• Keine Ergebnisse gefunden

Step 1. Define Alternatives and Value-Relevant Attributes

The Multi-attribute Utility Method

5.4 Practical Application of Multi-Attribute Utility Theory in the Domain of Housing

5.4.2 Example Two: Calculating Single-Attribute Utilities for Dwelling Attributes in a Large Sample

5.4.2.2 Step 1. Define Alternatives and Value-Relevant Attributes

We did not select alternatives (dwelling profiles) but provided respondents with a number of pre-selected attributes and attribute levels. The set of attributes considered are those that were used in the pilot study for the VROM module ‘Consumer behavior’

(Consumentengedrag) (Boumeester et al. 2008a; Jansen et al. 2009). In this study, 13 attributes were chosen on the basis of expert opinions and the literature that had shown these attributes to be important attributes influencing residential decision-making (Floor and van Kempen 1994, Goetgeluk 1997; Heins 2002; Boumeester et al. 2005). As explained before, the multi-attribute utility questions were part of a

Table 5.7 Respondents’

characteristics Gender (n = 2,043)

Female 1,087 (53%)

Age (n = 2,037)

Mean (std) 52 (13)

Range 19–89

Number of persons in household (n = 2,038)

One 298 (15%)

Two 896 (44%)

Three 291 (14%)

Four 368 (18%)

Five or more 185 (9%)

Children < 18 (n = 902)

None 291 (32%)

One 221 (25%)

Two 268 (30%)

Three or more 122 (13%)

Paid job (n = 2,036)

Yes 1,318 (65%)

Education (n = 1,979)

Lower 474 (24%)

Middle 716 (36%)

Higher 789 (40%)

larger survey into residents’ housing preferences (Huizenkopers in Profiel: Boumeester et al. 2008b). For this reason, we could not include all 13 attributes, because the interview burden for the respondents would be too high (fatigue, boredom, no more time, and so on). Except for residential environment, only attributes were included that pertained to characteristics of the dwelling. This choice was based on previous findings that dwelling characteristics are deemed to be more important than charac-teristics of the dwelling environment (Boumeester et al. 2008a). Furthermore, some dwelling environment characteristics may not be easy to influence, such as the compo-sition of the residents living in the neighborhood. However, residential environment (urban, suburban, and rural) was included in the survey, because this attribute has proven to be important to respondents (Boumeester et al. 2008a). For all attributes two to three attribute levels were chosen. The attributes and attribute levels are presented in Table 5.8. All analyses were performed using SPSS, version 14.0.

5.4.2.3 Step 2. Evaluate Each Alternative Separately on Each Attribute, That Is, Assess the Level of Satisfaction with the Attributes

The attribute level values were obtained directly with the use of rating scales.

Respondents were asked to indicate their likes or dislikes with regard to each level of every attribute on a scale with two anchors: on one side “extremely unattractive”

with an assigned value of 0 and on the other side “extremely attractive” with a value of 100. The questions were introduced by explaining these endpoints and stating that a higher appointed number was related to more attractiveness. Furthermore, the interviewer explained that the respondent had to take their current situation and household income as a starting point when answering the questions.

For example, when inquiring about living in a dwelling with three rooms, the following question was formulated: Please indicate on a scale anchored by 0

Table 5.8 Attributes and attribute levels

Dwelling type Purchase costs/rental costs

Apartment € 140,000/€ 338 per month

Terraced house/corner house € 220,000/€ 532 per month Semi-detached house € 300,000/€ 725 per month

Tenure Size of the living room

Rental house 20 m2

Owner-occupied house 30 m2

40 m2

Architectural style Number of rooms

Traditional 2

Innovative 3

Modern 4

Residential environment Backyard size/size balcony

Urban 5 m/4 m2

Sub-urban 10 m/7 m2

Rural 15 m/10 m2

Table 5.9 Mean values for attribute levels

Terraced house/corner house 57.4 29.4 2,034

Semi-detached house 67.1 27.5 2,033

“extremely unattractive” and 100 “extremely attractive” how you would value living in a dwelling with three rooms. Next, the respondent was asked about his/her evalu-ation of living in a dwelling with four rooms, and then with five rooms. This pro-cedure was repeated for each level of every attribute.

Table 5.9 shows the mean values for the attribute levels obtained in this study.

Note that in the current study, respondents were presented with questions on either

rental costs or purchase costs, depending upon their current living circumstances with regard to tenure. Similarly, only respondents with a backyard or a balcony were presented with questions on backyard and balcony size, respectively.

However, ultimately all respondents answered questions for the same eight attri-butes. The results show that respondents value the attribute level of owner-occupied houses the highest, followed by a traditional architectural style and a rural residen-tial environment. The attribute level of two rooms was valued the least, followed by a living room size of 20 m2 and monthly rental costs of € 725.

In general, the ranking of levels within a particular attribute is as expected, with more space and less costs related to higher preferences. An exception to this rule is purchase cost, because a dwelling of € 140,000 is not as highly valued as the more expensive houses. One would expect lower purchase costs to be evalu-ated more positively than higher purchase costs. This seems to suggest that respondents take other things into consideration, besides price, when evaluating the single attribute of purchase costs. For example, they might be worried about the size or the state of maintenance of such a relatively cheap dwelling. Thus, quality is implicitly included in the price. Price may reflect all kinds of qualities, such as maintenance, neighborhood amenities, image of the neighborhood, and so on. This finding agrees with that of Floor and van Kempen (1994) who notice that respondents more frequently pronounce a desire for a more expensive single-family owner-occupied dwelling than for a less expensive one. He argues that, apparently, respondents are aware that it is difficult to find a reasonable owner-occupied dwelling for a low price. Park et al. (1981) used a different method (the Decision Plan Net) but also observed an inconsistency with regard to the attribute of price. Their respondents were more than averagely satisfied with the attribute of price, even though they had paid more for their new home than they had previously found acceptable. Park and co-authors concluded that their subjects may not have perceived price as a separate dimension. When the other dimensions had appeared very satisfactory, the respondents’ perceptions of price might have been adjusted to reflect the net worth of the total entity. In this study the effect is not observed for rental houses; here the highest evaluation is for the dwelling with the lowest rental price. Presumably, the respondents expect price in a regulated rental market price not to be so strongly related to the quality of the dwelling and the neighborhood.