each value label. Table 12 details the findings from the second ANOVA.

Table 11: Variables and sample numbers within the second ANOVA

Table 12: Second ANOVA

R Squared = .072 (Adjusted R Squared = .063)

« Previous | Contents | Next »