Matching Task and Technology Characteristics to Predict mHealth
Tool Use and User Performance
A Study of Community Health Workers in the Kenyan Context
Maradona Gatara and Jason F. Cohen
Department of Information Systems, University of the Witwatersrand (WITS),
1 Jan Smuts Avenue, Braamfontein, 2000, Johannesburg, South Africa
Keywords: Mobile-Health, Community Health Worker, Information and Communication Technologies for
Development, Information and Communication Technologies for Community Health Workers, Healthcare
Service Delivery, Kenya, Task-Technology Fit, Use, User Performance.
Abstract: Equipping Community Health Workers (CHWs) in resource-constrained settings with mobile-health or
‘mHealth’ tools has the potential to improve healthcare service delivery. mHealth tool functionality must
however match CHW task needs before these tools are likely to have any significant impacts on CHW
performance. This paper contributes by drawing on Task-Technology Fit theory to test the extent to which a
match between CHW tasks and mHealth technology characteristics influences the performance of 201
CHWs using an mHealth tool in the counties of Siaya, Nandi, and Kilifi in Kenya. Results showed that the
interaction of paired task and technology characteristics did not always impact mHealth tool use and user
performance in the manner expected. When mHealth tool functions matched the task interdependence and
information dependency needs of the CHWs then CHW performance increased but CHW performance
decreased for some CHWs when mHealth functionality for time criticality and mobility was high.
Moreover, while information dependency had an independent positive effect on mHealth tool use, CHWs
came to depend less on the mHealth tool to support time criticality, interdependence, and mobility needs
when functional support was high. These findings have implications for the design and deployment of
mHealth tools.
1 INTRODUCTION
Many developing countries are deploying
Community Health Workers (CHWs) to deliver
lifesaving and high impact interventions at the
household level. One way in which CHWs can be
supported in their delivery of healthcare services is
by equipping them with supplementary mHealth
tools (Earth Institute, 2010; Liu, Sullivan, Khan,
Sachs and Singh, 2011). These technologies promise
an improvement in CHW performance by supporting
their needs to access health data at points of care,
coordinate and share health data with co-workers,
and to facilitate mobility as CHWs travel to various
household locations (Teo and Men, 2008; Junglas,
Abraham and Ives, 2009; Yuan, Archer, Connelly
and Zheng, 2010). Unfortunately, full-scale mHealth
deployment has been limited with many
unsustainable pilot projects that fail to scale up
meaningfully (LeMaire, 2011; Liu et al., 2011).
More substantive evidence is thus needed on how
these technologies can be designed to match with
CHW service tasks and improve their performance.
The purpose of this paper is to address this gap by
examining whether a ‘match’ between CHW tasks
and mHealth technology characteristics, influences
CHW performance. Task-technology fit theory
provides the theoretical underpinning for the study.
CHWs using an mHealth tool and operating in the
Siaya, Nandi and Kilifi counties in Kenya formed
the empirical context for the study. The results
reported in this paper are part of a larger project into
the use and impacts of mHealth on CHWs operating
in Kenya.
454
Gatara M. and Cohen J..
Matching Task and Technology Characteristics to Predict mHealth Tool Use and User Performance - A Study of Community Health Workers in the
Kenyan Context.
DOI: 10.5220/0005223504540461
In Proceedings of the International Conference on Health Informatics (HEALTHINF-2015), pages 454-461
ISBN: 978-989-758-068-0
Copyright
c
 2015 SCITEPRESS (Science and Technology Publications, Lda.)

2 BACKGROUND
2.1 Task and Technology
Characteristics in the mHealth
Context
CHWs perform various health monitoring,
promotion and referral tasks. These tasks are
characterised by varying degrees of rigidity in time
structuring e.g. promptly responding to medical
emergencies versus occasional follow-up care.
These tasks may also be interdependent (Teo and
Men, 2008), such that they require the coordination
and sharing of health data with practitioners in
clinics and hospitals. Moreover, CHWs need to
perform their tasks at varying locations (Yuan et al.,
2010), as CHWs must travel to different households
as the points at which they deliver healthcare
services. At these points of care, the CHWs tasks
may require them to access dynamic health data e.g.
on the location of medical and field supplies or
equipment needed during household visits (Yuan et
al., 2010). The ability of CHWs to perform their
tasks well can therefore be hampered by a lack of
understanding of temporal needs such as time
criticality. Spatial needs such as mobility, and
information sensitive needs such as interdependence
and information dependency, are equally important
for typical monitoring, promotion, and referral tasks
(Yuan et al., 2010). Thus, Time Criticality,
Interdependence, Mobility, and Information
Dependency emerge as critical task characteristics
relevant to CHW work.
Technologies comprise tool or system functions
that are intended to support users in the performance
of their tasks (Goodhue and Thompson, 1995). In
light of the critical CHW task needs outlined above,
mobile technologies should allow time-critical
notifications (for example via SMS) to be made to
remind CHWs when a task has to be performed
urgently (Yuan et al., 2010). mHealth tools can
better equip CHWs to coordinate and share
information with co-workers (Yuan et al., 2010), and
also offer location-sensitive functionality by
providing information to CHWs as they perform
their tasks and move between households (Junglas et
al., 2009; Yuan et al., 2009). mHealth tools should
thus support CHW needs for Time Criticality,
Interdependence, Mobility, and Information
Dependency (Earth Institute, 2010). These
complementary task and technology characteristics
considered relevant to the Kenyan context are
summarized in Table 1.
2.2 The Match between Task and
Technology Characteristics in the
mHealth Context
Task-Technology Fit (TTF) follows the premise that
a user will use a technology if it meets their needs by
providing adequate functional support for the user’s
task (Goodhue and Thompson, 1995; Dishaw and
Strong, 1998). Although TTF can be examined using
varied analytical schemes (Venkatraman, 1989) such
as fit as moderation or fit as mediation, this study
views fit as a theoretically defined match between
two related variables (Venkatraman, 1989; Dishaw
and Strong, 1998; Premkumar et al., 2005). In this
study’s context, this is the paired match between
task characteristics, i.e. Time Criticality,
Interdependence, Mobility, and Information
Dependency, which reflect CHW needs, and
technology characteristics, i.e. Time Criticality
Support, Interdependence Support, Mobility
Support, and Information Dependency Support,
which reflect mHealth tool functions. Fit as
Table 1: Task and Technology Characteristics.
Task (User Needs) Technology (Tool Functions)
Time Criticality
Degree to which user needs to
perform tasks urgently (Gebauer and
Tang, 2008)
Degree to which tool supports
urgency by providing timely
alerts (Wixom and Todd, 2005;
Junglas et al., 2009)
Interdependence
Degree to which user needs to
coordinate with co-workers (Teo and
Men, 2008)
Degree to which tool supports
coordination by pooling
information (Teo and Men, 2008)
Mobility
Degree to which user needs to move
between locations (Yuan et al., 2010)
Degree to which tool supports
user movement by tracking
locations (Yuan et al., 2010)
Information
Dependency
Degree to which user needs to access
point of care data (Yuan et al., 2010)
Degree to which tool supports
access to location-sensitive point
of care information (Yuan et al.,
2010)
MatchingTaskandTechnologyCharacteristicstoPredictmHealthToolUseandUserPerformance-AStudyof
CommunityHealthWorkersintheKenyanContext
455

Matching is depicted as a 4 x 4 matrix (see Figure 1)
where TTF occurs along the diagonals as an
interaction between complementary task and
technology characteristics (Dishaw and Strong,
1998).
Figure 1: Task-Technology Fit (TTF) as Matching: 4 x 4
Matrix (Based on Dishaw and Strong, 1998).
As per TTF theory (Goodhue and Thompson,
1995), Use and User Performance are consequences
of the fit resulting from a match between task and
technology characteristics. Use can be defined in a
number of ways, including as a dependency,
described as the extent to which the user comes to
rely on the technology (Junglas et al., 2009). User
Performance is defined in terms of the effectiveness,
efficiency, and quality with which the task is
performed. Amongst CHWs, this implies task
productivity, completion of tasks using minimal
resources to do the most in the least amount of time,
and whilst committing minimal errors with
improved decision-making in the reporting of typical
monitoring, promotion, and referral tasks. This study
posits that the match between complementary pairs
of task and technology characteristics will result in
increased Use and User Performance (Teo and Men,
2008). In other words, the better the fit as a match
between healthcare service task and mHealth tool
characteristics, the more likely it is that CHWs will
become dependent on the mHealth tool, and the
more likely it is that their performance will be
enhanced.
3 METHODS
3.1 Study Design
Data collection for this study formed part of a larger
research project into the use and impacts of mHealth
tools within the Kenyan CHW context. A structured
survey instrument was administered to CHWs in
Kenya, based in the county locations of Siaya,
Nandi, and Kilifi. The instrument was used to elicit
data from these CHWs on their task needs, their
perceptions of the mHealth tools deployed to them,
and their use of the mHealth tools in their work.
Data was also collected on the performance of
CHWs including both objective and perceptual
measures of performance. Only the data collected on
mHealth tool functionality, CHW task needs, CHWs
dependence on the tool and its perceived
performance impacts is reported in this paper.
3.2 Sampling Strategy
The research project relied on proportionate
stratified sampling with systematic random sampling
as the sampling strategy. In each county,
Community Health Units (CHU’s) made up of
CHW’s were targeted, and a proportional number
was obtained from lists of CHWs operating in each
of these units. The numbers drawn made up
sampling frames for each of the three counties i.e.
Siaya, Nandi, and Kilifi. Overall, 312 CHWs
constituted the sampling frame devised, and
subsequently participated in the project with 201
providing usable data for the purposes of this study.
3.3 Instrument Measurement
As reported elsewhere (Gatara and Cohen, 2014a;
Gatara and Cohen, 2014b), to measure each
construct, the instrument employed multi-item
scales. To capture respondents’ perceptions along
the four task and technology characteristics, use, and
user performance, 42 seven-point Likert scale items
were used. These measures were drawn from prior
validated instruments (Junglas et al., 2009; Gebauer
and Tang, 2008; Lin and Huang, 2008; Teo and
Men, 2008; Wixom and Todd, 2005; Yuan et al.,
2010), and then refined through pre and pilot testing.
Time Criticality items reflected CHW needs to
respond urgently, and start and finish tasks on time.
Interdependence items reflected CHW needs to
coordinate and share information with co-workers,
clinics, and hospitals. Mobility items reflected needs
of CHWs to perform tasks as they move from one
location to another. Information Dependency items
reflected CHW needs for access to information on
the location of medical supplies and equipment at
points of care such as households. Time Criticality
Support items reflected the mHealth tools rapid
response functions and provision of timely data to
CHWs. Interdependence Support items reflected the
mHealth tool’s coordination functions, in pooling of
data from co-workers, hospitals, and clinics.
Mobility Support items reflected the mHealth tools
HEALTHINF2015-InternationalConferenceonHealthInformatics
456

location tracking functions for CHWs on the ground.
Information Dependency Support items reflected the
mHealth tool’s functions to identify inventory i.e.
medical supplies and equipment at points of care.
The variables employed to measure Use comprise
three perceptual seven-point Likert scale measures.
The items reflect the degree to which CHWs became
dependent on using the mHealth tool Junglas et al.,
2009). User Performance measure comprised eight
seven-point Likert scale measures reflecting CHW
perception of effectiveness, efficiency, and quality of
care (Junglas et al., 2009; Torkzadeh and Doll,
1999).
3.4 Computation of Task-Technology
Fit as Matching
Task-Technology Fit was modelled using the
interaction approach (Venkatraman, 1989), where
Fit is modelled as a product of complementary pairs
of task and technology characteristics as expressed
in the following formula.
Time Criticality Fit, Interdependence Fit, Mobility
Fit, and Information Dependency Fit were each
modelled in this way. These dimensions of Task-
Technology Fit form the diagonals of the 4 x 4
matrix shown in Figure 1. The four Fit dimensions
were tested for their effects on Use and User
Performance. This was achieved using two models
each for Use and User Performance, as expressed by
the following equations.
The effects of each Fit dimension on Use and User
Performance were tested by comparing a regression
equation without Fit (model 0) to one with Fit
(model 1), and using an F test to measure the
significance of the change in R
2
values obtained
(Dishaw and Strong, 1998). These models were
tested separately for each fit dimension, and their
effects on Use and User Performance. Results and
analysis techniques are discussed in more detail in
the next section.
4 RESULTS
4.1 Response Rate and Profile
A total of 201 usable responses were retained for
analysis. Most CHWs as mHealth tool users were
aged between 25 and 34 years (51%). In addition,
the sample population comprised more female
(63%) than male (37%) CHWs. The majority (74%),
of respondents have been educated up to secondary
school level. Moreover, a sizeable number (79%) of
respondents have used mHealth tools for five
months or more.
4.2 Findings and Discussion
Initial Principal Component Analysis (PCA) of the
instrument measures was carried out, leading to the
removal of four User Performance items, three Time
Criticality items, two Interdependence items, and
one Mobility Support item. The remaining item
measures were then assessed for discriminant
validity, internal consistency reliability, and
unidimensionality (Hair, Hult, Ringle and Sarstedt,
2014) using Confirmatory Factor Analysis (CFA)
techniques. Overall, the measurement model
demonstrated adequate reliability and validity. Tests
of the base and Fit models could then proceed.
4.3 Base Models: Use and User
Performance
To test the base model (model 0) for Use as a
dependent variable, Partial Least Squares –
Structural Equation Modeling (PLS-SEM) was used
to obtain estimates for the model i.e. the path
coefficients representing the hypothesized
relationships between each of the four sets of task
and technology characteristics, and Use. To test the
significance of the path coefficients, a bootstrapping
procedure (using 500 samples; 201 cases) was run
(Hair et al., 2014). Table 2 summarizes the R
2
obtained after estimating the reflective base models,
with Use and User Performance as the dependent
variables.
4.4 Fit as Matching: Effects on
mHealth Tool Use
To test the Matching perspective, Fit, calculated as
the paired interaction between corresponding task
and technology characteristics, was included with
tests of the structural models. The inclusion of the
(model 0) Use/User Performance =

+ 
1
TaskCharacteristic
+ 
2
TechnologyCharacteristic + ,
(model 1) Use/User Performance = 
+ 
1
TaskCharacteristic
+ 
2
TechnologyCharacteristic + 
3
Fit + ,
Fit = f (Task x Technology)
MatchingTaskandTechnologyCharacteristicstoPredictmHealthToolUseandUserPerformance-AStudyof
CommunityHealthWorkersintheKenyanContext
457

Fit variable allows for the added effects of this
match between the task and the technology, and its
effects on Use to be estimated. For each interaction
model, an f
2
effect size (Hair et al., 2014) was
measured to assess whether the inclusion Fit had a
substantive impact on Use. The Fit term for the Time
Criticality, Interdependence, Mobility, and
Information Dependency models, each explaining
the dependent variable Use had small f
2
effect sizes
of 0.0028, 0.015, 0.071, and 0.031 respectively, with
Information Dependency Fit having the largest effect
(0.071) on Use. A bootstrapping procedure (using
500 samples; 201 cases) was run to test the
significance of the PLS estimates.
Table 2: PLS Results of the Base Model Predicting Use
and User Performance.
Base Model (0) R
2
Use
User
Performance
Time Criticality
0.177 0.160
Interdependence 0.080 0.149
Mobility
0.070 0.183
Information Dependency 0.170 0.113
Results were contrary to initial expectations. It was
found that CHWs who perceive high levels of
functional support provided for their Time
Criticality, Interdependence, and Mobility needs
depend less on the mHealth tool. This could be
explained by the concept of ‘over-fit’, which occurs
when technology provides excessive functional
support, causing ‘slack’ (Gupta, 2003). The
interaction term TimeC x TimeCSup has a significant
negative effect (p < .05) on Use (-0.159), showing
that high need users depend less on tool use than low
need users with increased functional support.
Similarly, the interaction term Inter x InterSup has a
significant negative effect (p < .10) on Use (-0.120),
showing that high need users depend less on the tool
than low need users when functional support is high.
The interaction term Mobil x MobilSup also has a
significant negative effect (p < .01) on Use (-0.266),
showing that high need users depend less on the tool
than low need users when functional support is high.
The interaction term InfoDep x InfoDepSup has a
significant negative effect (p < .05) on Use (-0.153),
where dependence on the mHealth tool is relatively
flat as the need for information dependency
increases, whereas at lower levels of functional
support, dependence on the mHealth tool increases
more steeply as this need increases. Information
Dependency however, is the only need that has an
independent effect on Use.
4.5 Fit as Matching: Effects on User
Performance
Interestingly, CHWs who perceive high levels of
functional support provided for their Time Criticality
and Mobility needs are less likely to provide
effective, efficient, and high quality care. This is
also attributed to over-fit, which may occur when the
mHealth tool provides excessive functional support
(Gupta, 2003). However, CHWs who perceive high
levels of functional support provided for their
Interdependence and Information Dependency needs
are more likely to provide effective, efficient, and
high quality care. With perceived lower levels of
functional support, CHWs will be less likely to
perform well. In areas of certain task needs, where
higher need users may be experiencing ‘under-fit’
(Gupta, 2003), it may be more difficult to match tool
support to their needs, whereas users who do not
even recognize they have a need, can nevertheless
perform better with a high functioning tool. The Fit
term for the Time Criticality, Interdependence,
Mobility, and Information Dependency models, each
explaining the dependent variable User
Performance, had f
2
effect sizes of 0.092, 0.048,
0.116, and 0.069 respectively, with Mobility Fit
having a medium effect size (0.116) signifying the
largest effect on User Performance. As was the case
for the Fit model predicting Use, a bootstrapping
procedure (using 500 samples; 201 cases) was run to
test the significance of the PLS estimates. The
interaction term TimeC x TimeCSup has a significant
negative effect (p < .01) on User Performance (-
0.287), showing that CHWs do not necessarily need
higher functional support, as this may in fact hinder
their performance by disrupting their established
workflows. By contrast, the interaction term Inter x
InterSup has a significant positive effect (p < .01) on
User Performance (0.203), showing high need users
perform better than low need users when the tool
provides high functional support. The interaction
term Mobil x MobilSup has a significant negative
effect (p < .01) on User performance (-0.295),
showing that higher need users are still satisfied that
they provide effective, efficient, and high quality
care without functional support. Lower need users
however feel they provide better care with higher
levels of functional support. The interaction term
InfoDep x InfoDepSup has a significant positive
effect (p < .01) on User Performance (0.248),
confirming that users with more tool support will
likely experience better performance outcomes.
HEALTHINF2015-InternationalConferenceonHealthInformatics
458

5 DISCUSSION AND
CONCLUSION
This study was based on the premise that by
matching CHW tasks and mHealth technology
characteristics, the impacts of mHealth tool use on
CHW performance in the Kenyan context could be
observed. We drew on task-technology fit theory to
define match as the degree to which CHW needs
reflected by task characteristics i.e. Time Criticality,
Interdependence, Mobility, and Information
Dependency, are supported by mHealth tool
functions reflected by technology characteristics i.e.
Time Criticality Support, Interdependence Support,
Mobility Support, and Information Dependency
Support. Aspects of this study are comparable to
other studies of mobile work support (Yuan et al.,
2010). Results provide important insights into how
CHW needs and mHealth tool functions influence
Use and User Performance. First, the study provides
substantive evidence that when mHealth tools are
designed to match required tasks they can enhance
CHW performance. Second, findings can be used to
inform design of mHealth tools to provide more
adequate functional support for the most critical user
needs. Third, by providing empirical insights on the
Table 3: PLS Results of the Interaction Models Predicting Use.
Path
Coefficients
t Values p Values Significance
Levels
90% Confidence
Intervals
Time Criticality Model R
2
= 0.199, f
2
= 0.028
TimeC  Use 0.099 1.65 0.10 * [0.00, 0.20]
TimeCSup Use 0.371 4.67 0.00 *** [0.24, 0.50]
Fit (TimeC x TimeCSup)  Use -0.159 2.09 0.05 ** [-0.29, -0.03]
Interdependence Model R
2
= 0.094, f
2
= 0.015
Inter  Use 0.042 0.77 0.44 NS [-0.05, 0.13]
InterSup Use 0.261 3.50 0.00 *** [0.25, 0.38]
Fit (Inter x InterSup)  Use -0.120 1.81 0.07 * [-0.23, -0.01]
Mobility Model R
2
= 0.132, f
2
= 0.071
Mobil  Use 0.014 0.28 0.78 NS [-0.07 0.10]
MobilSup Use 0.244 3.43 0.00 *** [0.13, 0.36]
Fit (Mobil x MobilSup)  Use -0.266 3.69 0.00 *** [-0.38 -0.15]
Information Dependency Model R
2
= 0.195, f
2
= 0.031
InfoDep  Use 0.192 2.54 0.01 ** [0.07, 0.32]
InfoDepSup Use 0.320 3.97 0.00 *** [0.19, 0.45]
Fit (InfoDep x InfoDepSup)  Use -0.153 1.96 0.05 ** [-0.28, -0.02]
Note: NS = Not Significant.
*
p < .10.
**
p < .05.
***
p < .01.
Table 4: PLS Results of the Interaction Models Predicting User Performance.
Path
Coefficients
t Values p Values Significance
Levels
90% Confidence
Intervals
Time Criticality Model R
2
= 0.231, f
2
= 0.092
TimeC  User Perf 0.156 1.79 0.08 * [0.01, 0.30]
TimeCSup Use Perf 0.268 3.57 0.00 *** [0.15, 0.39]
Fit (TimeC x TimeCSup)  User Perf -0.287 2.96 0.00 *** [-0.45, -0.13]
Interdependence Model R
2
= 0.188, f
2
= 0.048
Inter  User Perf 0.037 0.69 0.49 NS [-0.05, 0.12]
InterSup User Perf 0.385 4.42 0.00 *** [0.23, 0.53]
Fit (Inter x InterSup)  User Perf 0.203 2.90 0.00 *** [0.09, 0.32]
Mobility Model R
2
= 0.268, f
2
= 0.116
Mobil  User Perf 0.043 0.85 0.40 NS [-0.04 0.13]
MobilSup User Perf 0.411 6.70 0.00 *** [0.31, 0.51]
Fit (Mobil x MobilSup)  User Perf -0.295 3.86 0.00 *** [-0.42 -0.17]
Information Dependency Model R
2
= 0.170, f
2
= 0.069
InfoDep  User Perf 0.178 2.17 0.03 ** [0.04, 0.31]
InfoDepSup User Perf 0.277 3.35 0.00 *** [0.14, 0.41]
Fit (InfoDep x InfoDepSup)  User Perf 0.248 2.90 0.00 *** [0.11, 0.39]
Note: NS = Not Significant.
*
p < .10.
**
p < .05.
***
p < .01.
MatchingTaskandTechnologyCharacteristicstoPredictmHealthToolUseandUserPerformance-AStudyof
CommunityHealthWorkersintheKenyanContext
459

fit between healthcare service and mHealth tool
characteristics from a matching perspective, the
study findings can better inform mHealth tool use by
CHWs and enhance performance in their capture,
storage, transmission, and retrieval of health data
(Liu et al., 2011). In the areas of Information
Dependency and Interdependence, ‘Fit as Matching’
provides the best explanations for performance
outcomes. However, findings also indicate that just
because CHWs have needs does not mean that a
highly functional tool necessarily results in
increased dependency on Use or enhanced User
Performance. Similarly, just because CHWs do not
recognize a need, does not mean a high functioning
tool cannot influence their dependence on Use or
enhanced User Performance. The tool could be
compensating for those who have not recognized a
need and therefore have not already established
routines and coping mechanisms. However, for
those who have recognized a need, the tool may be
unimportant given already established preferred
practices. The study confirms that mobile
technologies could improve mHealth tool use and
CHW performance in low-resource community
household settings (Earth Institute, 2010). However,
designers should be cautious of excessive functional
support that may hinder CHW performance with
established routines, and that despite high mobility
and time criticality needs, an mHealth tool may not
always provide the best support. If function support
is excessive, users may depend less on the tool, and
its impacts may not be favourable at all levels of
need. These results can nevertheless add to the
growing interest in directly supporting CHWs at the
point of care. Future research may wish to consider
cost implications as instrumental to the successful
deployment of mHealth platforms in the Kenyan
context. Future work may also consider assessing
the match between CHW needs and mHealth tool
functions in other contexts and settings.
ACKNOWLEDGEMENTS
We sincerely thank Kenya’s Ministry of Health
(MOH) Division of Community Health Services
(DCHS) and all Community Health Workers
(CHWs) who participated in the study.
REFERENCES
Dishaw, M. T. and Strong, D. M. (1998) Supporting
Software Maintenance With Software Engineering
Tools: A Computed Task- Technology Fit Analysis.
The Journal of Systems and Software, 44, 107-
120.
Earth Institute, “One Million Community Health Workers:
Technical Taskforce Report”, 2010, http://
www.milleniumvillages.org/uploads/ReportPaper/1m
CHW_Technical TaskForceReport.pdf [last accessed
30
th
August 2014].
Torkzadeh, G. and Doll, W. 1999. The Development of a
Tool for Measuring the Perceived Impact of
Information Technology on Work. The International
Journal of Management Science, 27, 1, 327-339.
Gatara, M., and Cohen, J. F., 2014a. Mobile-Health Tool
Use and Community Health Worker Performance in
the Kenyan Context: A Task-Technology Fit
Perspective. In Proceedings of the South African
Institute for Computer Scientists and Information
Technologists SAICSIT Conference (Centurion,
Gauteng, South Africa, September 28 – October 01
2014).
Gatara, M., and Cohen, J. F., 2014b. The Mediating Effect
of Task-Technology Fit on mHealth Tool Use and
Community Health Worker Performance in the
Kenyan Context. In Proceedings of the 8th
International Development Informatics Association
IDIA Conference (Port Elizabeth, South Africa,
November 03-04, 2014).
Gebauer, J. and Tang, Y. 2008. Applying the Theory of
Task-Technology Fit to Mobile Technology: The
Role of User Mobility. Working Paper Abstract, 1-23.
Goodhue, D. L. and Thompson, R. L. 1995. Task-
Technology Fit and Individual Performance. MIS
Quarterly 19, 2, 213-236.
Gupta, S. 2003. Knowledge Management and
Performance: A Fit Perspective. In Proceedings of the
Ninth Americas Conference on Information Systems
(Tampa, Florida, USA, August 04-06, 2003).
Hair, J. F, Hult, G. T, Ringle, C. M. and Sarstedt, M. 2014.
A Primer on Partial Least Squares Structural
Equation Modeling (PLS-SEM), Sage, Thousand
Oaks, California.
Junglas, I., Abraham, C., and Ives, B. 2009. Mobile
Technology at the Frontlines of Patient Care:
Understanding Fit and Human Drives in Utilization
Decisions and Performance, Decision Support
Systems. 46, 534- 647.
LeMaire, J. “Scaling Up Mobile Health: Elements for the
Successful Scale Up of mHealth in Developing
Countries”, 2011 http://www.k4health.org/sites/sites/
default/files/ADA_mHealth%20White%20Paper.pdf
[last accessed 2
nd
May 2013].
Lin, T. and Huang, C. 2008. Understanding Knowledge
Management System Usage Antecedents: An
Integration of Social Cognitive Theory and Task-
Technology Fit. Information and Management, 45,
410-417.
Liu, A, Sullivan, S, Khan, M. Sachs, G. and Singh, P.
2011. Community Health Workers in Global Health:
Scale and Scalability, Mount Sinai Journal of
Medicine, 78, 1, 419-435.
HEALTHINF2015-InternationalConferenceonHealthInformatics
460

Teo, S. H. and Men, B. 1999. Knowledge Portals in
Chinese Consulting Firms: A Task-Technology Fit
Perspective. European Journal of Information
Systems, 17,1, 2008, pp. 557-574.
Venkatraman, N. 1989. The Concept of Fit in Strategy
Research: Toward Verbal and Statistical
Correspondence. The Academy of Management
Review, 14, 3, 423-444.
Wixom, B. H. and Todd, P. A. 2005. A Theoretical
Integration of User Satisfaction and Technology
Acceptance, 16, 1, 85-102.
Yuan, Y., Archer, N., Connelly, C. E. and Zheng, W.
2010. Identifying the Ideal Fit Between Mobile
Work and Mobile Work Support. Information and
Management, 47, 125-137.
MatchingTaskandTechnologyCharacteristicstoPredictmHealthToolUseandUserPerformance-AStudyof
CommunityHealthWorkersintheKenyanContext
461