Survey Weighting

A description of the methodology used for weighting the Departing Passenger Survey

Prior to weighting the survey records, various validation exercises are carried out to identify miscoding of flight numbers, destinations and routeings. Where possible these are corrected and the record retained. See the section on Survey Validation for more information.

The survey design is stratified by carrier, route and quarter for scheduled traffic, and route and quarter only for charter traffic. There may be multiple flight numbers within these stratification cells, but it is possible that there will only be one.

As interviews are generally conducted in the gateroom it is possible to target specific flight numbers. For scheduled flights, the intention is to cover as wide a range of flight numbers and routes as possible during the quarter. For charter flights, the focus is on sampling the route rather than individual flight numbers on the route.

The interviews are weighted firstly by matching the total airport passenger numbers to the stratification cells, and then by spreading these totals across the interviews achieved in each cell. Multi-sector routes are treated slightly differently to single sector routes to allow for the multiple destinations on a given flight number.

Only departing passengers are interviewed. In order to weight the interviews achieved such that they represent all the airport passengers, the departing passenger totals on each flight number are scaled up to represent both departing and arriving passengers. This is done with a constant scaling factor across the carrier-route in each month, which is then applied to each departing flight number within that carrier-route, such that the total of the scaled up departing flights on the carrier-route agrees to the total passengers travelling on the carrier-route. The various airport authorities provide uplift/discharge figures for each month.

If no interviews have been obtained on a given scheduled flight number, the stratification cell is extended to consider other flight numbers by the carrier to the same destination in the quarter.

If no interviews are recorded in the stratification cell (or extended cell) in the quarter of the flight departure, the search is extended to other quarters. If no interviews at all have been obtained in the year on the carrier-route (or just the route in the case of charter traffic), the population has to be dropped from the survey weights.

A Theoretical Example:

Survey Airport

Apt

Apt

Month

M

M

Quarter

Q

Q

Flight Type

Scheduled

Charter

Flight number

a

b

Matching criteria in period

Carrier & Numeric Flight Number & Sector = CFS(a)

Sector = S(b)

Sample size

 = Number of interviews in stratification cell

= n

For each CFS aggregate all the positive interviews obtained in quarter Q =n(a)

For each charter destination S aggregate all interviews obtained in quarter Q (irrespective of flight number) = n(b)

Population

Determine the scaling factor for scheduled departing flights by the carrier to the destination in the month.

Scaling factor for CS(a) =

[CS(a) scheduled departing population + CS(a) scheduled arriving population] / CS(a) scheduled departing population

Use this scaling factor to gross up the specific departing flight CFS(a) in each month

Population used for weighting =

POP(a) =

[CFS(a) x SF(CS(a))] summed over the quarter

Aggregate arriving and departing passengers for all charter operators in quarter Q from airport Apt to destination S(b).

Population used for weighting =

[(S(b) charter departing population + S(b) charter arriving population)]

POP(b) =

S(b) summed over the quarter

 

Final weight for each interview

POP / n

POP / n

 

A Practical Example:

Survey Airport

Gatwick

Gatwick

Month

June

June

Quarter

2

2

Flight Type

Scheduled Flight

Charter Destination

Flight number

BA1

TOM2

Matching criteria

Carrier & Numeric Flight Number & Sector = BA & 1 & CDG

Sector = AGP

Sample =

Number of interviews=

n

For flight BA1 to CDG aggregate all the positive interviews obtained in quarter 2, n = 20.

For charter destination AGP aggregate all interviews obtained in quarter 2, n = 50.

Population

Determine the scaling factor for departing flights by the BA to CDG in each month in quarter 2.

Use the scaling factors to gross up the passengers on departing flight BA1 in each month.

 

 

Arr (BA – CDG)

Dep (BA – CDG)

SF

BA1

Apr

10,000

10,000

2

505

May

12,000

15,000

1.8

400

Jun

9,800

14,000

1.7

600

 

 

Population used for weighting =

[BA1(pax) x SF(BA-CDG)] summed over the quarter

=  [(2x505)+(1.8x400)+(1.7x600)]

= 2,750

 

Aggregate all uplift and discharge figures for all charter operators in quarter 2 from airport Gatwick to destination AGP.

 

 

 

 

Arr (CS)

Dep (CS)

CS

Apr

1,000

1,000

2,000

May

1,200

1,500

1,400

Jun

900

700

1,600

 

 

Population used for weighting =

(Charter pax – AGP) summed over the quarter

= 2,000 + 1,400 + 1,600

= 5,000

 

 

Final weight for each interview

2750/20 = 137.5

5000/50 = 100

 

There may be times when interviewing cannot be carried out in gaterooms, but has to be transferred to the lounge area. When this happens a further weighting adjustment is required in order to account of the variation of the sampling proportions between shifts, and the differing flow rates of airside and landside passengers.