What time is it? Oh, 8 PM and time to get ready for work again. Tonight it is another overnight shift as house counsellor at a psycho-social rehabilitation facility. I hope for ten reasonably quiet hours, perhaps with all the clients asleep. Once medications are monitored and charts done and the clients settled down, maybe I can snooze a little... providing the fire alarm doesn't go off or clients' voices don't persuade them to do something self-destructive. I'll be back here on Saturday to work a "twenty-four" (a shift from 8 AM on Saturday morning until I am relieved at 8 AM on Sunday).

Saturdays are reasonably easy and I usually bring my portable computer and, using the data/fax line, log into the mainframe computer at the university to run statistical procedures. They are necessary to complete my part of the latest in the series of research studies I have been hired to help with this semester. Every weekday afternoon I work for 4 hours at the university library teaching faculty and staff how to use desktop and mainframe computers. One evening a week I teach a freshman introductory class in Criminal Justice.

All this activity puts me firmly in the ranks of the multiple job holders, a rapidly expanding section of the job market. Conforming to the stereotype of multiple job holders (Jamal, 1988), I work multiple part-time jobs as a matter of financial necessity and would much prefer to have one full-time, full-benefits job.

BACKGROUND

This paper will review some of the literature on multiple job holders to set the stage for reporting the results of an exploratory study of "moonlighters." While moonlighters represent approximately 5 percent of the total working population in the United States (Stinson, 1990), Jamal (1986) notes that there is a great deal of expense and difficulty involved in drawing a large enough sample from the general working population to include sufficient moonlighters willing to provide information. In an attempt to overcome this difficulty, we chose Usenet which provides a low cost solution with potentially high access to enough moonlighters to yield a "snapshot" of their characteristics.

At the time of this survey, our site subscribed to approximately 1200 different groups. The survey document was posted to a ten percent random sample of these groups.

What is Usenet?

The Usenet electronic bulletin board system was organized into over 1600 "newsgroups" at the time of this survey. Each newsgroup is devoted to the discussion of a single topic which ranges from sex to supercomputers, human to machine languages, pets to poetry, culture to computer architecture. These groups are open to all who have the technological resources (computer, modem, network availability) to access them. The newsgroup format has been likened to talk radio, with many "listeners" and relatively few vocal contributors. Some groups are moderated to ensure that the subject matter is restricted to its chartered topic, and to ensure that the participants remain relatively civil to one another. Most groups are not moderated and anyone can send a message. However, group members are quick to negatively sanction those who post messages on topics outside the realm of their particular group.

These newsgroups exist in the physical network of thousands of computers world wide that are tied together by fibre optic cable, phone line, and satellite. The original networks were created to enable scientists and researchers working at remote locations to access databases, software, and computer resources and to facilitate scholarly communication. Very early in the process it became evident that electronic mail was the most heavily used feature of these scholarly networks, as users expanded the range of their discussions beyond work to social discourse.

Sites, which include educational, government, military, or commercial organizations, can subscribe to all, a selection, or none of the newsgroups, usually at the discretion of the site administrator. The working definition of Usenet for the purposes of this research is "an informal, rather anarchic, group of systems that exchange 'news'" (Krol, 1992: 363). There is no way of estimating how many people read newsgroups every day. However, at the time of this survey, total volume of traffic on all the newsgroups regularly exceeded 20 megabytes (about 60,000 double-spaced pages) of information flow per day in over 1600 newsgroups. In May 1994 it is estimated that there are 180,000 Usenet sites in the entire net, with over 21 million user accounts. The number of users and newsgroups increase daily.

Research Questions

1. How does this Usenet sample compare on some demographic variables with a previously conducted United States Department of Labor survey?
2. Do professional, managerial, and technical personnel responding to this survey moon- light for the same reasons as a sample drawn from the general working population of the United States?
3. In this sample, is there a difference in the order of reasons for moonlighting given by women than is given by men?
4. Is there a difference in the order of reasons given by the foreign portion of the sample than is given by the United States respondents?

LITERATURE REVIEW

When the Bureau of Labor Statistics (Stinson, 1990) asked supplementary questions concerning multiple job holders on their Current Population Survey in May 1989, they defined a multiple job holder as an employed person who:

1. had a job as a wage and salary worker with two employers or more, or,
2. was self-employed and held a wage or salary job, or,
3. worked as an unpaid family worker on the primary job, but also had a secondary wage or salary job.

The primary job is defined as the one at which the individual worked the greatest number of hours.

In the United States, multiple job holders account for about 7.2 million jobs in our contemporary economy with a 26 percent increase over the past 5 years and a 52 percent increase in the past decade. The percentage of multiple job holders in the work force has risen from 4.9 percent in 1980, to 5.4 percent in 1985, to 6.2 percent in 1991. The number of men who hold more than one job has increased since 1985 by over one half million to 4.1 million.

Two-thirds of the increase of 1.5 million in multiple job holders can be accounted for by the sharp rise in the number of women holding multiple jobs. A record high was reached in May 1989 when the number of women holding multiple jobs reached 3.1 million and the rate at which they held multiple jobs reached 5.9 percent of women in the work force.

The highest percentages of multiple job holding was found among the 25-34 (6.5 percent) and 35-44 (7.1 percent) age groups. Married women were less likely to work multiple jobs than single women and married men were the most likely to work more than one job. The number of women moonlighters who were divorced, widowed or separated stood at a high of 7.2 percent of the working population, up from 5.7 percent in 1985.

The Bureau of Labor Statistics distinguishes between those who hold one full-time job and one part-time job and those who hold multiple part-time jobs. Of all multiple job holders 74.7 percent fit the one full-time, one part-time category, with that number representing 82.8 percent of the men and 64 percent of the women. Those who fit the category of one full-time and one or more part-time jobs will be called "moonlighters" for the purposes of this paper.

The reasons given in the Bureau of Labor Statistics report for multiple job holding are, first and foremost, financial, with 44 percent of respondents citing the need to meet regular household expenses or to pay off incurred debts. A desire to save for the future (16 percent) was closely followed (15 percent) by those who wanted experience in a different occupation or to build up their own business.

In a study of Arkansas teachers, Bell and Roach (1989) determined that teaching is, for many, a part-time job and that teachers lead most occupational groups in multiple job holding with almost 20 percent of teachers nationwide holding second jobs. Bell and Roach determined that second job holding for teachers does not refer to jobs worked when school is not in session but to concurrent jobs that bring the work week of many teachers to seventy or eighty hours. This is in sharp contrast to Jamal's remark that "The total workweek of moonlighters in most cases, is less than the normal workweek of a generation ago." (1988: 52).

Jamal (1986) notes the lack of systematic research on moonlighters. He surmises this could be for one of two reasons. First, the number of moonlighters is relatively small and it is expensive and difficult to draw a large enough random sample to contain in it a sufficient number of moonlighters to provide accurate information. His second reason is that, as popular belief holds, "moonlighters are generally socially withdrawn and economically deprived" (p. 977).

Eyler's book (1989), "The Executive Moonlighter: Building Your Next Career Without Leaving Your Present Job," is addressed to white collar and management persons and stresses the necessity of remaining in their full-time, benefit-paying jobs while they experiment with second careers, either as an option for or a supplement to their primary occupation. Eyler sees moonlighting as an ideal way of testing the waters for a solo business, or acquiring new skills and contacts as a transition step to new activities. He recommends that managers and executives prepare themselves for the possibility of a second career in an era of corporate downsizing and transnational capital flight.

In a 1983 article, Fraser points out that moonlighters are increasingly white collar workers, and while moonlighting jobs are easy to find in a booming economy, the second job can lead not only to extra money but to personal growth, and renewed belief in oneself in a slow-growth economy. Fraser also notes that "moonlight has led many executives to new careers; they test the waters before taking the plunge" (p. 53).

METHODOLOGY

A questionnaire was developed which was sent (posted) to a random, 10 percent sample of the over 1200 Usenet groups subscribed to by the University of Nevada, Las Vegas A list of newsgroups subscribed to by the university was printed out and starting with a randomly generated number the survey was sent to every 10th newsgroup.

Usenet was chosen for its convenient accessibility. While it is impossible to know the number of persons who are reached by the full range of newsgroups, it was estimated to be in excess of one half million persons world-wide at the time of this survey, with many persons reading multiple newsgroups. Most persons have access to the networks carrying Usenet through their work so this presented potential access to a large, international group of workers. It was hoped that this would provide a sufficient sample of moonlighters.

Subscribers to Usenet groups are those who have access to a networked computer and whose site subscribes to Usenet groups. For the most part these sites in the United States are businesses, military installations, government or educational institutions. A reasonable expectation is that these persons would be employed in white collar occupations, be of above average education and income, and be predominantly male. From personal observation, we have noticed that university faculty, staff, and computer professionals hold skills that are often rare in the general population, and they are frequently sought out to perform specialized tasks beyond their job descriptions.

Respondents to the survey can be characterized as a purposive and self-selected sample of those who had access to Usenet, read one or more of the ten percent sample of newsgroups surveyed, and had the technology and technical ability to respond.

Electronic survey turn-around time is often counted in hours. The first response was received within 2 hours of the questionnaire being transmitted and approximately 65 percent of the responses were received in the first three days. One hundred and nineteen complete and usable surveys were returned over a period of two and a half weeks.

The researcher sent a "thank you" reply to each respondent and followed up by asking them their opinion of being surveyed via electronic mail. The most frequent response from academics involved the question of sampling error as there was no possible way to determine how many persons actually received the survey. Those who did respond with completed surveys were very positive in their approach and helpful. Several long and interesting e-mail conversations ensued with various respondents, a situation unusual, if not impossible with the respondents to regular mail surveys. This frequently allowed the elaboration of answers and exploration of topics beyond those on the survey itself.

When messages are replied to, they include the name and electronic mail address of the sender. This raises issues of confidentiality as it is possible to determine immediately the name and location of the sender. A "reply" command will, in most electronic mail systems, format a response that is already addressed to the sender. When sending out the "thank you" notes it became evident that a number of respondents had found ways to "fake" their address so that the notes came back with the notation "user not known on this system." A quick check indicated that the responses were within the expected values. They had just found their own way to confidentiality. Three respondents chose to print out the questionnaire, remove all identifying lines and then post-office mail the surveys back. One respondent, an employee of a foreign national government, responded electronically, and specificity requested anonymity.

The survey also generated a vociferous and verbally abusive protest. These messages, known as flames, were directed to the researcher and to the local Usenet site administrator. His first inclination was to immediately suspend the researcher's Internet access privileges. However, permission to conduct the survey had been previously obtained from a senior member of the Academic Computing Center's administration. Angry respondents, many of them readers of technically oriented groups, were alarmed with the notion of "student" social researchers loose on the networks. They objected to the sanctity of their newsgroup's boundaries being transgressed by a non-member of the group, especially a stranger presenting a survey on their screen. Several list moderators sent back the survey without posting it, often with a standard, dispassionate note that indicated that the message was "off-topic" and perhaps belonged on another group.

Because the original sample list had been printed alphabetically, many lists having the same general topic and, therefore, overlapping readership, were clustered together. This would mean some persons saw the survey two or three times. Those who did appear to have become angrier each time. The researcher early discovered that it was not wise to respond to those who were "flaming" her. Instead she merely noted that approximately 20 percent of the total response to this survey were angry demands to cease and desist. Further research is indicated to investigate the vehemence of the protests and the sense of "private space being invaded" that pervaded the angry responses.

RESULTS

The response to this survey indicates the international audience that was reached via Usenet. The majority of respondents were from the United States, but eleven other countries are also represented (Table 1).

The larger proportion (81.5 percent, n=97) of the respondents were male, with 18.5 percent (n=22) being female. The mean age of respondents is 32.64 years with a range from 19 to 62 years, with women (mean = 35.72) being a little older than the men (mean = 31.94). Most moonlighters are found in the 25 to 34 age group (Table 2). The average educational level is high with 77.8 percent of respondents having a Bachelor's degree or higher. Women showed slightly more education than men (Table 3). Respondents from outside the United States and Canada showed marginally higher education levels (see last two columns in Table 3).

Income levels are also high (Table 4) with men earning somewhat more than women. There are obvious difficulties in trying to compare income across national boundaries. These involve the conversion of other currencies into US dollars and the relativity of income to standard of living. Overseas, dollar amounts in earnings considerably lower than those in the US can frequently purchase a comparable or higher standard of living. While US and Canadian incomes appear higher than those overseas (last two columns of Table 4), interpretation of this finding is problematical.

Hours worked on their primary job averages over 35 per week for both men and women (Table 5), and both men and women work the same mean number of jobs (2.45) (Table 6). Most of the respondents receive benefits from their primary employer. They work an average of 2.45 jobs with hours worked on second and third jobs averaging 11-15 hours per week. Most of the respondents (n=77) work in the field for which they were trained. Respondents are more likely to be married than single, however more women (31.8 percent) are divorced than men (12.6 percent) (Table 7).

As can be seen from Table 8 the most frequently cited reason for moonlighting by this Usenet sample was the desire to use existing skills. This was followed by needing the extra money and after that, a desire to provide service to others. Twenty-two respondents would prefer to be able to work just one full-time job. Finally only 6 respondents are required to work second or third jobs by their primary employer (this is often the case for public relations reasons or as part of a university's mission to serve their community).

No category was supplied for "saving for the future" nor was that included by respondents in the write-in "other" category. Other responses offered in this category are shown in Table 9.

Only five persons cited the need to reduce personal debt load. Many of the respondents indicate non-financial reasons, which are likely to come under the rubric of personal growth and development.

Table 10 compares reasons for moonlighting between men and women, and Table 11 compares United States and non-US responses for reasons to moonlight.

DISCUSSION

When the respondents to this survey were compared to moonlighters in the United States working population, the highest percentages of multiple job holders was found were in the 25-34 and 35-44 age groups. Stinson (1990) reported that women who moonlight were much more likely than men to work at multiple part-time jobs, and work fewer total hours on their multiple jobs, in the overall US population. Looking at the results in Tables 5 and 6, there is some evidence that this is true for the persons responding to this survey.

Married men were the most likely to moonlight in both this survey and the overall United States survey (Table 7). There seems to be little evidence to suggest those responding to this survey differed greatly from the general survey regarding moonlighting and marital status.

The literature reviewed indicates the reasons for moonlighting in a sample drawn from the total United States work force are primarily financial, followed by saving for the future, experience in a different occupational field or to build up a new business. In this survey of moonlighting workers, it was found that, unlike a sample drawn from the United States working population, these respondents did not cite financial concerns as their primary reason for moonlighting, but emphasized the opportunity to use existing skills. However, when the sample was split along gender lines, women, who are more likely to be divorced and to support more children than the men, cited needing the extra money as their primary reason for moonlighting.

Moonlighters in the sample were found to be, on average, in the higher brackets of both education and income, thus casting doubt on the popularly held notion that all moonlighters work because they are "financially deprived" (Jamal, 1986: 977) It was interesting that many of the respondents appear to moonlight for the personal satisfaction of using their skills to provide service to others. If you look at men's reasons for moonlighting, they follow exactly the rank order of the whole sample. However, the rank order of reasons women give for moonlighting shows a slightly different order than those for men. Women rank needing the extra money as their most frequent reason for moonlighting, with using existing skills second, gaining additional skills third, providing service to others, because moonlighting pays well and changing careers fields tied for fourth, because they were asked to and preferring a full-time job ranking fifth, to get benefits, sixth and because working part-time fitted their schedule, seventh.

When the United States respondents are compared with those from outside the United States there is a difference in the ranking of their responses. Foreign responses follow the full sample in ranking using existing skills as their primary reason for moonlighting, followed by providing service to others, then gaining additional skills. Needing the extra money comes fourth in their ranking as opposed to first among the United States residents.

LIMITATIONS

Mail Surveys

While there are considerable cost savings and the obvious appeal regarding being able to access a widely dispersed sample (in this case, moonlighters) using mail surveys, their limitations are also well documented. Even with well-designed questions and under ideal conditions, the performance of mail surveys are usually less than desired, (see for example, Rossi et al., 1983; Goyder, 1987; Fowler, 1988; Moser & Kalton, 1972). Disadvantages include such problems as: the ineffectiveness of mail as a way of enlisting cooperation, the inability to handle certain kinds of questions (e.g., open ended questions, questions that may be tedious or boring, "screen questions"), inappropriateness of use for spontaneous answers, no opportunity to probe beyond the given answer for clarification, and the answers can not be treated as independent given the respondent sees all the questions before answering any one of them. Another major difficulty of mail surveys, (compared to face-to-face interviews), is that of obtaining a representative sample. In this survey, the population was defined as Usenet users. The random selection of Usenet groups was an attempt toward finding a representative sample within this population.

The main problem with mail surveys is non-response. "The likelihood is that the non- respondents differ significantly from the respondents, so that estimates based on the latter are biased" (Moser and Kalton, 1972: 262). Mail questionnaires to the general public tend to result in an upward bias in social class composition and educational level, and response is correlated with interest in the subject of the survey. The only safe way to deal with non-response is to reduce it to a level sufficiently low so as to not cause serious bias. In this survey methodology, there is no way to calculate response rate given we don't know how many persons received the questionnaire.

Additionally, mail survey respondents may have more education than non-respondents because of the literacy requirements for filling out a survey. In this survey, Usenet users are probably more literate overall than the general population, and this may ameliorate some of this particular bias in respondents. (However, it may confound comparisons between Usenet participants and the general United States survey).

Other Limitations

Since this is the first time the Usenet moonlighters were surveyed via Usenet, there is no way to estimate changes from one year to another (as was done with the United States survey).

The Stinson (1990) report on moonlighting in the general United States population did not address the educational level of the respondents. To the extent that educational level affects moonlighting, and to the extent the Usenet population is different regarding education (Table 3), the comparisons between results would not be valid. Further research is suggested to analyze the source data of the United States data and, if possible, compare results between respondents with the same educational characteristics. The same limitation and recommendation is true for the income of moonlighters between surveys Table 4 for Income of this survey's respondents).

The category "Saving for the future", determined by the Bureau of Labour Statistics survey as an important reason, was not included in the choice of categories for moonlighting in this survey. This would have allowed this survey to conform more exactly to the national survey and render the results more comparable.

While the survey has been legitimated as a data collection technique, this is the first time we have used it for the electronic collection of information. We determined that, while the standard content guidelines continued to hold true, it was necessary to adjust the layout to accommodate display screens that display 22 lines down and 80 columns across. To ease completion, most questions provided categories for response, which does limit the statistical techniques that can be used.

As Jamal notes, it is difficult and expensive to draw a sufficiently large sample to contain enough moonlighters who will respond to a request to supply information. The response to this survey was comparatively small and in no way representative. This is especially true of the international portion of the sample and makes its results not generalizable.