Dan Cassino

One of my first apartments was about 200 meters from that underpass.

We occasionally get nonstandard flavors, but nothing like Japanese red bean paste, or prawn, or whatever, and definitely no vertical packs or multi-flavor packs like the Italians have.
We used to be a country, darn it.

But this… poorly designed. Probably unethical. Doesn’t give much in the way of actionable data. Disappoints voters before you even take office. And that’s assuming they fixed the problems, and are actually collecting data. As I said at the top, we all deserve better.

A real survey, run by the state, to track how well state orgs and programs are doing, with a validated sample and a panel design, would be a great idea. You can incorporate complaints, and have follow ups: see if you fixed a problem. You could see issues before they become crises.

See, you, the new governor, are asking thousands of people about their problems, and what you can do to help, then asking for their phone numbers and emails. Some of those folks are going to take you seriously, and expect an answer. When they don’t get it, you’ve upset them for no reason.

So maybe it’s just a confidence building measure. They’re not planning to analyze the data, just let people know that they’re listening (and get their phone numbers for a text list). But that’s real self-defeating (not to mention dishonest and unethical).

Maybe the plan is to toss the whole corpus into AI: but remember that you have those little nuggets of sensitive information strewn about. Sharing that with any external service is not only a bad idea, but “Go Directly to Jail” illegal.

And if you don’t have that manpower, if you’re not actually planning to do any analysis of the data, it’s completely unethical to collect it in the first place! You’re wasting respondents’ time for no reason!

But we’re still lowballing it. Because whenever you send out open ended govt surveys, no matter how much you warn respondents, some of them are going to give you sensitive information: SSNs and the like. You gotta scrub all those before you do anything, and have a plan to deal with that data.

That’s 30,000 minutes, or 500 man hours to code everything. Without any training, or building the codebook, you’re looking at 12 people working full time coding for a week. Add in training, validation, building the codes, etc, double that.

Of course, not everyone will answer every question, so let’s cut that down to 3 minutes (optimistically).
This was a convenience sample sent out to a huge pool of respondents. Again, let’s be conservative, and say that they get 10,000 responses.

After some reflection, I’m even more baffled by this survey. Seven open ended items. If you’re doing it right, hand coding everything across a huge number of potential response codes, say you can (with training) get it down to one minute per code, so seven minutes per survey.

Flavors aside, the structure of the package- a vertical orientation? They’re years ahead of us.

I had resigned myself to the fact that the US had fallen behind Japan in KitKat technology. But to find out that we’ve been overtaken by Italy really stings.

Spoiler: this is a spectacularly bad, (literally) broken survey, to the extent that I’m not sure what they were actually trying to do. Everyone involved deserves better.

The NJ public, and our Governor-elect, deserve better.

It’s a bad survey. It’s poorly designed. It has huge respondent burden. The population is ill defined. I can’t imagine what the analysis plan looks like. And it didn’t work on an iPhone, which, to my knowledge, is a pretty popular way for folks to take surveys online.

I’ll decline to make a judgment on which I think is more likely. Either way, this is really a wild way to start things off for a new administration. If they want a public priorities survey, surely they know someone who can do a competent job. If not, I’ll do it for free.

Two: they did a bad job on the survey because this is just a ploy to build up their text list, while making it look like they’re listening. They don’t have an analysis plan, because the plan is no analysis.
Even then, they did a bad job, because they couldn’t even collect my email or text.

Two options.
One: the Sherrill transition team just did a bad job on this survey. They slapped something together, didn’t test it, didn’t think about an analysis plan. Not ill-intentioned, just incompetence.

a close up of a man 's face with blood running down his face . ALT: a close up of a man 's face with blood running down his face .

Maybe it’s to do with the failure in the municipal field above? Ok, trying to fix, so I don’t lose the half hour I spent on my responses, but… the page resets.

Add a captcha, and we’re done. Phew!

What the junk do you mean “Calculations have failed. Please try to fill all from beginning.” ?

Oh, the race/ethnicity item is fine. I’d have MENA on there, and weird to be shy about asking for specifies now, but what evs.

a man with a mustache is wearing a suit and tie . ALT: a man with a mustache is wearing a suit and tie .

The problem is that they probably want sex (man, woman, nb, etc). But people misunderstand “gender,” so they’re going to get sexuality, trans status, “yes,” trolls saying “attack helicopter,” etc.
You’re telling me someone is going to be handcoding all these responses?

And now demos.
Honestly, I’ve never seen gender as *just* an open-ended item. I can see thinking of it as being open-minded; but it’s going to give bad outcomes (my next book is on gender measurement, so I’ve thought about this a lot).

On to the substantive questions. Which are all open-ended. Oof.
That’s a huge respondent burden. I hope they have a plan to read and code all the responses they’re getting. Seven long open-ended is rough; but I care about this stuff, so I spent about 30 minutes typing in responses.

Next, it asks if I’m ok with being contacted about my responses. But: if you give them your number for feedback, they’re adding you to their text list.
Not exactly frugging, but not great. I’m losing confidence in this survey by the second. Are we sure this isn’t just a way to build a text list?

Next question is about what group I most identify with. This is a weird one. “I identify as an Energy/Utility”? Like, PSEG is using this survey to talk to the governor? “State Agencies”?! What the junk?

On to geography questions. Weird that they’re called “identifying,” but ok.
And the survey is already broken. I put in my county (Essex), and there’s no options for municipalities. And it doesn’t accept typed entries.
But it’s not a required field, so whatever.