## EFFECTS ON DRIVER BEHAVIOR

The specific effects of speed limit changes on driver behavior include changes in the speed distribution, driver compliance, and close following.

Speed limit changes at the experimental sites ranged from a reduction of 20 mi/h (32 km/h) to an increase of 15 mi/h (24 km/h). However, as the posted limit was lowered by 20 mi/h (32 km/h) at only 2 sites, these locations were grouped with the 15-mi/h (24-km/h) reduction sites to obtain sample sizes appropriate for analysis. Similarly, as the speed limit was raised by 15 mi/h (24 km/h) at only 3 sites, the sites were grouped with the 10-mi/h (16-km/h) sites for analysis.

### EFFECTS ON SPEED

Free-flow vehicle speeds were collected simultaneously for 24-h periods at the experimental and comparison sites before and after the speed limit changes. In addition to total volume and free-flow volumes, the following spot speed statistics were summarized for analysis:
• Mean speed
• Standard deviation of speeds.
• Percentile speeds, ranging from the 1st to the 99th percentile.
• Percentage of vehicles exceeding the speed limit by 0, 5, 10, 15, and 20 mi/h (0, 8, 16, 24, and 32 km/h).
• Lower and upper limits of the 10-mi/h (16-km/h) pace.
• Skewness index.
The skewness index, a measure of the departure of the speed distribution from symmetry, is defined as:
```                                   2(P32 - P50)
SI= ------------
P93 - P7

where:
SI = Skewness Index

P7 - 7th percentile speed

P50 - 50th percentile speed

P93 = 93rd percentile speed

```
Previous studies of speed limit alterations primarily focused on changes in the mean and 85th percentile speeds. Shown in figure 8 are the before and after 85th percentile speeds for sites where speed limits were lowered. The before and after 85th percentile speeds for sites where speed limits were raised are shown in figure 9.

If lowering the speed limit reduces the 85th percentile speed, then the symbols shown in figure 8 would fall below the diagonal line. Conversely, if raising speed limits increases the 85th percentile speed, then the symbols shown in figure 9 should fall above the diagonal line. However, as shown in figures 8 and 9, the symbols appear to be uniformly distributed around the diagonals, irrespective of how much the speed limit was lowered or raised. Clearly, lowering or raising the posted speed limits at the experimental sites had little effect on the 85th percentile speeds. Thus, lowering the speed limit by 5, 10, and 15 or more mi/h (8, 16, 24 or more km/h) did not result in lower 85th percentile speeds. Raising the speed limit by 5, 10, and 15 mi/h (8, 16, and 24 km/h) did not result in higher 85th percentile speeds.

The average change in 85th percentile speed, as well as the largest change at any given experimental site, is shown in figure 10 for each speed limit group. For example, for the group of sites where speed limits were lowered by 15 and 20 mi/h (24 and 32 km/h), the average change in 85th percentile speed was a 0.1-mi/h (0.16-km/h) decrease. For this group, the largest decrease at any given site was 1 mi/h (1.6 km/h), and the largest increase was 2 mi/h (3.2 km/h). The data in figure 10 clearly illustrate that the 85th percentile speeds were not reduced even for large reductions in the posted speed limit. Conversely, the 85th percentile speeds did not increase at the sites where the speed limits were raised. Similarly, small changes in the 85th percentile speeds occurred at the comparison sites where speed limits were not altered.

The 85th percentile speed represents only one measure of the speed distribution. Shown in table 3 are the before and after average mean and average 85th percentile speeds for the experimental and comparison sites.

```
Table 3.  Before and after mean and 85th percentile speeds.

-----------------------------------------------------------------------
Experimental Sites

Speed            Average        Std.           Average        Std.
Limit           Mean Speed      Dev.          85th Speed      Dev.
Change     Before  After  Diff.  Diff.   Before  After  Diff.  Diff.
-----------------------------------------------------------------------
-15 & -20     42.1   42.2    0.1    0.9     49.1   49.0   -0.1    1.3
-10     42.7   42.7    0.1    1.0     50.0   49.9   -0.1    1.0
-5     43.7   43.7   -0.0    0.6     50.7   50.4   -0.3    1.0
+5     41.9   42.2    0.2    1.2     48.5   48.4   -0.2    1.4
+10 & +15     36.7   37.5    0.8    1.5     43.3   43.8    0.5    1.5
=======================================================================
Comparison Sites

Speed            Average        Std.           Average        Std.
Limit           Mean Speed      Dev.          85th Speed      Dev.
Group     Before  After  Diff.  Diff.   Before  After  Diff.  Diff.
-----------------------------------------------------------------------

-15 & -20     47.7   47.7    0.1    1.1     55.6   55.4   -0.1    1.1
-10     47.8   48.1    0.4    1.0     55.3   55.5    0.2    1.0
-5     46.2   46.4    0.2    1.0     53.1   52.9   -0.2    1.0
+5     40.5   40.4   -0.1    1.1     47.0   46.8   -0.2    1.1
+10 & +15     32.9   32.9   -0.0    1.1     39.5   38.8   -0.7    1.1
-----------------------------------------------------------------------
1 mi/h = 1.61 km/h

```
As can be seen in table 3, irrespective of how much the limit was raised or lowered, the differences in the group mean and 85th percentile speeds are quite small (less than 1 mi/h (1.6 km/h)). As indicated by the standard deviation of the differences, the largest change at any site was approximately 2 mi/h (3.2 km/h). It should be noted that the differences in speeds at the comparison sites where speed limits were not changed, are also small.

Unlike previous studies which only reported changes in the mean and 85th percentile speeds, the data collected for this study permit examination of the entire speed distribution. For example, shown in figure 11 is the cumulative frequency distribution of free-flow speeds for a site where the speed limit was lowered from 35 to 25 mi/h (56 to 40 km/h). As indicated in the figure. before and after speed differences along the entire distribution are quite small, which is typical of the findings at most of the experimental and comparison sites.

Because 24-h free-flow speed data were collected before and after speed limits were changed, the samples are large enough to permit examination of changes in the entire distribution of speeds. For example, the group with the smallest sample is the 9 sites where speed limits were lowered by 15 and 20 mi/h (24 and 32 km/h). For this group, the before period 50th percentile contains the speeds of over 8,900 vehicles. The 1st and 99th percentile each contain approximately 180 vehicles. The speed group with the largest number of sites (lowered by 10 mi/h (16 km/h)) contains over 61,000 vehicle speeds of the 50th percentile, and over 1,200 vehicles each in the 1st and 99th percentiles tiles.

The changes in the before and after percentile speeds for sites where speed limits were lowered and raised are shown in figures 12 and 13, respectively. Changes in the before and after percentile speeds for the comparison sites where speed limits were not changed are shown in figures 14 and 15, respectively. Finally, the net effects of the changes, obtained by subtracting the differences at the comparison sites from the differences at the experimental sites, is shown in figures 16 and 17. The data showing the changes in percentile speeds, as well as changes in the other speed statistics for the experimental and comparison sites, is given in appendix B.

As can be seen in figures 12 and 13, the average before and after change in any of the percentile speeds at the experimental sites was less than 1.5 mi/h (2.4 km/h) regardless of whether the speed limit was raised or lowered. At sites where the speed limits were lowered, percentile speeds below the 50th percentile speed tended to increase, and percentile speeds above the 50th percentile speed tended to decrease. However, as shown in figure 14, a similar trend occurred at the comparison sites where the speed limits were not changed. The net effects, shown in figure 16, indicate that when speeds were reduced by 10 mi/h (16 km/h), the slowest drivers (1st percentile) increased their speed approximately 1 mi/h (1.6 km/h). There were no changes in the highest speed drivers (99th percentile); however, when speed limits were lowered by 15 and 20 mi/h (24 and 32 km/h), there was approximately a 1-mi/h (1.6-km/h) increase in the 95th percentile speed.

At sites where speed limits were raised, generally there was a small increase in speeds below the 75th percentile (less than 1.5 mi/h (2.4 km/h)). The net effects, shown in figure 17, indicate that there was a small decrease in the 99th percentile speed when speed limits were raised by 10 and 15 mi/h V (16 and 24 km/h).

Other measures of the distribution of vehicle speeds include the standard deviation of speeds, the percent in the pace, the skewness index, and the coefficient of variation. These data are given in appendix B for each speed limit group.

For the sites examined in this study, there was a general reduction in the standard deviation of speeds ranging from 0.1 to 0.3 mi/h (0.16 to 0.5 km/h), irrespective of whether speed limits were raised or lowered. In addition, the percent in the pace increased by approximately 2 percent for all speed limit groups. There was very little change in the skewness index.

In order to compare speed variations between groups of sites with different speed zones, the coefficient of variation (the standard deviation of speeds divided by the mean speed) was calculated for each speed limit group. Generally, the coefficient of variation decreased by 1 percent for all speed groups, which implies that the distribution of speeds decreased slightly after the speed limits were altered.

It should be noted that statistical tests (i.e., t-test for means, f-test for variance, or Kolomogorov-Smirnow for shifts in distribution), applied to any of the data sets produces statistically significant results. The reason significant results are achieved is based on the fact that data at each site were collected for a 24-h period, which typically produces a large sample. When the data are combined for groups of speed limit changes, the samples are very large. With the large samples used in this data set, the statistical tests always indicate that the results are highly significant. Consequently, changes such as 1 mi/h (1.6 km/h) as mentioned above are statistically significant, but not practically meaningful.

### EFFECTS ON DRIVER COMPLIANCE

Driver compliance is defined as the percentage of vehicles traveling at or below the posted speed limit. While neither traffic engineering nor enforcement officials have defined specific thresholds for driver compliance it is generally felt that good compliance is achieved if 85 percent of the motorists drive at or below the speed limit. Average driver compliance at the experimental sites before the speed limits were lowered is shown in figure 18 Generally, the limits posted 45 mi/h (72 km/h) and greater provide acceptable driver compliance. However, at zones posted 40 mi/h (64 km/h) or lower, less than half the drivers were complying with the existing speed limits. As shown in figure 19, average driver compliance at the experimental sites before the speed limits were raised was very poor.

The change in compliance before and after speed limit alterations ere made at the experimental sites is shown in figure 20. As expected, at sites where the speed limits were lowered by 15 and 20 mi/h (24 and 32 km/h). The average compliance decreased by two-thirds. At sites where speed limits were lowered by 10 mi/h (16 km/h), there was approximately a 50 percent reduction in compliance. Conversely, at sites where speed limits were raised by 10 and 15 mi/h (16 and 24 km/h), there was a 50 percent increase in driver compliance. Even at this level, however, only about two-thirds of the motorists drove at or below the posted speed limits.

Noncompliance with the speed limit (the percentage of drivers who exceed the speed limit by various amounts) has also been used to describe the effects of speed limit changes. For example, shown in figure 21 is the percentage of drivers exceeding the speed limit at a site where the limit was lowered from 55 mi/h to 40 mi/h (89 to 64 km/h). The speed limit change at this site resulted in a 90 percent increase in drivers exceeding the posted speed limit.

The change in the percentage of drivers exceeding the speed limit at sites where speed limits were lowered and raised by various amounts is shown In figures 22 and 23, respectively. As expected, the percentage of drivers exceeding the speed limit was proportional to the amount the speed limit was altered. For the group of sites where speed limits were lowered in this study. there was a dramatic increase in the percentage of drivers exceeding the speed limit. Conversely, at sites where speed limits were raised, there was a dramatic reduction in the percentage of drivers exceeding the speed limit. The magnitude of these changes is illustrated in table 4. For example, at sites where the speed limit was lowered by 10 mi/h (16 km/h), the percentage of drivers exceeding the speed limit increased from 18.2 to 64.3 percent. Similar data showing the before and after percentages for drivers exceeding the speed limits by ~~~~ amounts is shown in appendix C.

```    Table 4.  Before and after changes in drivers exceeding
the speed limit at the experimental sites.

---------------------------------------------------
Speed Limit         Percent Exceeding Speed Limit
Change, mi/h           Before        After
---------------------------------------------------
-15 & -20           6.6          74.1
-10          18.2          64.3
-5          32.2          60.0
+5          74.6          44.9
+10 & +15          84.9          33.8
--------------------------------------------------
1 mi/h = 1.61 km/h
```
The change in driver compliance between the before and after periods for the comparison sites, where speed limits were not altered, was less than 1.5 percent. The net change (obtained by subtracting changes at the comparison sites from changes at the experimental sites) on drivers exceeding the posted speed limits is shown in figures 24 and 25. Because the changes at the comparison sites were small, the net changes at the experimental sites are quite similar to the raw before and after values.

Overall, altering the speed limits at the experimental sites had a dramatic effect on driver compliance and drivers exceeding the speed limit. After the speed limits were lowered, driver compliance was typically less than 40 percent At sites where speed limits were raised, compliance was generally less than 60 percent.

### EFFECTS ON CLOSE FOLLOWING BEHAVIOR

In conjunction with speed data collection, headway data were collected for a 24-h period at the experimental and comparison sites. Shown in figure 26 are the percentage of vehicles following at various headway intervals before and after the speed limit was lowered from 55 to 40 mi/h (89 to 64 km/h) at one of the experimental sites. Shown in figure 27 are headway data for a site where the speed limit was raised from 50 to 55 mi/h (81 to 89 km/h). The changes in headways at these sites are typical of the changes found at the other study sites.

Of particular interest in this study was the effect of speed limit changes on short vehicle headways, defined as a vehicle with a headway of 2 s or less. In general, as a result of raising or lowering speed limits, only small differences occurred in short vehicle headways. Shown in table 5 are the average before and after percentage of vehicles following at a headway of less than 2 s for each of the speed limit groups. Graphically, the change in headways of less than 2 s by speed limit change is shown in figures 28 and 29, respectively, for the experimental and comparison sites. The average change in headways of less than 2 s is less than 1 percent and appears to be related more to the increase in traffic volume than to the change in speed limit.

The data collected during this study indicates that raising and lowering speed limits has a small effect on vehicles with less than 2-s headways over a 24-h period. As can be expected, there was a greater percentage of vehicles with a headway less than 2 s during peak volume periods; however, the increase does not appear to be related to whether the speed limit was lowered or raised.

### REPEATED MEASUREMENTS

The results of speed observations have shown that variations in spot speeds at a point occur over time even in the absence of alterations in speed limits.[11-13] To provide a basis for examining the short- and long-term effects of speed limit alterations, repeated measurements were made at 11 experimental and comparison sites. Of these sites, speed limits were lowered at five and raised at six locations. Mean and 85th percentile speeds for two of the experimental sites are shown in table 6. A complete summary of the repeated measurements for all of the sites is shown in appendix D.

As shown in table 6, immediately after the speed limit was lowered from 55 to 45 mi/h (89 to 72 km/h) at one site, the mean and 85th percentile speeds decreased by 2 mi/h (3.2 km/h). One year after the speed limit change was made, the mean and 85th percentile speeds were similar to the immediate after values. At the time of the last measurement in May 1989, which was nearly 3 yr after the before measurement, the mean and 85th percentile speeds were only 1 mi/h (1.6 km/h) lower than the before period.

Also, as shown in table 6, the mean and 85th percentile speeds increased by 1 mi/h (1.6 km/h) immediately after the speed limit at another site was raised from 50 to 55 mi/h (81 to 89 km/h). In the following year, the mean and 85th percentile speeds increased by another mile per hour and remained stable when the last reading was taken in June 1988.

```
Table 6.  Repeated speed measurements at two experimental sites.

-------------------------------------------------------------------------------
Speed limit lowered from 55 to 45 mi/h
Limit
Before     Posted      -------------After--------------

Date                   6-25-86     9-07-86     10-1-86    7-15-87    5-03-89

Mean Speed, mi/h         53.0                    50.9       51.8       51.7

85th Percentile, mi/h    60                      58         58         59
===============================================================================
Speed limit raised from 50 to 55 mi/h
Limit
Before     Posted      -------------After--------------

Date                   8-3-86     9-23-86     1-11-87    2-25-88    6-26-88

Mean Speed, mi/h        54.8                    55.8       56.8       56.4

85th Percentile, mi/h   61                      62         63         63
-------------------------------------------------------------------------------
1 mi/h = 1.61 km/h

```
Similar results were obtained at the other sites where repeat measurements were taken. Shown in figure 30 are the average changes in the after 85th percentile speeds relative to the before 85th percentile speeds. The data for figure 30 were taken from appendix D. Although the number of sites is small, these data samples indicate that only small changes occurred in the 24-h 85th percentile speeds of free-flow vehicles. Furthermore, it appears that in a 2- to 3-yr after period, no more than a 1-mi/h (1.6-km/h) increase in the speed distribution occurs. It is plausible that this increase is due to natural variation in vehicle speeds and/or can be attributed to the accuracy of the speed data collection equipment.

### DISCUSSION OF SPEED EFFECT

Raising and lowering posted speed limits on the nonlimited access sites examined in this study produced minor changes in the speed distribution, but major changes in driver compliance. These results substantiate the findings of previous research and observational studies.

A number of studies have been conducted to examine the effects of changing posted speed limits on driver behavior. [14-25] A summary of the data obtained in published reports is shown in table 7 for urban and rural nonlimited access highways. Table 8 provides the results of recent operational studies.

While the number of sites sampled was small, the locations were confined to one regional area, and the majority of studies consisted of simple before and after measurements; nevertheless, the data indicate several trends. First, the magnitude of any change in speed due to a speed limit alteration was approximately 1 mi/h (1.6 km/h). Even in cases with large speed limit changes, the impact on the 85th percentile speed was small.

Studies conducted by Avery and Elmberg revealed that there were little changes, if any, in the mean, standard deviation, and percentage of vehicles in the pace when speed limits were established on the basis of the 85th percentile speed. [14,15] In this research, the few sites where speed limits were posted in the region of the 85th percentile speed also revealed little change in the speed distribution.

While previous research indicates little difference in most speed characteristics following a speed limit change, the effects on driver compliance are significant. For example, raising the speed limit by 5 mi/h (8 km/h) can increase compliance from 30 to 70 percent when the new limit is based on the 85th percentile speed. Conversely, lowering the speed limit by 5 mi/h (8 km/h) can decrease compliance from 70 to less than 50 percent.

Although the results of this study are similar to the findings reported in previous research, there are several major differences. First, this study used 24-h free-flow speed data to estimate speed effects at 100 sites located in 22 States, which provides the largest data base and regional distribution of sites. In addition, comparison sites were used to account for extraneous factors, such as weather conditions, changes in enforcement, etc., that could influence speeds in either the before or after periods.

The data from this study suggests that, on average, speed limits in the 22 States examined are set below the average speed of traffic. It appears that the wide range of procedures used to establish a safe and reasonable speed limit are the primary reason for this. Setting speed limits below the average speed of traffic has created widespread noncompliance with speed limits. For example, based on the total current total vehicle volume at the 100 experimental sites used in this study, it is estimated that at least 240,000 motorists exceed the existing posted speed limits every day. On a yearly basis, nearly 88 million motorists exceed the speed limits on just the 100 sites or 172 mi (277 km) of nonlimited access highway examined in this study. Clearly, an understanding of the effects of lowering and raising speed limits is needed. In addition, there is a need to examine the principles and practices used to set speed limits in the United States.

Table 7. Summary of reasarch studies on the effect of raising and lowering speed limits.
 #S = Number of Sections B = Before A = After AL = After Posting Lower Limit AH = After Posting Higher Limit
Author, Date Speed Limit Average 85th Percentile Speed
Location #SBA BALAH Remarks
Urban
St. Paul, Minnesota Avery, 1960 7 30 35 35.5 35.8 Compliance increased from 36 to 80 percent.
4 30 40 39.4 40.8 Compliance increased from 19 to 82 percent.

West Lafayette, Indiana Elmberg, 1960 1 35 30 38.4 38.5 Percentage of vehicles exceeding the speed limit by 5 mi/h or more increased from 10 to 38 percent.

St. Joseph Ogawa, 1962 1 30 35 30.6* 31.3* Speed limit revisions based on 85th percentile speeds. The mean speed increase in Orden is statistically significant.
Orden 1 40 35 32.2* 33.3*
Fithian, Illinois 1 35 40 34.9* 35.3*

Columbia, South Carolina Roberts, 1967 1 35 40 42.5 41.0 Compliance increased from 28 to 75 percent

10 California Cities Spitz, 1984 40 NA NA 39.9 40.3 The 85th percentile speed in the 27 comparison samples where no change in the limit was made increased from 40.1 to 40.8 mi/h.
10 NA NA 36.9 38.6

3 Urban Fringe Areas in Texas Dudek, 1986 6 55 45 55.9 54.0 Lowering the speed limit had little effect on vehicle speeds and speed variance

Rural
Illinois Kessler, 1959 30 30 40 38.9 38.4 Driver compliance increased from 39 to 85 percent.

Houston District, Texas Rowan, 1962 1 60 55 55.5 55.0 Speed limits were established in the after periods based on the 85th percentile speeds. A slight increase in the dispersion of speeds was reported after the new limits were established
1 60 50 53.7 52.0
1 60 45 47.3 46.4
1 60 40 41.5 39.9
1 60 35 40.2 38.9
1 60 30 31.2 33.2
Paris District, Texas 1 60 55 55.2 51.2
1 60 50 51.5 59.9
1 60 45 47.6 43.9
1 60 40 40.4 38.6
1 30 55 53.2 50.5
1 30 50 52.6 51.3
1 30 45 45.4 45.2
1 30 40 40.3 42.0
1 30 35 36.8 38.4

Operational studies
Massachusetts Oct. 1966 19
2-lane
30-45 20-35 47.5 48.3 After posting 10mi/h lower and higher limits than existing zones, no differences in 85th percentile speed were found
30-45 40-55 48.0
6
d-lane
50-55 40-45 60.7 60.3
60-65 60.7 61.0

Minnesota Jan. 1979 12
2-lane
45 30 NA 43.6
4
4-lane
40 30 42.0 40.3

Washington 1981-82 3 25 30 34.7 34.3
1 50 55 57.0 59.0
1 50 35 43.0 42.0
3 40 35 45.0 43.7

Michigan 1982 4 25 35 37.6 36.0 Compliance increased from 10 to 81 percent.
4 55 50 56.8 54.8 Compliance decreased from 73 to 56 percent.
4 55 50 57.8 56.0 Compliance decreased from 71 to 46 percent.
5 45 35 49.2 47.0 Compliance decreased from 56 to 7 percent.
 NA = Not Available * = Denotes mean speeds, not 85th percentile speeds
References: 14-25 respectfully

Next: Effects on Accidents | Prev: Site Characteristics | Contents