HOME
Presentation by Kiwamu MAKI at international conference
1)
Y.Nakamura, K.Maki, M.Inui. Relationship between the Classification of Urban Streetscape and Overall Evaluation, Proceedings of the 13th Conference of the International Association for People-Environment Studies (Manchester), pp.351-361, 1994
1. Introduction
The relationship between landscape constituents and overall evaluation is often explained for each category group of landscapes including those of office streets, residential streets, shopping streets, distant views and short-range views, natural landscapes, and artificial landscapes. For example, we often hear that office streets should be composed, while shopping streets should be livelier even if they are not very quiet. If this is true, then we need to present good landscapes for each street application.
To describe desirable streetscapes, townscape, etc. in such a manner, we should probably study the factors used for overall evaluation according to the application of each street. Some cases may require checking for the possibility of better classification than its present application. In this research, we investigated how people intuitively classify streetscapes, townscape, etc., in light of the objective stated above. We then studied the desirability of streetscapes, townscapes, etc. of each category group and the differences in the involvement level of the factors that people consider when evaluating the desirability.
2. Investigation by Interview 1: Classifying streetscapes, townscapes, etc.
We carried out the investigation through interviews to determine how people classify streetscapes, townscapes, etc. For this investigation, we prepared 125 photographs of streetscapes, townscapes, etc. which were taken in Tokyo and its environs (see Fig. 1) and showed them to 33 subjects (15 males and 18 females). The viewpoints on which they were based for classifying the photographs and the number of photographs of each category group were all left up to the subjects. The subjects were asked to subdivide the photographs by stages until they decided that they could no longer subdivide. Table 1 shows the investigation procedures taken. Every subject completed the classifying work within five steps.
Based on the results of the classification by the subjects, we prepared a matrix of the frequency of each photograph being classified into the same category group. Taking this as the similarity of streetscapes, townscapes, etc., we performed cluster analysis based on the furthest neighbor method. As a result, we obtained the clusters shown in Table 2. The majority of the subjects first classified the streets into shopping streets, residential streets, and office streets which belong to the left hand cluster of Table 2, then worked on the right hand cluster. The clusters thus obtained showed good matching with the processes through which most subjects went. The top group on the left is the category group based on the application of streets; as the level goes down from the top to the bottom, the number of impressions that the streetscapes, townscapes, etc. project and the factors constituting the landscapes increase.
From the results, we found that people first classify streetscapes, townscapes, etc. according to their applications, then they use the commonality of impressions and constituents for subdividing. The following describes the changes in the evaluation standard for each of the three categories based on the applications of streets which are commonly used by people and which are considered to most influence the commonality in the evaluation standard for streetscapes, townscapes, etc.
3. Investigation by Interviews 2: Extracting Evaluative Constructs
To find whether different categories have different constructs that are considered when the desirability of streets is evaluated, we carried out the investigation based on interviews by using the method suggested primarily by ??Sanui??. The investigation procedure used is as shown in Table 3. The stimulants presented were the same 125 photographs of streetscapes, townscapes, etc. as those used for the interview investigation 1. Each group contained 16 subjects, with approximately equal numbers of males and females in each group.
We totaled the frequencies of the words used to evaluate the streetscapes, townscapes, etc. which were obtained from the investigations. Table 4 shows the constructs that were used by three or more subjects. Further, to identify the connection among the evaluative constructs, we drew the network charts shown in Fig. 2. In the network charts, specific evaluative constructs (lower constructs) are shown on the right, while abstract evaluative constructs (higher constructs) are shown on the left. The lines with arrowheads connecting the constructs represent the relationship between the constructs, and it shows that the evaluative constructs at starting points are essential factors for satisfying the evaluative constructs at terminating points.
The frequencies at which the evaluative constructs are mentioned show that there are differences among the groups in terms of "organization," "excitement," and "lifestyle." Also, the evaluative construct "there is greenery" was mentioned very frequently, indicating that it is basically an important factor.
The characteristics of each category group are summarized as follows:
a. Office streets
The higher constructs were "unity" and "organization," the main lower constructs were "unity of buildings," "harmony of colors," "signboards are less conspicuous," "there are fewer objects," and "the roads are wide." Thus, a distinct system of describing the impressions has been obtained. The impression of "easy to walk along" has been connected primarily with "unobstructed view," "generous space," and "presence of greenery," suggesting that an emphasis is placed on the visual impression.
b. Shopping streets
The subjects expected shopping streets to be beautiful and composed. Many constructs they listed were related to "signboards" as with the case of office streets. Although the evaluative construct "looks exciting" was selected frequently, which streets appeal greatly vary from person to person, and it did not appear in the networks.
c. Residential streets
Many constructs such as "it shows the lifestyles of people living there" and "it looks like the neighborhood in which I used to live" which are related to lifestyle and human relationships were often selected. Also, the word "jumble" was often used in a positive meaning, which is considered to be associated with people's lifestyles and nostalgia.
Thus, it has been determined that, in evaluating streetscapes, townscapes, etc., superficial impressions such as "organization" and "unity" are important factors, or the factors connected with the images of the places, which are typically represented by "lifestyles of the people living there," are influential depending on the applications of the streets. It has been proved that the evaluative constructs for overall evaluation differ from one category group to another.
4. Experiments with Rating Scale, etc.: Relationship between Overall Evaluation and Evaluative Constructs
Next, to identify the differences in the association level among the evaluative constructs, we made 68 slides out of the 125 photographs used for the interview investigation and showed them to the subjects who then rated the landscapes using the same scale. The slides presented to the subjects included 25 scenes from residential streets, 22 scenes from shopping streets, and 21 scenes from office streets. The scale used for the rating was obtained by converting a total of 36 constructs of the evaluative constructs extracted in the interview investigation 2 and the overall evaluation, i.e., "desirability," into a 5-step rating scale (e.g., five-point rating scale). There were 50 subjects, 25 males and 25 females.
Analysis 1: Correlation among the evaluative constructs in a category group
Using Kruskal's multidimensional scaling, we attempted to show the levels of connection between rating scales for each category group. Each group was represented two- dimensionally under small stress values (see Fig. 3). Analyzing the results given in the charts, from which the connection of each scale was determined, shows that office streets and shopping streets, for example, exhibit a high level of correlation between "composure," "organization," and "harmony of colors" while residential streets exhibit a low level of correlation. Thus, the relationship between the scales varies according to the category group, and the scales which are closely correlated to the overall evaluation "desirability" also vary depending on groups. As far as correlation among the scales is concerned, the decisive factors in overall evaluation seem to vary according to category groups.
We cannot immediately conclude, however, that the influences exerted by evaluative constructs on the evaluation of streetscapes, townscapes, etc. differ from one category group to another. For instance, when we take the relationship between "composure" and "desirability," their correlation is relatively high for residential streets and shopping streets, while it is low for office streets. If, however, we assume that there are no great variations in the rating of scenes from office streets in the aspect of "composure," then even if the evaluation in terms of "composure" is closely related to "desirability," the error will significantly influence their correlation. In other words, if each category group has unbalanced characteristics or impressions, then the magnitude of a correlation coefficient may change even if the same evaluative constructs are used for defining overall evaluation. For this reason, we checked each category group for any unbalanced impressions by performing a factor analysis using un-grouped data of the whole and studying a chart bearing the factor scores.
Analysis 2: Analysis across the category groups
As a result of the factor analysis based on mean rating values, three factors represented by the scales of "composure," "cheerfulness," and "spaciousness" shown in Table 5 were extracted. We then determined factor scores for all scenes and placed them on the factor axes and found that some groups give similar impressions, and there is unbalance as shown in Fig. 4. Hence, we cannot decisively conclude that the evaluative constructs defining overall evaluation differ from one category group to another.
When we correlated the factor scores to overall evaluation, we found that the scores of the first and second factors are correlated to overall evaluation. When we expressed the rating in terms of "desirability" on the factor score chart, as shown in Fig. 4, the scores of "desirability" increased as the scores of the first and second factors increased. This trend was found to be constant in all category groups. We then carried out a multiple regression analysis with the two factors used as the explanatory variables. The correlation between the estimated values and the rating values was high at 0.944. Judging from the high level of correlation, it was decided there was no need for classification. The same multiple regression analysis was performed on all category groups and the analysis results were compared with the results of the whole. The comparison result indicated that the estimation accuracy stayed almost unchanged and the weighting also stayed the same for all category groups (see Table 6). This means that there is almost no difference among the category groups when we assume that the overall evaluation is defined by the evaluative constructs represented by the two factor axes.
5. Conclusion
Generally, desirable landscapes are often considered to differ according to the type of streets; this also applies to residential streets and office streets. It is true that the evaluative constructs extracted vary according to the type of street in the interview investigation, and the evaluative constructs closely correlated to overall evaluation vary according to the type of street also in the rating experiments. These phenomena, however, are very likely to result from the similarity of characteristics for each type of street. In this study, various evaluative constructs were analyzed along the two axes of "composure and unity" and "cheerfulness and excitement," and the influences exerted by these evaluative constructs on the evaluation of streetscapes and townscapes are considered not to vary with the type of street.
It is often necessary to focus on a particular street to implement research in order to obtain detailed data for creating desirable streetscapes, townscapes, etc. In such investigation and research, however, characteristics are extremely similar, and misleading results may be obtained. To acquire truly effective data, we need to implement research by carefully considering the whole streetscape.
前のページへ