Understanding Consumer Behaviour in
Information and Communication Technologies
(ICTs)
Alastair W Robertson
A.W.Robertson@lancaster.ac.uk
Department of Management Science
Lancaster University Management
School
Lancaster
LA1 4YX
United Kingdom
What are ICTs?
• ICTs are ‘gadgets’ that people can use to connect to
information and to communicate with one another.
– e.g. Computers and PDA’s.
– Focus here is limited to internet adoption.
90’s
9kbs
5.5M!
95
00
04
56 to
128kbs
512Mbs
2Mbs
Dig
i
Div tal
ide?
Broa
Gen dband
erat
i on
Ex p
l
inte osive
r
g ro w n e t
th
Milit
ar
60’s
Sim
plifi
ed
us e w eb
But why has research into ICT adoption become so important?
y Us
e
•
0?
Discussion highlights a fast changing
market
• What do market stakeholders need to know to be
able to forecast the market better?
• Why and Which consumers adopt technologies!
Knowledge of
consumer
behaviour
Develop
ment of
new
methodol
ogies, or
existing
technique
s reapplied
How to apply
this information
to produce
forecasts
The stakeholder positions
• Digital divide highlights missed revenue or
missed development opportunities and
cost saving.
– Marketers and Business Planners;
• Missed revenue: Untapped market
– Government and Regulators;
• Missed development opportunities: Countries
with less ICT may grow less.
• Missed cost saving: Those on ‘wrong’ side of
divide use government services more frequently.
Focus of the Research
• Human characteristics and ICT
Technology
acceptance
(perceptions)
2 Levels
of HC
Socio-Economic
(e.g. income, age)
Combine to
produce
consumer
groups with
unique ICT
characteristics
Segment using
perceptions,
confirm segment
validity using
socio-economic
Application of Human Characteristics
• Segmentation;
– measure of the digital divide?
• Model estimation;
– Application of choice modelling.
– confirmation of what drives the digital divide.
• Experimentally, applied to the diffusion modelling
process;
– Introduces idea that that segmental diffusion curves can be
estimated.
– Estimate of how digital divide may evolve over time.
Social Psychological Factors
• Massive literature on TAM;
– Google ‘technology acceptance model’
– A large number of test applications.
Fred Davis (1989);
Perceived
Usefulness
External
Variables
Attitude, Use
of Tech
Perceived
Ease of Use
Behavioural
Intention
Actual
Usage
• Adams (1992): Usefulness and Ease of Use
perceptions
– Applicable to diagnosis of user acceptance in
technologies in general
– Especially applicable when adoption is voluntary
• Igbaria et al (1996): TAM research justified due
to extensive expansion into ICTs by businesses,
but low final use
– Similar to residential ICT adoption?
Enjoyment and ICT adoption
• An obvious point,
– If computers become more enjoyable to use, their
adoption and usage will increase, Igbaria (1996)
– Perceived enjoyment distinct from U and EoU
– Three perceptions are measurable at the
consumer level
• EoU, U and E
First internet test of TAM, Teo (1999)
Usefulness
Ease of Use
Internet Usage
All perceptions
important, but
usefulness more
so….
Enjoyment
• Where next?
– New application of the TAM perceptions…
An application of TAM, Survey of UK
Households
– Extensive data collected from 1286 HHs.
– Data was weighted to minimise non-response bias.
Expected ICT
Utility
Enjoy,
Comp
Useful,
Comp
Easy,
Comp
Enjoy,
Net
Useful,
Net
Easy,
Net
Simple Application of Expected ICT Utility
Divide up the measure into arbitrary segments.
Merge ‘similar’ segments i.e. if demographically similar.
Measure known characteristics for each segment.
For each segment, measure their proportion in the data;
This is an estimate of the proportion of consumers in the UK of
this utility level.
100%
80%
60%
40%
20%
0%
Seg1
Seg2
Seg3
Seg4
Utility level
Age by Utility Level
Percentage within
Segment
NB: No
difference in
gender…
Percentage within
segment
Computer Adoption by Utility Level
80%
60%
40%
20%
0%
18 to 29
Seg1
30 to 39
Seg2
Seg3
40 to 49
Seg4
50 and above
Percentage within
segment
Educational Attainment by Utility Level
100%
80%
60%
40%
20%
0%
Seg1
Level 4
Level 3
Seg2
Level 2
Seg3
Other
Seg4
No Qualifications
Percentage within
Segment
Individual Disposable Income by Utility
Level
100%
80%
60%
40%
20%
0%
Seg1
Seg2
£7,500-£11,249
£18,750-£26,249
£33,750 and above
Seg3
Seg4
£11,250-£18,749
£26,250-£33,749
• Complex approach
– Incorporate expected ICT Utility with other strategies
→Stage 1, fairly common procedure
• Estimate consumer choice model
– e.g. logit
Factors drive
the choices…
ICT Choice
No internet
Narrowband Broadband
• Use the model to define segments via expected ICT utility
– Estimate segmental price sensitivities
How is logit output
interpreted?
Logit estimates ICT adoption probabilities given a set
of inputs, much like regression;
Pri ( j ) =
exp
n
( β1 x i 1 + β 2 x i 2 +...+ β k x ik )
∑ exp
( β1xi 1+ β 2 xi 2 +...+ β k xik )
j =1
Factor effects
facing
consumer β
Factor effects
facing
consumer xi
Where Pri ( j ) =
Probability
individual i
purchases
product j given xs
Technology Levels in the Home
Odds Ratio
Narrowband
Broadband
50
40
30
20
10
0
Tech 2 HH
Compare to
missing category
Tech 3 HH
Hi-tech HH
Technology Level
Odds Ratio
Household Educational Attainment
0.6
0.5
0.4
0.3
0.2
0.1
0
Level 3
Level 2
Narrowband
Other
No Qual.
Broadband
Presence of Children
Odds Ratio
3
2.5
2
1.5
1
0.5
0
Narrowband
Broadband
Internet Choice Elasticity: +0.2
Point: Overall effects stronger for
broadband than for narrowband
Insight: Demographic effects dissipate
over time…
Segmentation Results
Table 2: ICT utility segments
ICT Utility Level
Household
Description
Broadband
Price
Elasticity
Computer
Adoption
Low
(2.2%)
Low income and
educ., retired,
unemployed.
-1.77
8%
Low to Mid
(13%)
Moderately better
income, slightly better
educ. Blue collar.
-1.51
21%
Mid to High
(22.7%)
Good income and
educ., white collar,
possibly kids.
-1.42
65%
High
(62.1%)
High income and
educ., 25% with kids,
love technology.
-1.28
83%
→Stage 2, new and experimental procedure
• Apply price forecast to the model for each
segment
Broadband Price
Pt = exp
Forecast
( 4.04 − 0.22 t )
2015
2014
2013
2012
2011
2010
2009
2008
2007
2006
2005
2004
2003
2002
2001
£50
£40
£30
£20
£10
£0
2000
Price
Broadband Price Forecast
Segmental
Adoption
Probability
Work resulting segmental
adoption probabilities to the
diffusion process.


 1 − e−(ps +qs ).t 

Ns (t) = MProb(s| Pr)t 
1 + qs e−(ps +qs ).t 
 p

s


‘Moving’ Social System
Size N
Segmental
Innovation
and Imitation
Parameters
Low to Medium Utility
High Utility
High Utility
Adopt First:
Innovators
20
15
20
13
20
11
20
09
20
07
20
05
20
03
4
3.5
3
2.5
2
1.5
1
0.5
0
20
01
Number of New Adopters in Year
Segmental Diffusion Rate
Medium to High Utility
Aggregated Segments
Low Utility are
the Laggards
20
00
20
02
20
04
20
06
20
08
20
10
20
12
20
14
Number of
Households, Millions
30
25
20
15
10
5
0
Segmental Diffusion
Low to Medium Utility
High Utility
Broadband Actual
Medium to High Utility
Aggregated Segments
Bass
Segmental AEUD
Bass tends to
under perform
in presence of
limited data
20
14
20
12
20
10
20
08
20
06
20
04
20
02
25
20
15
10
5
0
20
00
Number of
Households, Millions
Model Comparison
Aggregate AEUD
Broadband Actual
Segmentation approaches: Simple versus
Econometric
Percentage
80%
62.1%
60%
42.1%
40%
13.0% 22.7%
20%
0%
47.6%
3.4%
6.9%
2.2%
Simple
Seg1
Which to
choose?
Econometric
Seg2
Seg3
Seg4
Recap.
• The presentation has introduced
expected ICT utility as a segmentation
variable for residential ICTs;
– Created from a sound theoretical
foundation.
– Applications go some way to confirm
validity of the measure.
Issues
• More testing needed to confirm validity;
– New survey is proposed for next year.
– Same HHs, two year interval.
– Track segmental shifts.
• Wider tests required;
– Different applications (e.g. wireless apps.).
– Different countries.
Future Possibilities
– Other applications may exist for this
variable also, especially if captured
regularly in time;
• Comparable measures across countries.
Expected ICT Utility it = f (Socio - economic factors, regulatory framework )it
Where i could be
individual or country…