Alfa İstatistik

What it does

• Data and hypothesis check

  It checks whether the hypotheses fit the data and removes personal data.

• Scale detection and scoring

  It detects scales automatically and verifies the scoring key with two separate models.

• Data preparation

  It handles missing data and outliers and calculates reliability coefficients.

• Assumption checks and test selection

  Before every analysis it checks the assumptions, then selects and runs the appropriate test.

• Four-layer validation

  Results pass through four layers of validation. If a validation layer fails, the system does not carry on silently; it stops the work.

• Results chapter

  It writes the results chapter with APA 7 tables as a Word document, in Turkish or English.

Outputs

You receive the results chapter (.docx), the cleaned and scored dataset, the analysis code (R/Python) and raw outputs. A decision log records what each decision was, why it was made and by whom. When supervisor feedback arrives, the work runs again as a new version, only from the affected stage.

Principles of trust

AI never produces a number; every calculation is done in Python and R code. The language model only suggests decisions, frames questions and writes text. No raw data is sent to the model, only column names and summary statistics.

How it is used

It is a self-service tool that runs on credit packages. Users can check the output themselves or, for additional credits, request a review by an Alfa İstatistik expert. Universities and research institutions get an institutional account with internal roles and a shared credit pool. The interface and reports are available in Turkish and English.

Who it is for

Master’s and doctoral students, academics, universities and research institutions.