. 2016 Sep 7;18(9):e239. doi: 10.2196/jmir.5531

Table 2.

Descriptive statistics, correlations, agreement, and Bland-Altman parameters for Fitbit One and ActiGraph GT3X+.^a

Statistic	Fitbit One		Jawbone UP
	Steps	MVPA	Steps	MVPA	Longest sedentary bout
Mean (SD)	9221 (3416)	17.0 (17.6)	8690 (4029)	75.5 (35.5)	87.1 (45.6)
GT3X+, mean (SD)	8497 (2878)	36.6 (25.0)	7511 (2692)	37.4 (22.8)	46.4 (9.8)
r /ρ (95% CI)^b	.85 (.80, .89)	.80 (.73, .85)	.75 (.67, .81)	.75 (.67, .81)	.19 (.03, .34)
ICC (95% CI)^c	0.90 (0.86, 0.93)	0.72 (–0.15, 0.90)	0.79 (0.72, 0.84)	0.56 (–0.20, 0.83)	0.08 (–0.12, 0.27)
Mean difference (SD)^d	0.2*x–916.1 (1820)	–0.4*x–9.2 (–19.2)	0.5*x–2491.3 (699.0)	0.5*x+10.6 (38.1)	1.7*x–72.3 (44.8)
95% Limits of agreement^d
Upper	(0.2*x–916.1)+3567.0	0.2*x+0.5	(0.5*x–2491.3)+5290.0	0.8*x+20.2	(1.7*x–72.3)+87.8
Lower	(0.2*x–916.1)–3567.0	–0.8*x–18.8	(0.5*x–2491.3)–5290.0	0.08*x+1.1	(1.7*x–72.3)–87.8

^aDays analyzed for Fitbit One: n=135; days analyzed for Jawbone UP: n=154.

^bCorrelations for steps were calculated using Pearson correlation coefficient (r). Correlations for MVPA and longest idle time were calculated using Spearman rank correlation coefficient (ρ) due to non-normally distributed data. All correlations were significant at P<.05.

^cAll agreements are significant at P<.001 except for longest sedentary bout (P>.99).

^dWhere the mean difference or limits of agreement were systematically biased, equations are presented, where x=a given value on the x-axis (mean of device and ActiGraph GT3X+ value).