A web-based digital assistant for sample size determination and stratified sampling design

An adequate sample size and an appropriate sampling technique are frequently mischosen by students and empirical researchers. This study reports the development and internal validation of the Research Sampling Calculator & Assistant, a web-based decision support application for sample size determination and stratified sampling design. The application was developed as a client-side single-page web application (React, TypeScript, Tailwind CSS, and Vite) that implemented five sample size formulas (Slovin; Lemeshow; Cochran with and without finite population correction; Krejcie and Morgan; and the Hair et al. indicator-based multivariate ratio), a rule-based sampling-technique decision engine, a proportional and disproportional stratified allocation algorithm with an automated minority-stratum safeguard, and a local Excel-based random sampling engine that performs all computation inside the user's browser without transmitting participant data to a server. Formula outputs were validated against published reference tables and independent manual calculations. The disproportional allocation algorithm was tested on a simulated three-stratum population. Outputs from all five formulas were in perfect agreement with reference values. The disproportional module was able to eliminate minority-stratum underrepresentation while maintaining overall target sample size. Structured feature comparison with G*Power, Raosoft, SurveyMonkey, and OpenEpi shows that the developed application is the only one among the five that combines multi-formula guidance, interactive sampling-technique decision engine, auditable stratified allocation, and privacy-preserving local sample execution in a single freely accessible interface. The application gives a practical, validated, and privacy-conscious contribution to diminish common errors in sample size and sampling design in empirical research.