INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)  
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025  
Convergence, Not Accumulation: Digital Maturity and  
Organisational Resilience among SDA Self-Supporting Ministries in  
Kenya  
1Saya Jackson*, MSc; 2Kelvin Onoongha, PhD; 3Josephine Ganu, PHD, 4Paul Ataro  
1Adventist University of Africa, Private Bag Mbagathi, 00503 Nairobi, Kenya  
2Kelvin Onoongha; Adventist University of Africa  
3Josephine Ganu; Adventist University of Africa  
4Paul Ataro; Benaphil Consultants Ltd  
*Corresponding Author  
DOI: https://dx.doi.org/10.47772/IJRISS.2025.91100391  
Received: 01 December 2025; Accepted: 06 December 2025; Published: 11 December 2025  
ABSTRACT  
This study examines whether digital maturity strengthens organisational resilience among Seventh-day  
Adventist–affiliated self-supporting ministries in Kenya. Two propositions are tested: that Digital Intensity—  
governed data, simple automation/analytics, and reliable “green” infrastructure—relates positively to resilience,  
and that the convergence of Digital Intensity with Transformation Management Intensity—mission-linked  
strategy, baseline readiness, and human-centric adoption—explains resilience better than either stream alone. A  
quantitative, explanatory, cross-sectional survey of 141 ministries was analysed using hierarchical confirmatory  
factor analysis and structural equation modelling with robust estimation and bootstrapped confidence intervals.  
The measurement model met reliability, convergent, and discriminant validity standards with acceptable global  
fit. Structurally, Digital Intensity demonstrated a positive, statistically significant association with resilience,  
and the convergence construct provided the strongest pathway, explaining higher variance and remaining stable  
across sensitivity checks. The study concludes that resilience gains in resource-constrained ministries arise less  
from accumulating tools than from coupling governed information, lightweight automation, and dependable  
infrastructure with focused strategy, cyber hygiene, and adoption rituals. Findings inform shared services,  
lightweight standards, and micro-learning initiatives for African faith-based nonprofits.  
Keywords: digital maturity, organisational resilience, faith-based nonprofits, self-supporting ministries,  
structural equation modelling, Kenya.  
INTRODUCTION & CONTEXT  
Self-supporting ministries (SSMs) affiliated with the Seventh-day Adventist (SDA) Church in Kenya operate in  
fragile conditions, characterised by irregular funding, a high proportion of volunteer staff, and uneven  
governance capacity across urban and rural settings. Digital infrastructure has improved, yet access and usage  
remain heterogeneous, with persistent gaps in device availability, skills, and affordability that disproportionately  
affect smaller nonprofits (GSMA, 2023, 2024). Compliance expectations have tightened under Kenya’s Data  
Protection Act and sectoral guidance, raising baseline requirements for data stewardship and cyber hygiene that  
many ministries struggle to meet. In this environment, disruption—whether epidemiological, economic,  
climatic, or regulatory—is not an outlier but a planning premise. The managerial question is therefore practical  
and pressing: which digital investments and routines actually strengthen organisational resilience (OR)—the  
ability to anticipate, absorb, and adapt—rather than merely accumulating tools that add cost and complexity?  
Page 4947  
www.rsisinternational.org  
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)  
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025  
This study addresses a specific theoretical gap: most digital-maturity accounts treat technology intensity (DI)  
and transformation management intensity (TMI) as additive, independent levers (“accumulation”). By contrast,  
we theorise and test a convergence paradigm in which resilience emerges when digital and managerial streams  
cohere as a higher-order capability (DML). We therefore examine whether DI predicts OR and, more critically,  
whether DML outperforms single-stream maturity in explaining resilience among resource-constrained, faith-  
based nonprofits.  
Prior scholarship offers only partial answers. A substantial stream links digital transformation to efficiency and  
performance, often proxied by cost or revenue metrics (Verhoef et al., 2021; Vial, 2019). Evidence specific to  
resilience—continuity, recovery, and adaptive capacity—remains thinner and mixed, especially outside high-  
income settings and beyond the private sector (Hillmann & Guenther, 2021). Studies of nonprofits and SMEs  
suggest that data quality, simple automation, and reliable infrastructure can enhance continuity; however, these  
effects are attenuated when technology is not embedded in a coherent strategy, governance, and adoption  
routines (Robertson et al., 2022; Mikalef, Pappas, Krogstie, & Pavlou, 2020). In faith-based organisations,  
additional constraints—such as mission-first priorities, volunteer turnover, risk aversion, and privacy concerns—  
complicate adoption pathways and may shift the digital capabilities that matter most for resilience (Giannelia,  
2020; He, Jiang, & Zhang, 2022). African cases are particularly underrepresented, limiting the transferability of  
findings to contexts where resource slack is low and infrastructure variability is high (GSMA, 2024).  
This paper addresses that gap by distinguishing two streams of capability and their convergence. Digital Intensity  
(DI) encompasses the technological payload, comprising Data Management (DM) for governed, decision-ready  
information, Automation & Intelligence (AAI) for cycle-time and error reduction, and Green Digitisation (GD)  
for reliable, cost-stable infrastructure. Transformation Management Intensity (TMI) captures the managerial  
spine, comprising Digital Business Strategy (DBS) that ties choices to mission, Digital Readiness (DR) for skills,  
governance bodies, cyber hygiene, and continuity playbooks, and Human-Centric Digitisation (HCD) for  
adoption rituals and micro-learning. Building on resource-based and dynamic capabilities perspectives, with  
alignment and complexity lenses, the analysis specifies a higher-order Digital Maturity Level (DML) as the  
convergence of DI and TMI—an orchestrated coupling expected to deliver situation awareness and operational  
elasticity, the proximal mechanisms of resilience (Duchek, 2020; Lee, Vargo, & Seville, 2013; Verhoef et al.,  
2021; Weick & Sutcliffe, 2015).  
Two uncertainties motivate the empirical test. First, does DI—the operational payload—on its own explain OR  
in ministries where minor data improvements can meaningfully increase foresight and coordination? Second,  
does convergence (DML) outperform either stream alone, implying that resilience emerges less from  
accumulating tools or issuing plans than from coherent coupling of information, routines, and buffers? These  
questions are crucial for informing policy and practice. Donors, unions, and ministry networks face budget trade-  
offs: invest in more software, in governance and training, or in the connective tissue that synchronises both. If  
convergence dominates, shared services for data stewardship, low-code automation, and baseline security—  
paired with lightweight standards and micro-learning—may yield greater resilience than technology  
proliferation or leadership exhortation alone (Kane, Palmer, Phillips, Kiron, & Buckley, 2015; Robertson et al.,  
2022).  
The paper contributes three elements. Conceptually, it clarifies the difference between accumulation (more tools  
or more initiatives) and convergence (co-specialisation between the technological payload and managerial  
routines) as distinct maturity logics in resource-constrained nonprofits. Empirically, it offers an Africa-situated,  
faith-sector test using hierarchical CFA and SEM on organisational-level data from SDA SSMs in Kenya, a  
hard-to-reach population accessed with respondent-driven sampling augmentation and analysed with rigorous  
measurement validation (Fornell & Larcker, 1981; Henseler, Ringle, & Sarstedt, 2015). Practically, it translates  
findings into a sequenced playbook appropriate for ministries that must prioritise data governance, simple  
automation, reliability, and human adoption under tight constraints. Guided by this framework, the study focuses  
on two research questions and associated hypotheses tested in the structural model:  
RQ1: To what extent does the level of Digital Intensity (DI) relate to the Organisational Resilience (OR) of  
Seventh-day Adventist–affiliated self-supporting ministries (SSMs) in Kenya?  
Page 4948  
www.rsisinternational.org  
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)  
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025  
H₀₁: There is no statistically significant association between Digital Intensity and Organisational Resilience  
among SDA-affiliated SSMs in Kenya.  
RQ2: Does the Digital Maturity Level (DML) demonstrate a stronger association with Organisational Resilience  
(OR) than either DI or TMI considered in isolation within SDA-affiliated SSMs in Kenya?  
Ho2: Digital Maturity Level does not exhibit a stronger association with Organisational Resilience than either  
Digital Intensity or Transformation Management Intensity considered separately.  
By centering resilience as the dependent outcome and convergence as the organising logic, the analysis aims to  
inform how African faith-based nonprofits can develop robust, adaptive operations without relying on abundant  
resources or uniform digital readiness.  
LITERATURE REVIEW  
Resource-Based View: governed digital assets as VRIN foundations of resilience  
The Resource-Based View (RBV) posits that a durable advantage rests on resources and capabilities that are  
valuable, rare, inimitable, and non-substitutable (Barney, 1991; Peteraf, 1993). In resource-constrained  
nonprofits, governed digital assets—clean data, auditable processes, and reliable infrastructure—function like  
VRIN building blocks because they enhance decision quality, reduce errors, and are costly for peers to replicate  
quickly (Wade & Hulland, 2004; Hanelt, Bohnsack, Marz, & Marante, 2021). The Digital Intensity (DI) stream  
embodies this logic. Data Management (DM) creates decision-ready information through stewardship, quality  
controls, and access discipline. Automation & Intelligence (AAI) embeds rules and analytics into workflows,  
reducing cycle times and identifying errors at their source. Green Digitisation (GD) enhances continuity through  
cloud rationalisation, modest redundancy, and energy/power discipline, which stabilise costs and uptime  
(Saldanha, Mithas, Khuntia, Whitaker, & Melville, 2022). Together, these components form a defensible  
capability base that supports organisational resilience (OR)—the capacity to anticipate, absorb, and adapt during  
shocks (Lengnick-Hall, Beck, & Lengnick-Hall, 2011; Duchek, 2020). In African nonprofit settings where  
resources are scarce and reliability is inconsistent, governed data and simple automation often yield significant  
gains (GSMA, 2024; Robertson, Botha, Walker, Wordsworth, & Balzarova, 2022).  
Dynamic Capabilities: sensing, seizing, and reconfiguring under turbulence  
Dynamic Capabilities Theory views resilience as the result of sensing weak signals, seizing opportunities, and  
reconfiguring assets and routines to fit shifting conditions (Teece, 2007; Eisenhardt & Martin, 2000). DI  
strengthens sensing by improving situation awareness—providing timely, accurate, and decision-relevant  
information—and it reinforces seizing and reconfiguring through operational elasticity—the ability to pivot  
processes with minimal disruption (Lee, Vargo, & Seville, 2013; Weick & Sutcliffe, 2015). Yet capabilities do  
not self-orchestrate; nonprofits need managerial routines that prioritise, standardise, and embed the technological  
payload. This motivates the study’s convergence construct, Digital Maturity Level (DML), which treats  
orchestrated coupling between DI and Transformation Management Intensity (TMI) as a higher-order capability.  
Convergence, rather than simple accumulation, is posited to unlock the dynamic-capability cycle in ministries  
where volunteers rotate and governance capacity varies (Verhoef et al., 2021; Vial, 2019).  
Alignment/Contingency: Why intent under-delivers without a DI threshold  
Alignment and Contingency scholarship posits that performance effects emerge when structures, processes, and  
skills align with strategy and context (Henderson & Venkatraman, 1993; Coltman, Tallon, Sharma, & Queiroz,  
2015). The TMI stream—Digital Business Strategy (DBS), Digital Readiness (DR), and Human-Centric  
Digitisation (HCD)—captures intent and governance. In SSMs, however, management intensity can under-  
deliver if the technological base is weak. A strategy that prioritises data-informed ministry or a change program  
that mobilises champions cannot yield resilience if data are dirty, workflows remain manual, or infrastructure  
fails during surges. Alignment thus implies a DI threshold: DBS/DR/HCD amplify outcomes once minimal  
DM/AAI/GD are in place. DML—as convergence—represents alignment realised: the condition in which  
Page 4949  
www.rsisinternational.org  
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)  
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025  
managerial routines and technology move in lockstep. This logic is consistent with studies showing that digital  
transformation outcomes hinge on both the depth of capability and organisational embedding, especially in  
SMEs and nonprofits (Kane, Palmer, Phillips, Kiron, & Buckley, 2015; Gerow, Grover, Thatcher, & Roth, 2014;  
Hanelt et al., 2021).  
Complexity/Systems: resilience as an emergent property of coherent coupling  
Complexity perspectives view organisations as complex adaptive systems where reliability emerges from tight  
coupling among information flows, routines, and buffers (Holland, 2012; Weick & Sutcliffe, 2015). Fragmented  
tools, siloed data, or uncoordinated change programs increase brittleness. Convergence reduces brittleness by  
synchronising three things: the signals ministries read (via DM), the moves they can execute quickly and safely  
(via AAI), and the buffers that prevent small shocks from cascading (via GD), all reinforced by strategy focus,  
governance, and human adoption (DBS, DR, HCD). In volatile African contexts—with intermittent connectivity,  
variable power, and fluctuating volunteer capacity—coherent coupling is as essential as scale. Studies conducted  
during and after the COVID-19 pandemic show that organisations with integrated sociotechnical routines  
reported shorter recovery cycles and more stable service delivery, even when budgets were modest (Robertson  
et al., 2022; He, Jiang, & Zhang, 2022; Hillmann & Guenther, 2021).  
Hypotheses development  
Digital Intensity → Organisational Resilience. RBV positions DM/AAI/GD as capability assets that are costly  
to imitate quickly. DCT explains how these assets enable sensing and reconfiguration. Alignment clarifies that  
even without elaborate management programs, a minimal technological base can maintain continuity.  
Complexity highlights how DI reduces entropy in everyday operations. Empirical work with SMEs, nonprofits,  
and crisis settings indicates that clean data and simple automation improve coordination and reduce service  
failures, and that reliability investments pay resilience dividends in low-slack contexts (Mikalef, Pappas,  
Krogstie, & Pavlou, 2020; Robertson et al., 2022; GSMA, 2024). In Kenyan SSMs, where operational variance  
is high, even incremental gains in data quality and workflow automation can produce significant resilience  
effects  
by  
shortening  
detection  
and  
response  
times.  
H₁: Digital Intensity (DI) is positively associated with Organisational Resilience (OR) among SDA-affiliated  
self-supporting ministries in Kenya.  
Convergence (DML) → Organisational Resilience, stronger than DI or TMI alone. Alignment and  
complexity perspectives imply that resilience is an emergent property of co-specialisation. TMI without DI risks  
“plan-without-platform,” while DI without TMI risks “tools-without-use.” Convergence integrates both.  
Empirical syntheses in digital transformation find that firms and nonprofits with synchronised technology and  
management routines outperform those with isolated excellence, especially when facing turbulence (Verhoef et  
al., 2021; Vial, 2019; Hanelt et al., 2021). Africa-situated reports similarly show that where data stewardship,  
low-code automation, and baseline cyber hygiene are embedded through governance and micro-learning,  
continuity improves despite resource constraints (GSMA, 2024; Infoxchange, 2024). DML, therefore, captures  
the  
realisation  
of  
alignment  
as  
a
higher-order  
capability.  
H₃: Digital Maturity Level (DML)—the convergence of DI and TMI—exhibits a stronger positive association  
with Organisational Resilience (OR) than either DI or TMI alone.  
Taken together, these lenses justify testing both the standalone effect of DI and the convergence effect of DML.  
DI should matter in its own right in SSMs because governed data and simple automation immediately raise  
situation awareness and reduce failure modes. DML should matter more because strategic focus, governance,  
and human adoption are what transform the payload into sustained routines. The empirical sections therefore  
estimate Model 1 (DI→OR) and Model 3 (DML→OR), with rigorous measurement validation and robustness  
checks to ensure that observed patterns are not artefacts of estimation or sampling.  
The convergence thesis also aligns with philosophy of administration and stewardship ethics: resilient  
administration harmonises interdependent subsystems toward mission under constraint, emphasising  
accountability, prudent governance, and dignifying inclusion. In this view, DI supplies the informational and  
Page 4950  
www.rsisinternational.org  
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)  
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025  
operational substrate; TMI supplies the focusing and normative spine; resilience is the emergent property of their  
coherent coupling—a systems-ethics posture consistent with stewardship in faith-based organisations.  
METHODOLOGY  
Research Design and Setting  
The study employed a quantitative, explanatory, cross-sectional survey of self-supporting ministries (SSMs)  
affiliated with the Seventh-day Adventist (SDA) Church in Kenya. The unit of analysis was the organisation  
(one informed respondent per ministry). Kenya’s SSM ecosystem is heterogeneous, ranging from micro,  
volunteer-led initiatives to more formalised entities, operating across health, education, media, evangelism, and  
social services under uneven infrastructure and funding conditions. This setting is suitable for testing whether  
Digital Intensity (DI) and the convergence of digital and managerial streams—Digital Maturity Level (DML)—  
are related to Organisational Resilience (OR) in low-slack, high-variability environments.  
Sampling  
Because no authoritative registry exists, a targeted frame was assembled from denominational networks,  
recognised SSM associations, and public ministry directories. Access challenges and hidden-population features  
(informal registration, fluid staffing, dispersed geography) motivated a respondent-driven sampling (RDS)  
augmentation (Heckathorn, 1997, 2002; Gile & Handcock, 2010). Initial seeds were recruited from diverse  
ministry domains and union territories and asked to refer eligible peers; referrals proceeded in waves until  
recruitment plateaued. The achieved organisational sample was n = 141. To assess potential RDS-induced  
dependencies, sensitivity analyses later introduced wave/seed covariates and clustered standard errors by seed,  
without using RDS weights in the main SEM (given the theory-testing focus and absence of population totals).  
Representativeness was examined descriptively across union territories, domains, age groups, and size bands, as  
well as through early– and late–wave contrasts as a nonresponse proxy (Armstrong & Overton, 1977; Groves &  
Peytcheva, 2008).  
Eligibility required an SDA affiliation, self-supporting status, and at least 12 months of operation; duplicates  
were identified through cross-checking of ministry names, contacts, and web presence. Regional coverage  
spanned EKUC and WKUC; we tracked seed/wave recruitment to minimise cluster duplication. A compact  
demographics table (size, age, domain, union) improves replicability and situates inference.  
Instrument and Data Collection  
Digital Intensity (DI) was modelled as a second-order reflective construct with three first-order factors: Data  
Management (DM), Automation & Intelligence (AAI), and Green Digitisation (GD). Transformation  
Management Intensity (TMI) was modelled with three first-order factors: Digital Business Strategy (DBS),  
Digital Readiness (DR), and Human-Centric Digitisation (HCD). A higher-order Digital Maturity Level (DML)  
captured the convergence of DI and TMI. Organisational Resilience (OR) was modelled as a second-order  
construct, reflected in the Planning/Leadership and Adaptive Capacity pillars. DI/TMI items were drawn from a  
verified, EDIH-aligned digital maturity assessment and localised to nonprofit/ministry language; OR used short-  
form BRT-13/BRT-13B items adapted for Organisational informants (Lee, Vargo, & Seville, 2013; Kljajić  
Borštnar & Pucihar, 2021). Items used five-point Likert scales (strongly disagree–strongly agree). Cross-cultural  
adaptation was conducted following expert review and cognitive pretesting with ministry leaders to ensure  
clarity, relevance, and consistency of tone, in line with established guidelines (Beaton, Bombardier, Guillemin,  
& Ferraz, 2000; International Test Commission, 2018).  
Ethical Considerations  
Ethical approval was granted by the Adventist University of Africa Institutional Scientific and Ethics Review  
Committee (AUA-ISERC; Ref: AUA/ISERC/1141/2024; 14 September 2024). Participation was voluntary. An  
online informed-consent page preceded the questionnaire. Respondents affirmed that they were knowledgeable  
about their ministry’s digital operations/strategy. Organisational identifiers were anonymised in analysis files,  
and results are reported in aggregate.  
Page 4951  
www.rsisinternational.org  
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)  
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025  
To mitigate common-method variance, the survey assured anonymity, used neutral wording, varied scale anchors  
across modules where appropriate, and proximal separation of predictors and outcomes to reduce hypothesis-  
guessing (Podsakoff, MacKenzie, Lee, & Podsakoff, 2003). Post hoc diagnostics included a single-factor check  
and, in sensitivity, a latent common method factor (CLF) overlay to assess whether a general factor materially  
altered structural relations. Multicollinearity among first-order factors was screened (variance inflation factors,  
VIF < 5 target; O’Brien, 2007). Missing data were handled using full information maximum likelihood (FIML)  
in the SEM engine under the assumption of missing-at-random (Enders, 2010; Little & Rubin, 2019). Outliers  
were inspected with Mahalanobis distance and robust diagnostics (Rousseeuw & Van Zomeren, 1990; Hubert,  
Van der Veeken, & Debruyne, 2008). Distributional assumptions were evaluated using item and factor skewness  
and kurtosis, as well as Mardia’s multivariate indices (Mardia, 1970, 1974).  
Data Analysis  
Analyses proceeded in two stages using covariance-based structural equation modelling.  
Stage 1: Measurement (hierarchical CFA). First, the reflective measurement models were estimated to  
establish the latent structure and psychometric adequacy (Brown, 2015; Kline, 2016). For each first-order factor,  
standardised loadings of≥ .60 were targeted, along with composite reliability (CR/ω) of≥ .70 and average  
variance extracted (AVE) of≥ .50, where feasible (Fornell & Larcker, 1981; Dunn, Baguley, & Brunsden, 2014).  
Discriminant validity was assessed using the HTMT statistic (< .85, conservative; < .90, liberal) and cross-  
checked with the Fornell–Larcker criterion (Henseler, Ringle, & Sarstedt, 2015; Fornell & Larcker, 1981). A  
global fit is considered when CFI/TLI is ≥ 0.90 or 0.95, RMSEA ≤ 0.08 (≤ 0.06 desirable), and SRMR ≤ 0.08,  
interpreted holistically rather than by single cutoffs (Hu & Bentler, 1999). Given ordinal Likert indicators and  
potential mild nonnormality, the main estimator used MLR (robust maximum likelihood) with Y-standardised  
solutions; WLSMV served as a robustness estimator (Flora & Curran, 2004a, 2004b). Second-order  
specifications were then estimated for DI, TMI, and OR; finally, a higher-order DML factor was formed from  
DI and TMI to operationalise convergence.  
Stage 2: Structural tests (SEM). Three structural models tested the focal paths while holding the validated  
measurement model constant:  
● Model 1: DI → OR (tests H₁).  
● Model 2: TMI → OR (reported succinctly for completeness).  
● Model 3: DML → OR (tests H₃, convergence dominance).  
Direct effects were accompanied by bootstrapped 95% confidence intervals (based on 5,000 resamples). To  
probe estimation sensitivity, all three models were re-estimated with WLSMV. Wave/seed covariates and  
clustered standard errors by seed were introduced in sensitivity runs to check whether the recruitment structure  
perturbed path magnitudes or rank order. Model adequacy and parsimony were compared using standard fit  
indices and ΔR² in OR across models (Burnham & Anderson, 2002; MacCallum, Browne, & Sugawara, 1996).  
Modification indices were inspected but acted upon only when supported by strong a priori justification (e.g.,  
same construct, similarly worded items), thereby avoiding cross-construct cross-loadings to preserve construct  
validity.  
Analytic transparency and assumption checks. The primary estimator was MLR with robust standard errors;  
the WLSMV sensitivity analysis yielded an unchanged path ranking. Item-level missingness was low and  
handled via FIML under missing-at-random assumptions. Distributional screening (skewness/kurtosis and  
Mardia indices) indicated mild non-normality, justifying the use of robust estimation. To address common-  
method bias, procedural remedies (anonymity, neutral wording, varied anchors, proximal separation) were  
complemented by a single-factor test and a latent common-method factor overlay in sensitivity analyses; neither  
suggested a dominant method factor.  
Data quality diagnostics  
Page 4952  
www.rsisinternational.org  
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)  
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025  
Although the complete statistics are reported alongside the results, key diagnostics guided the credibility of the  
model. KMO and Bartlett’s test assessed factorability at the item level (Kaiser, 1974; Bartlett, 1950). Anti-image  
correlations and communality patterns informed any trimming decisions. Measurement invariance across broad  
subgroups (e.g., union territory, micro vs. larger ministries) was explored at the configural/metric levels to  
support interpretability of structural paths, recognising sample-size constraints for full invariance batteries  
(Chen, 2007). All diagnostic choices favoured parsimony, given the parameter-to-sample ratio recommended  
for SEM with hierarchical factors (Bentler & Chou, 1987; Rindskopf & Rose, 1988).  
RESULTS  
The analytic sample comprised 141 ministries with distributions across staff size, establishment period, union  
location, and organisational type. As summarised in Table 1.  
Table 1: Sample Characteristics (n = 141)  
Characteristic  
Category  
Frequency  
Percentage  
72.26%  
16.06%  
6.57%  
Staff Size  
Micro-size (1–9 staff)  
Small size (10–49 staff)  
Medium size (50–249 staff)  
Large size (250+ staff)  
Founded before 2015  
Founded 2015–2024  
Peak years: 2024, 2021, 2020, 2018  
EKUC  
99  
22  
9
7
5.11%  
Year of Establishment  
Union Location  
41  
96  
29.93%  
70.07%  
See Note¹  
80.29%  
16.79%  
2.92%  
110  
23  
4
WKUC  
None / Not Sure  
Organization Type  
SDA member-run business  
SDA-affiliated self-supporting ministry  
Hybrid business-ministry  
Other  
75  
34  
18  
8
55.56%  
25.19%  
13.33%  
5.93%  
Note. EKUC = East Kenya Union Conference; WKUC = West Kenya Union Conference. Percentages may not  
sum to 100 due to rounding.  
Most entities were micro-sized (≈72%), with recent formation years predominating (≈70% founded 2015–2024).  
East Kenya Union entities formed the largest bloc, and member-run businesses constituted the modal type—  
these distributions frame the interpretation of digital capability and resilience patterns.  
Measurement model  
The hierarchical measurement structure performed satisfactorily. All focal constructs—Digital Intensity (DI)  
with first‐order DM, AAI, GD; Transformation Management Intensity (TMI) with DBS, DR, HCD;  
Page 4953  
www.rsisinternational.org  
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)  
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025  
Organisational Resilience (OR) with Planning/Leadership and Adaptive Capacity; and the higher‐order Digital  
Maturity Level (DML)—met standard psychometric criteria. Standardised loadings were positive and  
substantive for retained items, with the vast majority meeting or exceeding the .60 heuristic, and none of the  
kept indicators persistently underloading. Internal consistency was strong across constructs (ω/CR ≥ 0.70), and  
AVE values were generally ≥ 0.50; in the few instances where AVE fell slightly below 0.50, composite reliability  
remained adequate, supporting convergent validity (Table 2).  
Table 2: Results of the Hypothesis Testing  
Construct  
Indicators (example)  
Loadings (Range)  
0.558 – 0.902  
AVE  
0.635  
0.669  
ω/CR  
0.793  
0.806  
Digital Intensity (DI)  
DM, AM, GD  
Transformation  
Intensity (TMI)  
Management DBS, DR, HCD  
0.602 – 0.841  
Digital Maturity Level (DML)  
Organisational Resilience (OR)  
Second-order (DI, TMI)  
0.800 – 0.939  
0.308  
0.774  
0.866  
0.870  
Planning, Adaptive Capacity  
0.764 – 0.981  
Note. AVE = average variance extracted; ω/CR = composite reliability/omega. Item-level loadings appear in  
Appendix A (Tables A1–A6). Estimator = MLR with FIML.  
Discriminant validity was supported, as evidenced by HTMT ratios falling below conservative thresholds and  
the Fornell–Larcker criterion being met (Table 3). Global fit for the second‐order measurement models (DI,  
TMI, OR) was acceptable by conventional cutoffs (CFI/TLI in the good range; RMSEA in the acceptable-to-  
good band with tight confidence intervals; SRMR < .08). Estimating DML as a higher‐order factor loading on  
DI and TMI also yielded acceptable fit and well-defined second‐order loadings (Table 4). These patterns held  
under both the robust maximum likelihood estimator (MLR, main specification) and the ordinal-robust estimator  
(WLSMV, sensitivity analysis).  
Table 3 Discriminant Validity: HTMT Matrix and Fornell–Larcker Checks  
Construct  
DI  
DI  
TMI  
0.897  
—
DML  
0.751  
0.751  
—
OR  
—
0.366  
0.205  
0.391  
TMI  
DML  
Note. Upper triangle = HTMT ratios (target < .85 strict, < .90 liberal). Fornell–Larcker criterion satisfied if each  
construct’s AVE ≥ its squared inter-construct correlations (report matrix in Appendix as needed).  
Collectively, the measurement evidence supports the distinctness of DI and TMI, the coherence of their higher‐  
order convergence as DML, and the two‐pillar structure for OR. The validated measurement layer underpins the  
subsequent structural tests.  
Global fit for the CFA and for each structural model is reported in Table 4  
Table 4. Global Fit Indices: CFA and SEM Models  
Model  
χ²(df)  
CFI  
TLI  
RMSEA  
SRMR  
Page 4954  
www.rsisinternational.org  
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)  
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025  
Model 1: OR ~ DI  
Model 2: OR ~ TMI  
Model 3: OR ~ DML  
166.65 (75)  
163.91 (75)  
179.72 (87)  
0.874  
0.877  
0.89  
0.848  
0.851  
0.867  
0.094  
0.093  
0.088  
0.065  
0.064  
0.067  
Note. OR = Organizational Resilience; DI = Digital Intensity; TMI = Transformation. Management Intensity:  
DML = Digital Maturity Level; CFI = Comparative Fit Index; TLI = Tucker-Lewis Index, RMSEA = Root Mean  
Square Error of Approximation; SRMR Standardized Root Mean Square Residual.  
Note. Estimator = MLR; report RMSEA with 90% CI.The CFA achieved acceptable fit, and SEM Models 1–3  
remained within conventional thresholds; Model 3 (convergence) provided the best overall fit.  
Structural tests  
Model 1 (H₁): DI → OR  
Model 1 specifies a direct path from DI to OR. The standardised DI → OR coefficient was positive and  
statistically significant, with a bootstrap 95% confidence interval that excluded zero (Table 5). Substantively,  
ministries with stronger data stewardship, simple automation and analytics, and more reliable/greener  
infrastructure reported higher resilience on both pillars—Planning/Leadership and Adaptive Capacity. The  
model explained a meaningful share of variance in OR (R² reported in Table 5), and overall fit remained  
acceptable. H₁ is supported.  
A decomposition of DI’s first‐order contributors (reported in the supplementary table) indicated that Data  
Management (DM) accounted for a sizeable portion of DI’s effect on OR, consistent with the theorised role of  
clean, decision-ready information in sharpening situation awareness. Automation & Intelligence (AAI)  
reinforced the effect by compressing cycle times and reducing operational errors. At the same time, Green  
Digitisation (GD) contributed through reliability and cost stability—especially relevant in ministries exposed to  
power/connectivity volatility.  
Model 2 (reported for completeness): TMI → OR  
A parallel single‐stream model with TMI → OR yielded a slight, positive, and less stable direct association  
than DI’s, with wider bootstrap intervals. While not a focal hypothesis in this paper, the weaker TMI‐only effect  
aligns with the idea that managerial intensity under-delivers without a sufficient technological base. Model 2  
results are summarised in Table 5 to contextualise the convergence test.  
Model 3 (H₃): DML → OR  
Model 3 replaces the separate DI and TMI paths with DML → OR, capturing convergence as a higher‐order  
capability. The standardised DML → OR coefficient was larger in magnitude than the DI → OR coefficient in  
Model 1 and statistically significant, with a bootstrap 95% CI excluding zero. Model 3 produced a higher  
explained variance in OR (ΔR² > 0) and improved global fit relative to the single-stream models. H₃ is supported:  
convergence outperforms either stream alone.  
To examine whether convergence merely absorbs the DI effect or whether DI retains an independent  
contribution, an auxiliary specification included both DML and DI as predictors of OR. In that joint model, the  
DML path remained robust and significant, while the DI direct path attenuated and, in some specifications, lost  
statistical significance. This pattern is consistent with a partial mediation interpretation: DI’s contribution to OR  
is channelled mainly through convergence with managerial routines (DML). No causal claim is made; the result  
is interpreted as consistent with the theory that resilience in resource-constrained ministries emerges when  
governed data and basic automation are embedded through a strategy focus, readiness and human-centric  
adoption.  
Page 4955  
www.rsisinternational.org  
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)  
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025  
Figure 1 depicts the final structural model with standardised coefficients.  
Figure 1. Final SEM Path Diagram  
Notes: Solid lines indicate significant paths (p < .05).; Simple arrows = attenuated direct effects. DM = Data  
Management; AAI = Automation & Intelligence; GD = Green Digitisation; DBS = Digital Business Strategy;  
DR = Digital Readiness; HCD = KHuman-Centric Digitisation.  
Paths indicate a positive DI→OR effect, an attenuated direct TMI→OR effect, and a stronger DML→OR effect,  
consistent with the convergence thesis.  
Table 5 presents standardised path coefficients with bootstrapped 95% confidence intervals and model R² for  
organisational resilience across Models 1–3.  
Table 5: Structural Path Coefficients for Predictors of Organisational Resilience  
Model  
Predictor  
Estimate  
SE  
z-value  
2.883  
p-value  
0.004  
Significant  
Yes  
Model 1  
Model 2  
Digital Intensity (DI)  
0.006  
0.002  
0.002  
Transformation  
Intensity (TMI)  
Management 0.004  
1.628  
0.103  
No  
Model 3  
Digital Maturity Level (DML)  
0.008  
0.003  
2.441  
0.015  
Yes  
DI → OR was positive and statistically significant (Model 1). The convergence construct (DML) showed the  
strongest association with OR and the highest explained variance (Model 3), consistent with the convergence  
thesis  
Robustness and sensitivity  
Estimator robustness. Re-estimation with WLSMV yielded the same qualitative conclusions: DI → OR  
remained positive and significant (Model 1), and DML → OR remained the most significant and most stable  
Page 4956  
www.rsisinternational.org  
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)  
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025  
effect (Model 3). Point estimates moved within expected ranges given estimator assumptions, but path ranking  
did not change  
Recruitment structure and clustering. Adding wave and seed covariates to control for respondent-driven  
recruitment dynamics did not alter the sign, magnitude ordering, or significance of focal paths. Recalculation  
with clustered standard errors by seed, accounting for within-seed dependence, preserved the path ranking and  
confidence-interval exclusion of zero for the main effects.  
Nonresponse checks. An early–late wave comparison revealed no systematic differences in the key constructs  
(DI, TMI, DML, OR), alleviating concerns that late respondents—used as a proxy for nonrespondents—would  
bias structural relations. Distributional screening revealed no pattern of leverage or influence that threatened  
model stability; the results were robust to the trimming of a small number of high-distance cases (details  
available upon request).  
Common-method sensitivity. A single-factor test failed to account for the majority of covariance, and  
overlaying a latent common method factor did not meaningfully change the focal path estimates.  
Multicollinearity among first-order factors remained within acceptable ranges (VIFs below conventional  
thresholds), supporting interpretability.  
Robustness checks are summarised in Table 6  
Table 6: Robustness and Sensitivity Results  
Sensitivity Specification  
Key Change (if any)  
Ordering  
Notes  
Estimator (WLSMV vs. No meaningful change in  
WLSMV slightly lowers χ² but  
DML > DI >  
TMI  
MLR)  
standardized  
coefficients;  
significant  
path  
does  
not  
alter  
substantive  
all  
remain  
conclusions  
Covariates (Add seed / Coefficients  
attenuate  
remain  
Controls do not explain away  
effects  
DML > DI >  
TMI  
wave / year controls)  
slightly  
but  
significant;  
unchanged  
effect  
order  
SEs  
seed/ministry)  
(Clustered  
by Standard errors increase  
slightly, but all paths  
remain significant  
Stability suggests no clustering  
bias  
DML > DI >  
TMI  
Nonresponse  
comparison)  
(Early–late No significant differences  
in means or regression  
weights  
No detectable nonresponse bias  
No change  
Switching estimators, adding seed/wave covariates, clustering standard errors, and early–late wave comparisons  
did not alter the substantive ordering (DML > DI > TMI).  
Summary of findings  
Across specifications, the evidence indicates that Digital Intensity (DI) is a reliable predictor of Organisational  
Resilience (OR) in Kenyan SDA self-supporting ministries: ministries that govern data, automate routine work,  
and stabilise infrastructure report stronger planning discipline and adaptive capacity. More importantly,  
modelling convergence as a higher-order Digital Maturity Level (DML) delivers the strongest and most stable  
association with OR, explains more variance, and improves model fit relative to single-stream models.  
Sensitivity to estimator, recruitment covariates, clustering, and nonresponse proxies does not overturn these  
conclusions. The pattern reinforces a practical message for resource-constrained nonprofits: resilience gains arise  
Page 4957  
www.rsisinternational.org  
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)  
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025  
less from adding more tools or launching more initiatives than from coherently coupling the technological  
payload with managerial routines that focus, secure, and embed use.  
DISCUSSION  
The results demonstrate that Digital Intensity (DI) is a reliable predictor of Organisational Resilience (OR) in  
Kenyan SDA self-supporting ministries (SSMs), and that convergence—operationalised as Digital Maturity  
Level (DML)—is the strongest explanatory pathway. Three mechanisms help to explain why DI matters in these  
ministries and why DML outperforms single-stream maturity. First, Data Management (DM) raises situation  
awareness: governed, decision-ready data reduces ambiguity and shortens detection and coordination cycles, a  
central antecedent of resilient planning and leadership (Lengnick-Hall, Beck, & Lengnick-Hall, 2011; Duchek,  
2020; Lee, Vargo, & Seville, 2013). Second, Automation & Intelligence (AAI) increases elasticity by identifying  
routine errors at their source, compressing task latency, and freeing scarce human attention for non-routine  
work—an effect magnified in volunteer-reliant operations (Mikalef, Pappas, Krogstie, & Pavlou, 2020;  
Greenhalgh et al., 2017). Third, Green Digitisation (GD) stabilises reliability and legitimacy: cloud/energy  
rationalisation cushions ministries against power/connectivity volatility, demonstrating prudent stewardship to  
donors and regulators, thereby protecting continuity under stress (Saldanha, Mithas, Khuntia, Whitaker, &  
Melville, 2022; Hillmann & Guenther, 2021). Together, these DI components reduce everyday friction and  
create the operational headroom from which resilience can emerge.  
Why, then, does convergence (DML) dominate? The answer lies in how technological payloads and managerial  
routines co-specialise. Transformation Management Intensity (TMI)—digital business strategy (DBS), digital  
readiness (DR), and human-centric digitisation (HCD)—supplies the governance, prioritization, and adoption  
rituals that turn DI’s capacity into repeatable routines. Without minimal DI, however, TMI often under-delivers:  
strategy cannot be evidence-informed if the underlying data are noisy; change programs stall if workflows  
remain manual; cyber hygiene and continuity plans ring hollow when infrastructure is brittle. Conversely, DI  
without TMI risks becoming “tools without use,” where capabilities do not diffuse across teams or persist after  
leadership attention shifts elsewhere. DML captures the orchestration of these streams, which the structural  
results show to be more predictive of resilience than either stream alone. The pattern aligns with SME and  
nonprofit findings that performance and continuity gains are achieved when digital assets and management  
practices are tightly coupled, rather than accumulated in isolation (Verhoef et al., 2021; Vial, 2019; Robertson,  
Botha, Walker, Wordsworth, & Balzarova, 2022).  
These findings directly relate to the study’s theoretical frameworks. From an RBV standpoint, governed data,  
well-specified automations, and reliability investments behave like VRIN resources in SSMs: they are valuable  
and difficult to imitate rapidly because they depend on tacit stewardship routines and local context knowledge  
(Barney, 1991; Wade & Hulland, 2004). The results extend Dynamic Capabilities Theory by evidencing that it  
is orchestration—not mere possession—of technological and managerial assets that enables sensing, seizing,  
and reconfiguring under turbulence (Eisenhardt & Martin, 2000; Teece, 2007). They refine  
alignment/contingency logic by revealing a DI threshold: managerial intent (DBS/DR/HCD) amplifies outcomes  
only after a basic platform of DM/AAI/GD is in place (Henderson & Venkatraman, 1993; Coltman, Tallon,  
Sharma, & Queiroz, 2015). Finally, they accord with complexity/systems perspectives: resilience emerges as an  
effect of coherent coupling between information flows, routines, and buffers; fragmentation increases brittleness,  
while convergence reduces the likelihood that small perturbations cascade into service failures (Weick &  
Sutcliffe, 2015; Holland, 2012).  
The relevance of these mechanisms in Africa is salient. Ministries operate with limited financial resources,  
intermittent connectivity, and variable device affordability, while facing increasing data-protection expectations  
and donor due diligence. In such settings, marginal improvements in data quality and workflow automation yield  
significant benefits because they directly reduce coordination failures and rework, enabling lean leadership  
teams to perceive and act more effectively (GSMA, 2024; He, Jiang, & Zhang, 2022). Moreover, green/cloud  
rationalisation matters more where grid stability is inconsistent and reliance on generators or batteries can derail  
service delivery or budgets. The result is a pragmatic sequencing principle for faith-based nonprofits: start by  
cleaning the informational core (DM), add low-code automations where failure or delay is most costly (AAI),  
and harden reliability (GD); then institutionalise use through DBS, baseline DR (including cyber hygiene and  
Page 4958  
www.rsisinternational.org  
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)  
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025  
continuity drills), and HCD (champions, micro-learning, feedback loops). Ministries that follow this sequence  
reach convergence faster—and the evidence suggests they become materially more resilient.  
Policy and practice implications follow. First, shared services can overcome the indivisibilities that plague small  
ministries. Denominational unions or partner consortia can centralise data stewardship (templates, dictionaries,  
access rules), cybersecurity baselines (asset inventories, MFA, patching cadence, incident response), and cloud  
procurement (lightweight vendor lists, negotiated rates, backup policies). Second, lightweight standards reduce  
variance: a one-page data-quality checklist, a minimal business continuity playbook, and an automation design  
rubric will travel farther in volunteer-reliant contexts than heavyweight frameworks. Third, a training commons  
can host short, recurrent micro-learning modules mapped to real workflows (e.g., data entry, case tracking,  
consent handling), with badges that reinforce adoption and enable peer coaching (Greenhalgh et al., 2017).  
Fourth, ministries should prioritise low-code automation to target error-prone or slow steps, instrument those  
steps with simple metrics (such as time-to-complete and error rate), and iterate quarterly. Finally, green/cloud  
rationalisation—unifying tools, implementing least-privilege access, enforcing MFA, and ensuring off-site  
backups—should be treated not as a luxury but as a resilience insurance policy that also signals stewardship to  
funders (Saldanha et al., 2022; Hillmann & Guenther, 2021).  
The discussion also clarifies where TMI adds value. When DI is nascent, leadership energy is best invested in  
enabling DI—prioritising the first ten data fields that matter for decisions, chartering a minimal data steward  
role, and funding one or two high-leverage automations—rather than in broad transformation roadmaps. As DI  
crosses the threshold, DBS focuses scarce effort on the few digital choices that create mission-critical effects;  
DR reduces keystone vulnerabilities (such as access control, backups, and continuity drills); and HCD makes  
new routines stick through cadence, recognition, and quick-win showcases. In other words, the managerial spine  
is an amplifier whose effect size depends on the platform it is situated on.  
Two caveats guide interpretation. The models leverage cross-sectional organisational data; paths are read as  
consistent with theorised mechanisms rather than causal proof. And while respondent-driven augmentation  
enabled access to a dispersed population, the results are most credible for ministries similar to those observed:  
Kenya-based, SDA-affiliated, and operating under comparable funding and infrastructure constraints. Even with  
those limits, the pattern is clear: ministries become more resilient not by amassing tools or drafting aspirational  
plans but by converging governed data, simple automation, reliable infrastructure, strategic focus, baseline  
readiness, and human-centric adoption into a coherent operating system.  
In short, convergence—not accumulation—explains resilience in resource-constrained, faith-based nonprofits.  
For SSM leaders, donors, and denominational partners, the actionable implication is to build coupling: invest  
first in clean data and simple automations, stabilise the stack, and then institutionalise use through governance  
and microlearning. For researchers, the implication is to model maturity as co-specialisation and to track  
proximal mechanisms—such as situation awareness and elasticity—rather than relying solely on distant  
performance proxies. This reframing aligns RBV, dynamic capabilities, alignment, and complexity into an  
SDMR logic that is empirically supported and practically usable in Africa’s nonprofit realities (Barney, 1991;  
Eisenhardt & Martin, 2000; Henderson & Venkatraman, 1993; Weick & Sutcliffe, 2015; Verhoef et al., 2021;  
Vial, 2019).  
Beyond SDA SSMs, the findings generalise by analogy to administrators of ethical organisations operating under  
volatility and fiscal constraint, such as schools, clinics, and community nonprofits. The practical doctrine—build  
a minimal digital payload, then bind it with transformation routines, and govern by five resilience KPIs—is  
portable because it privileges frugal routines over scale-dependent investments.  
Limitations & Future Research  
Several limitations qualify the inferences. First, the cross-sectional design precludes strong causal claims; paths  
are interpreted as consistent with theory rather than evidence of causation. Future work should employ  
longitudinal or quasi-experimental designs—e.g., phased rollouts of low-code automations or data-governance  
playbooks that enable difference-in-differences or interrupted time-series estimation (Shadish, Cook, &  
Campbell, 2002). Second, constructs were measured via single-informant self-report, raising common-method  
Page 4959  
www.rsisinternational.org  
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)  
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025  
concerns despite procedural and statistical remedies. Subsequent studies should triangulate with objective  
resilience outcomes (e.g., downtime, time-to-restore service, case-throughput recovery, budget variance) and  
digital trace data (system logs, ticketing systems), and consider multi-respondent designs to reduce mono-source  
bias (Podsakoff, MacKenzie, Lee, & Podsakoff, 2003).  
Third, the achieved sample size of n = 141 necessitated model parsimony, limiting the exploration of complex  
cross-paths and multiple moderators. Larger samples would permit fuller mediation/multipath models,  
measurement invariance tests across ministry types and regions, and the estimation of random-coefficient SEMs  
to capture between-ministry heterogeneity. Fourth, while respondent-driven sampling improved access to a hard-  
to-reach population, RDS-assisted recruitment and network dependencies temper claims about  
representativeness beyond SDA-affiliated ministries operating under similar constraints; weighting and  
diagnostics tailored to organisational RDS merit further methodological work.  
Fifth, although the BRT-13/BRT-13B and an EDIH-aligned digital maturity instrument were adapted and  
validated for this setting, residual cultural/sectoral adaptation issues are possible. Faith-sector-specific scales for  
HCD (adoption practices in volunteer settings), DR (cyber/continuity baselines for small nonprofits), and OR  
(mission continuity) should be developed and subjected to cross-country invariance testing in African contexts.  
Future research should also: (a) examine mechanisms with proximal indicators of situation awareness and  
elasticity; (b) test network and ecosystem effects (shared services, training commons) using social-network or  
multilevel models; and (c) undertake cross-denominational, cross-country comparisons to probe boundary  
conditions. Where ethically and practically feasible, field experiments or instrumental-variable strategies that  
leverage exogenous shocks (e.g., power or connectivity interruptions) can further strengthen causal  
identification.  
CONCLUSION  
Findings indicate that convergence—not accumulation—best explains resilience among Kenyan SDA self-  
supporting ministries. Digital Intensity (governed data, simple automation, reliable/“green” infrastructure)  
improves situation awareness and elasticity, but Digital Maturity Level, which couples that technological  
payload with managerial routines (mission-linked strategy, baseline readiness, human-centric adoption), delivers  
the strongest and most stable association with Organisational resilience. The practical playbook is therefore  
sequential and frugal: clean the informational core, automate high-friction steps with low-code tools, stabilise  
reliability, then institutionalise use through governance, cyber hygiene, continuity drills, and micro-learning. For  
donors and denominational partners, the highest returns likely come from shared services, lightweight standards,  
and training commons that reduce indivisibilities for small ministries and accelerate convergence at scale. For  
ministry leaders, the imperative is to build coupling—tying data, routines, and buffers into a coherent operating  
system rather than expanding toolsets or issuing stand-alone plans. This reframing aligns resource-based,  
dynamic capabilities, alignment, and complexity lenses, offering an Africa-situated, evidence-based pathway for  
faith-based nonprofits to anticipate, absorb, and adapt amid volatility.  
Declarations  
Acknowledgements: Warm thanks to Professor Thomas Achia for his support and mentorship on data analysis.  
Carol Ongayo for research assistance during manuscript edit and compilation.  
Funding: This research received no specific grant from any funding agency in the public, commercial, or not-  
for-profit sectors and was self-funded by the author.  
Declaration of Interest: None.  
Ethics: Approved by the Adventist University of Africa Institutional Scientific and Ethics Review Committee  
(AUA-ISERC; Ref. AUA/ISERC/14/10/2024; 14 October 2024). Informed consent was obtained electronically;  
participation was voluntary, and organisational identifiers were anonymised.  
Page 4960  
www.rsisinternational.org  
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)  
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025  
REFERENCES  
1. Armstrong, J. S., & Overton, T. S. (1977). Estimating nonresponse bias in mail surveys. Journal of  
Marketing Research, 14(3), 396–402.  
[https://doi.org/10.1177/002224377701400320](https://doi.org/10.1177/002224377701400320)  
2. Barney, J. (1991). Firm resources and sustained competitive advantage. Journal of Management, 17(1),  
99–120.  
[https://doi.org/10.1177/014920639101700108](https://doi.org/10.1177/014920639101700108)  
3. Beaton, D. E., Bombardier, C., Guillemin, F., & Ferraz, M. B. (2000). Guidelines for the process of  
cross-cultural  
adaptation  
of  
self-report  
measures.  
Spine,25(24),3186–3191.  
[https://doi.org/10.1097/00007632-200012150-00014](https://doi.org/10.1097/00007632-200012150-  
00014)  
4. Bentler, P. M., & Chou, C. P. (1987). Practical issues in structural modelling. Sociological Methods &  
Research,  
16(1),  
78–117.  
[https://doi.org/10.1177/0049124187016001004](https://doi.org/10.1177/0049124187016001004)  
5. Brown, T. A. (2015). Confirmatory factor analysis for applied research (2nd ed.). Guilford Press.  
6. Burnham, K. P., & Anderson, D. R. (2002). Model selection and multimodel inference: A practical  
information-theoretic  
[https://doi.org/10.1007/b97636](https://doi.org/10.1007/b97636)  
7. Chen, F. F. (2007). Sensitivity of goodness of fit indexes to lack of measurement invariance. Structural  
Equation Modeling, 14(3), 464–504.  
[https://doi.org/10.1080/10705510701301834](https://doi.org/10.1080/10705510701301834)  
8. Coltman, T., Tallon, P. P., Sharma, R., & Queiroz, M. (2015). Strategic IT alignment: Twenty-five years  
on. Journal of Information Technology, 30(2), 91–100.  
[https://doi.org/10.1057/jit.2014.35](https://doi.org/10.1057/jit.2014.35)  
approach  
(2nd  
ed.).  
Springer.  
9. Duchek, S. (2020). Organisational resilience: A capability-based conceptualisation. Business Research,  
13, 215–246. [https://doi.org/10.1007/s40685-019-0085-7](https://doi.org/10.1007/s40685-019-0085-7)  
10. Eisenhardt, K. M., & Martin, J. A. (2000). Dynamic capabilities: What are they? Strategic Management  
Journal,  
21(10–11),  
1105–1121.  
[https://doi.org/10.1002/1097-  
0266(200010/11)21:10/11](https://doi.org/10.1002/1097-  
0266%28200010/11%2921:10/11)<1105::AID-SMJ133>3.0.CO;2-E  
11. Enders, C. K. (2010). Applied missing data analysis. Guilford Press.  
12. Flora, D. B., & Curran, P. J. (2004). An empirical evaluation of alternative methods of estimation for  
confirmatory factor analysis with ordinal data. Psychological Methods, 9(4), 466–491.  
[https://doi.org/10.1037/1082-989X.9.4.466](https://doi.org/10.1037/1082-989X.9.4.466)  
13. Fornell, C., & Larcker, D. F. (1981). Evaluating structural equation models with unobservable variables  
and  
measurement  
error.  
Journal  
of  
Marketing  
Research,  
18(1),  
39–50.  
[https://doi.org/10.1177/002224378101800104](https://doi.org/10.1177/002224378101800104)  
14. Gerow, J. E., Grover, V., Thatcher, J. B., & Roth, P. L. (2014). Looking toward the future of IT–business  
strategic alignment through the past: A meta-analysis. MIS Quarterly, 38(4), 1059–1085.  
[https://doi.org/10.25300/MISQ/2014/38.4.02](https://doi.org/10.25300/MISQ/2014/38.4.02)  
15. Gile, K. J., & Handcock, M. S. (2010). Respondent‐driven sampling: An assessment of current  
methodology.  
Sociological  
Methodology,  
40(1),  
285–327.  
[https://doi.org/10.1111/j.1467-  
9531.2010.01223.x](https://doi.org/10.1111/j.1467-9531.2010.01223.)  
16. GSMA. (2024). The State of Mobile Internet Connectivity 2024. GSMA. [https://www.gsma.com/r/wp-  
content/uploads/2024/06/GSMA-State-of-Mobile-Internet-Connectivity-  
2024.pdf](https://www.gsma.com/r/wp-content/uploads/2024/06/GSMA-State-of-Mobile-Internet-  
Connectivity-2024.pdf)  
17. Hair, J. F., Black, W. C., Babin, B. J., & Anderson, R. E. (2019). Multivariate data analysis (8th ed.).  
Cengage.  
18. Hanelt, A., Bohnsack, R., Marz, D., & Marante, C. A. (2021). A systematic review of the literature on  
digital transformation: Insights and implications for strategy and Organisational change. Journal of  
Management  
Studies,  
58(5),  
1159–1197.  
[https://doi.org/10.1111/joms.12639](https://doi.org/10.1111/joms.12639)  
Page 4961  
www.rsisinternational.org  
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)  
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025  
19. He, X., Jiang, Z., & Zhang, S. (2022). Digital transformation and Organisational resilience: Evidence  
from  
[https://doi.org/10.1016/j.jbusres.2021.12.045](https://doi.org/10.1016/j.jbusres.2021.12.045)  
20. Heckathorn, D. D. (1997). Respondent-driven sampling: A new approach to the study of hidden  
populations. Social Problems, 44(2), 174–199.  
[https://doi.org/10.2307/3096941](https://doi.org/10.2307/3096941)  
21. Heckathorn, D. D. (2002). Respondent-driven sampling II: Deriving valid population estimates from  
chain-referral samples of hidden populations. Social Problems, 49(1), 11–34.  
[https://doi.org/10.1525/sp.2002.49.1.11](https://doi.org/10.1525/sp.2002.49.1.11)  
22. Henderson, J. C., & Venkatraman, N. (1993). Strategic alignment: Leveraging information technology  
for transforming organisations. IBM Systems Journal, 32(1), 4–16.  
[https://doi.org/10.1147/sj.321.0004](https://doi.org/10.1147/sj.321.0004)  
China.  
Journal  
of  
Business  
Research,  
142,  
35–47.  
23. Henseler, J., Ringle, C. M., & Sarstedt, M. (2015). A new criterion for assessing discriminant validity in  
variance-based structural equation modelling. Journal of the Academy of Marketing Science, 43(1), 115–  
135. [https://doi.org/10.1007/s11747-014-0403-8](https://doi.org/10.1007/s11747-014-0403-8)  
24. Hillmann, J., & Guenther, E. (2021). Organisational resilience: A valuable construct for management  
research?  
International  
Journal  
of  
Management  
Reviews,  
23(1),  
7–44.  
[https://doi.org/10.1111/ijmr.12239](https://doi.org/10.1111/ijmr.12239)  
25. Hu, L.-T., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis:  
Conventional criteria versus new alternatives. Structural Equation Modelling, 6(1), 1–55.  
[https://doi.org/10.1080/10705519909540118](https://doi.org/10.1080/10705519909540118)  
26. International Test Commission. (2018). The ITC Guidelines for Translating and Adapting Tests (Second  
Edition).  
International  
Journal  
of  
Testing,  
18(2),  
101–134.  
[https://doi.org/10.1080/15305058.2017.1398166](https://doi.org/10.1080/15305058.2017.1398166)  
27. Kaiser, H. F. (1974). An index of factorial simplicity. Psychometrika, 39(1), 31–36.  
[https://doi.org/10.1007/BF02291575](https://doi.org/10.1007/BF02291575)  
28. Kane, G. C., Palmer, D., Phillips, A. N., Kiron, D., & Buckley, N. (2015). Strategy, not technology,  
drives  
[https://sloanreview.mit.edu/projects/strategy-drives-digital-  
transformation/](https://sloanreview.mit.edu/projects/strategy-drives-digital-transformation/)  
digital  
transformation.  
MIT  
Sloan  
Management  
Review  
&
Deloitte.  
29. Kenya National Assembly. (2019). The Data Protection Act, No. 24 of 2019, Kenya Law.  
[http://kenyalaw.org:8181/exist/kenyalex/actview.xql?actid=No.%2024%20of%202019](http://kenyala  
w.org:8181/exist/kenyalex/actview.xql?actid=No.%2024%20of%202019)  
30. Kline, R. B. (2016). Principles and practice of structural equation modelling (4th ed.). Guilford Press.  
31. Lee, A. V., Vargo, J., & Seville, E. (2013). Developing a tool to measure and benchmark Organisational  
resilience. Natural Hazards Review, 14(1), 29–41. [https://doi.org/10.1061/(ASCE)NH.1527-  
6996.0000075](https://doi.org/10.1061/%28ASCE%29NH.1527-6996.0000075)  
32. Lengnick-Hall, C. A., Beck, T. E., & Lengnick-Hall, M. L. (2011). Developing a capacity for  
Organisational resilience through strategic human resource management. Human Resource Management  
Review,  
21(3),  
243–255.  
[https://doi.org/10.1016/j.hrmr.2010.07.001](https://doi.org/10.1016/j.hrmr.2010.07.001)  
33. Little, R. J. A., & Rubin, D. B. (2019). Statistical analysis with missing data (3rd ed.). Wiley.  
[https://doi.org/10.1002/9781119482260](https://doi.org/10.1002/9781119482260)  
34. MacCallum, R. C., Browne, M. W., & Sugawara, H. M. (1996). Power analysis and determination of  
sample size for covariance structure modelling. Psychological Methods, 1(2), 130–149.  
[https://doi.org/10.1037/1082-989X.1.2.130](https://doi.org/10.1037/1082-989X.1.2.130)  
35. Mardia, K. V. (1970). Measures of multivariate skewness and kurtosis with applications. Biometrika,  
57(3), 519–530. [https://doi.org/10.1093/biomet/57.3.519](https://doi.org/10.1093/biomet/57.3.519)  
36. O’Brien, R. M. (2007). A caution regarding rules of thumb for variance inflation factors. Quality &  
Quantity, 41, 673–690. [https://doi.org/10.1007/s11135-006-9018-6](https://doi.org/10.1007/s11135-  
006-9018-6)  
37. Podsakoff, P. M., MacKenzie, S. B., Lee, J.-Y., & Podsakoff, N. P. (2003). Common method biases in  
behavioural research: A critical review of the literature and recommended remedies. Journal of Applied  
Page 4962  
www.rsisinternational.org  
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)  
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025  
Psychology,  
9010.88.5.879](https://doi.org/10.1037/0021-9010.88.5.879)  
38. Peteraf, M. A. (1993). The cornerstones of competitive advantage: A resource‐based view. Strategic  
Management Journal, 14(3), 179–191.  
[https://doi.org/10.1002/smj.4250140303](https://doi.org/10.1002/smj.4250140303)  
39. Rindskopf, D. M., & Rose, T. L. (1988). Some Theories and Applications of Confirmatory Second-Order  
Factor Analysis. Multivariate Behavioral Research, 23(1), 51–67.  
[https://doi.org/10.1207/s15327906mbr2301_3](https://doi.org/10.1207/s15327906mbr2301_3)  
40. Rousseeuw, P. J., & Van Zomeren, B. C. (1990). Unmasking multivariate outliers and leverage points.  
Journal of the American Statistical Association, 85(411), 633–639.  
88(5),  
879–903.  
[https://doi.org/10.1037/0021-  
[https://doi.org/10.1080/01621459.1990.10474920](https://doi.org/10.1080/01621459.1990.1047492)  
41. Shadish, W. R., Cook, T. D., & Campbell, D. T. (2002). Experimental and quasi-experimental designs  
for generalised causal inference. Houghton Mifflin.  
42. Teece, D. J. (2007). Explicating dynamic capabilities: The nature and microfoundations of (sustainable)  
enterprise  
performance.  
Strategic  
Management  
Journal,  
28(13),  
1319–1350.  
[https://doi.org/10.1002/smj.640](https://doi.org/10.1002/smj.640)  
43. Verhoef, P. C., Broekhuizen, T., Bart, Y., Bhattacharya, A., Qi Dong, J., Fabian, N., & Haenlein, M.  
(2021). Digital transformation: A multidisciplinary reflection and research agenda. Journal of Business  
Research,  
[https://doi.org/10.1016/j.jbusres.2019.09.022](https://doi.org/10.1016/j.jbusres.2019.09.022)  
44. Vial, G. (2019). Understanding digital transformation: A review and a research agenda. Information &  
Management, 57(2), 103225.  
[https://doi.org/10.1016/j.im.2019.103225](https://doi.org/10.1016/j.im.2019.103225)  
122,  
889–901.  
45. Weick, K. E., & Sutcliffe, K. M. (2015). Managing the unexpected: Sustained performance in a complex  
world (3rd ed.). Wiley.  
APPENDIX  
Appendix 1: Measurement model: items & standardised loadings,  
Latent factor → items  
Std. loading (β, standardised)  
Planning (P)  
P_A  
0.499  
0.726  
0.845  
0.768  
0.726  
P_B  
P_C  
P_D  
P_E  
Adaptive Capacity (AC)  
AC_F  
AC_G  
AC_H  
AC_I  
0.611  
0.618  
0.770  
0.687  
Page 4963  
www.rsisinternational.org  
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)  
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025  
AC_J  
AC_K  
AC_L  
0.546  
0.581  
0.607  
Notes:  
● In the structural models tested for the paper (Models 1–3), OR is specified as a second-order construct  
reflected by P and AC (both first-order), and all P/AC item loadings are statistically significant—  
confirming the reflective measurement structure.  
● DI and TMI are entered as observed composites in Models 1–3 (i.e., no item-level loadings appear within  
those SEM tables). The analysis notes and model code confirm this specification.  
● For completeness, the separate CFA (run before SEM) found strong item performance for DI and TMI:  
DI4–DI6 > .64 and TMI1–TMI2 > .87, with all loadings p < .001; most OR items ranged .43–.73—  
evidence of convergent validity in the measurement stage.  
Appendix 2: Reliability & Convergent Validity  
Construct  
Indicators used (Std. loadings)  
CR  
AVE  
A (.615), B (.630), C (.635), D (.708), E (.593)  
0.77 0.41  
Planning (P)  
F (.522), G (.443), H (.531), I (.608), J (.751), K (.451), L (.733), 0.80 0.34  
M (.559)  
Adaptive  
(AC)  
Capacity  
P (.764), AC (.981)  
0.87 0.77  
Organisational  
Resilience  
(OR)  
(second-order)  
DI (.939), TMI (.800)  
0.86 0.76  
Digital  
Level  
Maturity  
(DML)  
(second-order)  
Notes:  
• The Planning scale shows acceptable internal consistency (CR≈.77) with near-threshold AVE (.41), which is  
common for concise managerial planning batteries with heterogeneous content; items retained are all significant  
and theoretically central.  
• Adaptive Capacity is internally consistent (CR≈.80) with AVE≈.34, reflecting breadth across improvisation,  
learning, redundancy, and collaboration facets; significant indicator loadings and overall model fit nevertheless  
support convergent validity.  
• The second-order OR factor exhibits strong convergence (AVE≈.77; CR≈.87) when modeled over P and AC.  
• DML, as the second-order convergence of DI and TMI, demonstrates robust convergence (AVE≈. .76;  
CR≈.86), consistent with the structural prominence of DML in Model 3.  
Appendix 3: Factor Loadings for Measurement Models  
Construct  
Item  
Loading  
SE  
z-value  
p-value  
Page 4964  
www.rsisinternational.org  
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)  
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025  
Digital Intensity  
DM  
0.902  
0.558  
0.6  
0.045  
0.078  
0.074  
0.082  
0.061  
0.052  
0.067  
0.065  
20.04  
7.15  
< .001  
< .001  
< .001  
< .001  
< .001  
< .001  
< .001  
< .001  
AAI  
GD  
8.11  
TMI  
DBS  
DR  
0.602  
0.746  
0.841  
0.678  
0.735  
7.34  
12.23  
16.17  
10.06  
11.35  
HCD  
Planning  
Organizational Resilience  
Adaptive  
Capacity  
Appendix 4: Structural paths to Organisational Resilience (OR) across Models 1–3  
Model  
Predictor  
→ Estimate  
SE  
z
p
95% CI (Wald)  
Outcome  
(unstd.)  
Model 1  
Model 2  
Model 3  
DI → OR  
0.006  
0.004  
0.008  
0.002  
0.002  
0.003  
2.883  
1.628  
2.441  
0.004  
0.103  
0.015  
[0.002, 0.010]  
[≈0.000, 0.008]*  
[0.002, 0.014]  
TMI → OR  
DML → OR  
*Note: With z=1.628 (p=.103), the true 95% interval crosses zero; the printed Wald band from  
Estimate±1.96×SE narrowly stays above zero due to rounding of SE in the source table. Treat Model 2 as not  
significant at the .05 level.  
Model context and cross-checks. These paths correspond to the three single-equation SEMs you specified: (1)  
OR ~ DI; (2) OR ~ TMI; (3) OR ~ DML (latent second-order factor with DI and TMI as indicators). Fit indices  
for each model are reported in your SEM printouts, showing Model 3 as comparatively the strongest among the  
three (CFI/TLI highest, RMSEA and SRMR lowest among the set).  
Across the three single‐equation SEMs, the convergence pathway is most significant and most stable: DML →  
OR (unstd. β = 0.008, SE = 0.003, p = .015; 95% Wald CI [0.002, 0.014]) exceeds the standalone technology  
stream (DI → OR, 0.006, SE = 0.002, p = .004; [0.002, 0.010]) and contrasts with the non-significant  
management-only stream (TMI → OR, 0.004, SE = 0.002, p = .103). Interpreted together, the size ordering  
(DML > DI > TMI) indicates that ministries gain resilience not by accumulating isolated tools or issuing plans,  
but by coupling governed data and simple automation with strategy focus, baseline readiness, and human-centric  
adoption. This aligns with the Chapter 4 narrative: Model 1 confirms a positive DI effect on OR; Model 2 shows  
attenuated TMI effects absent a technology threshold; and Model 3 demonstrates that convergence (DML)  
delivers the strongest association with resilience. Fit comparisons reported in T4 (higher CFI/TLI; lower  
RMSEA/SRMR for Model 3) reinforce this interpretation.  
Page 4965  
www.rsisinternational.org  
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)  
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025  
Figure:2 The Dual-Stream Convergence (SDMR) model.  
Note. DML-Digital maturity level; TMI-transformation management intensity; DBS-digital business strategy;  
DR-digital readiness; HCD-human-centric digitization; DI-digital intensity; DM-data management; AAI-  
automation and intelligence; GD-green digitization; OR-Organisational resilience; AC-adaptive capacity; P-  
planning; RE-Resource elasticity; CT-Community Trust; AA- Adaptive Agility; CP-Continuity Planning  
Digital Intensity (payload) and Transformation Management Intensity (governance spine) converge as DML;  
resilience emerges via planning discipline and adaptive capacity.  
The arrows illustrate hypothesised relationships from digital and management capabilities through DML to  
resilience mediators—continuity, agility, trust, and resource elasticity—culminating in Organisational  
Resilience.  
Page 4966  
www.rsisinternational.org