Q and Trump's tweets

Q claimed that some Trump tweets, posted minutes after a drop, proved its authenticity (the "deltas"). Trump tweeted about 13 times a day, so coincidences are unavoidable. The question is whether there are more than chance produces.

Gap between each drop and nearby tweets

Tweets within 60 minutes of each drop, in 2-minute bins. A negative value means the tweet precedes the drop.

Show data

min	observed	expected (chance)
-60	107	94.73
-58	97	96.45
-56	94	95.79
-54	95	95.12
-52	86	95.08
-50	100	93.37
-48	82	96.17
-46	108	96.82
-44	111	95.57
-42	104	95.31
-40	108	94.31
-38	101	96.03
-36	103	93.92
-34	111	92.65
-32	76	94.32
-30	105	92.78
-28	95	96.08
-26	94	92.86
-24	91	95.02
-22	93	92.07
-20	84	90.72
-18	78	90.41
-16	93	93.88
-14	100	93.03
-12	96	92.94
-10	101	93.08
-8	94	92.11
-6	87	93.7
-4	99	91.94
-2	81	92.5
0	91	93.06
2	103	91.81
4	98	93.35
6	107	91.49
8	74	90.29
10	74	91.18
12	75	91.87
14	86	91.31
16	93	91.0
18	77	90.34
20	73	90.55
22	66	90.73
24	69	90.07
26	85	91.92
28	83	91.45
30	84	90.72
32	83	90.31
34	77	91.28
36	80	90.77
38	97	89.65
40	103	89.87
42	74	88.53
44	79	89.86
46	90	88.65
48	104	90.43
50	90	88.54
52	88	87.63
54	102	89.58
56	70	90.41
58	68	89.79

Baseline	Window	Direction	Observed	Chance	× median	p
local (±30 min–3 h)	1 min	tweet after drop	1.0%	0.9%	1.15	0.186
local (±30 min–3 h)	5 min	tweet after drop	4.2%	4.0%	1.05	0.227
local (±30 min–3 h)	10 min	tweet after drop	7.4%	7.3%	1.01	0.442
local (±30 min–3 h)	30 min	tweet after drop	16.5%	17.6%	0.94	0.995
local (±30 min–3 h)	60 min	tweet after drop	27.9%	28.8%	0.97	0.948
local (±30 min–3 h)	1 min	tweet before drop	0.7%	0.9%	0.84	0.887
local (±30 min–3 h)	5 min	tweet before drop	3.8%	4.0%	0.96	0.739
local (±30 min–3 h)	10 min	tweet before drop	7.4%	7.4%	1.01	0.463
local (±30 min–3 h)	30 min	tweet before drop	18.2%	17.8%	1.02	0.197
local (±30 min–3 h)	60 min	tweet before drop	30.9%	29.9%	1.03	0.041
days (±1–14 d)	1 min	tweet after drop	1.0%	0.8%	1.30	0.035
days (±1–14 d)	5 min	tweet after drop	4.2%	3.3%	1.28	0.001
days (±1–14 d)	10 min	tweet after drop	7.4%	6.1%	1.20	0.001
days (±1–14 d)	30 min	tweet after drop	16.5%	14.8%	1.11	0.001
days (±1–14 d)	60 min	tweet after drop	27.9%	25.1%	1.11	0.001
days (±1–14 d)	1 min	tweet before drop	0.7%	0.7%	0.99	0.532
days (±1–14 d)	5 min	tweet before drop	3.8%	3.4%	1.13	0.052
days (±1–14 d)	10 min	tweet before drop	7.4%	6.3%	1.18	0.001
days (±1–14 d)	30 min	tweet before drop	18.2%	15.4%	1.18	0.001
days (±1–14 d)	60 min	tweet before drop	30.9%	26.5%	1.17	0.001

Method

The actual share of drops followed (or preceded) by an original tweet within a window is compared with 1,000 draws where each drop is randomly shifted. The period runs up to January 8, 2021, with 4,953 drops and 46,694 original tweets (retweets excluded).

Two baselines. "Days" keeps the time of day but not the co-activity of heavy news days, when both Q and Trump post more. It overstates synchrony, and the excess shows up in both directions. "Local" keeps the activity of neighbouring hours and only breaks minute-level synchrony, so it is the one that judges the "deltas" claim. Highlighted rows: p < 0.05, uncorrected for multiple comparisons.

Q quotes Trump

59 drops link to a @realDonaldTrump tweet (63 found in the archive). The median delay between the tweet and its quotation is 114 min. Q reacts to tweets far more than it announces them.

13 drops mention a "delta" (Δ).