Table 5. Goodness-of-fit of the tweets model is influenced by disease incidence, access to computers, internet and social factors.

Cities were divided into two groups: high, for cities with model Dengue estimate explained deviance equal or higher than 60%, and, low, otherwise.

	Goodness-of-fit of the tweet model
Variable Description	High		Low		p value
	Mean	Std Error	Mean	Std Error
Population (n)	375,440.00	48,784.00	31,3882.00	10,1736.00	***
Gross Domestic Product per capita	34,599.00	1,793.00	28,611.00	1,844.00	**
IDHM Mean (2010)	0.77	0.003	0.74	0.004	***
IDHM Income (2010)	0.,75	0.004	0.73	0.005	*
IDHM Longevity (2010)	0.85	0.002	0.84	0.002
IDHM Education (2010)	0.70	0.004	0.67	0.005	***
Houses with personal computer (%)	52.46	0.79	47.10	1.13	**
Houses with internet (%)	42.96	0.80	37.48	1.01	***
Dengue cases, 2012 (n)	333.90	54.31	125.20	28.71	**
Dengue cases, 2013 (n)	6,419.00	1,093.00	1,553.00	255.10	**
Dengue cases, 2014 (n)	2,465.00	391.50	1412.00	511.00	**
Dengue cases, 2015 (n)	6369.00	868.40	2667.00	730.10	**
Dengue cases, 2016 (n)	4,293.00	1,084.00	1,506.00	305.50	**
Total Dengue cases (n)	19,880.00	2,851.00	7,164.00	1,669.00	**
Total Dengue Incidence (per 100,000 inhabitants)	6,604.00	400.80	3,976.00	355.90	***
Tweets, 2012 (n)	38.65	9.22	23.08	8.84	***
Tweets, 2013 (n)	378.90	94.69	134.20	56.58	***
Tweets, 2014 (n)	136.30	27.17	94.70	48.06	***
Tweets, 2015 (n)	316.10	60.03	218.20	129.70	***
Tweets, 2016 (n)	288.50	72.19	147.10	74.02	***
Total Tweets (n)	1,159.00	256.30	586.80	301.00	***
Twitter activity per 100,000 inhabitants	270.00	13.14	137.00	7.01	***
Linear Correlation (Dengue~Tweets)	0.67	0.01	0.43	0.02	***
Weeks with tweets (%)	55.60	1.72	35.89	2.13	***
Weeks with Dengue (%)	85.82	1.11	67.23	2.57	***
Mean relative error of model estimate	1.43	0.06	0.98	0.11	***

* p value <0.01

** p value <0.001

***p value <0.0001