Immobility measured by transport surveys, evolution of the

Immobility measured by transport surveys, evolution of the
questionnaire from 1994 to 2008
J-P Hubert, S. Roux
Chercheurs INRETS
9/7/2008
European conference on quality in official statistics
Rome, ISTAT, 8-11 / 07 / 2008
Overview
› Immobility in transport surveys, introduction
– Survey themes and definition of immobility
› Critics concerning the indicator’s quality
– Accuracy, consistency, and relevance to improve
› Changes in the 2008 questionnaire
– Immobility during a week with sliding observation day
› First attempt to compare 1994 data and 2008 partial data
– Methods, results by age- and activity-type-groups
› New indicators of immobility on a weekly basis
– Number of “mobile days” per week, alternate immobility rate
› Conclusions
Page 2
J.-P. Hubert, S. Roux
Q2008 Rome
9/7/2008
Introduction: the French national travel
household survey
› Main themes of “Enquête nationale sur les transports et les
déplacements”
– Socio-demographics
– Equipment at disposal for transportation
‐ Vehicles and licenses
‐ Public transport pass
‐ Parking facilities
– Travel behaviour
‐
‐
‐
‐
Regular trips from home to work / school (most household members)
Daily mobility on a reference day (one selected member “kish”)
Long distance mobility the last three months (one selected member “kish”)
Weekly diary for trips made with one selected vehicle
› One-year survey (April 2007 to April 2008), 18.000 households.
Page 3
J.-P. Hubert, S. Roux
Q2008 Rome
9/7/2008
Immobility in transport surveys, definition
› No trip made on the closest weekday before the visit,
i.e. the “reference weekday”
– In 1994, 16% of the individuals were immobile on the
reference weekday, because:
‐ No need to go out
‐ Temporary physical incapacity
‐ Permanent physical incapacity
‐ Need to stay at home
(somebody ill at home, work to do)
‐ Other
:
:
:
:
73%
10%
6%
5%
:
6%
Page 4
J.-P. Hubert, S. Roux
Q2008 Rome
9/7/2008
Critics concerning the indicator’s quality
› Accuracy: 2% of the 1994 sample is suspected of laziness or “soft
refusal”
– No trips declared on the reference weekday, Saturday, and
Sunday because “no need to go out”, three times.
› Consistency: discrepancies with Time-Use surveys (TUS)
– Comparisons made on French, Belgian, and British data show
large discrepancies: according to TUS, about 8% of the
individuals stay at the same place during a whole weekday,
according to NPTS, between 17 to 25%; though:
‐ Data on the place of activity in TUS are not beyond reproach;
‐ Belgian and British methodologies are based on self administrated paper
questionnaire (Be) or diary (UK).
› Relevance: one weekday is not enough
– Insufficient information on people who seldom go out of home
Page 5
J.-P. Hubert, S. Roux
Q2008 Rome
9/7/2008
Changes in the 2008 questionnaire
› In 1994, immobility is derived from one-day observations
– Did you make trips between 4 that day and 4, the day after? yes/no
› In 2008, immobility is observed on a sliding observation
weekday defined by a series of questions:
Last week, are there days when you have not been out ?
1)
2)
3)
…
7)
yesterday (the day, e.g. Monday)
the day before (the day, e.g. Sunday)
three days ago (the day, e.g. Saturday)
yes/no
yes/no
yes/no
seven days ago (the day, e.g. Tuesday)
yes/no
– The sliding observation weekday is the closest weekday when the
respondent was mobile. Immobility, except if it was during the whole
week, does not reduce the time of interview anymore.
– Daily mobility is collected for 95% of the individuals
› Weekly mobility profile now available for all individuals
Page 6
J.-P. Hubert, S. Roux
Q2008 Rome
9/7/2008
First attempt to compare 1994 and 2008 data
› No correction for non response on 2008 partial data
› Time period reduced to 10 months, March and April still missing for 2008
› Comparison on specific groups:
– Age-groups by day of the week
‐ 6-14
‐ 15-64
‐ 65-84
– Activity groups (15-64)
‐
‐
‐
‐
‐
‐
‐
Students
Unemployed
At home
Retired
Employed working at home
Employed working at various places
Employed working on a fix place (Females or Males)
– Scholars (6-14) during school or holidays periods
Page 7
J.-P. Hubert, S. Roux
Q2008 Rome
9/7/2008
Results:
+
1994 immobility rate
2008 immobility rate
95% confidence interval
With filter on questionable immobility in 1994
Page 8
J.-P. Hubert, S. Roux
Q2008 Rome
9/7/2008
Results:
Students
(>15)
Not working (<65)
Economically active at work (<65)
Children (6-14)
Retir: Retirement (<65)
W_swp: in several workplace
Ch_sc: school period
Stdnt
Not_W: Not economically
active at work (<65)
Whome: at home
Ch_ho: holyday period
Unemp: Unemployed
+
W1wpM: in one workplace (males)
W1wpF: in one workplace (females)
1994 immobility rate
2008 immobility rate
95% confidence interval
With filter on questionable immobility in 1994
Page 9
J.-P. Hubert, S. Roux
Q2008 Rome
9/7/2008
New indicators on a weekly basis
› Are the mobile persons on the reference weekday
always mobile during the week?
Number
of mobile
weekday
(%)
Mobile…
Stdnt
Retir
5
88
71
68
71
90
75
92
90
91
80
4-3
10
23
24
23
9
18
7
9
8
14
2-1
2
6
9
6
1
8
1
1
1
7
08 Immobility rate
Diff 08 / 94
20,0%
-1,8%
23,7%
-1,6%
5 - 14 years old indviduals
Mon Tue Wed Thu
Fri
Mobile…
Number
of mobile
weekday
(%)
7,7%
-2,5%
Not_W Unemp W_swp Whome W1wpM W1wpF Ch_sc
16,8%
-2,4%
8,0%
0,1%
22,5%
-5,4%
5,7%
1,7%
15 - 64 years old indviduals
Mon Tue Wed Thu
Fri
6,9%
1,3%
6,3%
-1,3%
Ch_ho
21,9%
0,4%
65 - 84 years old indviduals
Mon Tue Wed Thu
Fri
5
90
91
88
89
90
86
85
85
83
86
70
57
68
65
62
4-3
7
9
10
8
9
11
12
12
15
12
22
30
22
26
23
2-1
3
0
2
4
2
3
3
3
3
3
8
13
10
9
15
08 Immobility rate 11,2% 7,1% 17,4% 6,9% 5,6% 13,5% 9,5% 12,1% 11,7% 9,4% 34,0% 26,0% 31,2% 32,1% 32,5%
Diff 08 / 94
-0,3% 4,0% 0,4% -1,2% -2,2% 1,4% 1,1% 1,6% -0,5% -0,1% -5,6% -6,2% -6,0% 4,1% -1,9%
Page 10
J.-P. Hubert, S. Roux
Q2008 Rome
9/7/2008
Alternate immobility rate
› Two ways to measure the rate of immobility on each
weekday: the ranks do not change but difference can be
significant. On average, a difference of almost 0.4 point.
all week
days
reference
weekday
diff
Pros
Cons
Mo
16.9%
[16.4-17.4]
18.7%
[17.3-20.2]
1.8%
Tue
14.9%
[14.4-15.4]
13.8%
[12.5-15.0]
-1.2%
Wed
15.6%
[15.1-16.0]
16.7%
[15.4-18.1]
1.2%
Thu
15.2%
[14.7-15.7]
16.1%
[14.8-17.5]
1.0%
Fri
14.1%
[13.6-14.5]
13.4%
[12.3-14.4]
-0.7%
: smaller standard error, stable indicator (last 4 days to last 7 days)
: possible memory effect (less immobility on remote weekdays)
Page 11
J.-P. Hubert, S. Roux
Q2008 Rome
9/7/2008
Conclusion
› Accuracy and Consistency
– “Soft refusal” seems, at least partly, controlled by the
filter on three day-immobility without motive
– 1994 filtered data seem consistent with 2008 partial data
‐ Lower immobility rate in 2008 for elderly persons who are on
average younger, wealthier, and in better shape than in 1994
‐ Higher immobility rate in 2008 for working persons because of
more part-time work and more holydays
– A new issue: discrepancy between the two possible
estimators of immobility rate of almost 0.4 point.
› Relevance
– Weekly mobility profile is interesting to consider and
follow over time
Page 12
J.-P. Hubert, S. Roux
Q2008 Rome
9/7/2008
Thank you
[email protected]
[email protected]
INSEE
F340
18 Boulevard A. Pinard
75675 Paris cedex 14
France
Page 13
J.-P. Hubert, S. Roux
Q2008 Rome
9/7/2008
Appendix: Graphs taking into account the
2% of questionable immobile individuals
Page 14
J.-P. Hubert, S. Roux
Q2008 Rome
9/7/2008
+
1994 immobility rate
2008 immobility rate
95% confidence interval
Without filter on questionable immobility in 1994
Page 15
J.-P. Hubert, S. Roux
Q2008 Rome
9/7/2008
Students
(>15)
Not working (<65)
Economically active at work (<65)
Children (6-14)
Retir: Retirement (<65)
W_swp: in several workplace
Ch_sc: school period
Stdnt
Not_W: Not economically
active at work (<65)
Whome: at home
Ch_ho: holyday period
Unemp: Unemployed
+
W1wpM: in one workplace (males)
W1wpF: in one workplace (females)
1994 immobility rate
2008 immobility rate
95% confidence interval
Without filter on questionable immobility in 1994
Page 16
J.-P. Hubert, S. Roux
Q2008 Rome
9/7/2008