Anyone have experience with PSA Japan? I have a few friends who are superstitious about PSA JP quality control and send their JP cards to be graded in the US rather than locally.
2 Likes
I heard from several japanese shop owners that PSA Japan is supposed to be stricter. But It feels to me that this opinion is a mix of rumours going around and confirmation bias. So I would take that opinion with a big grain of salt.
2 Likes
Yea, I heard the same rumors. Was hoping to get outside perspective from someone who has experience grading with both.
Question about PSA. If you accidently forgot to list a card in your order but send it in any ways what happens?
they consider it as a nice tip when they hand you your psa 6
2 Likes
is this a serious answer?
it is not
1 Like
there is hope… I guess I’ll submit a ticket and see what happens hopefully they just return it
1 Like
If lucky you might be able to submit another order for it while they have it in hand but I would try calling on Tuesday morning to see if you can speak with someone vs waiting on email response
Way lower gem rates out of NJ. I think they are significantly stricter on centering. I’ve sent multiple submissions of same quality cards from the same sets. 87% gem rates from CA, 45% gem rates from NJ.
Across hundreds of cards, that’s far outside random variation.
3 Likes
It will went to “problem” order, and it will take a few week longer, and they will send you an email and ask if you want the card graded or not
2 Likes
PSA grades ~15 million cards annually. I’m sorry to say, but your sample size of a few hundred cards across a handful of months can’t say much about CA vs. NJ trends. Anecdotal experiences are helpful and should be shared, but I would hesitate to say that it’s far outside of random variation.
6 Likes
For your argument to be valid, that means that submissions ~100-200 cards have to be invalid for assessing grading standards. If the grading standard is really so variable that 40% (or 100%, depends how you want to count it) differing gem rates falls within “acceptable variance,” what does that actually say about PSA’s grading standards?
p<0.0001 lol, sure, let’s call that “the same standard…”
CA is my 3rd most recent return, NJ is 2nd. Here is that same return “NJ01” compared to my most recent, also from NJ:
Not significantly different, as one would hope. The same site applies the same grading standard across multiple submissions. Standards vary between sites. It is what it is.
There is a big difference between saying:
- My two submissions had different grades on average.
and
- These two sites grade differently on average.
Even if we assumed that every card between your two submissions were the exact same in terms of surface/corner/edge damage, centering, etc., your p-value would still not say anything about grading at the site. It applies to #1 and not #2. You are generalizing your single outcome to the entirety of their grading system at each site, which is a statistical fallacy.
Because you quote statistics, you probably also know that there is a huge difference between statistically significant and practically significant or meaningful. Your statistically significant difference doesn’t even reach a small effect size value (Cohen’s d = 0.0871; Small ~ 0.2, Medium ~ 0.5, Large ~ 0.8).
In other words, your p-value doesn’t say anything about the size of the difference, which happens to be minuscule. But of course you would know all of this because you threw stats at me. 
7 Likes
Yes, there is the issue that my complaint is only valid if the same quality cards are being graded. Luckily, I do also make that claim, considering that these submissions were the same cards from the same sets. Unfortunately there’s not really any way for me to prove this, but I’d argue that I don’t have an incentive to be lying. I want PSA to be consistent. They are, within a site. The problem is that as a customer I have no control of which site grades my submission.
Also this issue of size of difference, of course it’s going to be minuscule - we are looking at an average grade of 9.4 vs 9.7. Once again, this is an issue due to another factor beyond the customer’s control. Grades are not real decimal numbers, but integer values. The metric that actually matters to the customer is gem rate, what percent of cards came back with a 10 grade. Very minor differences in average grade return massive swings in the quantities of 9 vs 10, even more so when 90%+ are one of those 2. Obviously if one submission had a bunch of 1s and the rest 10s, it would be even more skewed, but that’s not what happened. You can actually see the standard deviations among these 3 submissions are all very similar.
In the end, this is only 3 total submissions: 1 CA, 2 NJ. I have 2 more submissions is in assembly right now, so hopefully I get at least 1 back from CA for additional comparison.
You have to remember that you are not submitting to “PSA,” but to a single (or two) graders out of dozens or hundreds. There is going to be inter-grader variability (i.e., variability between graders) and intra-grader variability (i.e., variability within the same grader over time).
When we send in submissions, we get a minuscule snapshot of a single grader during a single time. PSA’s job is to make sure that this experience can be consistent across multiple sites, time periods, and graders. Unfortunately, they cannot and will never reach perfect reliability.
I am shocked by how confident you are that your few anecdotal experiences generalize to an entire organization, rather than suggesting that you have had bad luck or that this difference is random variance and will normalize with a sufficiently large sample size.
6 Likes
I appreciate the attempted use of stats to demonstrate a difference in grading. This is the correct approach to really demonstrate the difference. As @Dyl mentioned, the effect size and statistical significance are different things. For a normie reading this, a very tiny p-value will tell you there is a difference between the two groups you are comparing but tells you nothing about how big that difference is.
But my primary issue with the analysis is that the statistics assume independent sampling of observations. But if we are comparing two submissions, the problem is there are a ton of uncontrolled variables, most importantly the grader. Effectively you have compared one grader to another and any detectable difference can be attributed to the difference in the two graders as much as CA vs NJ. Or even something as benign as the time of day or the weather outside are also uncontrolled.
Ideally, we would treat each submission as one data point. If we could accumulate hundreds of examples of submissions, we could get a more robust statistic
3 Likes
how does one know where your card is graded? Don’t we all just send to CA or can you choose NJ if you want?
1 Like
I want to say that I’ve had similar experiences with the NJ facility. My gem rate has been slashed significantly every time it returns from NJ.
1 Like
Look at the FedEx tracking of where the package was picked up from.
1 Like