Table 3.

Results of model comparison using PSIS LOO-CV as indicated by the ELDP difference from best model, and posterior means of parameters $\beta$ and ${\gamma }_k$. ATM=active transport measures.

	${\mathcal{M}}_1$	${\mathcal{M}}_2$	${\mathcal{M}}_3$	${\mathcal{M}}_4$	${\mathcal{M}}_5$
Population density (1000 pers./${\mathrm{km}}^2$)	1.102⁎	1.085⁎	0.811⁎	0.803⁎	0.833⁎
Cycling strongly associated	0.824⁎	0.832⁎	1.051⁎	1.065⁎	0.917⁎
Needs for ATM before Covid-19	0.440	0.499	0.972⁎	0.959	0.973
Quality of cycling infrstr. before Covid-19	0.082	−0.014	−0.135⁎	−0.134	−0.036
Mobility transition is priority		0.374
Infl. of parties		0.370
Political parties in favour of ATM		0.942
Infl. of interest groups		1.598
Interest groups in favour of ATM		0.618
Pol.parties infl:in favour		0.156
Inter. group infl:in favour		−1.585
Temp. ATM benefits are overest.			1.527⁎	1.609⁎	1.607⁎
Temp. ATM made permanent			1.641⁎	1.549⁎	1.681⁎
Benefit overestimated:made permanent			−0.252	0.306	−0.263
Temp. cycling infrastr. resource intensive			0.490	0.531	0.497
Covid-19 increased need for cycling infrstr.				0.021	0.034
Covid-19 opened window of opp. for ATM				0.478
Covid-19 window[k = 1]					1.261⁎
Covid-19 window[k = 2]					−0.283
Covid-19 window[k = 3]					−0.764
Covid-19 window[k = 4]					−2.156⁎
ELPD difference	−13.3	−15.2	−0.7	−3.1	0.0
sd(ELPD difference)	7.7	8.5	4.4	4.7	0.0
Rank	4	5	2	3	1

^*95% credible interval excludes zero.