In order to better interpret the multiple dimensions that
statsitically result from the ranking of stimulus pairs, respondents
are asked to list the criteria that likely influenced their judgment
while rating the pairs.
List of Probable Criteria
| #1 |
Population density |
| #2 |
Architectural and scenic beauty |
| #3 |
Xenophobia |
| #4 |
Mass transport |
| #5 |
Level of commercialization |
| #6 |
Travel advertisements |
| #7 |
Historical background |
| #8 |
Air Pollution |
Respondents are then provided with a blank table similar to
the one below and asked to fill the column headings with those
attributes which most probably influenced their decision. A rating
scale, by which each element of the stimulus set can be ranked,
is also provided.
Attribute Table
| |
Population density |
Architectural and scenic beauty |
Xenophobia |
Level of commercialization |
Air Pollution |
| Hong Kong |
|
|
|
|
|
| Seoul |
|
|
|
|
|
| Shanghai |
|
|
|
|
|
| Singapore |
|
|
|
|
|
| Tokyo |
|
|
|
|
|
| 0 |
10 |
20 |
30 |
40 |
50 |
60 |
70 |
80 |
90 |
100 |
| 0 = abence or weak
presence of attribute |
|
100
= strong presence of attribute |
From this exercise a table similar to the following is obtained.
Attribute Table
| |
Population density |
Architectural and scenic beauty |
Xenophobia |
Level of commercialization |
Air Pollution |
| Hong Kong |
80 |
90 |
80 |
90 |
80 |
| Seoul |
70 |
70 |
90 |
80 |
80 |
| Shanghai |
80 |
70 |
70 |
90 |
90 |
| Singapore |
80 |
60 |
70 |
80 |
70 |
| Tokyo |
90 |
70 |
90 |
90 |
60 |
From these data the basis for a regression analysis is obtained
and regressions run on the entire sample population. Only those
attributes that are common to a siginficant number of respondents
are used, so as to provide a large enough sample size.
Knowing how each element of the stimulus set loads on each
attribute can help the research interpret the dimensions and
perceptual map generated by the multidimensional scaling procedure.