Well, it’s a short duration event and it takes a little bit of sensing time. You probably reacted within 200 milliseconds or so. A GFCI is allowed to take almost 6 seconds to trip on a small amperage, but most trip way faster than that.
It probably felt a lot longer than that, lol.
edit:Amps not volts.
Just a thought. Maybe they used a metal junction box as a ground, but the junction box wasn’t actually grounded, or properly bonded to the grounding system.
I think you’re on the same path as me. Bad ground connection. Crazy thing is that the tester showed a good ground before I pressed the test button. But then again, I know sometimes the testers can be fooled.
A 3-light tester will often show a ground when it’s improperly grounded like I mentioned.
Yep. Tester failed to notify me that I was the good ground, lol.
It knew it could depend on you.
So, is your recommendation to the buyer to not hold on to the cover while testing the receptacle?
Interesting. Doesn’t look like you’re the first. I would love to hear one of our Sparky’s take on your experience…Glad you’re OK 
You can only get shocked if you were grounded somehow. How might that have occurred? This comment makes a valid point.
I would say so or the lack thereof a proper ground path for a miss-wired GFCI receptacle and enclosure.
A GFCI (Ground Fault Circuit Interrupter) outlet," when installed in a metal box," relies on the metal box’s grounding to provide additional safety.
I stop using 3 test bulb Sperry’s over a decade ago. Why-Reversed polarity bootleg ground and more. When I did, I used a non contact voltage tester in combination for outdoor GFCI anything in metal enclosures.
Observation. Exterior: Miswired GFCI. Below porch window. Recommend a licensed electrical further evaluate and correct. Act upon any recommendations therein.
Likely through my knee touching the ground as I knelt to inspect and test the receptacle, or through my feet, or through contact with the cover/box, if it is grounded.
Incomplete hypotheses.
False. You can get shocked if you are not grounded.
Current flow requires a - path- circuit.
Ground is a reference point.
Touch both the live/hot path and neutral/return conductors of a live electrical receptacle. This will cause a shock even if you are not touching the ground or grounded.
If there’s a voltage difference and a conductive path (like your body), current can flow, and you can be shocked.
I ran into a similar situation on a fire investigation. The insurance company didn’t hire us until after the same house had two fires. Even then, they didn’t hire us. The house had aluminum siding. Workers who were doing the restoration were being shocked and there were sparks coming from the siding. They refused to continue working. That’s when the insurance company finally hired us.
The problem was intermittent. It was so intermittent that the electricians on the job were baffled. Every time they tried to figure out what was going on, everything seemed normal.
I told the insurance company’s SIU guy that I had narrowed it down to a bedroom wall and I needed his approval to open the wall, which he agreed to.
What I found was a lag screw that was used to secure an aluminum four-season room to the house a year or so earlier. The lag screw was not in the stud. The threads were cutting into an NM cable in the wall next to a stud. When the sun came out and warmed the aluminum frame, the lag screw moved. It was acting like a switch.
The lag screw caused both fires and could have killed someone.
It is not necessary for a person to be grounded to be electrocuted. All that is necessary is to complete a circuit.
What a great story, George Forensic Electrical Consultant and Investigator.
One that should be posted frequently when discussing metal siding or even InterNACHI creating a special educational article partnered with yourself about inspecting metal siding and what to look for.
As always. Thanks.
Correct, I misspoke. I meant complete a path or circuit
Not your fault. I misspoke.
I should have accounted for that in my post.
Sorry David.
You do realize that if something is plugged into the GFCI and it got wet somehow, the receptacle should trip, but using your method the test button only, it might not trip when needed. Using the test button is the same as throwing a hairdryer into a bathtub. It should trip. If it does not trip using a tester then it fails in my book.
I have to agree. If it fails with a shunt to ground, it is basically useless.
The 3 light tester is not the recognized testing method. The test button is the appropriate method. You may be writing up a fully functional unit using your method, especially on 2 wire ungrounded wiring.
The 3 light testers will not trip an ungrounded GFCI device, so the test button is the only way to test those. If an ungrounded receptacle being tested isn’t identified with a placard stating “no equipment ground” I write it up as a receptacle that tested open ground.
If that were true, then why are GFCIs suitable replacements (with protection) for ungrounded outlets? It is not true.