Aljukhadar, M. Sénécal, S., and Daoust, C-E (2009), Information Overload Reexamined: Recommendation Agents as Consumer Response Heuristic and Effects on Choice Quality, Choice Confidence, and e-store Evaluation,” ACR 2009, Pittsburgh, PA.
1. A. INTRODUCTION
Given the internet’s low search costs, consumers are particularly exposed to elevated information loads during
online product choice and purchase. Because consumers are more loyal to e-stores that enable them to function
efficiently, and considering the increasing weight of e-commerce in today’s economy, developing an understanding of
how information-rich environments affect online consumer decision making is of crucial importance.
Many studies show that information overload has numerous undesired consequences on consumers’
choice and purchase, such as decline in choice quality and performance. Other researches indicate that consumer
decision support systems (e.g. recommendation agents) are decision heuristics that partly alleviate consumers ’
processing effort while maintaining an acceptable level of choice accuracy . They have the potential to reduce
information overload and search complexity while improving decision quality . Moreover, the effects of
information overload and recommendation agent use on retailers’ e-store perceived quality are not investigated in
the literature, although it has been showed to be a main driver of consumers’ behavioural intentions.
This innovative study consequently addresses several gaps in the information overload literature and examines
the joint effect of information overload and recommendation agent consultation on an important e-store
performance measure (e-store quality) as well as the effects on choice quality and consumers’ psychological states .
2. B. CONCEPTUAL FRAMEWORK AND THEORICAL BACKGROUND
INFORMATION BITS INFORMATION OVERLOAD
Alternatives / Attributes / Info Structure Perceptions
NEED FOR
COGNITION
E-STORE
EVALUATION
RECOMMANDATION AGENT Site quality dimensions
Consultation and Outcome
CHOICE
CONFIDENCE /
CHOICE QUALITY DIFFICULTY
Distance btw utility fct of Consumers’
optimal and actual choice psychological states
3. C. METHOD
1 – Pretest
(1) To execute manipulation checks, (2) to ensure relevance of product category, and (3) to determine
most important product attributes to be included in main experi ment
Method:
o 6 versions of the questionnaire (only 1 on 3 section differed throughout every version number)
o 77 participants from a snowball sample
Results:
o Manipulation checks successful (number of alternatives; attributes levels; info structure)
o Product category is relevant: high levels of experience and familiarity observed within participant base
o 35 out of the 45 considered product attributes included in main experiment based on attributes weights
2 – Laboratory experiment
Method:
o 466 participants
o Complete website created for a fictitious retailer
o Task consists in choosing one laptop with the option to consult a recommendation agent prior to final choice
o Recommendation was based on each participant’s utility function and represented optimal choice
o Retailer’s and laptop manufacturers’ names were not disclosed to prevent bias
o Presentation order of laptops was randomized for each participant to avoid presentation bias
4. C. METHOD (CONTINUED)
Variable manipulations:
o Manipulation levels were adapted from literature to ensure appropriate information load increments
o Participant randomly assigned to 1 of 18 experimental conditions (3 X 3 X 2)
6, 18 or 30 alternatives in product choice set
15, 25 or 35 attributes per alternative in product choice set
Disproportional (more diagnostic info) or proportional (less diagnostic info) distribution of attribute levels
across alternatives
Measures:
o Measurement of participants’ utility functions (based on personal attributes evaluation) and some of the
dependant variables occurred prior to task execution
o Scales were adapted/translated from existing reliable and valid scales
o Dependant variables:
perceived information overload (2 items)
choice confidence (3 items)
choice difficulty (1 items)
choice quality (distance between utility functions of optimal and actual choice )
need for cognition (18 items)
site information quality (3 items)
site interactivity (4 items)
o Control variables:
consumer product experience (3 items)
product category involvement (4 items)
sex, age, education, income
5. Effect of Recommendation Consultation Supported relation
D. RESULTS on Site Interactivity is not significant.
All other relations are supported.
Not supported relation
INFORMATION BITS INFORMATION OVERLOAD
Alternatives / Attributes / Info Structure Perceptions
NEED FOR
COGNITION
E-STORE
EVALUATION
RECOMMENDATION AGENT Site quality dimensions
Consultation and Outcome
CHOICE
CONFIDENCE /
CHOICE QUALITY DIFFICULTY
Distance btw utility fct of Consumers’
optimal and actual choice psychological states
6. E. CONCLUSION
This research has several novel findings and implications on research and practice . It contributes to the
information overload literature by integrating the traditional and structural met hods of product information
measurement and by investigating the relationship between information bits and overload perceptions. Results
show that consumers evolving online in an information-rich environment tend to employ information-
processing heuristics in response to information overload, with strong positive impact s on choice
quality. Indeed, this study is one of the first to establish a link between shopping strategy (use or non -use of a
recommendation agent) and performance (choice objective quality and e-store quality). It is also one of the first
studies to link a shopping outcome (final product chosen) to the e -store’s perceived quality (i.e., site interactivity).
Integrating a recommendation agent that considers consumers’ preferences and person alizes
product choice accordingly appears to be beneficial for both consumers (by improving choice quality and
confidence) and retailers (by enhancing the e-store quality and helping consumers make a choice of higher quality) .