Politique de Confidentialité

Privacy context for a browser-based statistics calculator needs more detail than a generic statement about respecting user information. People often paste small but meaningful datasets into quick tools: classroom grades, process yields, monthly returns, survey scores, or laboratory readings. Even when those values are not traditionally classified as highly sensitive, they may still reveal performance, health-adjacent behavior, personal history, or proprietary business information once they are tied to a person, group, or operational unit. That is why this site explains local calculation in plain language instead of assuming that every visitor already understands browser-side execution. Local execution means the mathematical operations happen on the device in the active browser session. The form fields receive the values, the script computes the result, and the output is displayed without requiring the dataset to be transmitted to the service as part of the normal workflow. For a user, that is a meaningful difference. A page that looks like any other website may in practice behave more like a local utility. Calling out that distinction reduces uncertainty and helps the visitor decide whether the tool is appropriate for exploratory work with real numbers. There is also a practical educational reason to explain this architecture carefully. Many people learning statistics are also learning basic digital privacy habits at the same time. They may know what standard deviation is but not what a browser stores, what a cookie does, or what information is normally present in a web request. By describing the calculation boundary clearly, the site supports statistical learning and digital trust at the same time. That is especially useful for classrooms, ad hoc workplace analysis, and self-directed study where no dedicated data-governance professional is present. The phrase anonymous analytics can be misunderstood if it is left unexplained. Some users hear analytics and assume the worst; others ignore the term entirely. A responsible privacy page separates operational metrics such as page views, device characteristics, and coarse usage patterns from the actual numerical content entered into a calculator. Those categories serve different purposes and carry different implications. Operational metrics may help maintain the site. Calculator inputs belong to the analytical task the user is performing and therefore deserve a much stricter boundary. Support interactions are one of the few paths where user-provided content can exceed the default privacy envelope. If someone emails a bug report and includes a screenshot or copied table, that communication may contain more information than the site ever processes automatically. A thoughtful policy should mention that reality because users tend to focus on the interface they can see and forget the voluntary disclosures they control outside the page itself. Advising users to sanitize examples is an inexpensive but meaningful form of risk reduction. Long-form explanation is not included here merely to satisfy a crawler or heuristic. Depth has a real usability function. A short promise such as we value your privacy does not answer the questions users actually have: Do my inputs leave my device? Are they stored anywhere? Is the result page itself a record? Are analytics the same thing as data capture? Does the site operator ever review my numbers? Can I safely test a real dataset? The policy tries to answer those questions directly because clarity is more useful than reassurance. The privacy discussion also reflects the kind of product this is. The site is not only a destination for arithmetic. It is a publishing surface for definitions, use cases, and interpretation advice. Readers come here to understand what a statistic means and how to apply it responsibly. If the site takes educational clarity seriously in one area, it should take the same approach in explaining how user-entered data is handled. That consistency helps the entire product feel more credible. Finally, the privacy model reinforces a healthy workflow for everyday analysis. Users can test assumptions quickly, compare sample and population modes, and validate a result without first moving the dataset into a heavier software environment. That convenience should not come at the cost of ambiguity about data handling. Clear boundaries make the tool more usable, more trustworthy, and more appropriate for the many lightweight analytical tasks that happen before formal reporting ever begins.