,. AUTOMOTIVE CRASH INJURY RESEARCH ' OF CORNELL UNIVERSITY CAT. FOR pupiic HEALTH The Consistency of ACIR Accident — Injury Relationships in Four States , by B. J. Campbell 316 EAST 61ST e@ NEW YORK 21, N. Y. '' ome ''a Automotive Crash Injury Research of Cornell University Robert A. Wolf Director B. J. Campbell Assistant Director Myron I. Macht Merwyn A. Kraft John W. Garrett Head, Field Branch Staff Scientist Head, Analysis Branch 316 East 61st Street New York 21, New York TE 8-86) '' <™™ ''THE CONSISTENCY OF ACIR ACCIDENT-INJURY RELATIONSHIPS IN FOUR STATES by B. J. CAMPBELL Approved? Fobed 0) Vo Robert A. Wolf Director June 1961 ''Public Health GIFT «t ''HE SbI4 1 a 0313 Neetrh Fe FOREWORD Lb hany 1 / OUTLINE SUMMARY 2 CHAPTER 1 - INTRODUCTION 3 CHAPTER 2 - METHODS 4 Section 1 - Selection of Factors 4 A. Injury Criteria 4 B. Aecident Factors 5 1. Listing of 5 2. Fractioning of (orthogonal set) 6 Section 2 - Determination of Accident-Injury Relationships 7 Section 3 - Comparison of States for Consistency 7 A. Possible Outcomes 7 B. Expected Outcomes 8 Section 4 - Summary of Methods 10 CHAPTER 3 - RESULTS 12 Section 1 - Comparison of 39 Accident Sub-Factors with Injury Using Fatal vs. Non-Fatal Injury Criterion 12 Section 2 - Comparison of 39 Accident Sub-Factors with Injury Using Dangerous-Fatal vs. Remainder Criterion 19 Section 3 - Distribution of Strong and Weak Factors 20 CHAPTER 4 - DISCUSSION AND CONCLUSION 23 APPENDIX 1 - List of 39 Sub-Factors 24 2 - List of Seven Conditions 26 3 - Strong and Weak Factor Distribution 28 4 - Summary of Findings 35 5 - Details of Statistical Tests 38 699 ''''FOREWORD Automotive Crash Injury Research (ACIR) of Cornell University is engaged in the study of occupant injury causes in automobile accidents. Such investigation contributes to efforts of automotive safety engineers to alter components of automobiles in such a way as to minimize injury they may cause during a crash. Data for this research are reported on special forms developed by ACIR and completed by state highway patrolmen, city policemen, and physicians. The completed accident reports are available through cooper- ation of state and local police organizations. The medical reports are made available through the efforts of state medical societies, state health departments and state hospital associations. It is only by virtue of the exceptional cooperation by members of all these organizations that this body of data is available for study at ACIR. ACIR was supported during the research period of this report by funds from the Office of the Surgeon General, Department of the Army, following recommendations of the Armed Forces Epidemiological Board; from the Automobile Manufacturers Association; and from the National Institutes of Health of the United States Public Health Service. The author wishes to express appreciation to his colleagues; to Mr. Boris Tourin, formerly of ACIR; and to Dr. Irwin Bross, ACIR statistical consultant for helpful advice concerning this paper. Thanks are also due to other members of the ACIR staff for performing and check- ing the extensive calculations. ''''SUMMARY Accident-injury relationships were computed using data from four states in order to demonstrate that given relationships were consistent from state to state. Thus, if the relationship were strong in one state, it should be strong in others; if it were weak in one, then it should be weak in others. A total of 39 accident sub-factors were chosen in such a manner that they constituted a statistically independent set of compar- isons. Some of these factors were known to have a strong relationship to injuries and others were known to have a weak relationship. Two injury criteria were established and the 39 sub-factors were related to each. Depending on direction and magnitude of relationships in the four states, sub-factors were listed as falling under one of seven Conditions. Conditions I and II indicated strong relationships consistent among the states. Conditions III, IV, and V indicated weak relationships consist- ent among the states, or null relationships. Conditions VI and VII indicated significant inconsistencies. It was stated that in order for ACIR data to be regarded as acceptable, the present study (1) must show that sub-factors fall mostly under Conditions I-V, and not under VI and VII, and (2) must show that sub-factors under I and II come primarily Prom factors previously found to be strong, and that sub-factors under III, IV, and V come primarily from factors previously found to be weak. The study results confirmed both stipulations. All sub-factors fell under Conditions I-V. There were no instances of significant in- consistencies (Conditions VI or VII). Moreover, sub-factors emerging strong (Conditions I and II) came primarily from those previously found to be strong. Sub-factors emerging weak (Conditions III, IV, and V) came primarily from those previously found to be weak. It was concluded that accident-injury relationships in ACIR data are consistent from state to state and that ACIR findings have general applicability. ''''THE CONSISTENCY OF ACIR ACCIDENT-INJURY RELATIONSHIPS IN FOUR STATES CHAPTER 1 INTRODUCTION ACIR findings deal with relationships between certain accident factors and injury. Since data for these studies were collected in a total of 19 states, it is most important to determine (1) whether data from each state adequately represent events in that state, and (2) whether the accident-injury relationships are consistent from state to state. A recent ACIR publication deals with the first question and indi- cates that the data are adequately representative,* but the second ques- tion remains. The ability of ACIR to define injury problems and to evaluate attempted solutions is predicated on consistency of accident- injury relationships in various parts of the country. Obviously, use- fulness of the findings would be placed in question if the data indicated, for example, that greater injury was associated with higher speed in one state, while in another state greater injury was associated with lower speed. It is the purpose of this paper to test many accident-injury relationships using data from four states, in order to estimate inter- state consistency with respect to magnitude and direction. *A Comparison of Automotive Crash Injury Research Samples with Complete State Data, ACIR, February 1961 (B. J. Campbell). ''ra Bl see) i ‘ : ; A te, ate? % ‘ Ps - - ‘ : - +f ‘ to - - - . i - tae - . * . . tore Se 35 agap rGieve ape ete: .: id ests c* + a sg . 1 y ! it : ed j q r Pia ’ - : A x iy 2 ‘ a . J . oe - 5 . ) 2 oy t vy ‘ 7 t 7: + . + « kok. a oe P : s { “ root I ; ' i i 7 ; : 5 * a toro. Mo ny : - par ‘ ‘ : J Adie yi rer . “ - = - oid 3 “ . 4 4 tea ks ‘ —y ine. “oe - : ’ %J fiitw ec f diettews Wa: hua i Pan ''CHAPTER 2 METHODS ACIR data collected in four states were used to compute accident- injury relationships; the four states were Arizona, California, North Carolina and Texas. Data from California and North Carolina were col- lected according to Random Sample Plans, while data from the other two states were collected according to Administrative Sample Plans. These sampling methods were described in the paper mentioned in Chapter 1. Comparison of accident-injury relationships for inter-state consistency was done in three steps listed immediately following and then described in more detail in subsequent sections of this chapter. Step 1: Selection of accident and injury factors to be tested for relationship. Step 2: Determination of relationship between each accident factor and each injury criterion using data from each of the four states. Step 3: Comparison of accident-injury relationships among the four states for consistency of magnitude and direction. Section 1 Selection of Factors A. Injury Criteria The analytical procedures utilized required dividing the injury scale into two parts. The first injury criterion involved classifying occupants as to whether or not they were fatally injured. ACIR fatals are defined as those who died within 24 hours after the accident. The second injury criterion involved combining the fatal category with the "dangerous injury" category. In ACIR terms, a dangerous injury is one that could result in death after the first 24 hours, despite prompt medical care (detailed criteria are used by ACIR Accident Analysts). This combined "dangerous-fatal" category constitutes one part of the in- jury continuum and all lesser degrees of injury constitutes the other part. The following diagram illustrates the point at which the injury spectrum was divided for each of the two injury criteria. Fatal vs. Non-Fatal none, minor, Dangerous-Fatal vs. | Remainder none, minor, ...... | dangerous, fatal '' ''ay Thus, two injury criteria are defined and each of these is compared with numerous accident factors selected by the procedures described below. B. Accident Factors ACIR data contain very many accident factors suitable for relation- ship testing against injury, and from this large group, ten were select- ed. Five of these factors had previously been shown to have a powerful relationship to injury, and the other five had been shown to have a weak relationship.* Thus, the research was designed to permit observation of the state-by-state consistency of strong accident-injury relationships, as well as the state-by-state consistency of weak (or zero) relationships. The ten factors are listed as follows: A. Strong Factors 1. Applicable Impact Speed: The speed at impact of the vehicle that principally determined severity of the acci- dent. 2. Area of Severest Impact: The part of the car sustaining the principal forces of the accident. 3+ Accident Severity: A five point rating scale taking into account speed and damage. 4, Seated Position: Specification of the seat occupied by the person. 2+ Ejection: Whether or not the occupant was thrown from the car during the accident. B. Weak Factors 6. Occupant Age. 7. Occupant Height. 8. Occupant Sex. 9. Occupant Weight. 10. Year of Manufacture: The production year model of the car involved in the accident. After selection of the ten factors, it was necessary to dichotomize them as was done with the injury criteria. For example, the Applicable *Accident factor strength was studied in the preliminary phase of the ACIR Basic Science Program (unpublished). In this study, accident fac- tors were ranked according to the amount of injury variance for which they accounted. ''''he Impact Speed Scale (Strong Factor 1) could be divided into two parts such as "up to 39 mph" and "40 mph and over". If each factor were thus dichotomized, then the relationship between the ten accident factors and the two injury criteria could be computed and a total of 20 comparisons would be available. Such a procedure would not be very efficient, how- ever, because most of the accident factors can be divided meaningfully into more than two parts. For example, data were available in sufficient quantity to permit a four-way division of Applicable Impact Speed and still to have useable frequencies in each category. The scale was divided as follows: 1. O-19 mph 2. 20-39 mph 3. 40-59 mph 4. 60 mph and over With the data divided into four parts, there are many two-way comparisons possible, whereas if simply dichotomized, the factor could be used only once. Now, for example, the "0-19 mph" category could be compared to the "60+" category; the "20-39 mph" category could be com- pared to the "40-59 mph" category; or all categories up to 59 mph could be compared to the "60+" category. There are several other possible com- binations. Although dividing the scale into several parts increases the number of possible comparisons, and more efficiently uses the data, another problem is introduced. This is the fact that some combinations would be such close duplicates of others that they would tend to "tell the same story". The desirable compromise is to select several unique combina- tions -- ones that are not duplicates.* After considering various re- quirements of the situation, the following three sub-factors where chosen for Applicable Impact Speed: 1. O-19 mph vs 60+ mph 2. 20-39 mph vs 40-59 mph 3. 20-39 mph+ 40-59 mph vs 0-19 mph+ 60 mph Using the same general procedure, combinations of values for the other nine factors (page 5) were selected and finally a total of 39 nearly independent sub-factors were chosen (listed in Appendix 1). Six- teen of the sub-factors were from the five strong factors and 23 were from the five weak factors. Each of the 39 sub-factors was compared against each injury criterion; thus, there were 78 opportunities to com- pare the four states for consistency of magnitude and direction of acci- dent-injury relationship. *It was advisable to deal with a set of statistically independent (orthogonal) comparisons. Since there is more than one such set, it was desirable that the one selected be the most meaningful in terms of the data under study. The set finally selected was nearly orthogonal. The author wishes to express appreciation to Dr. Irwin Bross (ACIR statistical consultant) who formulated this analysis. ''''Section 2 Determination of Accident-Injury Relationships The second step consists of determining the relationship between each accident sub-factor and each injury criterion for each of the four states. The 2 x 2 Chi square test was used, and completing the analyti- cal program required calculation of 312 statistical tests (39 sub-factors, 2 injury measures and 4 states). The 2 x 2 Chi square gives two kinds of information. One pertains to the degree of confidence that the test reflects a true relationship, and the other pertains to the direction of relationship. All tests were calculated in the same format, illustrated as follows: Non-Fatal Fatal Total 4% Fatal Sub-Factor A 90 10 100 10.0 Sub-Factor B 170 30 200 15.0 260 40 300 * The direction of association is always shown in the final column. A "plus" means that the bottom percentage is larger and indicates a posi- tive correlation. A "minus" indicates negative correlation. If the Chi square for all four states carried a "plus" sign, then all four of the associations would be in the same direction. If the signs varied among the states, then the direction of association would appear to vary. Whether the "plus" or "minus" represents a true relationship or just a sampling fluctuation depends on the magnitude of the relationship. If the Chi square value exceeds the one percent level, the relationship is accepted as significant. Otherwise, the apparent relationship is at- tributed to chance fluctuations. Section 3 Comparision of States for Consistency Since so many statistical tests were involved, interpreting the data would be difficult in the absence of some ground work; therefore, two questions are listed immediately following and are discussed there- after: A. What are the possible outcomes of the 78 comparisons? B. What outcomes would be expected in view of the presence of strong and weak factors? A. Possible Outcomes The outcomes depend on both magnitude and direction of the rela- tionship. As far as direction of relationship is concerned, all four states could be the same, but they could also differ. For example, higher speed might be associated with higher injury in all four states, ''''but it is also possible that higher speed could associate with lower in- jury in some states and higher injury in others. Magnitude of relation- ship can vary in the same way. A relationship might be strong enough to be statistically significant in all four states, but it is also possible that none would be significant, or that some would be significant and some would not. Considering jointly the aspects of magnitude and direc- tion, it can be seen that possible outcomes include the full range from (1) all four relationships significant and in the same direction, to (2) all four relationships significant but two in one direction and the other two in the opposite direction. In other words, outcomes could range from a high degree of interstate consistency to a high degree of inconsistency. Specifically, the possible outcomes of the state comparisons are divided into seven categories. These are described as follows and are summarized in Table 1. The first possible outcome (called Condition I) is the one in which there is perfect agreement. That is, in all four states, the accident-injury relationship is significant and the direction of association is the same. Condition II also indicates consistency but somewhat less in degree. This covers the situation in which the direc- tion of association is the same for all four states and some, but not all, are significant. Conditions III and IV also indicate consistency but in this case a consistent lack of relationship. Condition III is defined as the situation in which direction of association is the same in all four states, but none is significant. Condition IV is the situation in which none is significant and direction of association is not the same for all. Condition V includes opposition of direction and one or more relationships significant but not significantly opposite each other. For example, a relationship might be positive and significant in one state, positive but not significant in two others, and negative but not significant in the fourth. Conditions VI and VII indicate significant inconsistency. Condi- tion VI includes opposition in direction and some significantly opposed. In Condition VII, all are significant and there is opposition of direc- tion. Table 1 on the following page summarizes the foregoing (Appendix 2 describes the Conditions in greater detail). B. Expected Outcomes Although it is possible for sub-factors to fall under any of the seven Conditions, the meaning of the data depends on the manner in which the 78 groups of tests are actually distributed among them. It is dif- ficult to specify exactly how the tests will be distributed, but some general predictions are reasonable. First, data falling under Conditions I and II would indicate strong consistent relationships among the states. This is true because data from all four states would indicate the same direction of relation- ship and some or all would be significant. Such would argue well for ''''Direction of Condition Association I Same for all 4 states II Same for all 4 states III Same for all 4 states IV One or two in opposite direction Vv One or two in opposite direction VI One or two in opposite direction VII One or two in opposite direction Table 1 The Seven Conditions Example* Negative Correlation Positive Correlation Magnitude of Not Not Association Significant Significant Significant Significant All significant XXXX One or more but not XX xX all significant None significant XXXX None significant XX XX One or more but not xX xX xX all significant - does not include those significantly opposite At least two xX xX xX xX significant and opposite All significant XX xX *As can be seen from the definition, there are other possible arrays meeting the conditions. **Each "X" represents one of the four states. ''''10 adequacy of ACIR data, particularly if sub-factors falling under Condi- tions I and II came from those previously defined as strong factors. Second, data falling under Conditions III, IV, and V would indicate sub-factors with weak relationships to injury or those with no relation- ships (null relations) to injury. That is, if a sub-factor showed no significant relationship to injury in one state, there might be inter- state consistency in that the other states would also show no signifi- cant relationship. Such would also argue well for consistency of ACIR data, particularly if sub-factors falling under Conditions III, IV, and V came primarily from factors previously defined as weak. Finally, data falling under Conditions VI or VII could indicate significant inconsistency among states. That is, in one state there might be a significant positive relationship between a sub-factor and injury, while in another state the relationship might be significant but opposite in direction. Such findings would seriously question ACIR data as consistent and of general applicability. The exact distribution of sub-factors under the various Conditions requires knowledge not obtainable; however, by making certain assumptions regarding power of the statistical tests, etc., it is possible to esti- mate the situation and to construct a model which predicts the distribu- tion of sub-factors among the Conditions. This is discussed in the necessary detail in Appendix 3. Based on the foregoing and on the ex- panded treatment in Appendix 3, the following is set forth: ACIR data may be regarded as consistent and adequate for general use only if both of the following stipulations are met: 1. The great majority of sub-factors are distributed under Conditions I-V and very few or none are under Conditions VI and VII. 2. The sub-factors falling under Conditions I and II are primarily from those previously defined as strong, and the sub-factors falling under Conditions III, IV and V are primarily from those previously defined as weak. In the Discussion Chapter, the results are evaluated in light of these stipulations. Section } Summary of Methods Accident-injury relationships were computed using data from four states in order to demonstrate that given relationships were consistent from state to state. Thus, if the relationship were strong in one state, it should be strong in others; if it were weak in one, then it should be weak in others. A total of 39 accident sub-factors were chosen such that they constituted a statistically independent set of comparisons. Some of these factors were known to have a strong relationship to '' at ''11 injuries and others were known to have a weak relationship. Two injury criteria were established and the 39 sub-factors were related to each. Depending on direction and magnitude of relationships in the four states, sub-factors were listed as falling under one of seven Conditions. Conditions I and II indicated strong, consistent relationships. Condi- tions III, IV, and V indicate weak or null relationships. Conditions VI and VII indicated significant inconsistencies. It was stated that in order to be acceptable, ACIR data (1) must fall mostly under Conditions I-V and not under VI and VII and (2) must show that sub-factors under I and II come primarily from factors previously shown to be strong and sub-factors under Conditions III, IV and V come primarily from factors previously shown to be weak. ''''i2 CHAPTER 3 RESULTS The Results Chapter is divided into three sections as follows: 1. Comparison of 39 sub-factors with injury using the fatal vs. non-fatal criterion. 2. Comparison of 39 sub-factors with injury using the dangerous- fatal vs. remainder criterion. 3. Distribution of strong and weak factors. For simplicity, only one sub-factor under each Condition is pre- sented in Section 1, and in Section 2 an even more brief summary is given. Data concerning all 39 sub-factors are listed in Appendix 4, Section 1 Comparison of 39 Accident Sub-Factors With Injury Using the Fatal vs. Non-Fatal Injury Criterion Sub-Factors under Condition I: This represents the situation in which a significant accident-injury relationship exists in all four states and the direction of association is the same for all. Condition I suggests consistency among states, and the findings with respect to ejection illustrate it (the ejection sub-factor is No. 16 listed in Appendix 4a). In each of the four states, occupants injured fatally and those not injured fatally are classified according to whether or not they were ejected. Table 2 shows these data. ''''Table 2 Fatalities as a Function of Ejection 13 Arizona California Non Non Fatel Fatal WN b Fatel Fatal WN b Not Ejected 569 11 580 1.9 768 12 780 1.5 Ejected 131 27 158 17.1 120 17 «:137.”—=—s:12..4 Total 700 38 =: 738 + 888 29 917 + X°(1df)=55.6 p
    .05 X2(1df)=4.01 02 p<.05 North Carolina Texas Not Not Killed Killed W b Killed Killed WN b All Occupants 764 20 784 2.6 476 29 505 5.7 Drivers Alone 131 13 144 «9.0 108 7 115 6.1 Total 895 33 928 + 584 36 620 + X2(1af)=13.1 p<.0l X2(laf)=.0l1 p>.05 '' ''Sub-Factors under Condition IIT: 15 Condition III describes the situation in which all four associations are in the same direction, but none is significant; data corresponding to this Condition support the hypothesis of consistency by suggesting that if an accident-injury relationship is weak in one state, it will be weak in other states. Condition III is illustrated in Table 4 which shows the frequency of fatalities according to occupant height. quency of fatality among taller persons seemed somewhat greater, but the tests results were uniformly not significant. In all, 15 sub-factors fell under Condition III (see Appendix 4b). Table 4 In each of the states, fre- Fatalities as a Function of Height Arizona California Hetgnt Killed Killed No $ Killed Kea HO o - 48" 39 2 41 4g 45 0 45 0 4g - 73" 606 38 6459 779 2T 806 ~—- 33 Total 645 Lo 685 + 82h 27 851 + x2(laf)=.01 p>.05 X2(laf)=.66 p>.05 North Carolina Texas Not Not Height Killed Killed W b Killed Killed W b Oo - 48" 29 2 ST = 3 59 3 58 5.2 4g - 73" 737 32 769 4.2 542 36 578 6.2 Total 792 34 826 + 597 39 636 + X@(ldf)=.01 p>.05 X@(ldf)=.0006 p>.05 The first three Conditions pertained to situations in which the direction of apparent association was the same in all four states. The next four Conditions describe situations in which the direction of asso- ciation varies among the states. ''''16 Sub-Factors Falling under Condition IV: This is the situation in which none of the relationships is significant and in which the direc- tion of association varies among the states. This Condition should in- clude situations in which there is a random fluctuation around a true zero association and it supports the hypothesis of consistency by show~- ing a consistent lack of relationship. Table 5 illustrates Condition IV by comparing the relationship between fatalities and area of car damage in the four states. In two states, the fatality percentage is higher in the case of front area impacts, while in the other two, it is lower in this area. The direction of difference is opposite in two states, and none of the associations is significant. There are 12 comparisons that fall under Condition IV (see Appendix 4b). Table 5 Fatalities as a Function of Area Struck Arizona California Kitiea Killed WN z Kittea Killed WN b Compartment 19 2 81 2.5 178 10 188) 5.3 Front 267 a5 282 65.3 391 8 399 2.0 ‘Total 346 17 -363—t—«‘a+ 569 18 587 - X°(1df)=.6 p> .05 X°(1df)=3.7 p>.05 North Carolina Texas Not Not Killed Killed WN b Killed Killed W b Compartment 116 3 Wg 2.5 129 12 141 = 8.5 Front 304 19 403 4.7 257 19 276 6.9 Total 500 22 522 + 386 31 417 - X2(1df)=.6 p>.05 X2(1af)=.2 p> .05 ''_ ''Sub-Factors Falling under Condition V: 17 Condition V is one in which there are differences in direction of apparent association and one of the associations is significant. Table 6 shows an example of data which conforms to Condition V - a comparison of fatalities by age of person killed. significant. In one of the states, the association is positive and In another the association is positive but not significant. In the remaining two states, the direction of association is opposite to the significant one, but not significant. to Condition IV, but may seem less supportive of the hypothesis of con- sistency because divergence of trend among the states is more extreme. Two comparisons fit Condition V (see Appendix 4b). Table 6 Thus, Condition V is similar Fatalities as a Function of Age of Person Killed Arizona California Not Not Age Killed Killed WN b Killed Killed W b 45-54 69 0 69 0 70 Z 75 6.7 35-44 91 12 103 11.7 112 3 115 2.6 Total 160 12 172 + 182 8 190 - X@(1df)=6.9 px.Ol X2(1df)=1.0 p>.05 North Carolina Texas Not Not Age Killed Killed WN b Killed Killed W b 45-5) hg 4 53. «7.5 58 2 60 3.3 35-4h 5 8 123 6.5 77 8 85 9.4 Total 164 12 176 - 135 10 145 + X°(ldf)=.00 p> .05 X2(1df)=1.2 p>.05 '' ''18 Sub-Factors Falling under Conditions VI and VII: Conditions VI and VII cover situations which contradict the hypothesis of consistency because they involve significant disagreement among the states. None of the sub-factors fell under Condition VI and none fell under Condi- tion VII. Summary of Sub-Factors Falling under Conditions I-VII: All 39 of the sub-factors (using the fatal vs. non-fatal injury criterion) fell under Conditions supporting the hypothesis of consistency. Three sub- factors fell under Condition I and seven under Condition II; 15 fell under III and 12 under IV; 2 fell under V. There were no instances of sub-factors falling under Conditions VI or VII; thus, there were no significant inconsistencies. Table 7 summarizes the number of sub- factors under each Condition. Table 7 Distribution of 39 Sub-Factors According to Seven Conditions (Fatal Injury Measure) Direction of Magnitude of Number of Condition Association Association Sub-Factors I Same for all 4 states All significant 3 II Same for all 4 states One or more but not all significant 7 III Same for all 4 states None significant 15 IV One or two in opposite None significant 12 direction Vv One or two in opposite One or more but not direction all significant - 2 does not include those significantly opposite vI One or two in opposite At least two signifi- direction cant 0 vit One or two in opposite All significant 0 direction _ ''''19 Since only one sub-factor under each Condition was actually shown in the Results, a summary of the analysis of all 39 sub-factors for both injury criteria is presented in Appendix 4a, 4b, and 4c. Moreover, information permitting reconstruction of any test in the study is present-~ ed in Appendix 5. Section 2 Comparison of 39 Accident Sub-Factors With Injury Using The Dangerous-Fatal vs. Remainder Criterion This part of the analysis involved retracing the analytical pro- cess just described in Section 1, this time using a somewhat different injury criterion. Instead of using fatal injuries, the classification included a combination of those injured to a dangerous degree plus those injured fatally. This was done to check the possibility that accident-injury relationships may be influenced grossly by the dividing point of the injury scale. The manner in which the sub-factors were distributed among the Conditions using the dangerous-fatal injury criterion was similar to the results obtained using the other injury criterion; therefore only a sum- mary is presented here. The full analysis is summarized in Appendix le and 4e and the tests can be reconstructed from Appendix 5. As before, there were no instances of Conditions VI or VII. There was, however, a somewhat greater number of sub-factors falling under Condition V (seven instead of two as in the first analysis). None of the 39 sub-factors showed significant disagreement when the dangerous-fatal measure was used, just as there was no significant dis~ agreement when the fatal measure was used. Table 8 summarizes the distribution of 39 sub-factors among the seven Conditions using the dangerous-fatal injury criterion. (This is comparable to Table 7 for the other injury criterion. ) ''''20 Table 8 Distribution of 39 Sub-Factors According to Seven Conditions (Dangerous Fatal Injury) Direction of Magnitude of Condition Association Association Frequency i Same for all 4 states All significant 4 TZ Same for all 4 states One or more but not all significant 8 III Same for all 4 states None significant 8 IV One or two in opposite None significant 12 direction V One or two in opposite One or more but not direction all significant - 7 does not include those significantly opposite VI One or two in opposite At least two signifi- direction cant and opposite 0 VII One or two in opposite All significant O direction Section 3 Distribution of Strong and Weak Factors In Sections 1 and 2 of this Chapter each sub-factor was listed as falling under one of the seven Conditions. It was stated that Conditions I and II would be expected to reflect strong, consistent relationships; Conditions III, IV and V would be expected to reflect weak, consistent relationships and zero relationships. Conditions VI and VII would be expected to reflect strongly inconsistent relationships. Earlier, in the Methods Chapter, it was stated that previous research had indicated certain factors as bearing a stronger relationship to injury than others. It follows that sub-factors which analysis showed as falling under Conditions I and II should more frequently be from the 16 sub- factors previously defined as strong. Sub-factors falling under Condi- tions III, IV and V would be expected to come primarily from the 23 sub- factors previously defined as weak. ''''el Table 9 summarizes analysis using the fatal vs. non-fatal injury criterion which revealed ten sub-factors strongly related to injury. Of the ten, nine had been predicted to be strong (90%). Analysis also showed that 29 sub-factors had a weak or zero relationship to injury. Twenty-two of the 29 (76%) had been predicted to be weak. Analysis re- vealed no inconsistent relationships and, of course, it was assumed previously that none was inconsistent. Table 9 Correspondence of Factor Strength with Predicted Strength Using Fatal vs. Non-Fatal Injury Measure Sub-Factor Defined Earlier as: Strong Weak Sub-Factor Emerged as: Strong (Conditions I & IT) 9 4. Weak (Conditions III, 7 22 Iv, & V) Contradictory (Conditions VI & VII) 0 0 Table 10 shows the same arrangement based on use of the dangerous- fatal vs. remainder injury criterion. This analysis revealed that twelve sub-factors had a strong relationship to injury. Of the 12, eight (67%) had been correctly predicted to be strong. The analysis also indicated that 27 sub-factors had a weak or zero relationship to injury. Of the 27, 19 (70%) had been correctly predicted to be weak. No sub-factors turned out to be inconsistent and, of course, none had been so predicted. ''''Table 10 22 Correspondence of Factor Strength with Predicted Strength Using Dangerous-Fatal vs. Remainder Injury Measure Sub-Factor Emerged as: Sub-Factor Defined Earlier as: Strong Weak Strong (Conditions I & II) 8 4 Weak (Conditions III, Iv, & V) 8 19 Contradictory (Conditions VI & VII) 0 0 ''''23 CHAPTER 4 DISCUSSION AND CONCLUSION Earlier it was stated that ACIR data could be regarded as con- sistent and adequate for general use only if two stipulations were met. The data will be discussed in terms of how well the stipulations are met. Stipulation 1: The great majority of factors must fall under conditions I-V and few may fall under Conditions VI or VII. The Results show that all of the 39 sub-factors tested against each of the injury criteria fell under Conditions I-V. No sub-factor fell under Conditions VI or VII. Thus, there was no instance of a significant inconsistency. Stipulation 2: Sub-factors falling under Conditions I and II must come primarily from factors previously found to have a strong relationship to injury. Sub-factors falling under Conditions III, IV, and V must come primarily from factors previously found to have a weak relationship to injury. The Results show that most of the sub-factors which emerged as strong were those previously found to be strong. Most of the sub-factors which emerged as weak were those previously found to be weak. On the basis of this research, it is concluded that accident-injury relationships in ACIR data are consistent from state to state and that ACIR findings have general applicability. ''''II. til. VI. VII. eh APPENDIX 1 LIST OF 39 SUB-FACTORS Applicable Impact Speed (20mph categories from 0 to 60+) ls 0-19) vs. (60+) 2. (20-39) vs. (40-59) 3. (20-59) vs. ( 0-19) + (60+) Area of Severest Impact (compartment, front, fender, rear and rollover) - compartment vs. front rear vs. any fender rear, any fender vs. front, compartment 7. front, compartment, rear, any fender vs. rollover 4 2 6 Accident Severity (1) Minor, (2) Moderate, (3) Moderately Severe, (4) Severe, (5) Extremely Severe + Extreme. 8. minor (1) vs. extremely severe + extreme (5) 9. moderate (2) vs. severe (4) 10. minor (1) + extremely severe + extreme (5) vs. moderate (2) + severe (4) 11. minor (1) + moderate (2) + severe (4) extremely severe + extreme (5) vs. moderately severe (3) Seated Position: (1) driver alone, (2) driver with passengers, (3) center front - CF, (4) right front - RF, (5) rear. 12. driver with passenger(s) vs. RF 13. CF vs. driver with passenger + RF 14. rear vs. driver with passenger + CF + RF 15. driver with passenger + CF + RF + rear vs. driver alone Ejection (ejected - not ejected) 16. not ejected vs. ejected Occupant Age: (1) up to 14, (2) 15-19, (3) 20-24, (4) 25-34, (5) 35-4k, (6) 45-54, (7) 55-64, (8) 65 and over. 17. 65+ vs. 55-64 18, 45-54 vs. 35-44 19. 25-34 vs. 20-2) 20. 15-19 vs. up to 14 21. 55-65+ vs. to 19 22. 35-54 vs. 20-34 23. (to 19) + (55+) vs. 20-54 Sex of Occupants (male and female) eh. female vs. male ''''VIII. IX. X. 25 Height of Occupants: (1) up to 36 inches, (2) 37-48, (3) 49-60, 25, 26. 27. 28. 29, (4) 61-66, (5) 67-72, (6) 73 and more. 49-60 vs. 73+ 61-66 vs. 67-72 (49-60) + (73+) vs. 61-72 37-48 vs. up to 36 up to 48 vs. 4O+- Weight of Occupants: (1) up to 99 pounds, (2) 100-124, 30. 31. 32. 33. Year of 34. 35. 36. 37. 38. 39. (3) 125-149, (4) 150-174, (5) 175 and more. up to 99 vs. 100-124 150-174 vs. 175 and more up to 124 vs. 150 and more (up to 124) + (150 and more) vs. 125-149 Manufacture: (1) pre-1952, (2) 1952, (3) 1953, (4) 1954, (5) 1955, (6) 1956, (7) 1957. 1952 vs. 1954 1953 vs. 1952 + 1954 pre-1952 vs. 1952 + 1953 + 1954 1955 vs. 1956 1955 + 1956 vs. 1957 through 1954 vs. 1955 and later ''''26 APPENDIX 2 LIST OF SEVEN CONDITIONS Condition I: The situation in which all the accident-injury rela- tionships are significant and in the same direction for all four states. Data falling under this condition support the hypothesis of consistency of accident injury relationships. This is logical because if a given relationship is present in one state, it should be present in the others. Condition II: The situation in which some (one or more, but not all) of the relationships are significant, and all are in the same direc- tion. Data falling under this condition support the hypothesis. This follows the same logic as Condition I, although here not all relation- ships have the same strength. Condition III: The situation in which the accident-injury rela- tionship is not significant in any of the four states, but the apparent direction of relationship is the same for all four. Data falling under this condition support the hypothesis of consistency. This is logical because if a given variable shows a trend (though so weak as not to be significant) in one state, then the same should hold true in the others. Condition IV: The situation in which none of the relationships is significant and one or two are opposite to the others in direction. Data falling under this condition support the hypothesis. Thus, if there is a complete absence of a relationship, test results should fluctuate randomly around zero. Condition V: The situation in which one or more (but not all) relationships are significant, and one or two are opposite in direction to the others. Data falling under this condition are consistent with the hypothesis. This Condition is similar in logic to Condition IV, except that there is somewhat more variation in the results. This Condition does not include the situation of two or more significant relationships which are opposite to each other. Condition V might, for example, include such as the following: State Direction Magnitude J + Not significant 2 Significant 3 - Not significant 4 - Not significant Condition VI: The situation in which at least two results are significant and are in opposite directions. Data falling under this Condition contradict the hypothesis, because they would indicate that an accident circumstance which produced significantly greater injury in one state, produced significantly less injury in another. ''r ''eT Condition VII: The situation in which all four results are sig- nificant and one or two are opposite to the others in direction. Data falling under this Condition would contradict the hypothesis for the game reason as given for Condition VI. ''''28 APPENDIX 3 STRONG AND WEAK FACTOR DISTRIBUTION In the text, seven Conditions are listed and each sub-factor is described as falling under one or the other of the Conditions. A discus- sion is desirable in order to illustrate the manner in which the sub- factors might be expected to be distributed smong the Conditions. This will be done by presenting three hypothetical situations. 1. First consideration is given to an hypothetical situation in which there is a large number of sub-factors, each of which bears a close relationship to injury, and all of which have the same direction of asso~ ciation, regardless of the state in which the sample was gathered. If moderate sized samples are used, a test such as Chi square would be quite powerful and the probability of detecting the relationship would be high (i.e., the probability of obtaining a significant result). The proba- bility of a Type II error would, correspondingly, be quite low.* If such a situation exists and a given sub-factor is tested in four states, one of the two most likely outcomes would be Condition I (all four significant and in the same direction). If the process were repeated many times, sometimes the relation- ship in one or more states would not be significant (Type II errors). This would result in Condition II (some states significant and some not, but all in the same direction). If none of the four states were sig- nificant, the sub-factor would fall under Condition III but the probabil- ity of a Type II error in all four states would be very small. The probability of having one or two states opposite to the others in terms of direction of association would also be quite small. Thus, the chances of having a sub-factor to fall under Conditions III, IV, V, VI or VII would be negligible. When relationships are strong and consistent, a concentration of sub-factors under Conditions I and II would be expected. The exact division would be a matter of "how strong is strong?" At a higher level of strength, almost all would fall under I and few under II. Ata somewhat lower level of strength, there would be more under II than under I (but still virtually none under any of the other Conditions). A computation illustration of this situation is included at the end of this Appendix. 2, Second consideration is given to an hypothetical situation in which there is a large number of sub-factors, each of which has no *A Type II error is the probability of rejecting the alternate hypothe -~ sis when it is true. Most statistics books have a discussion of Type I and II errors. ''''29 relationship to injury, and this null situation is existent in all four states. If repeated tests were performed, the results would usually be non-significant. If the one percent level is selected (i.e., the risk of a Type I error is set at .01), then one time in 100 an apparently significant test would be obtained. Also, in such a null situation, samples would fluctuate randomly around zero; thus, the apparent direc- tion of association would be equally often positive and negative. If the four states were compared repeatedly under these circum- stances, the most typical outcome would be Condition IV (no tests sig- nificant, and some positive and some negative in direction). About one time in eight the direction of all four would be the same (just as all of 4 coins would be heads or tails once in eight tosses), and the result would be Condition III. Rarely, one of the tests would show up signifi- cant (a Type I error) and then the result would be Condition V or II depending on whether or not the states happened to be in the same direc- tion. Condition V would be rare but Condition II would be extremely rare because, not only would a test have to be significant, but simultan- eously all four would have to be in the same direction. Thus, when relationships are null, a concentration of sub-factors would be expected under Condition IV. A smaller number would appear under III, and rarely a finding would appear under V or II. Note that there is virtually no overlap in the Conditions under which sub-factors would fall if there is a strong relationship as com- pared to a null relationship. A computational illustration of this situation is included at the end of this Appendix. 3. Third consideration is given to an hypothetical situation in which there is a large number of sub-factors, each of which bears a weak relationship to injury and all of which have the same direction of asso- ciation regardless of the state in which the sample was gathered. In this situation, the power of the tests would be lower (i.e., the Type II error would be made more often), and tests would show up insignificant more often than in the first illustration. The consistency of direction of the weak relationship would, however, tend to result in all four states agreeing in apparent direction. There would, nevertheless, be some occasions of disagreement in direction. In this as opposed to the null situation, there would be fewer instances of Condition IV and more of Condition III. That is, there would still be many instances in which none of the tests was significant, but in the weak situation they would tend to be in the same direction, whereas in the null situation they would tend to fluctuate randomly in direction. Thus, in the situation of weak but consistent relationships, there would be a concentration under III, some under II, IV, and V. A computational illustration is included at the end of this Appendix. ''''30 Since the data on hand actually contain a mixture of strong and weak variables, it would be expected that findings would be scattered under all five Conditions. The exact allocation is impossible to pre- a@ict because it depends on the number of strong and weak factors as well as the magnitude of the strength or lack of strength. It would be expected, however, that those factors emerging as strong should come largely from those defined previously as strong. The same should be true for the weak factors. This is discussed in the Results section of Part II of the paper. '' t. ''ae 31 Computational Illustrations ‘4 Strong Sub-Factors Given: A large number of accident-injury relationships which are strong in all four states and consistent in direction. The power of the test is .75 (i. e., the probability of detecting a relationship). The probability of a test indicating the wrong direction of association a- proaches zero. There are 16 ways in which the results of the four states could come out with respect to significance (binomial theorem). Let a "plus" indi- cate that the state showed a significant relationship (for this example, the probability of a “plus" is .75). Let a "minus" indicate that the state showed no relationship (probability .25). State 1 2 3 4 + + + + All states Pr aaye = + + + - probability: (.75) 2316 + + - + + + - - + “ + + + - + - + ~ - + + - - - Some states significant - + + + probability: 4(. 75)3(. 2 = .4ee pS: CBP - + - + 75 025 )I= .O4 e+ & & 680 - - + + - - + - - - - + No states edyais ieee - - - - probability: (.25)4* = .ook The above model gives the probabilities for various Conditions as far as significance is concerned. As for direction of association, in this example, the probability of the correct direction is 1.00. There- fore: ''tT. ''32 Probability of: Joint Condition Direction Significance Probability I all sig. 1.00 (.75)4 = .316 132 same dir. II some sig. 1.00 -680 .68 same dir. III none sig. 1.00 (.25)4 = .ook .00 same dir. IV none sig. «OO 00 some opp. Vv none sig. 00 00 some opp. VI some sig. 00 00 some opp. VII all sig. 00 00 some opp. 1.00 Thus, for the situation of strong relationships, a concentration would be expected under I and ITI. 2. Null Relationships Given: No accident-injury relationship in any state. Type I error set at .01 (probability of obtaining an apparently significant result). Probability of a positive direction of relationship is .50 (the proba- bility of all four in the same direction would be 2(.0625) = .125 -- that is, the sum of the probabilities of all being positive and all being negative). Using the same 16 possible combinations as shown in the first exam- ple, but substituting the new values (.99 and .01) gives the following: probability of all being significant (.01)4 = —>.00 probability of some being significant 4(.01)3(.99) = —».00 — 6(.01)*(.99)2= .0006 4(.o1) (.99)3 -0388 -039 probability of none being significant (.99)4# = -961 Now, adding the probabilities concerning direction, the Conditions may be specified: '' i ''33 Probability of: Joint Condition Direction Significance Probability I All sig. £125 (.o1)* = 0 .00 all same II Some sig. wie 039 00 all same III None sig. 6125 (.99)4 = .961 .12 all same IV None sig. 875 961 By some opp. xv Some sig. »875 -03 : some Opp. .034% ¥VI Some sig. -875 -00 . some opp. VII All sig. 875 0 00 some opp. Thus, for the null condition, most sub-factors would be under Condition IV. A few would be under III and a negligible number would be under V. 3. Weak Relationships Given: A large number of weak accident-injury relationships, which are, nevertheless, consistent in direction of association. Power of test is .25. Probability of a test indicating the correct direction is .90. (The probability of all four Lee sag the correct direction and, there- fore, the same direction is (.90)* = .6561. To this must be added the probability of all being in the incorrect direction because these too would be in the same direction (.10)* = .0001. Sum = .6562) *¥Note Conditions V and VI are identical as far as their binomial proper- -ties are concerned. That is, they both are characterized by some sig- nificant tests and some not, and they both have some opposite in direc- tion. There is, however, an important distinction in terms of the meaning of the data. That is, Condition VI contains the additional proviso that two of the significant tests must be opposite in direction.. (Condition V, on the other hand, requires that all significant tests be consistent in direction.) A separate evaluation was made to partial out the exact probabilities of Conditions V and VI. This was done by evaluating, in detail, two binomials: one corresponding to significance (.O01 and .99), and the other corresponding to direction of association (.50 and -50). ''ty ''34 The evaluation of significance combinations is as follows: (.25)4 = .0039 probability of all significant probability of probability of some significant none significant 4(.25)3 4(.25) (.75) = 6(.25)°( 275 )e= (.75)3= +2109 4219 .6796 (.75)4 = .3164 Adding the factor of direction of association results in the fol- lowing: Condition I All sig. same dir. II Some sig. same dir. III None sig. same dir. IV None sig. some opp. *V Some sig. some opp. *VI Some sig. some opp. VII All sig. some opp. Direction 66 66 Probability of: Significance —» .00 -680 +320 032 —> .00 Joint Probability 00 045 eel ell 18 we > .05 -00 In the weak situation, there is a concentration under II, with cases under III, IV, and V and a negligible number under VI. The following table summarizes the expected distribution according to the three models illustrated: Condition Strong I 032 II .68 III 00 IV 00 Vv 00 VI -00 VII 00 Null 00 00 12 8h 03 00 00 Weak -00 45 eel ell 18 205 -00 *Note: The probability of V and VI had to be determined separately, as - before. ''''APPENDIX ha SUMMARY OF FINDINGS FATAL DANGEROUS FATAL Arizona California No. Carolina Texas Arizona California No. Carolina Texas Magni Magni- Magni- Magni- Magni- Magni: Magni- Magni- Factors tude Direc. tude Dir. tude Dir. tude Dir. Condition tude Dir. tude Dir. tude Dir. tude Dir. Condition de NS + Ss + NS + NS + 2 NS + Ss + Ss + NS + 2 Impact Speed 2. NS + Ss + Ss + NS + 2 NS + NS + NS + NS + 3 36 Ss + NS + Ss + Ss + 2 Ss + s + Ss + s + a: 4h. NS + NS - NS + NS - 4 NS + Ss - NS + NS - 5 Area of be NS + NS oO NS 0 NS + 3 NS + NS - NS + NS + 4 Severest Impact 6. NS + NS + NS + s + 2 NS + s + NS + NS + 2 Te NS + NS + NS + NS 0 3 Ss + S$ + NS + NS + 2 8. S + s + s + s + 1 s + s + s + s + 1 Accident 9. S + iS} + s + s + A Ss + s + s + Ss + 2 Severity 10. NS - NS + NS - NS + 4 NS - s + NS - NS + 5 ll. s - NS - NS - NS - 2 NS - NS + NS + NS + 4 12. NS + NS + NS + NS + 3 NS + NS + NS + NS + 3 Seated 13. NS + NS + NS + NS + 3 NS + NS + NS + NS + 3 Position 14. NS + NS + NS + NS + 3 NS + NS + NS + NS + 3 15. NS + Ss + s + NS + 2 NS + NS + s + NS + 2 Ejection 16. s + 8 + s + _S + 1 s + s + s + s + A. ij. NS + NS + NS + NS + 3 NS - NS = NS + NS + y 18. Ss + NS - NS - NS + 5 Ss + NS - NS - NS + 5 Age of 19. NS + NS - NS - NS - 4 S + NS + NS - NS + 5 Occupants 20. NS + NS - NS - NS + y NS - 8 - NS - NS + 5 el. NS - NS - NS - NS - 3 S - Ss - NS - Ss - 2 22. NS - NS - NS - NS - 3 NS - NS - NS - NS - 3 236 NS + NS + NS + NS + 3 __NS + NS + s + NS - 5 Sex ek. NS + NS + NS + NS - 4 NS + NS + NS + NS + 3 256 NS + NS + NS + NS 0 3 NS + NS + Ss + NS + 2 Height of 26. NS + NS + NS + NS + 3 s + NS + NS + NS + 2 Occupants 27. NS + NS + NS + NS + 3 NS + NS + NS + NS + 3 28. NS - NS 0 NS - NS + 4 NS - NS 0 NS + NS + 4 29- NS + NS + NS + NS + 3 _NS + NS + NS + NS + 3 30. NS - NS + NS - NS - h NS + NS + NS + NS - 4 Weight of 31. NS + NS - NS + NS + 4 NS + NS - NS + NS + 4 Occupants 32. NS + NS + NS + NS + 3 s + NS + s + s + 2 33. NS - NS - _NS - NS + 4 NS - NS - _S - NS + 4 3h. NS + NS 0 NS + NS + 3 NS - NS - NS + NS + k Year of 35-6 NS - NS - NS + NS + 4 NS - NS + NS + NS + 4 Mfg. of Car 36. s + NS + NS + NS + 2 s + NS + NS - NS + 5 37. NS - NS + NS + NS + 4 NS + NS - NS + NS - 4 38. s - NS + NS - NS + 5 NS - NS + NS - NS + 4 39. NS - NS + NS - NS + 4 NS - NS - NS + NS + 4 Se '' ''15. 36. 5. Te 12. 13 e 14. 17. 21. 22. 23. 25. 26. 27. 29. 32. 3h. k, 10. 19. 20. 2h. 28. 30. 31. 33. 356 37° 396 18. 38. APPENDIX hb FATAL Factors Accident Severity: minor vs. extremely severe + extreme moderate vs. severe Ejection: not ejected vs. ejected Applicable Impact Speed 0-19) vs. (60+) a vs. (40-59) 20-59) vs. (0-19) + (60+) Area of Severest Impact rear, any fender vs. front, compartment Accident Severity minor + moderate + severe + extremely severe + extreme vs. moderately severe Seated Position driver with passenger + CF + RF + rear vs. driver alone Year of Manufacture pre-1952 vs. 1952 + 1953 + 1954 Area of Severest Impact rear vs. any fender front, compartment, rear, any fender vs. rollover Seated Position: driver with passenger(s) vs. RF CF vs. driver with passenger + RF rear vs. driver with passenger + CF + RF Occupant Age 65+ vs. 55-64 55-65+ vs. to 19 35-54 vs. 20-34 (to 19) + (55+) vs. 20-54 Height of Occupants 49-60 vs. 73+ 61-66 vs. 67-72 (49-60) + (73+) vs. 61-72 up to 48 vs. 4O+ Weight of Occupants up to 124 vs. 150 and more Year of Manufacture 1952 vs. 1954 Area of Severest Impact Compartment vs. front Accident Severity minor + extremely severe + extreme vs. moderate + severe Oceupant Age 25-34 vs. 20-2) 15-19 vs. up to 14 Sex of Occupants female vs. male Height of Occupants 37-48 vs. up to 36 Weight of Occupants up to 99 vs. 100-124 150-174 vs. 175 and more (up to 124) + (150 and more) vs. 125-149 Year of Manufacture 1953 vs. 1952 + 1954 1955 vs. 1956 through 1954 vs. 1955 and later Occupant Age 45-54 vs. 35-4) Year of Manufacture 1955 + 1956 vs. 1957 36 Condition ''vo ''3. 8. 16. 6. Te 15. 21. 25. 26. 32. 12% L's 14, 22. ak, 27. 29. De ll. 17. 28. 30. 31. 33. 3h. 35-6 37 38. 39. h, 10. 18. 20. 23. 36. APPENDIX he DANGEROUS + Factors Applicable Impact Speed: (20-59) vs. (0-19) + (60+) Accident Severity: minor vs. extremely severe + extreme moderate vs. severe Ejection: not ejected vs. ejected Applicable Impact Speed: (0-19) vs. (60+) Area of Severest Impact: rear, any fender vs. front, compartment front, compartment, rear, any fender vs. rollover Seated Position: FATAL driver with passenger + CF + RF + rear vs. driver alone Occupant Age: 55-65+ vs. to 19 Height of Occupants: 49-60 vs. 73+ 61-66 vs. 67-72 up to 124 vs. 150 and more Applicable Impact Speed: (20-39) vs. (40-59) Seated Position driver with passenger vs. RF CF vs. driver with passenger + RF rear vs. driver with passenger + CF + RF Occupant Age: 35-54 vs. 20-34 Sex of Occupants: female vs. male Height of Occupants (49-60) + (73+) vs. 61-72 up to 48 vs. 49+ Area of Severest Impact: rear vs. any fender Accident Severity: minor + moderate + severe + extremely severe + extreme vs. moderately severe Occupant Age: 65+ vs. 55-6 Height of Occupants 37-48 vs. up to 36 Weight of Occupants: up to 99 vs. 100-124 150-174 vs. 175 and more (up to 124) + (150 and more) vs. 125-149 Year of Manufacture: 1952 vs. 1954 1953 vs. 1952 + 1954 1955 vs. 1956 1955 + 1956 vs. 1957 through 1954 vs. 1955 and later Area of severest Impact compartment vs. front Accident Severity minor + extremely severe + extreme vs. moderate + severe Occupant Age 45-54 vs. 35-4h 25-34 vs. 20-2) 15-19 vs. up to 14 (to 19) + (55+) vs. 20-54 Year of Manufacture pre-1952 vs. 1952 + 1953 + 1954 37 Condition ''''APPENDIX 5 FREQUENCIES ON WHICH 312 TESTS WERE BASED 38 ''''aRoctOog 43 = Eure 10 a * mo +B ct f FRPP FWNH FWP FReRYP FUNE FRePH FWRH FURH FUR FATAL VS. ALL OTHERS Fatal Total Injuries None-F atal Top Bottom Top Bottom ._Row Row Row Row Q 2 25 = 165 0. $ 78 90 Oo 8 816 3h 163 0 20 8 138 1 11 167 307 i, —e- 36 278 oO 18 27). . ),38 - Zz 11 185 26h 12 48623) 7h S190 15 9 62h 168 18 461606712 ~—Sss«197 14 20g 146 2 15 81 = 282. 10 8 188 399 3 19 119 403 12 19 Uy1 276 a 2 50 35 0 0 56 hig OQ 0 LS 37 0 1 58 39 2 17 85 363 0 18 105 587 O 22 82 522 1 31 97 417 19 22 Ls 290 18 11 692 213 22 1 60h 350 32 9 Sih Uh 0 7 4 27 1 3 186 19 0 6 38 27 2 nM 90 16 2 2h 211 153 Q 27 358 8h 1 17 357 150 lL 2 266 126 7 26 59 36h h 617 205 2: 6 18 65 507 6 25 106 392. xe 2.8 6.) 2oh © 28 4eO 929 1.6 2320 320 11.9 21.0 hed Fatal» Total Dangerous Non-Fatal Top Bottom Top Bottom p Row Row Row Row - 2 4 25 165 02-01 1 23 78 90. - OO 33 34, 163 @° OO... 32... 8-138. | ~ 6 20 £4367 307 05-02 21 26 346 278 eOL 19 8647 27, 438 - 9 23 185 26) e0L 260) 3 47h 190 - 7 2; 62h 168 201 66 33 712 197 e0l 32 ai hug 146 - kh 2 81 6.282 = 25 29 188 399 #- il1 39 119 8403 - 20 30 Wl 276 ‘= Q 3 500s 35 “ L Q 56 9 - 1 3 4537 ~ 3 58 3839 - 3 32 85 363 - 1 5h 105 587 -~ 4 SO 82 522 05-02 5 50 97 17 - 35 he bug 290 - 55 28 692 213 - 54 6 £604 350 - 55 18 5, bh eOL 1 . 8 32 27 Ol 2 6 186 19 e0l 1 10 38 27 Ol 3 7 90 16 el kh YO aL 153 e001 IL 33 358 8h, 0113 35 357 150 el h 36 26 126 - 9 Ih 59 = 36h - 8 kh 205 dhe - lm 28 65 507 - 10 O £106. 392 FATAL-DANG. VS. ALL OTHERS x2 206 182 G9 lel 1.3 1.9 205 129 41.0 66 Bel 2107 lel de? : 20 o7 203 ee) 05 02 220 702 bad 302 Tel he? 307 6.0 3520 11.0 2125 46.8 95.05 5 oh 6e2 207 05-02 02-01 201 e201 eO1 O01 201 e01 OL 02-01 '' = aE Pr =e Se oer ee ee a tp eS ope oe ea Ea Fo rage > _ a a a ea IN ITS RET EEE Le TO JA". By" Sunt Ta iscot wfesat Were te ee ae ee oe fetsTenot evoreaied - % . motion get totto® gol | a 7 - OS. das as 0 gh. —" fOer~ S,0f. 00 7 ay fo" o£ oo. fa "Ye: fe 9 é me ~ of BEF . 8° fo 8 ~ fof for §ar ; 4 - Gel’ BY2 .oue oS 8 - te Cab MPs Via cr - C.of © dds 282 fs Le ; Oe fst NeL “ : fORSQ 3.8 9 3d” We = £08 Yer ° Sey £Oe TLS Ofr Qil - -—3 or... dof 88S Bk S020 Sed ORE, BBE KS aS » Ox, POs er ef EL - Te VS fr oe as GS .e€ Of ¢ fr CG * : * 1 coe) TE os PL L. oc tm OVW FS PH co P a) A mn ? Tt or 5 + te Ta =. 5 } § SF t £05 O42 RE ons 3. 5 [0s Ga it 1S BE OL J mo S 7 ~~ : Geol Get xs tate TDs, ject Ro Sen gE ic Me eae) Off Pit Be ef IC, i. Pid Age ass OE foes 39 Ski gos iif O.F oo. tah te iy "ba ‘ wor wok’ woil wok q Pie SO" — Ce ey PTO, *6 “ sai 433 - 8.8 Deal ° 24? * “ De ° - OL Ge re} « e QAHHTO WIA .2% JATAD & Lede i-srt0¥ gebrubaL mottod gol motsgeg got wofl. wou woh = wot we 2 wy z 2 fstoT isdst-:. @ z 9 af és ES = 0.9 Oe? ay. 8 Gi far He eft: BEF 5 OS 5 Baim & ike yes. Jor: £f i STS , fet . AY = Has: § g8L re F . ng sy > wf wih — ¥ — Of iT “§ of of B6f. 53 e ac 3¢ Jer SET, Oe BEE Oye’ Oh. GS ue: g S6S Lis eo 8 L. VEE Bot a OL 2: of Off. et €&.€ Ags’ opts er Sig ue 243 g To" XE Gua on Ne “35 > “. £30 28 VE y c . “9 / <, ” ; A aor of 3 4S ae: se ‘ss’ oOo. € Crit ¥ ‘ + ote fe. om e OLS Ade 8s er f Res” 8a rE er $0 Oct (02 alr $3e = fit ARS § Seo * Vs Sf. . o of a aur € £8 S68 &. O &€ GL. 5¢@ yh" $ u, ef ms ue os £ shu. Bk, Yi 3 $ Gar. = FE ia Se osy Jd o £ of pos 23 3S r JZ Slo BGs v4 & § Wot Fe at o €£ See cage 2g oo a wf MPA “ee ''FATAL VS. ALL OTHERS Fatal Total Injuries None-Fatal Top Bottom Top Bottom Row Row Row Row 423, 25 " @ crm cH i e@HOoctaQ gw 1 7 2 22 7 647 23h 3 & u 572. 379 h» 31 8 498 17 1221 13 12 210 18) 2 9 331 25 3 8 8 270 8=2h8 h 13 12 185 157 i 2 2 63 39h 2 1 20 67 576 3 1 16 89 518 4 1 25 50 632 wl 3 = 27 138 457 2 L 2h 135 643 3 3 17 177. +607 h 3 2 113 392 151 30 8 595 101 2 22 7 778 = 96 3 20 13 78h yh hk 29 7 505 115 161 11 27 580 158 2 12 HN 780 137 3 20 16 788 171 h 22 19 554 106 171 862 ~—(CO4& 23.043 2 1 5 i? 71 3 OO 3 21 22 h uw 6 2335 wi 0 12 69 103 2 5 3 75 #115 3 k 8 53 123 ho 2 8 60 85 191 7 9 157 132 & 6 4 9 130 3 9 #5 211 208 h 8 6 125 107 20 1 2 3 119 76 2 h 1 172 «112 3 5 2 20) 99 4 h 6 102 #110 x2 Pp 02-01 seang arse peur re FATAL=DANG. VS. AL% OTHERS Fatal- Total Dangerous Non-F atal Top Bottom Top Bottom Row Row Row Row x2 53 #19 423 25 3 52 29 647 23k 30h 59 «= 572 379 oS 50 22 498 =-1h7 2.3 23 «22 210 18) 20 30 ©6330 331 =2h5 le2 23-28 270 386.28 «8 2L 21 185 157 2 5 4s 63 39h ol 3 60 67 576 1.8 3 Sl 89 518 302 1 2 50 = 3h2 37 9 50 138 ©6457 1.8 9 63 135 643 1.0 2 5h 177 607 5 8 43 113392 lel 59 15 595 10l le? 72 «+12 778 96 e7 66 33 78 ly 2563 51 16 505 115 1.0 31 he 580 158 605 47-37 780 137 5902 58 2 788 171 4207 hl 32 554-106) he? h 8 23.43 Pe) 9 9 7 71 5 3 21 8622 «2 10 23 35 lel 1 3 h 20 69 103 503 1 12 75 #125 ek 10 16 53 123 6 5 12 60 85 06 8 18 157 132 Se 12 «lb 149 =130 03 30 «25 211 208 3 ll UW 125 107 o7 9 3 119 76 5 16060 1 i72 6i@ Tel 15 3 20), 99 15 9 10 102 110 & OO» oh eee! oe 'S 05-02 ''£06 rrr £0» SD eo ie €e8S LsdoT Late Tea it. meFtod wosl ass HES Ayo O28 LOL ae Ctr° > . IW rai (4 te fv a. SV¥iRS OD AU Es eo sei OFF yor a Sit ee OIL = tr fe, qoT motsod wos wok £Sii OL V1 a” es sv2 is Ae1) SS "Ose 6 6Ss ll. fee Oz, OYS ¢ 88° ehi £s ES ay tO 0 638 id od oy BER. ce. efse ga Tra gid Cif £4 22 nw 2 “TA. € a x i _ a, Sf£* (oa: ngs Hh pets | fe e0e 9 ~~ oa ‘ Ss a OUy. re. ar oe foe ae oe att 325 OG Yue « £o e * Ses -¢ OF ei £ 63 0g aah as be CA. ery Cikde & SVL £ e fos SOL. OL UATATATAT ~fatea: evotegns(. qo wo Im™ uns ~ o ka Qu fv b3 +9 b% mmr ns mMmMorm 226s: i ia<¢4 Sy A Sea Ue Ve 0 * « ys t~ “ Wns cy => *.8 . ON Taye ar —5 2 t. * >" Bs , . anmiTd its sev Atay dgto? Lage aiiowt got” inode - motsod wor les: Wye. gilt eae ere ese 25 ser Ger 80S TOL Sit te. oft ate WOs: €Su yua Sf C a 2s. outt ors Set See ak, GEE SYL LOS SOL fotdh net wo ” , \ we ful Sw CR.GO Fee LD fen Oe be py Oirtiied. thi “ got’ yon" eg: co os 67 EU LP: tes 2 fs “> . Cor se: te.” TP tO be eM Be. wes mM 1 In © cir lors tain SPA LSE: TPO. Tamia er mmvr af . i ‘ay > al: FL Sr ¢ ''8 OH OctOaO gS A Octem tM ny i) ™ nm —~ ND ~ — @ SJ On Wr = WwW FwWNRHP FUNKE FWNH FWNP FWP FWNP FWNHY FWPH FWNHP FUpH nN XO w oO * FATAL WS. ALL OTHERS Fatal Total Injuries None«F atal Top Bottom Top Bottom Row Row Row Row 6 5 66 195 6 5 118 28) 3 7 43 303 7 10 58 212 12 16 172 289 8 10 190 279 12 Uy 176 19 10 14 1y5 232 11 28 261 61 1. 3=«(18 402 69 10 26 36 595 17 2h 270 377 6 35 214 520 10 19 343 50 6 30 2h6 696 15 26 232 15 0 3 28 8628 0 2 52 6 0 1 27) ok 0. 0 31 8632 8 27 205 383 9 16 317 391 5 26 183 515 13 23 197 318 3 35 56 588 2 25 98 708 pi 31 71 698 0 36 63 515 1 1 ly, 27 0 0 20 «25 L i 2h = 33 iL 2 31 27 2 38 hl 6h O 27 45 806 2 32 57 769 a 36 58 578 2 2 56 «89 0 77 140 2 i 86 100 h 3 83 85 S FATAL-DANG. VS ALL OTHERS Fatale Total Dangerous Non-Fatal Top Bottom Top Bottom Row Row Row Row 12 12 66 195 18 17 118 = 28 h 18 43 303 13 19 58 «212 2h 26 172 289 23 26 190 279 26 55 176 =19 17 25 W5 232 2h 50 261 461 35 uo 02 69 220681595 32 2 270 377 16 61 214 520 29 55 343 5h0 21 ~2«~=-82 246 696 21 «453 232 = 1 3 28 28 1 6 52 6 0 8 27 Lh 0 h 31 32 14 5h 205 383 31 hl 317 391 Ly 68 183 515 20 hh 197 318 h 68 56 6588 7 72 98 708 8 82 71 698 4 6h 63 515 1 i ly 27 0 0 20 25 1 2 2h 33 1 3 31 27 2 72 ya 644 0 be 45 806 3 90 57 «769 4 68 58 578 2 h 56 89 1 12 77 140 h 5 86 100 6 in 83 85 x2 Tee 79 o3 667 203 7 ol 20 3 06 liel 20 205 05 1.6 1.7 03 3.0 309 2s3 Pp ''eee Ory tee ~ ¢ 1) EG~8G, -” — ~~ PMMHTO Lie BV -DMACMATAY efstad 0. he Le fadoT Lets G0 modtod Wood eer {5s ” £0£ S£s 8S ess +. RL ~ ae c. Ede 3 worl ad BLS. a vf “ORE oy alt AO L 2 ei abe OTs des Cc yd es NS fe! aos o TLE. fn es Tel. ‘ Smit ta Al (ut. & —1 Sy 0) 0-4. 7u Q’ GC wos “4&1 w Friis Rain cy ray > ’ Wstv fui Mf Ot. bs iu esorsaned got motsod got wo ? ye . Sy Yat fos Oak 2 . Ws Oweo 1M Oo. cmmano wing ae > « am © a“ _ « ''''ES2h9Ebe0? a Aa tayea8 ON ''