NCJ Number
53157
Journal
Police Research Bulletin Issue: 31 Dated: (AUTUMN 1978) Pages: 23-29
Date Published
1978
Length
7 pages
Annotation
THE PROCEDURE FOR ANALYZING AND PRESENTING STATISTICS ON VEHICULAR ACCIDENTS FROM MONTHLY STATISTICS IS DESCRIBED AND ILLUSTRATED.
Abstract
TO ILLUSTRATE THE PROCEDURE OF ANALYSIS AND PRESENTATION, DATA ARE USED ON NONINJURY ROAD ACCIDENTS OCCURRING EACH MONTH IN DERBYSHIRE (ENGLAND) FROM 1971 TO 1976. THE STATISTICS VARY CONSIDERABLY FROM MONTH TO MONTH, WITH THE PEAK ACCIDENT PERIOD OCCURRING CONSISTENTLY AT THE END OF THE YEAR. IT IS ALSO OBSERVED THAT THE GENERAL LEVEL OF ACCIDENT FREQUENCY ROSE DURING 1971 AND 1972, BUT DID NOT CHANGE APPRECIABLY THEREAFTER. IT IS CONCLUDED FROM THESE OBSERVATIONS THAT EACH MONTHLY TOTAL IS THE COMBINED RESULT OF THREE INFLUENCES: (1) THE GENERAL LEVEL OF ACCIDENT FREQUENCY, OR TREND; (2) SEASONAL VARIATION; AND (3) RANDOM VARIATION, MEANING ANY DEVIATION NOT ACCOUNTED FOR IN THE OTHER TWO INFLUENCES. THE PROCEDURE FOR SEPARATING AND MEASURING THE INDEPENDENT EFFECTS OF TREND AND SEASONAL VARIATION ARE INDICATED, WITH THE PURPOSE OF MINIMIZING, BUT STILL MEASURING, THE EXTENT OF RANDOM VARIATION. IT IS ADVISED THAT TREND IS BEST ILLUSTRATED BY THE MOVING ANNUAL TOTAL, THAT IS, THE TOTAL FOR THE LAST 12 MONTHS, RECALCULATED EACH MONTH. HAVING ESTABLISHED THE TREND THE SEASONAL VARIATION CAN BE MEASURED BY MAKING AN ESTIMATE OF THE EXPECTED NUMBER OF INCIDENTS FOR A PARTICULAR MONTH, ASSUMING THERE WAS NO MONTHLY VARIATION, AND COMPARING THIS WITH THE ACTUAL FIGURE RECORDED. THE EXTENT OF RANDOM VARIATION IS MEASURED TO DETERMINE IF THE CALCULATED PATTERN GENUINELY REFLECTS A REAL UNDERLYING PATTERN OR IS THE RESULT OF RANDOM CHANCE. THE MEASURING OF RANDOM VARIATION ALSO INDICATES HOW CLOSELY THE MEAN MONTHLY FACTORS ARE LIKELY TO CORRESPOND TO THE TRUE PATTERN, ASSUMING THIS CAN BE MEASURED. GRAPHS TO BE USED IN THE PRESENTATION OF THE STATISTICS ARE ILLUSTRATED. DATA ARE PROVIDED. (RCB)