A motor vehicle safety scale for young drivers
Earl H. Blair
Abstract: This study seeks to develop a brief instrument with validity and reliability to measure college students’ attitudes and behavior with regard to motor vehicle safety. The scale was administered to a sample of 158 undergraduate students. Item analysis showed high discriminative power of items and adequate internal consistency of each subscale and the total scale. Factor analysis led to retaining five components on which 17 items were loaded. The five components accounted for 70.8% of the total variance of the motor vehicle safety, and each component demonstrated adequate internal consistency. This study has launched a solid step in the validation process of the motor vehicle safety scale for young drivers.
Motor vehicle deaths claim the lives of a number of the nation’s youth. During the year 1998, nearly 7,500 young drivers between ages 18 to 24 years died in a vehicular crash (National Safety Council [NSC], 2001). Although the current motor vehicle death rate is lower than any rate recorded during the past 35 years (NSC, 2001), the cumulative years of productive life lost from accidental death is far too great a loss for the nation. A nations youth is one of its most important assets; therefore, preventing or reducing premature deaths from motor vehicle crashes is a national priority.
It seems that the majority of vehicle crashes can be avoided by minor changes in driver behavior. For example, it is estimated that the use of seat belts saved over 11,000 American lives in 2000 (National Highway Transportation Safety Administration [NHTSA], 2001). However, research has shown that failure to use seat belts is most common among young drivers and males (Jonah, 1990; Kim & Kim, in press; Preusser, Williams, & Lund, 1991; Wells, Williams, & Farmer, 2002). This is affirmed by the fact that the crash rate and resultant fatalities are greater for younger drivers. The death rate due to motor vehicle crash for 16-19-year olds was more than four times the rate for drivers ages 30-69 (McCartt, Leaf, Farmer, Ferguson, & Williams, 2001). Male drivers are involved more frequently in fatal motor vehicle accidents than females (Bever, 1996).
A leading cause of motor vehicle injuries appears to be human error (Bever, 1996). A variety of reasons are given for young adults’ overrepresentation in motor vehicle crashes (Begg & Langley, 2001). Research indicates contributing factors include inexperience, inattention, poor risk perception, impulsivity, and a propensity for thrill-seeking, and sensation-seeking (Jonah, 1986). Other contributors involve risky driving behaviors such as driving at excessive speed, dangerous overtaking, close following, driving after drinking, or driving after using drugs (Williams, 1998).
In some cases, young adults tend to believe they are ‘invincible’ and have the attitude “it won’t happen to me.” Studies have found that younger drivers tend to estimate their own risk of being involved in a crash as lower than their peers, or older drivers (Matthews & Moran, 1986). Studies also indicate that most people think their own driving skills and behaviors are “above average.” This was confirmed in an informal show of hands in the classroom when approximately 90% of the students indicated they believed they were “above average drivers.”
The main key to the reduction of motor vehicle accidents is education to enhance safety awareness and to induce actions conducive to motor vehicle safety. The college setting provides a good opportunity for motor vehicle safety education. Considering that vehicular crash is the leading cause of death for college students, emphasis must be placed on the importance of sale driving education. However, the lack of an instrument designed for measuring young drivers’ attitudes and practice related to motor vehicle safety has been an obstacle in effectively designing, implementing, and evaluating driver safety education programs. Therefore, this study was conducted to construct a valid and reliable instrument to measure college students’ attitudes and practice with regard to motor vehicle safety.
In developing an instrument for college students, we established two principles. One was to develop a minimal number of question items based on a parsimonious rule as long as it reveals evidence of reliability and validity of the scale. Classical testing model indicates that the more items there are in a scale designed to measure a particular construct, the more reliable the measurement is. Therefore, in terms of internal consistency reliability, it is better to have a significant number of items in a scale. However, in an attempt to increase the utility of the instrument with reasonable evidence of validity and reliability, a scale was developed with a minimal number of items.
The other principle we established was to include items that represent various important constructs that presumably affect motor vehicle safety considerably rather than adhering to the dichotomized concepts of attitude or behavior in the scale development process. Human nature is complex in that attitudes cannot predict all related behaviors. Also, many times, measurements of behaviors do not address an undistorted whole picture. As a result, attitudinal and behavioral items are included in the scale. However, the consideration of attitudinal or behavioral nature of items was not given until various important constructs of motor vehicle safety in young adults were condensed into six constructs that represent the scale.
The instrument for this study is a summated rating scale of the Likert-type with a five-point format. Based on the Theory of Reasoned Action (Montano, Kasprzyk, & Taplin, 1997) and related literature, a large pool of statements appropriate to a table of specifications was developed for the motor vehicle safety instrument. The table served as a valuable means for content validity and direction for item construction and selection. It included six major constructs of motor vehicle safety: 1) speed limit violation, 2) seat belt usage, 3) driving under the influence of drugs or alcohol, 4) perceptions regarding risky driving behavior, 5) attitudes regarding compliance with rules, and 6) defensive driving. From the pool, 18 items (see Appendix A) were selected after a jury of experts reviewed the statements. The initial scale was pilot tested in a small class of students and specific statements were refined for better understanding.
In terms of attitudinal or behavioral nature of question items, it included 12 attitudinal items and 6 behavioral items. Items 1 through 7 had the alternatives of “strongly disagree” to “strongly agree.” In items 8 through 12, the alternatives were reversed. Items 13 through 18 had the alternatives such as “never,” “seldom,” “occasionally,” “most times,” and “always.” To establish score values for responses, the investigators identified each statement as being either favorable or unfavorable toward motor vehicle safety. Each response option for a statement was then assigned a value ranging from one to five points, depending on whether it was a favorable or unfavorable statement toward motor vehicle safety–five points being assigned to the most favorable response to vehicle safety construct.
The scale was group administered to a sample of 158 undergraduate students at a Midwestern university enrolled in elective courses. Although it was not a random sample, the subjects were reasonably representative of general undergraduate students. Ages of the subjects ranged from 18 to 42 years, with a great majority being between 18 and 22. In terms of academic standing, about 20% of the students were freshmen, 15% sophomores, 24% juniors, and 38% seniors (missing 3%). Approximately 65% were females and 35% were males.
The obtained scores for each item and the total scale were subjected to statistical analysis to investigate the following research hypotheses relating to the validity and reliability of the scale:
1) Each item is positively correlated with the summated attitudinal subscale score, the summated behavioral subscale score, and the total scale score.
2) The summated attitudinal subscale score is positively correlated with the summated behavioral subscale score.
3) Each item shows a discriminative power (DP) between the top 27% of respondents’ scores and the bottom 27% (The mean score made for each item by the top 27% of respondents is significantly higher than the mean score for each item made by the bottom 27% of the respondents in terms of total scale score).
4) Factor analysis of the scale reveals all the six underlying constructs for motor vehicle safety which were identified before the data analysis.
5) Each component identified by the factor analysis is positively correlated with the total scale score.
6) Each attitude-dominant component is positively correlated with the summated behavioral subscale score and each behavior-dominant component is positively correlated with the summated attitudinal subscale score.
To examine the above hypotheses, the following statistics were computed: 1) coefficient of correlation between variables of interest; 2) DP coefficient, the difference between the mean item scores made by the top 27% and the bottom 27% of respondents in terms of total scale score; 3) t-ratio of the DP coefficient; 4) frequency of response to each alternative for each item; and 5) the Cronbach alpha.
To provide evidence of construct validity and to reveal underlying constructs, the investigators submitted the data for factor analysis using the Statistical Package for the Social Sciences, windows version 11.0 (SPSS Inc., September 2001). In terms of factor extraction, Principal Component Analysis was adopted rather than Principal Factor Analysis since the investigators were more concerned about the set of factors that can account for all the variability in an item rather than the least number of factors that only account for the common variability in an item with other items. As for rotation strategy, Varimax method was used to maximize variance of the squared loadings of a factor on all the items in a factor matrix.
Approximately 99% of the total subjects responded completely to all the items of the scale. The minimum and maximum possible scores that can be obtained per subject by the scale are 18 and 90 respectively. The actual scores ranged from 23 to 85. The mean score was 69.17 and the standard deviation was 10.43.
As Table 1 shows, the Pearson product-moment correlation coefficient between each of the 18 items and the total scale score was round to be a positive value and significant at the .001 level.
Also, each item was positively correlated with the summated attitudinal subscale score (items 1-15, p < .001; items 17-18, p < .01; item 16, p < .05) and with the summated behavioral subscale score (items 1-2, 6-7, 11-18, p < .001; item 10, p < .01; items 3-5, 8-9, p < .05). The first hypothesis was held tenable. Each item in the scale was found to measure the subjects' attitudes and behavior toward motor vehicle safety in the same direction, as did the total of all items, the total of attitudinal items, and the total of behavioral items in the instrument. This indicates high internal consistency for the entire scale. The reliability coefficient of the whole 18-item scale was .86 by Cronbach alpha.
Also, the Pearson product-moment correlation coefficient between the summated attitudinal subscale score (items 1-12) and the summated behavioral subscale score (items 13-18) was found to be .40, which was significant at the .001 level. Thus, the second hypothesis was retained. This implies that the attitudinal subscale is significantly related to the behavioral subscale in the same direction. The Cronbach alpha of the attitudinal subscale and the behavioral subscale was .86 and .65, respectively.
Table 1 also shows the evidence of the discriminative power of each item. Some researchers suggest comparing the mean item scores made by the top 25% and the bottom 25% of respondents in examining the discriminative power of items (Garson, n.d.). However, a conservative criterion of 27% (Torabi, 1991) was used. The t-values for all the 18 items in the scale were significant at the .001 level, retaining the hypothesis that the mean score made for each item by the top 27% of respondents is significantly higher than the mean score for each item made by the bottom 27% of the respondents in terms of total scale score. This implies that the scale has a discriminative power between respondents who show positive attitudes and behavior toward motor vehicle safety and those who do not.
In terms of magnitude of the discriminative power, items 9-11 showed largest DP coefficients (1.84-1.89) while items 15-17 showed smallest DP coefficients (0.64-0.81). This indicates three possibilities. First, the three behavioral items (15-17) might not be worded effectively enough to distinguish those who have positive attitudes and behavior toward motor vehicle safety and those who do not. Second, in considering the three items are located at the end of the questionnaire (item 18, the last item, also showed a low DP coefficient of 0.95), respondents’ fatigue or rush to finish the survey might have influenced the discriminative power. In other words, sequence effects might have diluted the discriminative power of the items. Lastly, the respondents might have similar behaviors measured at items 15-17, relative to their attitudes.
A close examination of the items seems to support the third possibility. The respondents appear to have similar behaviors for those measured at items 15-17. While most items had standard deviations (SDs) of 1.00 or higher, the three items had SD of less than 1.00 (item 15: M = 2.20, SD = 0.84; item 16: M = 3.78, SD = 0.99; item 17: M = 4.20, SD = 0.84). A notable finding was no respondent answered “never” to the item 15 (“How often do you exceed the posted speed limit?”). Forty-five percent of the respondents reported that they exceed the posted speed limit most times and 21% always. Even among the respondents who scored the top 27%, which means they had positive attitudes and behavior overall toward motor vehicle safety, 47% reported they exceed the posted speed limit most times or always.
Another notable finding was 20% of the respondents reported they drink and drive occasionally and 18% reported they drive after using drugs either illicit or prescribed occasionally. According to the Harvard School of Public Health College Alcohol Study conducted nationwide in 1999 (Wechsler, Lee, Kuo, & Lee, 2000), 44% of colleges students were binge drinkers (defined as the consumption of five or more drinks in a row for men and four or more for women) and 23% were frequent binge drinkers in 1999. Another national study reports that the prevalence of past 30-day marijuana use rose from 12.9% to 15.7% between 1993 and 1999 among college students (Gledhill-Hoyt, Lee, Strote, & Wechsler, 2000). Strote, Lee, and Wechsler (2002) also reports the prevalence of ecstasy use is rapidly increasing, an increase of 69% between 1997 and 1999.
It is interesting that a difference was found in the driving behavior held by male and female subjects. When the mean scores for these two groups were subjected to the two-tailed t-test, the self-reported driving behavior of the female group was round to be significantly more favorable toward safe driving than was the behavior of the male subjects at the .01 level. Even though the female group also appeared more favorable toward safe driving than the male group in terms of both the total scale score and the attitudinal subscale score, the mean difference was not significant at the .05 level.
The factor analysis using the extraction method of Principal Component Analysis (PCA) produced six factors with eigenvalues greater than 1 (Kaiser criterion), which accounted for 73.6% of the total variance.
The rotation of factor loadings using Varimax rotation technique produced a much clearer pattern of factor loadings. The total 18 items were reduced to six specific factors in a more interpretable manner. Each of the 18 items was uniquely loaded on only one of the six factors except item 5 and item 7 that were loaded on two factors (using a factor loading of .40 as a cut-off point). The rotated factor matrix is presented in Table 2, where items that constitute each factor are underlined.
In case of PCA, each retained component must have at least two substantial loadings (Zwick & Velicer, 1986). As shown in Table 2, even though all the six factors contained at least two substantial loadings, factor 6 contained only one unique item (item 16). To determine the number of components to be retained, the investigators further examined the magnitudes of six eigenvalues after rotation, Scree plot, and Cronbach alpha for each factor. The eigenvalues after rotation were 2.94, 2.93, 2.34, 2.04, 1.72, and 1.30 from factor 1 to factor 6. The largest percentage drop of eigenvalues occurred between factor 5 and factor 6, which indicated that factor 6 should not be retained. Scree plot, one of the most accurate methods for determining the number of components to retain (Zwick & Velicer, 1982, 1986), confirmed retaining five factors that accounted for 66.4% of the total variance.
Lastly, Cronbach alpha was calculated for each factor. As shown in Table 3, factor 1 through factor 4 showed high levels of internal consistency (.82, .84, .80, and .84) and factor 5 showed a marginally acceptable level of internal consistency (.61). However, factor 6 showed a low internal consistency (.45). Thus, five components from factor 1 to factor 5 were retained. Revised factor analysis with items loaded on the retained five components produced the same factor structure that accounted for 70.8% of the total variance.
Each component was examined and labeled as follows: Five items (items 3-4, 10-11, 15) were loaded only on component 1 and they apparently measured perceptions and behavior related to speed limit. Thus, component 1 was labeled as “speed limit violation.” Component 2 was accounted for by five other items (items 1-2, 5-7). These items measured perceptions regarding risky driving behavior; therefore, component 2 was named “perceptions regarding risky driving behavior.” Component 3 was explained by three other items (items 8-9, 12). These items dealt with attitudes regarding compliance with rules. As a result, component 3 was termed “attitudes regarding compliance with rules.” Component 4 was identified with two other items (item #13-14) that clearly measured seat belt usage and, consequently, was named “seat belt usage.” Three items (item #7, 17-18) accounted for component 5 and obviously measured attitude and behaviors related to driving under the influence of drugs or alcohol. Thus, component 5 was named “driving under the influence of drugs or alcohol.”
The fourth hypothesis that factor analysis of the scale reveals all the six underlying constructs for motor vehicle safety was not tenable. The factor analysis led to retaining five components among the six constructs that were identified before the data analysis. Even though factor 6 contained both items that were designed to measure “defensive driving,” it failed to show internal consistency.
As Table 3 shows, the Pearson product-moment correlation coefficient between each of the retained components and the total scale score was found to be a positive value and significant at the .001 level ([r.sub.1T] = .74, [r.sub.2T] = .81, [r.sub.3T] = .67, [r.sub.4T] = .53, [r.sub.5T] = .64). Also, each attitude-dominant component (speed limit violation, perceptions regarding risky driving behavior, and attitudes regarding compliance with rules) is positively correlated with the summated behavioral subscale score ([r.sub.1B] = .32, [r.sub.2B] = .40, [r.sub.3B] = .26). Each behavior-dominant component (seat belt usage and driving under the influence of drugs or alcohol) is positively correlated with the summated attitudinal subscale score ([r.sub.4A] = .31, [r.sub.5A] = .53). The fifth and sixth hypotheses were held tenable. Each of the five components in the scale was found to measure the subjects’ attitudes or behavior toward motor vehicle safety in the same direction, as did the total of all items, the total of attitudinal items, and the total of behavioral items in the instrument.
To examine the relationship among the five retained components, bivariate correlation coefficients were calculated.
As shown in Table 4, the largest correlation was found between component 2 (perceptions regarding risky driving behavior) and component 5 (driving under the influence of drugs or alcohol) (r = .59, p < .001). This implies that college students' behavior of driving under the influence of drugs or alcohol is closely related to their perceptions regarding risky driving behavior measured by five attitudinal items, 1 and 2 and 5 through 7. The next largest correlation was found between component 1 (speed limit violation) and component 2 (perceptions regarding risky driving behavior) (r = .48, p < .001). This implies that college students' attitude and behavior regarding speed limit violations are again closely related to their perceptions regarding risky driving behavior. Even though the corretation between component 1 (speed limit violation) and component 4 (seat belt usage) was significant at the .05 level (r =. 19), the magnitude of the relationship between component 1 and component 4 was the weakest among the five components.
DISCUSSION AND CONCLUSIONS
This study was designed to develop a brief instrument to measure college students’ attitudes and behavior with regard to motor vehicle safety. Extensive analyses of data yielded favorable results of validity and reliability of the proposed scale. Evidence of content validity and construct validity were provided through the use of the table of specifications, review by a jury of experts, and factor analysis. Especially the six factors identified through the factor analysis revealed a clear structure of underlying constructs regarding motor vehicle safety, which virtually corresponded to the table of specifications even though one factor was not retained. The internal consistency of the items of the total scale, attitudinal subscale, and behavioral subscale was provided through correlation between each item score and the total scale score and each subscale score. The Cronbach alpha of the total scale, attitudinal subscale, and the behavioral subscale was .86, .86, and .65, respectively. Also, the correlation between the attitudinal subscale and behavioral subscale was found to be highly significant, implying the attitudinal subscale is significantly related to the self-reported driving behavior of subjects. The scale was also found to have a strong discriminative power. Each of the 18 items showed a significant discriminative power between respondents who show positive attitudes and behavior toward motor vehicle safety and those who do not.
All the hypotheses were held tenable except one hypothesis that factor analysis of the scale reveals all the six underlying constructs for motor vehicle safety, which were identified before the data analysis. The construct “defensive driving” did not reveal itself as a reliable component of the scale. The causes might be 1) under-development of the items that measure defensive driving, 2) lack of importance or relevance of the “defensive driving” dimension in young drivers’ motor vehicle safety, or 3) inappropriate wording of question items.
All three causes may have played a role in the failure of retaining the defensive driving component. Since item 5 that was originally designed to measure defensive driving related perception turned out to be more associated with component 2 (perceptions regarding risky driving behavior), at least one or two items that measure defensive driving construct needed to be included in the scale. Also, in item 16, “How often do you practice defensive driving techniques?” the respondents might not have fully understood what defensive driving techniques were even though three examples of defensive driving techniques were given in item 5. Therefore, it is suggested that a future study to validate this scale include more items measuring defensive driving construct and provide more examples of defensive driving techniques in each relevant item.
It is obvious that this scale does not include all possible dimensions related to motor vehicle safety. Invincibility, over-assessment of individual driving skills, and general propensity to risk-taking are some examples of dimensions that are hOt directly measured by the proposed scale. The instrument was condensed into six dimensions that would account for considerably large proportion of variance in motor vehicle safety in young adult population. The finding that the retained rive components of the scale accounted for as much as 70.8% of the total variance of the motor vehicle safety provides evidence that the purpose of this study was accomplished.
Another merit of this scale is its potential for predictive purpose. It measures hOt only beliefs and attitudes but also several critical behaviors with regard to motor vehicle safety. Therefore, it is more likely to provide unbiased assessment regarding college students’ driving safety than an attitudinal scale or a behavioral scale alone. As demonstrated in this study, attitudes are not a sufficient precursor to behavior. Even though there was no significant gender difference in the attitudes toward motor vehicle safety, there was a significant difference in the behavior between male and female subjects. Therefore, a scale that measures both attitudes and behaviors is more likely to yield an undistorted picture of the construct being measured, which, in turn, is conducive to the predictive usage of the scale.
Prior to wide dissemination of the scale, further research needs to be conducted to affirm the findings of this study and validate the scale. Especially, it is suggested to perform a confirmatory factor analysis with a second group of college students preferably recruited by a probability sampling method. Despite these limitations, the various statistical analyses and item analyses indicate that this study has launched a solid step in the validation process of the motor vehicle safety scale for young drivers.
Appendix A. Survey Questionnaire
Directions: Please rate statements one to seven on your answer sheet using the 5-point scale below.
1 = Strongly Disagree, 2 = Disagree, 3 = Neutral, 4 = Agree, 5 = Strongly Agree
1. Everyone should always wear safety belts when driving motor vehicles.
2. Young adults (19-24) should always wear safety belts when riding as a passenger in motor vehicles.
3. Exceeding the posted speed limit is one of the major causes of motor vehicle crashes.
4. Exceeding the posted speed limit is one of the major causes of motor vehicle related deaths.
5. Most vehicle crashes could be avoided if drivers practiced defensive driving techniques, for example, aiming high in their vision, resisting distractions, and looking at the big picture.
6. Anyone who drives under the influence of alcohol is practicing a very risky behavior.
7. Anyone who drives under the influence of drugs (other than caffeine or mild pain relievers) is practicing a very risky behavior.
Please rate the questions eight to 12 using the 5-point scale below.
1 = Strongly Agree, 2 = Agree, 3 = Neutral, 4 = Disagree, 5 = Strongly Disagree
8. Laws requiring vehicle safety belt use are very helpful in reducing injuries.
9. Laws requiring vehicle safety belt use are very helpful in reducing fatalities.
10. Driving the posted speed limit will help prevent motor vehicle injuries.
11. Driving the posted speed limit will help prevent motor vehicle deaths.
12. Young adults should be able to drink two or three beers during one hour and drive safely.
Please rate questions 13 to 18 using the 5-point scale below.
1 = Never, 2 = Seldom, 3 = Occasionally, 4 = Most Times, 5 = Always
13. How often do you wear safety belts when driving motor vehicles?
14. How often do you wear safety belts when riding as a passenger in motor vehicles?
15. How often do you exceed the posted speed limit?
16. How often do you practice defensive driving techniques?
17. How often do you drink alcohol and drive a motor vehicle afterwards?
18. How often do you use drugs (prescription and/or illegal) and drive a motor vehicle afterwards?
Table 1. Item Values of the Motor Vehicle Safety Scale for Young
Drivers by Internal Criteria and Distribution of Responses by Option.
Item TS AS BS Lower Group
No. r r r t (DP)
1 .68 *** .68 *** .37 *** 6.75 (1.57) ***
2 .72 *** .73 *** .39 *** 8.14 (1.70) ***
3 .53 *** .58 *** .19 * 7.21 (1.48) ***
4 .54 *** .59 *** .18 * 7.83 (1.46) ***
5 .45 *** .46 *** .20 * 4.49 (0.91) ***
6 .66 *** .67 *** .33 *** 5.55 (1.36) ***
7 .66 *** .68 *** .30 *** 7.11 (1.72) ***
8 .58 *** .64 *** .17 * 7.85 (1.74) ***
9 .63 *** .69 *** .19 * 8.39 (1.84) ***
10 .66 *** .71 *** .25 ** 9.07 (1.84) ***
11 .71 *** .76 *** .25 *** 10.56 (1.89) ***
12 .49 *** .48 *** .29 *** 6.06 (1.46) ***
13 .53 *** .32 *** .79 *** 5.66 (1.26) ***
14 .47 *** .27 *** .73 *** 5.99 (1.38) ***
15 .34 *** .25 *** .39 *** 4.70 (0.81) ***
16 .33 *** .18 * .54 *** 3.95 (0.81) ***
17 .37 *** .21 ** .59 *** 3.61 (0.64) ***
18 .38 *** .24 ** .56 *** 4.44 (0.95) ***
% of Subject Response
No. SA A N D SD
1 73 15 4 4 4
2 61 23 7 6 3
3 8 44 18 24 6
4 11 39 25 22 3
5 40 43 9 8 1
6 77 15 2 2 4
7 57 23 10 4 6
8 34 29 13 17 8
9 35 29 12 15 9
10 19 39 22 16 5
11 20 34 27 18 3
12 37 22 17 17 7
13 60 30 3 4 3
14 50 29 10 6 4
15 0 6 28 45 21
16 23 44 22 7 3
17 44 35 20 1 1
18 49 26 18 4 3
Note. TS = Total scale score. AS = Attitudinal subscale score. BS =
Behavioral subscale score. DP = Discriminative power coefficients. r =
Correlation coefficient. t = Statistical t-value. SA = Strongly agree
(“Always” for items 13-18); A = Agree (“Most times” for items 13-18);
N = Neutral (“Occasionally” for items 13-18); D = Disagree (“Seldom”
for items 13-18); SD = Strongly Disagree (“Never” for items 13-18).
* p <.05. ** p <.01. *** p <.001.
Table 2. Distribution of Items by Factor Loadings and Communalities
Rotated by Varimax Method.
Item No. Factor 1 Factor 2 Factor 3 Factor 4 Factor 5
1 .18 .82 .12 .37 -.06
2 .24 .76 .21 .39 -.12
3 .84 .12 .02 .08 -.02
4 .80 .18 .01 .04 .00
5 .18 .41 .06 -.13 -.03
6 .15 .81 .07 .03 .28
7 .21 .70 .17 -.14 .40
8 .15 .12 .92 .10 -.05
9 .22 .14 .92 .08 -.00
10 .72 .33 .23 -.07 .17
11 .73 .27 .36 -.01 .09
12 -.03 .10 .59 .06 .35
13 .04 .17 .14 .86 .22
14 .07 .12 .06 .86 .10
15 .51 -.20 .04 .26 .01
16 -.02 .04 .02 .24 .09
17 .07 .06 -.02 .26 .80
18 .03 .12 .10 .04 .78
Item No. Factor 6 Communalities
1 .03 .85
2 .08 .85
3 -.08 .73
4 .01 .68
5 .65 .65
6 .15 .78
7 .13 .76
8 -.02 .89
9 -.04 .91
10 .02 .71
11 .05 .75
12 .22 .53
13 .07 .84
14 .11 .79
15 .33 .48
16 .80 .71
17 -.02 .71
18 .07 .63
Note. Extraction Method: Principal Component Analysis
Table 3. Internal Consistency and Correlation Coefficients of Retained
Components with Total Scale Score and Attitudinal or Behavioral
Internal Scale Subscale
Consistency Score Score
Component Label alpha r r
1 Speed limit violation .82 .74 *** .32 *** (B)
2 Perceptions regarding .84 .81 *** .40 *** (B)
risky driving behavior
3 Attitudes regarding .80 .67 *** .26 *** (B)
compliance with rules
4 Seat belt usage .84 .53 *** .31 *** (A)
5 Driving under the .61 .64 *** .53 *** (A)
influence of drugs/
Note. r = Correlation coefficient. (A) denotes correlation coefficient
of each behavior-dominant component with the summated attitudinal
subscale score and (B) denotes correlation coefficient of each
attitude-dominant component with the summated behavioral subscale
score. *** p<.001.
Table 4. Intercorrelation Matrix of Retained Components.
Component 1 Component 2 Component 3 Component 4
Component 1 1.00 .48 *** .34 *** .19 *
Component 2 1.00 .37 *** .33 ***
Component 3 1.00 .23 **
Component 4 1.00
Component 1 .30 ***
Component 2 .59 ***
Component 3 .29 ***
Component 4 .28 ***
Component 5 1.00
Note. * p <.05. ** p <.01. *** p <.001.
Begg, D., & Langley, J. (2001). Changes in risky driving behavior from age 21 to 26 years. Journal of Safety Research, 32, 491-499.
Bever, D. L. (1996). Safety: A Personal Focus. Boston: McGraw-Hill.
Garson, G. D. (n.d.). PA 765 Statnotes: Online textbook (North Carolina State University). Retrieved September 20, 2002, from http://www2.chass.ncsu.edu/garson/PA765/standard.htm
Gledhill-Hoyt J., Lee, H., Strote, J., & Wechsler, H. (2000). Increased use of marijuana and other illicit drugs at U.S. colleges in the 1990s: Results of three national surveys. Addiction, 95(11), 1655-1667.
Jonah, B.A. (1986). Accident risk and risk-taking behavior among young drivers. Accident Analysis and Prevention, 18(4), 255-271.
Jonah, B.A. (1990). Age differences in risky driving. Health Education Research, 5(2), 139-149.
Kim, S. & Kim, K. (in press). Personal, temporal and spatial characteristics of seriously injured crash-involved seat belt non-users in Hawaii, Accident Analysis and Prevention.
Matthews, M. L., & Moran, A. R. (1986). Age differences in male drivers’ perception of accident risk: the role of perceived driving ability. Accident Analysis and Prevention, 18(4), 299-313.
McCartt, A. T., Leaf, W. A., Farmer, C. M., Fersguson, S. A., & Williams, A. F. (2001). Effects of Florida’s graduated licensing program on the behaviors and attitudes of teenagers. Journal of Safety Research, 32(2), 119-131.
Montano, D. E., Kasprzyk, D., & Taplin, S. H. (1997). The theory of reasoned action and the theory of planned behavior. In Glanz, K., Lewis, F. M. & Rimer, B. K. (Ed.), Health Behavior and Health Education (pp. 85-112). San Francisco, CA: Jossey-Bass Inc.
National Highway Traffic Safety Administration. (2001). The Facts: To buckle up America (DOT HS 808 866). Retrieved October 9, 2002, from http://www.nhtsa.dot.gov /people/injury/airbags/buckleplan/seatbeltsafroamerican/index.htm
National Highway Traffic Safety Administration. (1994). Traffic safety facts 1993: a compilation of motor vehicle crash data from fatal accident reporting systems and the general estimates system (DOT HS 808 169). Washington, DC: U.S. Government Printing Office.
National Safety Council. (2001). Injury Facts, 2001 Edition. Itasca, IL: Author.
Preusser, D. F., Williams, A. F., & Lund, A. K. (1991). Characteristics of belted and unbelted drivers. Accident Analysis and Prevention, 23(6), 475-482.
Strote J., Lee, J. E., & Wechsler, H. (2002). Increasing MDMA use among college students: Results of a national survey. Journal of Adolescent Health, 30(1), 64-72.
Torabi, M. R. (1991). A Cancer Prevention Behavior Scale. Health Values, 15(3), 12-21.
Wechsler, H., Lee, J. E., Kuo, M., & Lee, H. (2000). College binge drinking in the 1990s: A continuing problem. Results of the Harvard School of Public Health 1999 College Alcohol Study. Retrieved October 10, 2002, from http://www.hsph.harvard.edu/cas/rpt2000/CAS2000rpt2.html
Wells, J. K., Williams, A. F., & Farmer, C. M. (2002). Seat belt use among African Americans, Hispanics, and Whites. Accident Analysis and Prevention, 34(4), 523-529.
Williams, A. (1998). Risky driving behavior among adolescents. New perspectives on adolescent risk behavior. Cambridge: Cambridge University Press.
Zwick, W. R., & Velicer, W. F. (1982). Factors influencing four rules for determining the number of components to retain. Multivariate Behavioral Research, 17, 253-269.
Zwick, W. R., & Velicer, W. F. (1986). Comparison of five rules for determining the number of components to retain. Psychological Bulletin, 99(3), 432-442.
HEALTH EDUCATION RESPONSIBILITY AND COMPETENCY ADDRESSED
Responsibility IV-Evaluating Effectiveness of Health Education Program
Competency A-Develop plans to assess acjievement of program objectives
Sub-comptency 8-Develop valid and reliable intruments
Earl H. Blair, Ed.D., is an Associate Professor, Mohammad R. Torabi, Ph.D., is Chancellor’s Professor and Chairperson and Dong-Chul Seo, Ph.D., is a Lecturer in the Department of Applied Health Science at Indiana University. Graham F. Watts, Ph.D., is with the Duval County (FL) Health Department. Address all correspondence to Mohammad R. Torabi, Ph.D., Department of Applied Health Science, Indiana University, Bloomington, IN 47405. PHONE: 812-855-3936; E-MAIL: firstname.lastname@example.org.
COPYRIGHT 2004 University of Alabama, Department of Health Sciences
COPYRIGHT 2004 Gale Group