The Estes method, step by step
In the Estes method one person wears noise-isolating headphones fed directly by a spirit box and a blindfold, and says aloud whatever they hear, while the others ask questions the listener cannot hear. Standard settings are FM, 200 ms dwell, never below 100 ms, volume high.
Where it comes from
The method takes its name from Estes Park, Colorado, where it was developed by Karl Pfeiffer, Connor Randall and Michelle Tate around 2016 during sessions at the Stanley Hotel. Its spread through podcasts and television made it the most widely used spirit box protocol today, and the one most apps and devices now ship a preset for.
What it is for
Spirit box sessions have an obvious weakness: the person listening also hears the questions, so hearing an "answer" in the fragments is partly expectation. The Estes method separates the two roles. The listener cannot hear the questions, so any correspondence between a question and what they call out cannot come from the listener knowing what was asked. It does not make the fragments anything other than radio; it removes one specific bias from how they are interpreted, and that is why serious investigators adopted it.
The setup
- The listener wears closed, noise-isolating headphones plugged straight into the spirit box, and a blindfold. They do not know the questions.
- The questioner(s) sit apart and ask questions at normal volume. They note what the listener says and when.
- Settings: FM band, 200 ms dwell, never below 100 ms, volume high enough that room sound cannot leak in.
- The listener says every word they hear, immediately, without filtering for relevance. Filtering is the questioners' job, afterwards.
- Record everything: the box's output and the room, so that what was heard can be checked against what was asked.
Running it with SpiritBox
The app has an Estes preset: 200 ms dwell with the screen fully black, so the only thing left is audio; tap the screen to leave. Headphones on, start a recording before the preset, and the session log will give the questioners something the original method lacked: for every fragment the listener calls out, the log shows which station was audible at that second — a station that can be looked up in the public directory and listened to afterwards. A "yes" at 00:41 that turns out to be a Hungarian sports commentator is still a "yes"; you just know where it came from.
Mistakes that spoil a session
- Open-back or leaky headphones, so the listener hears the questions.
- Dwell below 100 ms: the audio turns into noise and the listener stops hearing anything.
- The listener editing what they say. The method only works with everything spoken aloud.
- No recording. Without it, the matching of answers to questions is memory, and memory favours hits.
Questions
Why FM and not AM?
FM carries more talk and music at higher audio quality, which gives more speech-sized fragments. AM gives a narrower, older sound; the P-SB7 and the app both offer it, and some investigators prefer it.
How long should a session last?
Practitioners usually run 10 to 20 minutes per listener and then swap roles. Longer sessions tire the listener and the number of called-out words drops.
Does the method prove anything?
It controls for one bias — the listener knowing the question. It does not change what the box produces, which is radio fragments and noise. Treat it as a cleaner way to run the experiment, not as evidence in itself.