Presentation is loading. Please wait.

Presentation is loading. Please wait.

A New Approach to Measure Preferences of Users in Built Environments: Integrating Cognitive Mapping and Utility Models Benedict Dellaert Erasmus University.

Similar presentations


Presentation on theme: "A New Approach to Measure Preferences of Users in Built Environments: Integrating Cognitive Mapping and Utility Models Benedict Dellaert Erasmus University."— Presentation transcript:

1 A New Approach to Measure Preferences of Users in Built Environments: Integrating Cognitive Mapping and Utility Models Benedict Dellaert Erasmus University Rotterdam Theo Arentze Oliver Horeni Harry Timmermans Eindhoven University of Technology

2 Why study users preferences an decisions?
Consumer research increasingly important for real estate decision making to predict choice behavior to improve product design and services to communicate more effectively

3 Example – preferences shopping locations
Which type? Which attributes important? Why important? Centre of small city Centre of large city Centre at the edge of city

4 Example – preferences housing locations
City centre – which attributes? – why important? Outside centre – which attributes? – why important? Country side – which attributes? – why important?

5 Measuring mental representations
Traditional Laddering (face to face, qualitative) Gutman 1982, Reynolds and Gutman 1988 Meaning Structure Method Coolen and Hoekstra 2001 Association Pattern Technique / Hard laddering Audenaert et al. 1998, Ter Hofstede et al. 1999 CNET (Causal Network Elicitation) Arentze et al 2008; Horeni et al. 2014 Attribute-Benefit surveys Ashok et al. 2002, Chandukala et al. 2002, Dellaert and Stremersch 2005, Luo, Kannan and Ratchford 2008

6 Example - Choosing a city-trip destination
Tourist destination choice Attributes Benefits Utilities

7 Problem and objective Cognitive mapping methods offer rich information
which attributes are important for which reasons But, the information is qualitative no quantitative information about the importance of attributes and values of preferences Purpose of the present study extending means-end chains analysis with a method to estimate importance weights for attributes directly from mental representations

8 Theory and model Cognitive activation of components in mental representation is based on utility and cognitive cost framework Akin to information search theory Additional components are included only if expected gain of evaluating alternatives exceeds the cognitive costs

9 Model of activation Basic building block is a D–A–B chain
Decision alternative Attribute Benefit Value of including a D–A–B chain Gain: reduced risk of making incorrect decision Costs: mental effort required for activation in evaluation

10 Parametrization of the model
SDA SAB α c Decision alternative Dispersion Cognitive cost (threshold) Si Strength of influence of one node on another α Utility of benefit in decision D Dispersion across the range of the variable SDA x SAB x α x D Gain of adopting a given chain in the MR c Gain has to be greater than a threshold of cognitive costs to include D-A-B chain in mental representation.

11 Utility model of activation
Net utility of activating a DAB chain in situation n Need to standardize the dispersion (scale) per attribute across alternatives Probability of activating a DAB chain follows binary logit specification

12 Data collection – choice task
Mental representation of decision where to shop, when to shop, and which transportation to use

13 Data collection – survey method
Representative sample from a Dutch online panel, 1070 responses, collected over two waves Average age 43.1 years, 55% female, and 35% had earned a bachelor’s degree or higher Minimum frequency for each of the AB and DA links set to 30 (approx. 5% occurrence probability) 108 AB links and 41 DA links meet this criterion Implies: 196 observations per individual of yes/no occurrence for DAB chain Also observed chosen shopping alternative

14 Results Some illustrative DAB chains
Shopping location (e.g., Down town) (D) – Crowdedness in the store (A) – Shopping pleasure (B) Transportation mode (e.g., Car) (D) – Number of bags to carry (A-S) – Ease of Travel (B)

15 Link strength DA (example))
Results - estimation Link strength DA (example)) Attributes Decision alternatives Value t-value available time to shop timing of shopping 1 shopping location 0.90 -5.51 weather transport mode 0.84 -8.39 0.82 -9.55 number of bags to carry 0.78 -14.41 0.75 -15.55 durability of bought products required time to shop departure time flexibility of the transport mode 0.81

16 Link strength AB (example))
Results - estimation Link strength AB (example)) Benefits Attributes Value t-value travel pleasure number of bags to carry 1.24 8.22 departure time flexibility of the TM 1.23 7.58 crowdedness on the way 1.21 4.39 travel time 1.09 2.89 capacity of the transport mode 1.01 0.22 weather 1 accessibility of the store 0.99 -0.07 ease of shopping available time to shop 1.20 8.06 1.02 0.78 crowdedness in the store 0.67 required time to shop opening hours

17 Benefit activation and cognitive costs (example)
Results - estimation Benefit activation and cognitive costs (example) Benefits Value t-value travel pleasure 1.94* 33.83 ease of shopping 3.66* 61.21 shopping comfort 3.43* 36.01 shopping pleasure 3.26* 33.01 attractivity of the shopping environment 2.50* 18.11 diversity in product choice 4.41* 46.40 shopping success 3.62* 39.71 ….. Cognitive costs 2.29* 57.18 Individual error component 0.72* 38.59

18 Results – deriving attribute utilities
Attribute utility values and willingness to travel (example) Attribute Value (utility) (minutes travel time) crowdedness in the store -72.99 40.0 crowdedness on the way -67.27 36.9 departure time flexibility of the TM 54.70 30.0 accessibility of the store 50.01 27.4 simplicity of the travel route 46.81 25.7 time to find a parking lot -37.27 20.4 atmosphere in the shopping location 37.00 20.3 familiarity with the shopping location 35.39 19.4

19 Results – deriving attribute utilities
Attribute and benefit utility impact rankings

20 Conclusions This study has shown how the utility of attributes can be derived from individuals’ means–end chains of decisions This bridges the gap between qualitative insights on the basis of means–end chain or laddering research quantitative insights based on conjoint choice experiments or discrete choice models The new method most useful under market conditions in which supply is relatively complex, highly stable over time, and narrow in terms of the number alternatives that any given customer will consider

21 Conclusions The proposed approach based on means–end chains provides
a viable way to generate in-depth insights about consumer attribute consideration that can be quantified in terms of consumer decision utility and used as input for discrete choice experiments

22 Thank you!


Download ppt "A New Approach to Measure Preferences of Users in Built Environments: Integrating Cognitive Mapping and Utility Models Benedict Dellaert Erasmus University."

Similar presentations


Ads by Google