No large scale statistical studies pertaining to astrological patterns in accidents have been reported in the literature. Therefore, there is no specific model from which to extract specific methodologies or experimental design. This chapter will present the methods and procedures employed in this study.
This study examines the relationship between date of birth and date of injury for people injured on the job, therefore it is considered to be a correlational research. Because the injuries have occurred prior to the study, the design also can be termed as ex post facto (after the fact). The limitations of such a design are obvious: There is no control over the variables. However, it is unlikely that any researcher will embark upon a design that will control events such as birth and injury of a person. Therefore, as Smith (1981:388), states, "hypothesis testing normally is preferred to ex post facto interpretation. There are times however, when researchers have no choice but to interpret data ex post facto, for instance, serendipitous findings or lack of theory", or as Rosenberg (1968) states:
Two sets of variables were obtained for each subject: His/her date of birth, and his/her date of injury. In an attempt to find an astrologically predictable pattern, no causative assumptions are made, i.e., it is not the contention of the present researcher that the Sun causes an injury by its transit to the natal position. It only indicates a time when accidents are more likely to happen, the cause of which we do not know.
The subjects for this study were drawn from public records. In the State of California, employers are required to file an injury report with the State. The information on the form includes, among other facts, the date of birth of the injured party and the date of the injury. It may also include the type of injury (which part of body) and the exact time of the injury, but this is not uniformly reported. The reports of fatal injuries are released only by numbers, and data on these subjects could not be obtained, therefore Sample A and Sample B of this study do not include any fatal injuries.
The subjects for the first sample (Sample A), totaling 523 individuals, were obtained from the records of three independent medical-legal transcribers who transcribed the initial medical reports of physicians following the injury. The subjects were referred to the physicians by attorneys, representing them for a Workers' Compensation claim. The records were kept by the transcribers through the years 1987-1991, and included the following information: The subjects name, social security number, date of birth, date of injury and the Workers' Compensation Appeals Board number (the actual filing number with the State). No records were kept as to the part of body injured. Note was taken whether it was a physical injury or a result of "Continual Stress and Strain". (For a sample record page, see Appendix A). One of the transcribers copied only the date of birth and the date of injury for each subject, and omitted to state whether it was a male or a female, therefore, Sample A will not be analyzed according to gender.
Sample B subjects (609), were also injured in a work related injury and filed a Workers' Compensation claim. These records were directly obtained from a Los Angeles clinic that evaluates Hispanic workers injured on the job. A clinic staff member copied the information required by this study from the patients files according to this researcher's specifications that were also provided to the medical-legal transcribers who provided the data for Sample A. (see below). This sample provides a sub-division of male and female patients.
It should be noted, that the researcher did not, at any time, see any of the subjects, or have any control over their selection for either treatment by the clinics or inclusion in this sample, other than the requirements for inclusion that will be discussed later in this chapter. Nor did any of the medical-legal transcribers or the clinic staff know at any time what was the hypothesis to be tested or that this study has anything to do with astrology. They were told that a statistical study was being conducted about work related injury patterns, that required the above information from their files.
For the purpose of further replication of this study, if access could be gained to the Workers Compensation Appeals Board files, this type of information can be copied from their records. Insurance companies also keep records of hundreds of thousands of claims, and if access to these files could be gained, another replication is possible. The benefit of securing subjects from the above Workers' Compensation claimants' pool is that all of them suffered an injury severe enough to render them disabled for at least three months, whereas State files are likely to include all and any injury, including minor ones that required outpatient treatment and may have incurred no loss of work days. Although it was not possible to get precise percentages of seriously injured workers who filed claims, conversations with several attorneys and physicians indicate that some 95% or more of all seriously injured workers do file Workers' Compensation claims in the State of California, therefore the fact that the subjects in samples A and B in this study are taken from the population of claim filing workers is not seen as a confounding variable.
The criteria for inclusion in both samples (Sample A and Sample B) are as follows:
No control group was used in this study for obvious reasons: There is no way to look for a lack of injury pattern if the injury has not occurred. There is no point in assigning subjects to a control group if all we know about them is their date of birth. On the other hand, replication of the study with different samples is a reasonable approach to provide additional strength to the design. For that purpose, Sample B (consisting of Hispanic, mainly Mexican and some El Salvadorean subjects) was analyzed. Sample C, consisting of 55 recorded dates of birth and dates of injury obtained from C.E.O. Carter's book (1932) was analyzed as a replication sample. Another replication sample (Sample D) consisting of birth dates and completed suicide dates obtained from Lester's (1987) study served as a control group. (See below under "Replications").
In order to verify that the research samples accurately represent the population of work-injured people, a comparison was conducted with the statistics available through the California Department of Industrial Relations (1989) on age distribution of injured workers. (See Appendix B).
Hypotheses and Questions:
Under the null hypothesis, there is no expected relationship between a person's date of birth and the date of an accident he/she may be involved in. The present study hypothesizes, however, that such a relationship does, in fact, exist, and it is predicted and explained by astrology. It is expected that significantly more people are injured when the transiting Sun forms a hard aspect (conjunction, square or opposition) to the natal Sun than would be expected by chance. In other words, it is expected that people would tend to get injured significantly more frequently around, three months before or after, or six months following their birthday.
In addition to the research hypothesis stated above, this study attempted to answer the following questions:
The data for this study is non-parametric in nature. It is dichotomous, and contains only two variables for each subject: Date of birth and date of injury. There was no manipulation of subjects and there was no intervention by an experimenter, as the data, as discussed above, is ex post facto. So, parametric statistical tests are of no value in this case. The statistical tool to examine the research hypothesis is, of course, a measure of probability. The Chi-Square test for Goodness of Fit was used on all results, as it was possible to compare expected values with observed values.
No horoscopes were calculated for any of the samples. (Sample A is the first sample where there is no separation between male and female subjects, and Sample B is the replication sample with the Hispanic subjects that has separate male/female categories). The only positions that were looked at were the following:
The above computations were performed on each sample separately, and then combined.
Sample B was treated exactly like sample A in the manner described above. The only difference was the additional division into male/female subjects. The same computations were performed on all subjects and additional comparisons were made between the all male and all female groups.
Sample C was obtained from C.E.O. Carter's book, The Astrology of Accidents (1932), reviewed earlier. It was possible to extract from the many examples given in the book, 55 subjects who were severely injured or killed in accidents, and for whom Carter provided both a date of birth and a date of injury. He did this only for these 55 subjects, although he did not look at the dates of the accidents for transits, progressions or any other astrological techniques. This sample was coded like the main samples, for Sun sign separation, day separation and different orbs.
Sample D was obtained from David Lester (1987), who was contacted by this researcher during the process of reviewing the literature for clarifications on his findings. He had in his data files the dates of suicide for all of his subjects, and provided his data for further analysis. The sample consists of 206 completed suicide cases (totals for 1982, in the City of Philadelphia), who had full dates of birth and the City records of their suicide date. (Incomplete birth dates were eliminated from his sample).
The reason for using this sample was to provide a control group for the present study. It is obvious that suicide is not an accidental occurrence, as people think about it for some time prior to attempting it and many times leave a note stating the intention. Suicide also usually involves depression, despair, etc., which astrologically are not signified by the Sun. Astrological theories would predict aspects involving other horoscope factors which are beyond the scope of this study. Therefore, it was expected that completed suicides would not produce the effects expected by this researcher for accidents, i.e., it was not expected that there would be a large concentration of suicides occurring around the time of the subject's birthday, three months later, six months later and nine months later. Rather, it was expected that from the transits of the Sun to its natal position, this sample would produce a random distribution. Therefore, the exact same computations and calculations that were performed for Samples A and B were also done for Sample D.
The standard procedure followed by research in the social sciences uses a randomly selected group of subjects from a larger population, which is compared with the subject population of the study. The randomly selected group is called a control group and is compared to the subject group of the study to determine whether the study population is similar or different from the control group.
In studying the planetary patterns in the sky, it is possible to measure precisely the amount of time that any given pattern is present. So, the presence of a given pattern for a subject population or sample group of subjects can be compared with the actual situation in the sky, eliminating the need for a randomly chosen control group, which we hope will approximate the general population. (Pottenger & Vail, 1986). This method determines the probability of any given aspect occurring in any given period of time between any pair of planets. This method is being used for astrological research involving aspects, the angular separation between planets. A computer analysis called Frequencies for Aspect Research (Pottenger, 1990) was used on all subjects and all dates involved in this study to determine whether the expected values in the Chi-Square tables are indeed actually astronomically present in the sky. In other words, this analysis was used to determine whether the percentages used for the different orbs of aspects in all samples as the expected values in the Chi-Square tests are justified astronomically. (See Appendix D for a sample page of the computer analysis printout).