Visions of recovery: a cross-diagnostic examination of eating disorder pro-recovery communities on TikTok

Table 1 Engagement and demographic data and group differences of posts

	ANA	ARFID	BED	MIA	ORTHO	Total	Significance
Posts (n)	50	47	50	45	49	241
Fans (median, IQR)^	8,161^a 14,159.3	4,024^a 21,937.0	14,600^a 57,966.5	7,256^a 27,377.5	13,200^a 32,324.5	9,113 28,096.5	F (1,236) = 2.63 p = 0.035
Plays (median, IQR)*	233,850^a 429,100.0	1,288^c 4,637.0	11,200^b 29,936.5	5,689^b 24,894.5	10,600^b 13,929.5	12,800 152,491.0	F (1,236) = 47.48 p < .001
Female-presenting (n, %)*	46 92.0%	33 70.2%*	47 94.0%	42 93.3%	43 87.7%	211 87.6%	χ² (4, 241) = 19.35, < .001
White-passing (n, %)^	47 94.0%	38 80.9%	40 80.0%	42 93.3%	38 77.6%	205 85.1%	χ² (4, 241) = 10.48, p = 0.033

Analysis of Variance (ANOVA) was run on log(10) transformed Fan and Play data, as these were highly positively skewed. Superscripts indicate which groups are significantly different from one another per Tukey post-hoc tests. Chi-square analyses were run on the demographic data followed by cell-comparisons based on standardized adjusted residuals
ANA #anarecovery. ARFID #arfidrecovery, BED #bedrecovery, MIA #miarecovery, ORTHO #orthorexiarecovery
*Statistically significant group differences at p < .05 level after application of the Benjamini–Hochberg procedure
^marginally significant group differences at p < .1. After the Benjamini–Hochberg procedure, the highest p-value considered statistically significant was p = 0.016, and the highest p-value considered marginally significant was p = 0.078. For cell comparisons using Z-scores, a significant p-value was set to p < .05. We only assessed cell comparisons in cases when the omnibus test was significant after the Benjamini–Hochberg procedure. Fishers Exact Tests were used if any expected cell count was less than 5

ISSN: 2050-2974