Table 2. Results of pairwise Wilcoxon signed rank tests (p-values) for the various materials as specified in questions Q9a-9d (Pipe, Small ball, Big ball, Hard wood), 8d (Bare chain) and Q10a-10d (Short chain, Chain hanging too high, Chain on the floor, Branched chains), sorted by average score and Friedman mean rank.

	Branched chains	Chain on the floor	Hard wood	Pipe	Bare chain	Short chain	Small ball	Big ball	Chain hanging too high	Mean Rank	Level 1	Level 2	Level 3	Level 4	Level 5
Branched chains		0.000	0.001	0.000	0.000	0.000	0.000	0.000	0.000	8.21	a
Chain on the floor	0.000		0.062	0.011	0.004	0.000	0.000	0.000	0.000	6.84		b*
Hard wood	0.001	0.062		0.364	0.623	0.003	0.007	0.003	0.000	5.61		b#	C
Pipe	0.000	0.011	0.364		0.697	0.037	0.024	0.009	0.000	5.53			C
Bare chain	0.000	0.004	0.623	0.697		0.017	0.017	0.013	0.000	5.42			C
Short chain	0.000	0.000	0.003	0.037	0.017		0.455	0.068	0.000	4.26				d*
Small ball	0.000	0.000	0.007	0.024	0.017	0.455		0.168	0.000	4.18				d*#
Big ball	0.000	0.000	0.003	0.009	0.013	0.068	0.168		0.000	3.71				d#
Chain hanging too high	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000		1.24					e
Count (n)	28	28	27	27	22	28	27	26	28
Count > 5.5	14	9	6	6	4	3	2	1	0
Average	5.07	4.36	3.74	3.52	3.32	3.08	2.80	2.52	1.34

Count (n) gives the number of respondents providing a welfare score (scale 0 to 10). Count > 5.5 are numbers of respondents giving a score that is higher than 5.5, which was specified as the cut-off point for acceptability. Cells marked green represent statistically significant pairwise comparisons (p<0.05). Light green: trend (p<0.1); Yellow: not significant. Levels 1–5 show groups of enrichment materials, sorted from high to low, where different characters (a-e) indicate levels of statistical significance, and where different symbols (*, #) indicate statistical trends (so d*, Short chain, tends to differ from d#, Big ball (p = 0.068), but not from d*#, Small ball).