1995 Nationwide Personal Transportation Survey

Corrections and additions to the revised data sets:

DAYTRIP File Changes: The DAYTRP95 data set has been renamed DAYTRP95_2

Variable: DAYNGHT2

Changes: New variable, revised DAYNIGHT.

Stats : 98 121 records
AM 137,541 records
PM 271,363 records

Background: The DAYNIGHT variable was mis-coded on some records.

Variable: DRVR_FLG

Changes: This variable was re-coded from ‘01' meaning yes to ‘02' on some records. ‘01' indicates that the sample person drove on the trip.

Stats :

01 = 252,574 records
02 = 156,451 records

Background: DRVR_FLG was inaccurately coded ‘01' for trips other than personally operated vehicles (TRPTRANS modes ‘01' through ‘08'). For records where the TRPTRANS variable was not ‘01' through ‘08' but show the DRVR_FLG as being ‘01', that DRVR_FLG variable was changed to ‘02' .

Variable: DWELTIM2

Changes: New variable comparable to DWELTIME except that DWELTIM2 is the time spent at the destination of the current record and is calculated using the revised STRTTIM2 variable. This correctly coincides with the trip purpose of the destination, so if dwell times were estimated by purpose, the analysis would be straightforward. Negative dwell times were set to missing.

Stats:

VALUES	DWELTIME	DWELTIM2
-540 - < 0	11,246 records	0 records
0	14,386 records	14,404 records
1-1250	295,392 records	295,556 records
missing	88,001 records	99,065 records

Background: The dwell times were calculated for the minutes spent at the destination of the previous trip, before starting the current trip (the record on which the public use data set posted dwell time). This variable is missing for each person's first trip of the day, and when the start time of the trip (STRTTIME) or the minutes in travel (TRVL_MIN) were not determined.

When performing an analysis of the dwell times at destinations by purpose, the programmer needs to carry the purpose of the previous trip forward, or the results are nonsensical. It made sense to create a more accurate dwell time to the correctly coincide with the purpose of the destination, which is the most common way to analyze dwell time.

The negative dwell times occurred because of respondent's mistakes in relaying information about the start time of the previous trip, and the total minutes the trip took, in relation to the start time of the next trip. Of the 321,024 records with calculated dwell times 11,246 were negative. If these were included in an analysis, the average dwell time would be 117 minutes, if they are excluded, the average is 122 minutes.

Variable: DWELSEC2

Changes: New variable corresponds to DWELTIM2.

Background: This variable is the DWELTIM2 in seconds. It is anticipated that most people using this data set will be using the SAS software package. As SAS internally uses time variables in seconds, this was provided for ease in use to generate time calculations.

Variable: DWEL2_HM

Changes: New variable corresponds to DWELTIM2.

Background: Shows DWELTIM2 in easy to read HH:MM format.

Variable: STRTTIM2

Changes: New variable, revised STRTTIME. There were some inconsistencies in the original STRTTIME.

Stats :

VALUE	STRTTIME	STRTTIM2
0	825 records	0 records
1-2359	408,080 records	408,079 records
2400	0 records	825 records
9998	114 records	0 records
9999	6 records	0 records
missing	0 records	121 records

Background: STRTTIME was mis-coded as ‘1099'. When examining the trip records before and after it was found that the actual time should have been coded as ‘1059', ‘1200', ‘159' or ‘9998'. The STRTTIME and STRTTIM2 variables are the trip begin times in military time format.

Variable: TRPMILES

Changes: This variable gives the distance in miles of the recorded trip. Actual distance was coded from 0 - 1200 miles. Less than a mile is re-coded on the enclosed dataset:

9,338 records coded as one block or less (9996) are re-coded as .1
22,265 records coded as less than half a mile (9997) are recoded as .5 .

Stats :

< 1 mile	68,703 records
1-1200 miles	333,595 records
9998	6,615 records
9999	112 records

Background: Trips of less than a mile were supposed to be coded as either 9996 (less than one block) or 9997 (half a mile). In the Public use dataset, some trips were coded as .5 for half a mile or less, and some as 9997. The changes were made to consistently code these variables and to eliminate unnecessary code for estimating miles.

Variable: TRPNUM2

Changes: New variable compares to TRPNUM.

Background: TRPNUM2 is to be used to chronologically reorder the trips within each person's records. Resorting the file by HOUSEID, PERSONID and TRPNUM2 enables a user to more accurately examine trip chaining.

PERSON File Changes: The PERSON95 data set has been renamed PERS95_2

________________________________________________________

Variable: PTCRIME and PTNTCLN

Changes: Data labels were reversed in the public use data set.

Background: These two public transit variables elicit the respondent's perception of whether worry about crime on public transportation (PTCRIME) and whether the transit stations and vehicles not being clean (PTNTCLN) was a "Large Problem", a "Small Problem" or "No Problem".

Variable: YEARMIL2 new variable comparable to YEARMILE.

Changes:

69,990 unweighted drivers in the dataset
4,580 drivers reported driving zero miles in the previous year
3,519 zero-miles drivers reported driving on the travel day (these are recoded as ‘miles not reported' or 9998).
1,061 zero-miles drivers remaining .

Stats :

VALUE	YEARMILE	YEARMIL2
0	4,580 records	1,061 records
1-200,000	61,138 records	61,138 records
999994	25,194 records	25,194 records
999998	4,422 records	7,941 records
999999	26 records	26 records
weighted mean	12,389.05	13,316.78

Background: Numerous data users had questioned the earlier annual average miles driven because there were declines in per driver VMT between 1990 and 1995 in virtually all age/gender categories other than men 65 or older. This seemed incongruous, given the overall strong increase in travel during this time. Upon checking, we found that in 1990 only 2 percent of the drivers reported driving no miles during the year, while 9 percent of drivers reported driving no miles in 1995. Of the 9 percent, a significant number indicated that they actually did drive, either on their assigned Travel Day or as the primary driver of one of the household vehicles. Because we believe that the report of ‘no miles' is in error for these drivers, these zero-values were changed to ‘miles not reported.' After this edit, only about one and a half percent of all drivers remained in the "no miles category." The new estimates of vehicle miles of travel in each age group for 1995 are shown in the following table.

The revised data show modest increases of generally less than 10% for most age/gender groups. The big exception is the 16-19 year-old group, where miles declined between 1990 and 1995. This is probably the result of changes in the survey weighting process between 1990 and 1995, which resulted in a large increase in the number of persons age 16-19. Of course, with more individuals in this teenage group in 1995, the average miles per driver would decline. Other factors at work may also include delayed licensing laws and/or higher auto insurance premiums for young drivers.

For men, the most dramatic increases in travel were for those 65 and older. Younger men, namely those 20-54 may finally be reaching saturation in their travel. Women's travel shows a very different pattern, with declines in the youngest group (16-19), consistent increases of 7 to 8 percent for those 20 through 64, and no change in average travel for those 65 and older.