Available online at www.sciencedirect.

com

ScienceDirect
Available online at www.sciencedirect.com
Procedia Computer Science 00 (2018) 000–000
www.elsevier.com/locate/procedia
ScienceDirect
Procedia Computer Science 124 (2017) 536–543

4th Information Systems International Conference 2017, ISICO 2017, 6-8 November 2017, Bali,
Indonesia

The Moderation Effect of Age on Adopting E-Payment Technology

Anggar Riskinanto, Bayu Kelana, Deliar Rifda Hilmawan*
Sekolah Tinggi Ilmu Manajemen dan Ilmu Komputer ESQ, TB Simatupang Kav. 1, Jakarta Selatan, 12560, Indonesia

Abstract

E-payment is one of the fin-tech solutions that has gained users from different generations. This is unique as it is known that
distinctive generations may adopt technology differently. This study aimed to assess on how the adoption of an e-payment system
differs on different generations using the quantitative method. We used TAM model moderated by age as the research model. Data
was collected through an online questionnaire and successfully gathered 532 responses from application users. The findings showed
that only perceived ease of use has a positive effect on perceived of usefulness, that was moderated by age. This result may provide
a new perspective on how e-payment being adopted on Indonesian users.
© 2017 The Authors. Published by Elsevier B.V.
Peer-review under responsibility of the Conference Program Chairs.

Keywords: e-payment, TAM, partial least square

1. Introduction

Technology has become the part of today’s people life even when they don’t realize it. The application of
technology stretching from basic things, such as electricity to the more sophisticated as in financial technology (fin-
tech). The simplicity and speed of this technology have led people to adopt it in their everyday life. This has been
augmented with the invention of the smartphone, allowing people to use applications of fin-tech (eg. e-payment)
directly from their own hand.
However, the use of technology may be different for some generation. For example, older generation may perceive
technology or new inventions as hard to use, while the younger one might accept it with ease [1]–[3]. This phenomenon
raises a question on how different generations perceive new technology, such as e-payment in their daily lives in

* Corresponding author. Tel.: +6221-2940-6999; fax: +6221-2940-6999

E-mail address: anggar.r@esqbs.ac.id

1877-0509 © 2017 The Authors. Published by Elsevier B.V.

Peer-review under responsibility of the scientific committee of the 4th Information Systems International Conference 2017 .

1877-0509 © 2018 The Authors. Published by Elsevier B.V.

Peer-review under responsibility of the scientific committee of the 4th Information Systems International Conference 2017
10.1016/j.procs.2017.12.187
 Anggar Riskinanto et al. / Procedia Computer Science 124 (2017) 536–543 537
2 Riskinanto, Kelana, Hilmawan / Procedia Computer Science 00 (2018) 000–000

Indonesia. This paper will try to find the effects of age or generation on the technology acceptance, especially in
electronic payment.
This study used Technology Acceptance Model (TAM) with age or generation as a moderating variable. We use
this model as it has been the most frequently used to explain the behavioral intention of technology usage. Several
studies have been made, focused on several areas such as banking, education, and manufacturing. On contrary, this
study aims for users in consumer electronic payment. The findings of this study hopefully will serve as a guideline for
companies intend to develop a similar solution.

2. Theoretical Background

2.1. Fintech

Fintech is an acronym for financial technology, which can be interpreted as a way of payment through an electronic
or information technology. It is the convergence of financial service and information technology that provide
innovative services offered to people. Two most known solution in worldwide may be Apple Pay and Google Wallet
[4]. Indonesia also has several solutions regarding fin-tech and one of them is Paytren, which developed by PT. Veritra
Sentosa Internasional. An Android based mobile application, it allows its user to make a different type of payments
of bills, such as electricity, cell phone, gas, etc. Paytren may be seen as an emerging solution as it provides alternative
to the current complex way of paying bills.

2.2. Technology Acceptance Model (TAM)

TAM was developed by Davis [5] and Davis, et al [6] based on Theory of Reasoned Action (TRA) [7]. This model
added two constructs into TRA model, which is perceived usefulness and perceived ease of use. The first TAM model
consists of six main constructs, which the first one is perceived ease of use. This construct is defined as a measure
where the user in the future perceive a system is free from error [5], [8], [9]. The second contruct is perceived ease of
use, which could be measured through indicators of clear and easy to understand, also easy to master [10]–[12]. The
third one is perceived usefulness. It is expressed as a measure where the use of a technology is believed to bring
usefulness to the user who uses it. This construct can be measured with the indicators of productivity increase, effective
and faster work [11], [12]. The fourth construct is attitude toward using. This construct is perceived as an attitude of
the user in accepting or rejecting the use of a technology [8]. It is also could be explained as an acceptance of a user
to an information technology. The fifth construct is behavior intention. This could be described as an intention of users
to do a certain behavior. Several studies show that behavioral intention is a good predictive of the user’s technology
usage [5]. The last construct is actual use. This construct is defined as an external psychometric response that can be
measured by a user based on an actual use [5].
TAM itself has been revised by Venkatesh and Davis [13], being referred as TAM2. The revised model did not
include attitude towards using and added new variables such as experience and subjective norm.

3. Research Model and Hypotheses

The research model used in this study was adopted from the research conducted by Rigopoulus [10]. The model
itself is a modified version of TAM. We have added a moderating effect and name it as age. This moderating variable
is added to see the effect of different generations on adopting e-payment technology. The detail of our research model
could be seen in Figure 1.
538 Anggar Riskinanto et al. / Procedia Computer Science 124 (2017) 536–543
Riskinanto, Kelana, Hilmawan / Procedia Computer Science 00 (2018) 000–000 3

Fig. 1. Research Model.

Based on the literature review, we proposed hypotheses that intended to validate our study. These hypotheses are:
H1: Attitude towards using has a positive effect on actual system use, moderated by age.
H2: Perceived ease of use has a positive effect on actual system use, moderated by age.
H3: Perceived ease of use has a positive effect on attitude towards using, moderated by age.
H4: Perceived ease of use has a positive effect on perceived of usefulness, moderated by age.
H5: Perceived of usefulness has a positive effect on actual system use, moderated by age.
H6: Perceived of usefulness has a positive effect on attitude towards using, moderated by age.

4. Analysis and Results

A survey-based approach is needed for this study as it is intended to test the research model that has been proposed.
For this study, we use questionnaire created in Google Form and sent the link to all Paytren users that located in
Indonesia. The items in it used a 4-point Likert scale from “strongly disagree” to “strongly agree” and it is intended
to avoid neutral responses.
We have distributed the questionnaire from March to June 2017. During that period, we collected 532
questionnaires and used them for our analysis. From the questionnaires, we found no missing data in it so it can be
analysed further without the need to apply missing data replacement. Based on the collected questionnaires, we build
a demographic information listed in Table 1.

Table 1. Demographic Characteristics of Respondents.

Variables Categories Frequency Percent
Gender Male 269 50.6%
Female 263 49.4%
Year of Birth 1946-1966 5 0.9%
1965-1976 52 9.8%
1977-1998 448 84.2%
1999-2012 27 5.1%
Education K12 school 332 62.4%
Junior college 41 7.7%
University 139 26.1%
Master 20 3.8%

From Table 1, we could see that the numbers between male and female respondents are almost similar, with a
percentage of 50.6% and 49.4% respectively. Furthermore, the composition of generations of Paytren users in our
respondents have consisted mostly of age between 1977 to 1998. This so-called Generation Y [14] has the biggest
percentage of all with 84.2%. We could also assume that over the half of respondents still have the basic education
with a percentage of 62.4%, while the rest have higher education.
For this study, we implemented partial least square (PLS) method to help with analyzing the data. PLS is one of
structural equation modeling (SEM) technique that could be used to evaluate the relationship within model [15]. PLS
 Anggar Riskinanto et al. / Procedia Computer Science 124 (2017) 536–543 539
4 Riskinanto, Kelana, Hilmawan / Procedia Computer Science 00 (2018) 000–000

is chosen because of the ability to analyse on small sample sizes [15]–[17]. It also could work better if it has a
prediction objective, the model itself is complex, and the phenomenon is new or changing [16]. We have used
smartPLS application [18] for its simple navigation and features offered.
The analysis in PLS is divided into two steps: measurement model analysis and structural model analysis [19]. The
first measurement would implement confirmatory factor analysis (CFA) to investigate the reliability and validity of
the latent variables. Whilst the second measurement would test the hypotheses by examining path coefficients and
their significance.

4.1. Measurement Model Analysis

The analysis for measurement model would provide a confirmatory assessment of reliability, convergent validity,
and discriminant validity [19]. The first step to measure the model was investigating the value of loading factor of
model’s variables. This can be done by running the PLS algorithm function in smartPLS. After running this, we found
that one of the variables (ASU06) was below the requirement of 0.4 [20] with the value of 0.021. Based on this finding,
we removed the variable and run the test once again. The result of the second outer model test is described in Table
2.
Table 2. Measurement Model Results.
Construct Item Loadings AVE CR Cronbach’s 
ASU ASU01 0.877 0.706 0.923 0.895
ASU02 0.874
ASU03 0.849
ASU04 0.783
ASU05 0.815
ATU ATU01 0.888 0.807 0.954 0.940
ATU02 0.916
ATU03 0.918
ATU04 0.860
ATU05 0.908
PEOU PEOU01 0.874 0.779 0.955 0.943
PEOU02 0.891
PEOU03 0.899
PEOU04 0.864
PEOU05 0.906
PEOU06 0.859
POU POU01 0.870 0.775 0.954 0.942
POU02 0.889
POU03 0.835
POU04 0.916
POU05 0.912
POU06 0.858

From Table 2, we could see that the Cronbach’s  values range from 0.895 (for ASU) to 0.943 (for PEOU). All
values have met the lowest value of 0.7 according to Nunally [21]. We also have assessed the individual item reliability
by examining the factor loadings of all variables with their respective constructs. The result showed that all variables
range from 0.783 for ASU04 to 0.918 for ATU03, which met with the minimal value of 0.7 [22]. As for final
measurement of our model’s reliability, we checked the composite reliability. The result showed that all constructs
met the requirement of 0.7 [23], [24]. We thus concluded that the reliability of our research model was acceptable.
Another measurement that needs to be taken is convergent validity which refers to the degree of agreement of
measuring the same concept [15]. In this study, the convergent validity could be assessed by examining the values of
average variance extracted (AVE). As seen in Table 2, the AVE for all constructs was above 0.5, which met the
requirement [24]. This indicates that there is a good convergent validity for the scales.
As for the final measurement of our research model, we assessed the discriminant validity. This could be done by
measuring the variance shared between the construct and other constructs [24], [25]. The result of the computation is
shown in Table 3.
540 Anggar Riskinanto et al. / Procedia Computer Science 124 (2017) 536–543
Riskinanto, Kelana, Hilmawan / Procedia Computer Science 00 (2018) 000–000 5

Table 3. Fornell-Larckers Criterion.

ASU ATU PEOU POU
ASU 0.840
ATU 0.809 0.898
PEOU 0.794 0.841 0.882
POU 0.819 0.891 0.853 0.880
Note: Square root of AVE is on bold

As seen in Table 3, all the square roots of AVE were larger than their corresponding coefficients of correlation
with other factors. This indicates that all constructs meet the discriminant validity test [23]. From these results, we
concluded that our analyses have met both convergent and discriminant validity.

4.2. Structural Model Analysis

The second step in analysing using the PLS method is by assessing the structural model. This is done by running
the bootstrap resampling technique [26] with 5000 iterations to ensure stability [20], [23]. This provides analyses on
hypotheses and constructs’ relationship based on examination of standardized paths. The result of our assessment is
displayed in Table 4 and Fig. 2.

Table 4. Hypotheses Testing Results

Hypothesis Relationship Std. Beta Std. Error t-value Decision
H1 ATU -> ASU 0.280 0.073 3.860*** Supported
H2 PEOU -> ASU 0.264 0.063 4.201*** Supported
H3 PEOU -> ATU 0.298 0.052 5.798*** Supported
H4 PEOU -> POU 0.853 0.021 40.693*** Supported
H5 POU -> ASU 0.345 0.076 4.536*** Supported
H6 POU -> ATU 0.636 0.051 12.402*** Supported
*p < 0.10; **p < 0.05; ***p < 0,01

Fig 2. Structural Model Results.

Our research model explains 72.76% of variations in perceived of usefulness, 81.81% in attitude towards using,
and 71.85% in actual system use. A path analysis was also performed to test all the hypotheses that we have specified
before. Table 3 showed that all hypotheses are supported with t-value ranging from 3.860 to 40.693. Perceived ease
of use (ß = 0.853; p < 0.01) was significant in explaining perceived of usefulness. Both perceived of usefulness (ß =
0.636; p < 0.01) and perceived ease of use (ß = 0.298; p < 0.01) were significant in explaining attitude towards using.
Perceived of usefulness (ß = 0.345; p < 0.01), attitude towards using (ß = 0.264; p < 0.01), and perceived ease of use
(ß = 0.280; p < 0.01) were significantly explaining actual system use.
Further testing was performed to assess the effect of age or generation on all constructs. This moderator effect will
explain if a variable has affected the relation’s direction or strength between independent and dependent variables
 Anggar Riskinanto et al. / Procedia Computer Science 124 (2017) 536–543 541
6 Riskinanto, Kelana, Hilmawan / Procedia Computer Science 00 (2018) 000–000

[27]. The result of our analysis showed in Table 5. It can be concluded that not all hypotheses are supported. Only the
relationship between perceived ease of use (ß = 0.039; p < 0.10) to perceived of usefulness that being affected by age
or generation with t-value of 1.947.

Table 5. Moderator Analysis Results.

Hypothesis Relationship Std. Beta Std. Error t-value Decision
H1 ATU * AGE -> ASU 0.030 0.046 0.652 Not Supported
H2 PEOU * AGE -> ASU -0.015 0.044 0.339 Not Supported
H3 POU * AGE -> ATU 0.035 0.047 0.730 Not Supported
H4 PEOU * AGE -> POU 0.039 0.020 1.947* Supported
H5 POU * AGE -> ASU -0.011 0.054 0.208 Not Supported
H6 PEOU * AGE -> ATU 0.005 0.049 0.094 Not Supported
*p < 0.10; **p < 0.05; ***p < 0.01

We could see that most hypotheses are rejected except for H4 in Table 5. This indicated that age doesn’t affect most
of the constructs relationship. However, the relationship of perceived ease of use has a positive effect on perceived of
usefulness, that moderated by age. Although there is a relationship, we could also see that it can be considered a
relatively weak relationship with p-value of only 0.10.

5. Discussion

The main purpose of this study is to understand the contributing variables of the acceptance of e-payment
technology among Indonesian users, adapting the TAM, moderated by age. Various significant aspects of the results
will be discussed as followed.
Firstly, this study explains that all dependent variables were being significant related to their independent variables.
This finding supports the TAM, proposed by Rigopolous, with the new contributions of the TAM in e-payment
acceptance in Greece [10]. However, Rigopolous did not analyze variance of its variable. In the other research, Jaradat
analyze the variance of extended TAM in mobile payment adoption in Jordan [28]. In Jaradat’s study, the variance of
Attitude Towards Using (ATU) and Perceived of Usefulness (POU) explained in the model is 41% and 61%
(respectively), in an analysis of a mobile payment adoption in Jordan. In our case, for an e-payment adoption in
Indonesia, such as ATU, the variance explained in is 81.8% and POU is 72.8%, higher than the values found by
Jaradat.
Secondly, Rigopoulos proposed statistically six significant relationships for TAM in e-Payment in Greece [10].
Among them, the relationship between Perceived Ease of Use (PEOU) and Perceived of Usefulness (POU) variables
has the strongest relationship in the model. Then, the relationship of Attitude Towards Using (ATU) and Actual
System Use (ASU) variables has the weakest relationship in the model. In Indonesia, this research found the different
result in the weakest relationship, which is between PEUO and ASU variables. We believe that these differences in
these two research-works may be due to cultural differences, as these works were carried out with samples from
banking customer in Greece and e-payment application user in Indonesia.
Finally, the power of all variable relationship in this study was decreased after moderated by age. All variables
shown have no significant relationship, except for the relationship between Perceived Ease of Use (PEOU) and
Perceived of Usefulness (POU). Compared to Jaradat and Rigopoulus, this study gives a new insight that TAM,
moderated by age, on e-payment adoption among users in Indonesia.

6. Conclusion, Limitation, and Further Research

In this study of e-Payment adoption in Indonesia, moderated by age, the most significant variables are Perceived
Ease of Use (PEOU) and Perceived of Usefulness (POU). This result is not different from study in Greece, without
age moderation.
A large sample in Generation Y is one of the limitations of the study. The distribution of the survey instrument was
limited to one particular user setting, at one point in time, and was therefore limited for broad generalizations. Further
research that represents different age groups with a balanced composition is advised.
542 Anggar Riskinanto et al. / Procedia Computer Science 124 (2017) 536–543
Riskinanto, Kelana, Hilmawan / Procedia Computer Science 00 (2018) 000–000 7

References

[1] K. E. Olson, M. A. O’Brien, W. A. Rogers, and N. Charness, “Diffusion of technology: Frequency of use
for younger and older adults,” Ageing Int., vol. 36, no. 1, pp. 123–145, 2011.
[2] Age Uk, “Technology and Older People Evidence Review,” 2011.
[3] B. Y. M. Anderson and A. Perrin, “Tech adoption climbs among older adults,” 2017.
[4] W. Lumpkins and M. Joyce, “Near-Field Communication: It Pays: Mobile payment systems explained and
explored,” IEEE Consumer Electronics Magazine, vol. 4, no. 2. pp. 49–53, 2015.
[5] F. Davis, “Perceived usefulness, perceived ease of use, and user acceptance of information technology,” MIS
Q., vol. 13, no. 3, pp. 319–340, 1989.
[6] F. D. Davis, R. P. Bagozzi, and P. R. Warshaw, “User Acceptance of Computer Technology: A Comparison
of Two Theoretical Models,” Manage. Sci., vol. 35, no. 8, pp. 982–1003, 1989.
[7] M. Fishbein and I. Ajzen, “Belief, Attitude, Intention and Behaviour: An Introduction to Theory and
Research,” Read. MA AddisonWesley, no. August, p. 480, 1975.
[8] F. D. Davis, “User Acceptance of Information Technology: System Characteristics, User Perceptions and
Behavioral Impacts,” Int. J. ManMachine Stud., vol. 38, no. 3, pp. 475–487, 1993.
[9] Y.-S. Wang, Y.-M. Wang, H.-H. Lin, and T.-I. Tang, Determinants of user acceptance of Internet banking:
an empirical study, vol. 14, no. 5. 2003.
[10] G. Rigopoulos and D. Askounis, “A TAM Framework to Evaluate User’s Perception towards Online
Electronic Payments,” J. Internet Bank. Commer., vol. 12, no. 3, pp. 1–6, 2007.
[11] D. Gefen, E. Karahanna, and D. W. Straub, “Trust and TAM in Online Shopping: An Intergrated Model,”
MIS Quaterly, vol. 27, no. 1, pp. 51–90, 2003.
[12] N. Yahyapour, “Determining Factors Affecting Intention to Adopt Banking Recommender System,”
Technology, 2008.
[13] V. Venkatesh, F. D. Davis, and S. M. W. College, “Theoretical Acceptance Extension Model : Field Four
Studies of the Technology Longitudinal,” Manage. Sci., vol. 46, no. 2, pp. 186–204, 2000.
[14] A. B. Acar, “Do intrinsic and extrinsic motivation factors differ for Generation X and Generation Y?,” Int. J.
Bus. Soc. Sci., vol. 5, no. 5, pp. 12–21, 2014.
[15] M. Wang and T. Yang, “Asia Paci fi c Management Review Investigating the success of knowledge
management : An empirical study of small- and medium-sized enterprises,” Asia Pacific Manag. Rev., vol.
21, no. 2, pp. 79–91, 2016.
[16] W. W. Chin and P. R. Newsted, “Structural equation modeling analysis with small samples using partial
least squares.,” Stat. Strateg. Small Sample Res., no. January 1999, pp. 307–341, 1999.
[17] H. Wold, “Soft modelling: the basic design and some extensions,” Syst. under Indirect Obs. Causality-
structure-prediction, pp. 1–54, 1982.
[18] A. (2005). S. 2. 0. M. H. S. R. from http://www. smartpls. co. Ringle, Christian M., Wende, Sven, Will,
“SmartPLS 2.0.M3.,” Hamburg: SmartPLS., 2005. .
[19] J. Anderson and D. Gerbing, “Structural Equation Modeling in Practice: A Review and Recommended Two-
Step Approach,” Psychol. Bull., vol. 103, no. 3, pp. 411–423, 1988.
[20] J. F. Hair Jr, M. Sarstedt, L. Hopkins, and V. G. Kuppelwieser, “Partial least squares structural equation
modeling (PLS-SEM),” Eur. Bus. Rev., vol. 26, no. 2, pp. 106–121, 2014.
[21] J. C. Nunnally and I. H. Bernstein, “Psychometric theory,” PsycCRITIQUES, vol. 24, pp. 275–280, 1979.
[22] J. F. J. Hair, G. T. M. Hult, C. Ringle, and M. Sarstedt, A Primer on Partial Least Squares Structural
Equation Modeling (PLS-SEM), vol. 46, no. 1–2. 2014.
[23] W. W. Chin, “Issues and Opinion on Structural Equation Modeling,” MIS Q., vol. 22, no. March, pp. vii–
xvi, 1998.
[24] C. Fornell and D. F. Larcker, “Evaluating Structural Equation Models with Unobservable Variables and
Measurement Error,” J. Mark. Res., vol. 18, no. 1, p. 39, 1981.
[25] D. Compeau, C. A. Higgins, and S. Huff, “Social cognitive theory and individual reactions to computing
technology: A longitudinal study.,” MIS Q., pp. 145–158, 1999.
 Anggar Riskinanto et al. / Procedia Computer Science 124 (2017) 536–543 543
8 Riskinanto, Kelana, Hilmawan / Procedia Computer Science 00 (2018) 000–000

[26] J. Henseler, C. M. Ringle, and R. R. Sinkovics, “The use of Partial Least Squares Path Modeling in
International Marketing,” Adv. Int. Mark., vol. 20, no. 2009, pp. 277–319, 2009.
[27] R. Baron and D. Kenny, “The moderator-mediator variable distinction in social psychological research,” J.
Pers. Soc. Psychol., vol. 51, no. 6, pp. 1173–1182, 1986.
[28] M.-I. R. M. Jaradat and K. M. S. Faqih, “Investigating the Moderating Effects of Gender and Self-Efficacy
in the Context of Mobile Payment Adoption: A Developing Country Perspective,” Int. J. Bus. Manag., vol.
9, no. 11, pp. 147–169, 2014.
[29] H. Kusuma and D. Susilowati, “Determinan Pengadopsian Layanan Internet Banking: Perspektif Konsumen
Perbankan Daerah Istimewa Yogyakarta,” Jaai, vol. 11, no. 2, pp. 125–139, 2007.
[30] K. Eriksson, K. Kerem, and D. Nilsson, “Customer acceptance of internet banking in Estonia,” Int. J. Bank
Mark., vol. 23, no. 2, pp. 200–216, 2005.
[31] Y. Mohd Yusoff, Z. Muhammad, M. S. M. Zahari, E. S. Pasah, and E. Robert, “Individual Differences ,
Perceived Ease of Use , and Perceived Usefulness in the E-Library Usage,” Comput. Inf. Sci., vol. 2, no. 1,
pp. 2–9, 2009.

Appendix A. Questionnaire Items

Constructs Items No. Source

Perceived Ease of Paytren is easy to learn. PEOU01
Use (PEOU) Paytren is easy to control. PEOU02
Paytren is clear and easy to understand. PEOU03
[5], [10]–[12]
The use of Paytren is flexible. PEOU04
Paytren id easy to master. PEOU05
The overall use of Paytren is easy. PEOU06
Perceived of Paytren makes personal transactions faster. POU01
Usefulness (POU) Paytren supports my performance. POU02
Paytren can increase my productivity. POU03
[5], [11], [12]
The use of Paytren is convinient to save time POU04
Paytren can make work easier. POU05
The overall use of Paytren is helpful for me POU06
Attitude Towards Paytren is fun to use ATU01
Using (ATU) Paytren is a good idea ATU02
Paytren is needed to be used to be used to
ATU03 [12], [29]
support transaction activities.
Everyone should use Paytren ATU04
Paytren is a wise idea. ATU05
Actual System Paytren is used repeatedly ASU01
Usage (ASU) I use Paytren more than any other payment
ASU02
service.
I use Paytren for business transaction. ASU03 [10], [30], [31]
I use Paytren for personal transactions ASU04
I use Paytren for all transactions. ASU05
I use Paytren for certain transactions only. ASU06