An online reaction-time score is an end-to-end measurement. It begins when software schedules a visual change and ends when the browser receives an input event. Between those points are several queues, physical transitions, and timing boundaries.

The result therefore contains human reaction plus device and software latency. Understanding the chain helps explain why the same person can score differently on two computers.

The measurement chain

A simplified visual-click test includes:

  1. JavaScript schedules the signal.
  2. The browser prepares a new frame.
  3. The operating system and graphics system present it.
  4. The display scans and pixels begin changing.
  5. The user sees the signal and initiates movement.
  6. The mouse or touch device detects the action.
  7. The input travels through hardware and the operating system.
  8. The browser receives an event and reads the clock.

Every step can add delay or variation.

Display refresh timing

A display updates at discrete intervals. At 60Hz, refresh opportunities are about 16.67 ms apart. At 120Hz, they are about 8.33 ms apart. If the browser finishes a frame just after one opportunity, it may wait for the next.

Higher refresh rate reduces this waiting interval but does not guarantee that the browser, graphics system, or panel uses every opportunity correctly. Confirm the active configuration with the refresh-rate test.

Pixel response and processing

After a frame is sent, pixels need time to reach the new value. Display scaling, noise reduction, motion interpolation, and other processing can add buffering. A low-latency or game mode may disable some processing, although names and effects vary by manufacturer.

An online test cannot separate scanout, processing, and pixel response from human perception. It observes only the combined outcome.

Input-device latency

A mouse samples its switches and reports data at intervals. Wireless transport, power saving, USB scheduling, internal debounce, and operating-system queues contribute to the time before the browser sees a click.

A higher polling rate reduces the maximum wait between reports, but total latency also depends on firmware and the complete connection. Touch screens follow a different sampling and gesture-recognition path, so mouse and touch results should not be compared as if they were equivalent.

Browser and system scheduling

JavaScript shares resources with page rendering, extensions, background applications, and the operating system. Heavy work can delay a timer or event handler. Browsers also reduce work in background tabs, which is why a reaction test must remain visible.

The timer used by the page can be precise while the scheduling around it varies. Precision in the displayed number does not eliminate uncertainty in when the signal became visible.

Network latency usually does not matter

Once a simple reaction test has loaded, it can run entirely in the browser. No server round trip is needed between the signal and click. Internet speed is therefore not part of the timing path unless the site sends every trial to a server before displaying a result.

Mini Test Lab measures and scores reaction rounds locally.

Estimate device influence through comparison

You cannot subtract one universal device-latency number, but you can run a controlled comparison:

  1. Use the same person, browser, input method, and test.
  2. Warm up on both configurations.
  3. Alternate five-round sets rather than completing all tests on one device first.
  4. Record at least five set averages per configuration.
  5. Compare the medians and the spread of results.

Alternating reduces the effect of practice and fatigue. A consistent difference across several sets is more meaningful than two personal-best scores.

Reduce avoidable latency and variation

  • Select the intended display refresh rate.
  • Keep the test on the primary, directly connected display.
  • Close heavy background applications and unnecessary extensions.
  • Avoid remote desktop and video streaming paths.
  • Connect input devices directly during diagnosis.
  • Use the same button and hand for every comparison.
  • Keep battery-saving modes consistent.

Do not over-correct the score

Subtracting a guessed monitor or mouse delay creates a number that looks cleaner but may be less accurate. Device components interact, and published laboratory measurements may not match your settings or connection.

For personal tracking, preserve the raw five-round average and record the configuration. For broader interpretation, read what is a good reaction time. The honest description is not “my nervous system reacted in exactly 212 ms.” It is “this browser-and-device setup measured a 212 ms average under these conditions.”