Table 2.

Coding scheme for content analysis of Twitter and Tumblr between-platform differences.

Code		Twitter (N = 353)	Tumblr (N = 356)	χ2
Location where post originated	United States	78 (22.1%)	33 (9.3%)	ns
	United Kingdom	19 (5.4%)	12 (3.4%)	ns
	Canada	6 (1.7%)	5 (1.4%)	ns
	Germany	5 (1.4%)	5 (1.4%)	ns
	Other	19 (5.4%)	9 (2.5%)	ns
	Unknown	225 (64%)	292 (82%)	ns
Gender	Male	52 (14.7%)	20 (5.6%)	p < 0.001
	Female	249 (70.5%)	196 (55.1%)	ns
	Organization (charity or business)	15 (4.2%)	2 (0.6%)	p < 0.001
Identifiable	Full name provided	112 (31.7%)	15 (4.2%)	p < 0.001
	Identifiable from photo	116 (32.9%)	90 (25.3%)	ns
Original post written by user		213 (60.3%)	47 (13.2%)	p < 0.001
Nature of post	Pro-ana	91 (25.8%)	49 (13.8%)	p < 0.001
	Anti-ana	95 (26.9%)	175 (49.2%)	p < 0.001
	Pro-recovery	38 (10.8%)	50 (14%)	ns
Body images and nature of image	Thinspiration	32 (9.1%)	44 (12.4%)	p < 0.001
	Challenged norm around thinness as the ideal	1 (0.3%)	95 (26.7%)	p < 0.001
	Body image shared (unclear purpose)	4 (1.1%)	19 (4.8%)	ns
Emotion expressed	Negative emotion	28 (7.9%)	23 (6.5%)	ns
	Positive emotion	7 (2%)	3 (0.8%)	ns
	Mixed emotion	1 (0.3%)	9 (2.5%)	p < 0.01
Support offered and nature of support	Pro-ana support	3 (0.8%)	3 (0.8%)	ns
	Anti-ana or pro-recovery support	29 (8.2%)	35 (9.8%)	ns
	Support offered (unclear purpose)	2 (0.6%)	3 (0.8%)	ns
Raising awareness	Whether the post appeared to aim to raise awareness of issues around ED in others.	35 (9.9%)	53 (14.9%)	p < 0.05
Challenging norms	Whether the post appeared to be aiming to challenge social norms, e.g., pressure to be thin.	46 (13%)	134 (37.6%)	p < 0.001
Humor	Whether the post referred to ED in a humorous manner.	96 (27.2%)	64 (18%)	p < 0.01