As emphasized throughout this work, it was impossible during the course of the research to obtain a large enough or representative enough sample of Death Metal fans. As a result, the findings are not highly reliable or generalizable. However, as a result of the extensive interviews conducted and of my comprehensive experience of the Death Metal scene, I do believe firmly in the validity of those theories and insights that have been expressed here. The purpose of this study was not to draw definitive conclusions about fans of Death Metal, nor to establish an explanation for violence in society or violence in entertainment. Instead, I hoped only to provide an open window into the scene, which as small as it might be, permits outsiders to glance in and to question the assumptions and prejudices which they may already hold. It is my hope that someone more capable than myself will gain a new perspective in peeking through this window, and will undertake more extensive and scientific research that will ultimately defend the metal community and perhaps help to justify the existence of Death Metal to some of its most powerful opponents.
As mentioned earlier, much of the data in this study was collected via lengthy surveys. Although these surveys were pre-tested, certain potentially problematic aspects of the survey instrument itself were discovered later in the course of the research. For instance, questions 11 and 13 ask about one’s relationship with his or her parents. These questions are inherently double-barreled because in both instances, the survey asks about “parents” without differentiating between mothers and fathers. This is particularly significant in a population where rates of divorce are perceived to be high, and the difference between feeling close to one parent and feeling close to both parents should have been noted. Because it was not, the responses to question 13 were not exhaustive and the accuracy of responses to this question suffers as a result. Question 11, which asks for an open response to the question “Did you feel close to your parents when you were growing up?” enabled respondents to differentiate between mothers and fathers. Unfortunately, my established coding process did not allow for this differentiation because I classified each response as either “close” or “not close” to both parents. For those respondents who reported closeness to only one parent, I alternated between “close” and “not close” responses in entering the data codes for analysis. In future studies, greater accuracy and validity can be obtained through the differentiation of relationships with fathers and with mothers.
Question number 21 was particularly problematic because it lacked the quality of exhaustiveness. In reviewing question 21, one may notice immediately that the option of keeping taxes precisely as they are is not mentioned. This oversight should be avoided in future studies. Luckily, it would not significantly affect the results because it is only one of many questions used to formulate an ideological score for respondents, and the score can still be calculated with ease even when question 21 is left blank.
The other major problem with the survey instrument lay in the questions that presented scenarios and possible reactions in order to measure respondents’ tendencies toward violent behavior. Responses were coded only as: nonviolent; violent only when physically provoked; or violent when physically or verbally provoked. In some situations, respondents reported nonviolent reactions to physical provocation and violent reactions to verbal provocation. This was unexpected and proved a barrier to the coding process. I dealt with the problem by assigning numerical values to the different responses on an interval scale, where 1 was nonviolent, 1.5 was nonviolent yet hostile, 2 was violent when physically provoked, and 3 was violent when verbally provoked. For each question, I recorded the respondent’s answer as one of these numbers. I then found the mean of the numbers recorded for each respondent. Although this may have provided a reasonable portrait of the violent tendencies of given persons, the measurements were highly imprecise and too arbitrary. An accurate and meaningful average value cannot be calculated using such rough interval measures. However, I felt constrained to deal with the data as if it were ratio data in order to develop a recordable and measurable statistic. Such a procedure is improper, and in future studies, a more accurate and more measurable means of quantifying violent tendencies should be developed and utilized.
In a general criticism of the survey instrument employed in this study, I affirm that the length of the survey is certainly its most undesirable factor. Actual rate of response and completion rate could not be calculated because of the snowballing methods which took place beyond the researcher’s eyes, and because with surveys sent out via e-mail it is impossible to tell how many persons actually saw the survey and how many began but did not complete it. It is likely, however, given the length of the survey, that response and completion rates were low. Because of the quantity and variety of information that I wished to obtain for this analysis, I felt as though I could not sacrifice any data to make the survey shorter and more manageable. On the other hand, had the survey been shorter, it is likely that more persons would have completed it, and that a larger and more representative sample would have been obtained. I thus had to mediate between the goals of in-depth analysis and generalizability, and depth of analysis took priority in this particular study partly because generalizability was simply not a realistic goal to begin with. The limited resources and time for the research process precluded it from becoming more generalizable and representative. However, for the exploratory purposes of this study, a deeper look into the characteristics of a smaller number of individuals seemed most desirable.
Other than these factors, I can think of no more specific criticisms of my methodology and data collection that could benefit researchers conducting studies in this area. In general, the weaknesses of this project are grounded in the inordinately small sample size and my inability to obtain a reliably representative sample. Future researchers are encouraged to find means of data collection that are less likely to have a class bias than the e-mail methods utilized in this study. In the best of worlds, all data would be obtained from a sample of Death Metal shows across the nation. Although this sample could never be truly random, it would be more representative and more generalizable than any findings of this study.
In conclusion, I would like to point out that it is not as difficult to obtain a large and representative sample of Death Metal fans as it may initially appear. Had I not been bound by financial and temporal constrictions, I would have attempted to do so, and I hope that future researchers will. The amount of cooperation and the great assistance that I received from so many metal fans indicates that an in-depth, more scientific study of the Death Metal scene is a realistic possibility and a worthy endeavor.