Data analytics adoption in Nepali SMEs: Survey dataset and analysis outputs (N=396)
Description
This dataset contains the survey responses behind a study of data analytics adoption in small and medium enterprises and its association with the effectiveness of business decision-making. Responses were gathered from 396 owner-managers, managers, and senior staff of SMEs across 3 cities of Nepal (Kathmandu, Pokhara and Bharatpur), using a structured questionnaire administered through KoboToolbox. The questionnaire measured nine determinants drawn from the technology-organization-environment framework, namely relative advantage, compatibility, complexity, top management support, organizational readiness, IT capability, competitive pressure, government support, and vendor support, together with data analytics adoption and decision-making effectiveness. All 43 measured items use a seven-point Likert scale running from 1 (strongly disagree) to 7 (strongly agree). The file also records firm and respondent characteristics, including sector, firm size, firm age, respondent position, and the analytics tools currently in use. Every field that could identify a respondent or firm has been removed. The accompanying codebook gives the exact wording and coding of each variable.
Files
Steps to reproduce
Questionnaire items were adapted from established TOE, technology adoption, and decision-making scales, then reviewed for face validity before administration. The instrument was built in KoboToolbox and administered to owner-managers, managers, and senior decision-makers of SMEs across 3 cities of Nepal (Kathmandu, Pokhara and Bharatpur). 407 responses were collected. Three firms above the 250-employee SME ceiling and seven that were ineligible on inspection, either non-SME entities or firms outside the target activity band, together with one straight-line response that gave an identical answer to every item, which left 396 valid cases. Likert responses were coded as integers from 1 to 7. Complexity items (CX1 to CX3) were retained as recorded rather than reverse-coded; the expected negative association is carried by a negative path coefficient in the structural model. The measurement and structural models were estimated in SmartPLS using partial least squares structural equation modelling, with bootstrapping used for significance testing. The higher-order TOE constructs were specified through a two-stage disjoint approach, with data analytics adoption modelled as a mediator and decision-making effectiveness as the outcome.
Institutions
- Pokhara UniversityGandaki Pradesh, Pokhara