| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From the Department of Economics, Pomona College, Claremont, California.
Address correspondence and reprint requests to Gary Smith, Pomona College, Claremont, CA 91711. E-mail: gsmith{at}pomona.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
Methods: The California Department of Health Services mortality database was used to identify the birth year and age at death of decedents with positive or negative initials, as defined by the original study and as chosen by a new survey.
Results: There is no substantial or statistically significant relationship between either set of initials and longevity when decedents are grouped by birth year, either for the original study period 1969 to 1995 or for the longer period 1905 to 2003.
Conclusion: These data do not indicate that mortality is affected by one’s initials.
Key Words: initials • mortality • longevity • names
Abbreviations: AAD = average age at death.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
Christenfeld et al. (7) report that a person’s 3-letter initials are also important; specifically, that people with positive initials (such as ACE or VIP) live much longer than do people with negative initials (such as PIG or DIE). The idea that longevity is affected by one’s initials has been dubbed "monogrammic determinism" by Corliss (8), who publishes a bimonthly newsletter of "observations and facts that challenge prevailing scientific paradigms."
If this theory were true, then parents should choose their children’s initials carefully, and doctors and health researchers should consider initials in their diagnosis, treatment, and understanding of diseases. Is this person’s health at risk because of his initials? Can this woman prolong her life by changing her name in order to change her initials? Has our study of, say, heart disease controlled for the confounding influence of initials? However, there are theoretical and statistical issues that make it especially important to see if independent data confirm this proposed relationship between initials and mortality. This study uses mortality data for the years 1905 to 2003, with decedents grouped by birth year to attempt to replicate the results previously reported using mortality data for 1969 to 1995 with decedents grouped by death year.
The Original Study
Christenfeld et al. (7) use the following 12 positive initials: ACE, GOD, HUG, JOY, LIF, LIV, LOV, LUV, VIP, WEL, WIN, and WOW; and 19 negative initials: APE, ASS, BAD, BUG, BUM, DED, DIE, DTH, DUD, HOG, ILL, MAD, PIG, RAT, ROT, SAD, SIC, SIK, and UGH. They report that their lists were compiled by consulting an electronic dictionary of 3-letter English words and adding several near-words, such as LUV. They then confirmed their choices by asking 39 undergraduates to label these 31 initials and 9 neutral initials as positive, negative, or neutral.
Decedents Grouped by Death Year
Christenfeld et al. (7) use California mortality data for the years 1969 to 1995 for white non-Hispanic decedents, separated by gender. Their primary analysis groups decedents by death year and compares the average age at death (AAD) for those with positive or negative initials to the AAD of decedents with neutral initials. They report that, in comparison to their controls, males with positive initials lived 4.5 more years, males with negative initials lived 2.8 fewer years, females with positive initials lived 3.4 more years, and that there was no difference for females with negative initials. A 4.5-year difference in life expectancy is larger than the difference between the United States and Venezuela and almost as large as the difference between the United States and Algeria (9). There are plausible explanations for observed AAD differences among the United States, Venezuela, and Algeria. We do not have a comparable scientific explanation for why initials might have such large effects on life expectancy.
Decedents Grouped by Birth Year
Christenfeld et al. (7) also examine data for decedents with positive and negative initials whose birthdates and death dates were both between 1969 and 1995. For each of these subjects, they compared the subject’s age at death with the AAD for people with neutral initials who were born on the same day as the subject. They report that most decedents with negative initials died earlier than the AAD of the controls (187 versus 127) and that most decedents with positive initials died later than the AAD of their controls (18 versus 15).
Problems With the Original Study
Decedents Grouped by Death Year
Retrospective studies have many possible pitfalls. Here, one serious problem is that the frequency of 3-letter initials changes over time. Suppose, for example, that mortality rates are constant and identical for people with positive and negative initials but that there are more people with negative initials in later birth cohorts than in earlier cohorts. If so, people with negative initials, whether dead or alive, will tend to be younger than people with other initials.
Christenfeld et al. (7) acknowledge that this is a problem and report that the ratio of decedents with negative initials to all decedents increased by 40% between 1969 and 1995 (which would decrease the AAD) and that the ratio of decedents with positive initials to all decedents was approximately the same in 1969 and 1995. They attempt to control for the increased popularity of negative initials by identifying neutral initials that increased in popularity by comparable amounts between 1969 and 1995.
However, it is the popularity of initials in every year that matters, not just the first and last years. Suppose that initials do not affect mortality and consider, for example, a straight line drawn between the frequency of initials in 1969 and 1995. There are many possible patterns for initials in the intervening years, including (a) above the line for the first half and below the line for the second half; and (b) below the line for the first half and above the line for the second half. Even though the frequencies are the same at the end points, there will be extended periods of time when the AAD will be higher for (a) than for (b) and extended periods when the reverse is true. Thus we cannot reliably assess the effects of changing popularity just by looking at 1969 and 1995, and we cannot confidently identify a control group simply by looking at the frequencies of initials in 1969 and 1995.
Matters are further complicated by the fact that Christenfeld et al. (7) estimate changes in the popularity of initials by comparing the frequency of initials for decedents grouped by death year, which is a flawed measure of changes in the popularity of initials at birth. This is an example of the more general problem of survivor bias in case-control studies (10). Here is a simple example. Suppose that 80% of the population, regardless of initials, die at age 50 and 20% die at age 100; their AAD is 60 years. Now suppose that 100 people were born in 1870 with positive initials and 80 people were born in 1920 with positive initials; in all other years, there were no positive initials. In 1920, 80 people with positive initials die and their AAD is 50; in 1970, 84 people with positive initials die and their AAD is 61.9 years. When decedents are grouped by death year, we observe an increase in the number of deaths with positive initials (from 80 to 84), even though there was a decrease in the number of births with positive initials (from 100 to 80). We also observe an increase in the AAD (from 50 to 61.9), even though there has been no change in mortality rates and there was an increase in the number of deaths of people with positive initials.
How do we circumvent this problem? By grouping decedents by birth year. If 2 groups have the same mortality rates, the omission of living persons will reduce the level of the AAD, but the expected value of the AAD over any horizon will be identical for the 2 groups. For example, if we look at 2 groups with the same mortality rates who were born in 1900, the expected value of the AAD increases as we expand the horizon from, say, 25 to 50 to 100 years; however, the expected value of the AAD is the same for both groups whether we look at horizons of 25, 50, or 100 years.
Decedents Grouped by Birth Year
The Christenfeld et al. (7) analysis of decedents whose birthdates and death dates were both between 1969 and 1995 has the virtue of grouping decedents by birthday. However, none of these people lived past age 26, and most died at very young ages. If so, these data will not tell us much about the relationship between initials and mortality. In addition, these data are likely to be positively skewed, with means that are much larger than the medians.
It would be better to look at a data set that includes people who lived more than 26 years and to compare the AAD for decedents with positive and negative initials grouped by birth year.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
A Survey of Good and Bad Initials
The positive and negative initials chosen by Christenfeld et al. (7) are not necessarily those perceived by others to be the best and worst initials. We compiled an independent list of positive and negative initials by using all 26 (3) possible 3-letter combinations for initials to chose 100 candidates, including the 31 used by Christenfeld et al (7). We gave 46 students and faculty this list of 100 initials, together with these instructions:
Hi, I’m compiling a list of 3-letter initials that might be considered "positive" or "negative"; for example, Gary Nance Smith would have the initials GNS, which doesn’t seem to have positive or negative connotations. What I would like you to do is look at the attached list and select the 10 initials that you would be happiest to have (the 10 best or 10 most positive) and the 10 initials you would be unhappiest to have (the 10 worst or the 10 most negative). Thanks for your help.
We tabulated the total number of votes received by each set of initials. The top 12 positive initials (in order) were ACE (32), ICE (18), JOY (17), VIP (17), CEO (16), WOW (16), GEM (14), FLY (13), FOX (13), HIP (12), WIT (12), and WIN (11). The top 19 negative initials (in order) were ASS (29), KKK (27), FAG (22), DIE (21), GAY (21), ZIT (20), FUK (18), PIG (16), DUM (15), RAT (15), SOB (13), TIT (12), GAS (11), FAT (11), BAD (10), POX (10), HOR (9), BUM (8), and SIN (8). Only 5 of the 12 positive initials (ACE, JOY, VIP, WOW, and WIN) used by Christenfeld et al. (7) were among the top 12 vote-getters; only 6 of their 19 negative initials (ASS, DIE, PIG, RAT, BAD, and BUM) were among the top 19 vote-getters. Four of their initials (LOV, LUV, SIC, and UGH) received only 1 vote from the 46 persons surveyed, and several others received 2 votes.
A Statistical Comparison of Average Ages at Death
The California Department of Health Services (12) maintains a mortality database back to 1960 that identifies initials, gender, date of birth, date of death, and race or ethnicity. They also have a mortality database for 1905 to 1959 that identifies initials, gender, date of birth, and date of death, but not race or ethnicity. For the years 1969 to 1995 studied by Christenfeld et al. (7), white non-Hispanic decedents who had any of the 12 positive or 19 negative initials they selected were grouped by birth year. For the longer time period 1905 to 2003, decedents with these initials were grouped by birth year, and decedents with the initials selected by our survey participants were also grouped by birth year. One analysis used all decedents; another excluded children who died before they were 5 years old. We looked at white, non-Hispanic decedents for 1960 to 2003 and all races for 1905 to 1959 because of the absence of race information. In practice, the early years have few usable data because the recorded date of birth is usually "unknown."
One tricky issue is females, who often change their initials during their lifetime. Are they most affected by their initials at birth or by their initials at death? If they care strongly about their initials, is there self-selection bias in the women who choose to marry men with fortunate or unfortunate names and then change their names to acquire positive or negative initials? There is no practical way to deal with these issues, other than to examine males and females separately and to recognize that female data are a less reliable test of the theory.
For each birth year t, the AAD was calculated for those with positive initials (AADt+) and for those with negative initials (AADt–. If there were at least 5 decedents with positive initials and 5 decedents with negative initials, then the paired difference in the AAD was calculated for that birth year:
|
|
For example, for male decedents born in 1900, there were 66 people with positive survey initials and 98 people with negative survey initials. The AAD was 78.28 years for those with positive initials and 75.48 years for those with negative initials, a difference of 78.28 to 75.48 = 2.80 years.
The null hypothesis is that mortality rates are not affected by initials, so that the expected value of each paired difference is zero:
|
|
The nonparametric Wilcoxon signed-rank test for paired differences tests the null hypothesis that the median of the paired differences is zero (13). With more than 25 observations, the probability distribution of the Wilcoxon test statistic is well approximated by the normal distribution, which can then be used to calculate p values.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
|
1969 To 1995 Decedents Grouped by Birth Year
We next reexamined the 1969 to 1995 mortality data used by Christenfeld et al. (7) with the decedents grouped by birth year, rather than by death year as in the original study. Figures 1 and 2 show the male and female paired differences using their lists of initials and, as in the original study, no restrictions on the number of observations or the age at death. The figures show that the deviations are (approximately) as likely to be positive as negative and, overall, are about the same average size.
|
|
|
Young Decedents Grouped by Birth Year
Our reanalysis of the Christenfeld et al. (7) study of decedents whose birthdates and death dates were both between 1969 and 1995 found that most died at very young ages— indeed, in the first year of life. For those with positive initials, the median age at death is 0.2 years and the mean age at death is 3.9 years; for those with negative initials, the median age at death is 0.4 years and the mean age at death is 4.3 years. It is not plausible that infant deaths are affected by their feelings about their initials.
These data are also positively skewed, with means that they are much larger than the medians. It is consequently misleading to count (as do Christenfeld et al. (7)) the number of decedents who died before or after the AAD for the control group. Even if we were to compare 2 identical mortality distributions that are skewed in this way, we would find that the age at death for most decedents in either group is less than the AAD for decedents in the other group. Christenfeld et al. (7) report that most people with positive initials (18 out of 33) lived longer than the average person in their control group. This is truly surprising because the distribution is highly skewed and the median age at death of people with positive initials was 0.2 years. We could not replicate this finding. Instead, we found that only 14 out of 33 people with positive initials lived longer than the average person in their control group.
A New Study of Decedents Grouped by Birth Year
Mortality data for the years 1905 to 2003 were used to calculate the AAD for decedents grouped by birth year, using both the initials in the original study and the initials selected by our survey participants. For the initials used by Christenfeld et al. (7), there was a total of 2678 males with positive initials, 4792 males with negative initials, 1098 females with positive initials, and 7387 females with negative initials; for the survey initials, there was a total of 4201 males with positive initials, 6485 males with negative initials, 2798 females with positive initials, and 4533 females with negative initials.
Larger data sets generally reduce the p values (14), and this is true here. However, Table 2 shows that, for either set of initials, these data do not support the results reported by Christenfeld et al. (7). The observed differences in AADs are neither substantial nor statistically persuasive, and the mean and median differences in AADs are negative for males, the opposite of the direction reported by Christenfeld et al. (7).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
One problem inherent in any study of initials and mortality is that language usage and slang expressions change during the lifetimes of many, if not most, decedents. Some current slang, like ICE and FLY, was not used 50 or 100 years ago; some slang that was popular 50 or 100 years ago is not commonly used today. Someone born in the early part of the twentieth century might have considered GAY to be positive in the first half of the century and negative in the second half. The only practical way to deal with this problem is to analyze a variety of positive and negative initials.
The primary analysis in the original 1969 to 1995 study grouped decedents by year of death; however, AAD calculations for decedents grouped by death year can be misleading if the frequency of initials changes over time. When these 1969 to 1995 decedents are grouped by year of birth, there are neither substantial nor statistically significant differences between the AAD of decedents with positive and negative initials. The original study of decedents grouped by birthday who were born and died between 1969 and 1995 is potentially more reliable but weakened in practice by the fact that most of the people who died in this brief interval were infants; in addition, the statistical measure used by this study is flawed because mortality data are skewed.
The most natural test of whether there is a statistical relationship between initials and longevity is to group the decedents by birth year and to compare the AAD of those with positive and negative initials. If 2 groups have the same mortality rates, the expected value of the AAD over any horizon will be identical for the 2 groups. California decedents grouped by birth year do not replicate the results reported for decedents grouped by death year, either for the original study period 1969 to 1995 or for a longer period 1905 to 2003. These data do not indicate that mortality is affected by one’s initials.
We are very grateful to the editors and reviewers for their careful, detailed, and extraordinarily helpful suggestions.
|
NOTES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
Received for publication November 23, 2004; revision received May 31, 2005.
DOI:10.1097/01.psy.0000181283.51771.8a
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  N. J.S. Christenfeld DON'T YET NAME YOUR CHILD P.I.G.: REPLY TO MORRISON AND SMITH Psychosom Med, October 1, 2007; 69(8): 823 - 824. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
