Jointly published by Akadmiai Kiad, Budapest and Springer, Dordrecht

Scientometrics, Vol. 69, No. 2 (2006) 000000

Age likes some years

A case study for ages more prone to death
AHMED F. SIDDIQI
School of Business & Economics, University of Management & Technology, Lahore (Pakistan)

A person can die at any age. It is an omni-spoken common saying. Is it really true? Are all ages equally prone to die? Does there exist some predictable pattern that may conjecture the incidence of death? These are the questions that are attempted here in this article. Literature is replete with cohort dependant age distributions and pyramids that focus, and are adjusted, primarily for the living persons. The current article is using a cohort free group of people and focuses exclusively on age at death to rummage for some pattern in these ages. A statistical investigation is made of the life span of human beings of previous two centuries. The life span, or age, distribution is revealed to be a quadric modal in nature, refuting the prevailed myth that all ages are equally susceptible to death.

Introduction Every soul shall have a taste of death (Qur'aan [21.35]). These are the words, the Holy Quraan, divine revelation Muslims believe from God, uses to explain that every human being on the face of earth is to die. But when? At what age? Are all ages equally susceptible to death? If not, which are the ages that are more, or less, susceptible? Is it possible that a trend, or a probability distribution, can be observed for the ages, which can be used for prediction and forecasting for the incidence of death? These are the questions that are being asked and answered in this article.
Received January 12, 2006 Address for correspondence: AHMED F. SIDDIQI School of Business & Economics, 11-Aibak Block, Garden Town, Lahore 54700, Pakistan E-mail: ahmedfsiddiqi@gmail.com 01389130/US $ 20.00 Copyright 2006 Akadmiai Kiad, Budapest All rights reserved

A. F. SIDDIQI: Age likes some years

Human life and age has always been the center of research. How can it be made long? How can the process of aging be delayed? Research Literature is replete with all sorts of struggles to de-enigmatize these and related issues. But little is available to study human age with reference to some dangerous or safe years. Actuarial literature does present some probability-based models (CHARLESWORTH, 1980), but primarily all such models are cohort based (see COALE & DEMENY, 1983; NEWELL, 1988 for details) and then adjusted for living populations (see COX, 1972; KEYFITZ, 1985; LEE et al., 1988 for details). More often, these models are developed to see population growth, and growth rates, segmentation in population with respect to these growth rates, life expectancies, and other such characteristics of a group of people belong to same cohort. There also exists some die-hard assumptions (see COALE & DEMENY, 1983; LUTZ et al., 2004 for details), in addition to cohorts, for their justifiable application. All these nice models and insightful techniques are failing in answering the sole question of the article; Are all ages equally susceptible to death? If not, which are the ages that are more, or less, susceptible? Even the basic sources of research, like Encyclopedia of Britannica, or Encyclopedia of Americana do not have any direct, or even indirect, reference to this facet of human age. QURASHI & SHAH (1983) and QURASHI & SHAH (1984) however, made attempts to model the creative span, total life span, and month of expiry of early Muslim scientists and some eminent Pakistani scientists of the last hundred years. A similar analysis was made by QURASHI (1993) for eminent Pakistan scientists died between 1970 and 1992. Although, these attempts have serious statistical shortcomings, the very idea is taken from these references. The present author, in search of answers, conducted a thorough sample survey to investigate the relationship, if there exist any, between human age and occurrence of death. The survey where establishes the modular relationship, it also unveils more susceptible-to-death ages and contrives a trend for some rough prediction about death. The Hypothesis All ages are not equally vulnerable to death. This is quite a general statement. It has been established that infant mortality rates, child mortality rates and adult mortality rates are significantly different (VIERCK & HODGES, 2003) so the scope of this hypothesis is reduced to adult mortality only. A qualified hypothesis is thus stated as: all ages beyond 30 years are not equally vulnerable to death. Establishing Hypothesis A survey is conducted to gather data for an empirical testing of the hypothesis. Generality and scope of the hypothesis is asking for a very large volume of data which

Scientometrics 69 (2006)

A. F. SIDDIQI: Age likes some years

is limiting the chances of using primary data in limited resources. Even for a secondary data, accuracy and coverage are important issues soliciting for an intelligent sample design. Following are some of the dimensions of such a survey. Target population Very nature of the hypotheses is asking for a wide variety of people to be included in the investigation. I am including people from all the four corners of the globe, from all the cultures, and from all walks of life but limiting the list of people to celebrities only. This limit is used just for an ease in data accumulation. Another limit is on age that I am interested only in person older than 30 years. Further, I am interested only in natural deaths and not accidental/unnatural deaths. Such a limit is even more important here as rate of unnatural death is much higher among celebrities. Methodology A sample of 5000 celebrities, of previous two centuries, of five different professions, i.e. show-business (named category # 1 say C1), medicine (named category # 2 say C2), academics (named category # 3 say C3), science (named category # 4 say C4), and law (named category # 5 say C5) is taken from different sources like Encyclopedia Britannica (19931998), Encyclopedia Americana (1992, 1993, 19951998), Macmillan Encyclopedia (1998), Encarta Multimedia Encyclopedia (2004), Grolier Multimedia Encyclopedia (2000) and their dates of birth and death are recorded to study their ages. There should not be any ambiguity in defining these categories as these definitions are quite common sensual. However for sake of understanding, I am using these categories in the sense as has been defined in Encarta 2004. These celebrities are then classified into five classes, according to the five specified categories, such that each class has about 1000 celebrities. 26 strata are made, for each of the 5 classes, according to 26 alphabetic characters (A through Z). A simple random sample is, then, selected from each stratum of a size proportional to the available celebrities in the stratum to have a sample of size 60 for the particular class. Such a peculiar scheme is used here to free the selected sample from any sort of geographical and time bias that may exist in the accumulated data. SPlus is used to draw these simple random samples. Thus, five proportional stratified random samples of 60 celebrities each, of previous two centuries and of five different categories are the result. It is thought that the results on the celebrities can be generalized for all human beings. In the following elucidation, this point shall also being proved. Furthermore, the celebrities who have been murdered or had an unnatural death are not included. This exclusion has to be taken as the rate of being murdered in celebrities is understandably quite high as compare to common populace.

Scientometrics 69 (2006)

A. F. SIDDIQI: Age likes some years

Table 1 is showing compiled results, of this recording giving number of celebrities dying in different age groups. The 1st column is showing different age groups while the column 2 onward are showing the number of deaths in different categories. The last column gives the number of deaths recorded from a sample taken from all of the 5000 selected celebrities. It should be noted here that ages are compiled using a group difference of 5 years. The tabular presentation is giving only a vague idea of their distribution. Figure 1 is a shaded area frequency curve (drawn with the help of MS Excel) for this table (Table 1), giving a cumulative effect for all the categories.
Table 1. Data on life-span for five different categories of celebrities Age groups Less than 30 31 35 36 40 41 45 46 50 51 55 56 60 61 65 66 70 71 75 76 80 81 85 86 90 91 95 Above 96 Totals C1 1 1 0 1 1 2 7 2 9 7 13 9 5 1 1 60 C2 1 3 3 1 1 3 1 5 7 12 8 5 7 1 2 60 C3 0 0 1 1 2 1 5 6 12 7 9 5 7 2 2 60 C4 0 0 1 1 2 5 9 0 7 8 10 8 7 1 1 60 C5 0 0 1 1 4 6 4 1 10 6 7 7 9 4 0 60 Whole 1 2 2 1 2 3 5 2 9 7 9 7 7 2 1 60

Figure 1. Shaded area frequency curve for the ages, when the group difference is 5. Thorny peaks are indicated

Scientometrics 69 (2006)

A. F. SIDDIQI: Age likes some years

Here are some of the main features of the above shaded area frequency curve, drawn for Table 1. 1. All classes of the celebrities behave almost identically, as the up-down formation is almost identical for all the individual curves. In other words, the death pattern in all the investigated five professions is almost the same. So, it can be said that the age at death is indifferent to profession. Numbers of celebrities dying in different age groups are not the same and all classes of celebrities are corroborating this fact. This is establishing the considered hypothesis that all ages are not equally vulnerable to death. Thorny peaks are observed for some age groups (especially for the groups (56-60), (66-70), (76-80) and (86-90) where most of the classes show abnormal conic pattern), indicating a high vulnerability for deaths of these ages. So, there exists dangerous ages. These peaks are observed for approximately the same age groups in all the classes. There are some safe ages, especially after the thorny peaks.

2.

3.

4. 5.

So, it can be inferred that the human age is not equally distributed over age groups, beyond 30 years, but there are some groups are more vulnerable than others. This characteristic is observed in all the five categories of celebrities. As the death pattern is indifferent to the profession as established in above stated feature #1, so the results may easily be extended and applied on common populace.
Table 2. Frequency distribution of ages of celebrities of five different fields (Group difference is 4 years) Age groups Less than 30 31 - 34 35 - 38 39 - 42 43 - 46 47 - 50 51 - 54 55 - 58 59 - 62 63 - 66 67 - 70 71 - 74 75 - 78 79 - 82 83 - 86 87 - 90 91 - 94 95 Totals C1 1 1 0 1 1 2 3 4 1 1 8 7 9 8 7 4 1 1 60 C2 1 2 4 0 1 1 4 2 1 4 7 10 5 4 6 5 1 2 60 C3 0 0 1 1 1 2 1 5 3 5 10 6 4 7 5 5 2 2 60 C4 0 0 0 0 0 2 5 7 2 1 8 7 10 4 9 3 1 1 60 C5 0 0 1 2 1 3 6 3 2 2 8 6 5 5 6 7 3 0 60 Whole 1 1 2 1 1 2 3 4 2 5 6 8 7 6 4 5 1 1 60

Scientometrics 69 (2006)

A. F. SIDDIQI: Age likes some years

Table 1 is arranging these numbers of deaths for a group of size 5 and there may exist some grouping error. To eradicate the grouping error in the data, let us arrange the data again in the age groups with a difference of four years, instead of five years. Table 2 is showing the results. Following is a shaded area frequency curve (contrived with the help of MS Excel 97) for this table, giving a cumulative effect for all the categories.

Figure 2. Shaded area frequency curve for the ages (Group difference is 4 years)

Table 3. Comparison of thorny peaks Giving ages with maximum vulnerability to death Thorny peaks (in years) Peak 1 Peak 2 Peak 3 Peak 4 58 68 78 88 56 68 76 84 57 68 77 86

Group difference = 5 years Group difference = 4 years Mean

Explicitly, the modified frequency distribution, and the corresponding shaded area frequency curve, for human ages tells the same story as for the graph with an age difference of five years. Again, there are four thorny peaks observed for almost same age groups, or ages, indicated in Figure 1 and Table 3.

Scientometrics 69 (2006)

A. F. SIDDIQI: Age likes some years

Conclusion The survey was conducted to look into the veracity of a prevalent myth that the chances of dying are equal at all ages, especially beyond 30 years. The survey revealed that these chances are not equal, but some ages are considerably more vulnerable to death and some are less. The ages for these can be given as 571, 68, 771, and 861 years (refer to Table 3) for these four peaks identified in the present analysis. These compare with the peaks found in an earlier analysis of a small sample of eminent Pakistan scientists (died between 1970 and 1992) published by QURASHI (1993), which are 51 years, 63 years, 77 years, and ~85 years. Giving agreement to within +3 years as the average, this tends to confirm further our hypothesis that persons at these special ages are highly susceptible to death. Many reasons may be attributed for these four highly vulnerable ages. There may be genetic underpinnings, physiological explanations, or some mythical justifications but one thing is clear that human body undergoes some serious transformations, or adjustments, at these four ages. Some of us survive these and some unfortunately not. References
CHARLESWORTH, B. (1980), Evolution in Age-Structured Populations. Cambridge. Cambridge University Press. COALE, A. J., P. DEMENY (1983), Regional Model Life Tables and Stable Populations. New York, Academic Press. COX, D. R. (1972), Regression models and life-tables. Journal of the Royal Statistical Society. Series B (Methodological) 34 (2) : 187220. KEYFITZ, N. (1985), Applied Mathematical Demography. New York, Springer-Verlag. LEE, R. D., W. B. ARTHUR, G. RODGERS, INTERNATIONAL UNION FOR THE SCIENTIFIC STUDY OF POPULATION (1988), Economics of Changing Age Distributions in Developed Countries. Oxford, Clarendon Press. LUTZ, W., W. C. SANDERSON, S. SCHERBOV (2004), The End of World Population Growth in the 21st Century : New Challenges for Human Capital Formation and Sustainable Development. London, Earthscan. NEWELL, C. (1988), Methods and Models in Demography. London, Belhaven Press. Qur'aan (21.35), Every Soul Must Taste of Death, and We Try You with Evil and with Good, for Ordeal. And Unto Us Ye Will Be Returned. The Meaning of the Holy Quran (Translation). M. M. PICKTHALL, UBSPD Publishers. 3.185. QURASHI, M. M. (1993), Scientometric studies on Muslim scientists; Part I: A case study for eminent scientists of Pakistan died between 1970 & 1992. Science & Technology in Islamic World, 11 (2) : 7588. QURASHI, M. M., S. M. A. SHAH (1983), Scientometric studies on Muslim scientists; Part I: A case study for eminent scientists of Pakistan. Proceedings of Pakistan Acedemy of Sciences, 21 (1) : 2536. QURASHI, M. M., S. M. A. SHAH (1984), Scientometric studies on Muslim scientists; Part II: A case study for eminent scientists of Pakistan (Died between 1960 & 1992). Proceedings of Pakistan Acedemy of Sciences, 20 : 111. VIERCK, E., K. HODGES (2003), Aging : Demographics, Health, and Health Services. Westport, Conn., Greenwood Press.

Scientometrics 69 (2006)