. 2014 Nov 14;16(11):e250. doi: 10.2196/jmir.3532

Table 4.

Correlations between tweet rates and emergency department ILI^a rates by city.^b

	1. All tweets	2. Non-retweets	3. Retweets	4. Fisher’s z transformation^c	5. Tweets without a URL	6. Tweets with a URL	7. Fisher’s z transformation^d	8. Total number of tweets
	r	r	r	P	r	r	P
Boston	.23	.47	−.004	<.001	.03	.41	<.001	17,370
Chicago	.51^e	.54^e	.23	<.001	.59^e	.45^e	<.001	21,655
Cleveland	.68^e	.87^e	.39	<.001	.62^e	.58^e	.005	7152
Columbus	.62^e	.54	.61	.018	.62^e	.47^e	<.001	3288
San Diego	.80^e	.92^e	.40^e	<.001	.88^e	.79^e	<.001	8002
Seattle	.72^e	.71^e	.67^e	.001	.62^e	.71^e	<.001	9735

^aILI: influenza-like illness

^bCorrelation coefficients of all tweets and tweet categories with emergency department ILI rates for each city. Comparisons between tweets and ILI began in Weeks 40-41 (weeks starting September 29, 2013 to starting October 6, 2013) as ILI data became available by city and ended in Week 9 (ending March 1, 2014).

^cThis column displays the P values from Fisher’s z transformation comparing the correlation coefficients of non-retweets to retweets.

^dThis column displays the P values from Fisher’s z transformation comparing the correlation coefficients of tweets without a URL to tweets with a URL.

^eSignificant correlation coefficient (P<.05).