A controlled comparison instead of a single winner
This scroll wheel comparison test describes several dimensions and preserves protocol limits. It never combines incompatible delta units or turns a short browser sample into a universal ranking of two devices.
How to use the Scroll Wheel Comparison Test
Configure and lock
Choose 10, 15, or 30 seconds, direction rules, input mode, axis, and optional warm-up. Add local labels if they help you remember the setups.
Run A, then repeat as Run B
Use the same motion in the same capture area. Run B inherits the locked Run A protocol; restart the comparison if any configuration must change.
Review dimensions, quality, and units
Read rate, median and p95 intervals, variability, direction consistency, candidates, charts, and flags together. Save or export only when you choose.
What You Can Compare When You Compare Mouse Scroll Wheels
The scroll wheel comparison test can compare two mice, two modes on one mouse, a browser setting before and after a change, two browsers, or two deliberate scrolling techniques. A mouse wheel A B test counts trusted `wheel` events received inside the capture area and summarizes event rate, event spacing, peak one-second density, direction consistency, timing variability, and opposite-direction candidates. These are outcomes from the selected browser protocol, not measurements of an internal encoder or USB report stream.
Run labels such as Work Mouse or Gaming Mouse are private notes stored only when you save the summary. The interface does not infer a device name. Results are presented side by side, and difference notes describe which run was higher or more variable on a particular dimension. There is no single Winner because a higher event rate, lower timing variability, and fewer direction changes are separate observations with different uses.
How to Keep Run A and Run B Fair
Keep duration, direction mode, axis, warm-up, browser zoom, window focus, input mode, system settings, and gesture style constant. When you compare two scroll test runs, the wizard locks Run B to Run A’s configuration so an accidental control change cannot create an invalid comparison. If you need to change a setting, restart the whole comparison and record both runs again.
A three-second warm-up can help establish motion before the measured period, but use it for both sides or neither. One-direction mode is appropriate when you can repeat one requested action. Free scrolling permits intentional changes and therefore does not treat the minority direction as an opposite candidate. If a tab is hidden or the window loses focus, the run pauses or receives a quality limitation rather than quietly continuing.
Why Event Rate Is Not Polling Rate
Events per second counts `wheel` messages delivered by the browser during the measured duration. Hardware polling rate describes a different layer of the input pipeline. Device firmware, drivers, operating-system acceleration, browser coalescing, event dispatch, main-thread scheduling, and the user’s motion stand between those layers. The peak one-second count is likewise a browser event-density summary, not a polling-rate measurement.
A higher browser-received rate may make a plotted trace look denser, but it does not automatically make a device faster, healthier, or better. Median interval describes typical spacing, p95 shows a slower tail of intervals, and MAD describes robust timing variability. Reading them together is more informative than treating EPS as a hardware specification.
When Raw Delta Values Can Be Compared
Raw delta totals are shown only as comparable when both runs use the same protocol, axis, direction rule, and one matching `deltaMode`. Pixel, line, and page delta units are different categories. The tool does not multiply them into a supposed common physical distance. When units differ, the comparison keeps unit-independent measures such as event count and timing but displays a clear warning that raw delta totals are not directly comparable.
Even matching delta modes do not guarantee the same physical scale across devices or browsers. Raw totals can be useful within a tightly controlled same-environment experiment, especially when checking a settings change, but they remain browser values. Display charts may scale values for readability without changing stored raw values.
Comparing Notched and Free-Spin Wheels
A standard notched wheel often creates compact groups associated with deliberate steps, while free-spin behavior can create a longer, denser event sequence. Select the correct mode yourself and use the same choice for both runs. Comparing notched Run A with free-spin Run B can be an intentional exploration, but it answers a mode question rather than a like-for-like device question.
Opposite-direction candidates are meaningful only in a controlled one-direction run. Momentum, bounce-like sequences, intentional reversal, or software mapping can all produce a minority sign. The comparison reports candidates and quality; it does not say an encoder is damaged.
Comparing Browsers and Settings
To compare browsers, keep the device and physical action as consistent as possible, then label each browser locally. To compare scroll speed after a setting change, keep every other control fixed. Differences may reflect event dispatch, smoothing, settings, extensions, load, or focus rather than the mouse. For before-and-after settings tests, change one setting at a time and note it in the labels before saving or exporting.
A browser page cannot guarantee identical system conditions, so repeat the complete A/B protocol. One pair can reveal a pattern worth checking; repeated pairs help show whether that pattern persists. Use the same scroll wheel comparison test when you compare mouse scroll wheels again. Do not create an unsupported global average or percentile from your local runs.
Understanding Session Quality
Session quality describes whether the sample supports the requested comparison. Limited sample size, a hidden page, lost focus, mixed delta units, a long pause, an unexpected zoom modifier, protocol changes, or frequent direction changes can limit interpretation. A quality flag is not a mouse-health score.
When quality is limited, repeat the affected side by restarting the full comparison so Run A and Run B remain aligned. The local result center also checks test type, protocol version, duration, direction mode, delta mode, and serious flags before comparing saved summaries. It will explain an incompatibility instead of forcing a misleading contest.
Scroll Wheel Comparison Test questions
Does a mouse wheel A B test name a winner?
No. It describes rate, timing, direction, and quality separately because one number cannot represent every scrolling characteristic.
Can I change Run B settings?
Not inside a locked comparison. Restart the entire protocol so both sides use the new settings.
Is events per second the mouse polling rate?
No. It is the rate of browser-delivered wheel events during this run.
Why is raw delta comparison disabled?
The delta units or another required protocol field differ. Pixel, line, and page values must not be treated as one physical scale.
Are my A and B labels uploaded?
No. Labels stay in the browser unless you actively include them in a local download and share that file yourself.