ESM Figure 1. Percentage stair climbing in normal weight and overweight commuters over successive weeks of a calorie-based intervention.

ESM Table 1.

Average traffic and the percentage of traffic in the early period (08:00-08:59) across intervention weeks.

Wk 1 / Wk 2 / Wk 3 / Wk 4 / Wk 5 / Wk 6 / Wk 7 / Wk 8 / Wk 9 / Wk 10
Average traffic
(Rank) / 74.0
(2.5) / 74.0
(2.5) / 77.1
(6) / 71.4
(1) / 81.3
(8) / 77.6
(7) / 82.1
(9) / 75.9
(5) / 84.6
(10) / 74.5
(4)
%traffic in early period
(Rank) / 68.5
(8) / 70.5
(10) / 66.9
(6) / 58.6
(1) / 58.7
(2) / 60.8
(4) / 67.1
(7) / 63.2
(5) / 68.6
(9) / 59.8
(3)

This supplementary figure depicts stair climbing in normal weight and overweight commuters over successive weeks of the study. While it shows the overall pattern of the data averaged across longer time intervals in the paper, i.e. an increase in the percentage stair climbing for the overweight commuters, there is a lot more variability. The table below the figure presents the average traffic in each week and the percentage of the traffic occurring in the earlier morning period (08:00-08:59) during which stair climbing is more likely. These data have been ranked across weeks from 1 (low) to 10 (high) for each of the variables (see below).

It is important to note that the raw percentages presented in this figure are uncorrected for the influence of other, relevant variables summarised in the table. Effects of pedestrian traffic volume on stair climbing are almost ubiquitous; increased traffic is almost invariably associated with increased stair climbing [1]. At busy times, a pedestrian approaching the choice-point may find the escalator full and choose the stairs as the quicker alternative. Importantly, in travel contexts such as stations, the pulsatile nature of trafficflow as many disembarking travellers seek to leave the station at the same time results in large magnitude effects on stair climbing. For example, in a busy station with up to 300+ passengers leaving each train, the stairs became saturated with passengers such that 45% chose the stairs as traffic rates reached 150-200. This represented an increase of some 25% in absolute terms relative to the least busy trains [2]. The net outcome is that any spontaneous variation in traffic flows between weeks will not be accounted for in a figure depicting the raw percentages. Similarly, this figure is not corrected for any variation between weeks in the percentage of the passengers travelling in the earlier morning period (08:00-08.59) relative to the later one (09:00-09:59), which had a large magnitude effect on overall rates of stair climbing in this data set, or the ratio of males to females, which had less of an influence here. The point is that greater variation is to be expected when data are depicted over shorter time periods as spontaneous variations in the usage of the station between weeks will have greater leverage at smaller sample sizes. To provide a simple graphic illustration of the potential magnitude of the effects of these uncontrolled variables, the ranks for each week have been summed across the two uncontrolled variables and the resultant sum plotted for each week on the figure (broken line). As can be seen, the variability in stair climbing, in part, matches the effect of these uncontrolled variables that reflect natural variation in the use of the site across weeks.

One further point is worth noting from the data presented in the table. During week four, i.e. the second week of the point-of-choice prompt, average traffic and the percentage of it occurring in the earlier period were at the lowest point for both variables. Consequently, the ‘spontaneous’ variations in usage of the station during week four would be biased against stair climbing and may, in part, contribute to the absence of an overall effect for the point-of-choice prompt alone in this data set.

1. Webb OJ, Eves FF, Kerr J. A statistical summary of mall-based stair climbing interventions. J Phys Act Health. (in press).

2. Eves FF, Lewis AL, Griffin C. Modelling effects of stair width on rates of stair climbing in a train station. Prev Med. 2008; 47: 270-272