Thread by @lxeagle17

Wow, even for midterms, this is a new one.

@EchelonInsights poll of #TXGov (813 LV, 8/31-9/7):

@GregAbbott_TX: 48%
@BetoORourke: 46%

Biden underwater at 41-58 in TX (47-52 nationally)

netchoice.org/wp-content/upl…

@EchelonInsights @GregAbbott_TX @BetoORourke Okay, now some other good insights from this.

#AZSen and #AZGov:

@KariLake: 40%
@KatieHobbs: 50%

@bgmasters: 37%
@CaptMarkKelly: 52%

(773 LV, Biden approvals at 40-59)

#GASen and #GAGov:

@ReverendWarnock: 50%
@HerschelWalker: 40%

@staceyabrams: 48%
@BrianKempGA: 47%

(751 LV, Biden approvals: 43-55)

@ReverendWarnock @HerschelWalker @staceyabrams @BrianKempGA Generic ballot:

Democrats: 46%
Republicans: 46%

(1228 LV)

look, don't ask me, okay? I have no idea either except to say I'm very, very skeptical that the generic ballot is tied while O'Rourke is only down two in Texas or while Kelly has a 15 point lead. Those two things cannot both be true at the same time.

Let's keep going, because this is too wild to not.

@JohnFetterman: 57%
@DrOz: 36%

@JoshShapiroPA: 55%
@dougmastriano: 36%

(828 LVs, Biden at 46-54)

@JohnFetterman @DrOz @JoshShapiroPA @dougmastriano #OHSen and #OHGov:

@JDVance1: 39%
@TimRyan: 45%

@MikeDeWine: 54%
@nanwhaley: 35%

(831 LV, Biden at 39-60)

And now for the last one: #FLSen and #FLGov:

@marcorubio: 50%
@RepValDemings: 41%

@RonDeSantisFL: 52%
@CharlieCrist: 42%

(815 LV, Biden at 40-58).

@marcorubio @RepValDemings @RonDeSantisFL @CharlieCrist My take to wrap it up: when something flies in the face of all the other evidence, something is probably off, and I would not be uncomfortable saying that I find the results here very, very odd and don't think they mirror the final results or even the current state of the races.

@marcorubio @RepValDemings @RonDeSantisFL @CharlieCrist That said, I don't think they're worth discarding either; Echelon is a reputable pollster. I'd throw them in the average and move on. I'm admittedly trying to figure out *how* these numbers came in to existence, but whatever.

also props to them for releasing all this data -- a lot of people would have spiked the survey instead of publishing the outliers.