David Lesniaski, chair; Tim Cherubini, Charles Coldwell, Richard Griscom, Pat Fisken, Michelle Koth, Renee McBride, Cindy Richardson

This article summarizes some of the significant findings about the MLA garnered through the results of the Working Group survey distributed over the spring and summer of 1997. Though the data now is a few years old, we believe the essential conclusions still hold.

The Working Group began this project several years ago. Originally the focus was to be on new members and on placement activities. However, we realized that we had little solid information on the organization as a whole. This situation led us to develop the questionnaire. We had several objectives in mind: to create a "statistical portrait" of the organization in the late 1990s, to use the power of standard statistical procedures to test whether or not certain commonly held assumptions about members’ activities and about the organization were indeed true, and, finally, to see if there were any reasons some individuals let their memberships lapse, or did not join MLA even though their interests would seem to coincide with those of members.

This report is in two large sections. This narrative section discusses some of the significant results of the survey, while the appendix presents a summary of the actual data compiled from the questionnaires. The two sections complement each other. The data summary presents the responses for each question, and, in some cases, narrative or graphical interpretation of the data. Such a summary is valuable for the wealth of detail it contains. However, the narrative report draws together results from many different questions and areas of the report, and presents the results of various statistical analyses. Only through such analyses is it possible to test hypothesis or assertions about MLA member characteristics and come to any conclusions about the significance and meaning of the data. Hence these two sections really should be read together.

The phrase "statistically significant" occurs many times throughout the article, for different statistical tests were used to test hypotheses or assertions about MLA members’ characteristics. It is not necessary to know the details of these tests in order to read and understand this article, but a sense of what information certain statistical tests can provide will give a better basis for considering the implications of our findings. This section is a very cursory look at statistics and hypothesis testing and is no substitute for a careful study of statistical reasoning and methodology.

By way of an example, consider this hypothesis: Female librarians in the U.S. earn more than male librarians in the U.S. Suppose further that we did a survey by questioning a random sample of librarians, and came up with the following result: the average salary of female librarians was $45,000 and that of male librarians was $44,500. Would it be prudent to conclude that we were right?

If we looked at the actual data, hoping for a trend, suppose the results were as follows:

…and so on.

This data seems pretty clear. But suppose the data looked more like this:

and so on. Suppose also the averages came out to $45,000 and $44,500 respectively, but the data when graphed looked this way:

It’s hard to tell from this picture just what to conclude—while the average salary of women is higher than for men (the black line represents the average) the men’s salaries seem to be higher (the lowest male salary is higher than the lowest female salary; the highest male salary is higher than the highest female salary). Just looking at averages doesn’t tell the whole story.

Suppose, since the numbers are so close, we did another sample, and came up with this result: female, $45,500; male, $45,250. The numbers are different—females still are earning more than males, but by a smaller margin than in the first sample. Moreover, the average salary for males in the second sample is higher than that for females in the first sample.

By this point, things look confounding indeed. There are statistical tests which provide a way out of this quandary.

Note that we took a sample of all librarians by doing a survey. Obviously surveying all U.S. librarians would be an impossible task, so taking a random sample would make sense. The absolute size of the sample is important, but most surveys question from a few dozen to a few hundred individuals. Many political polls, for example, survey only a few hundred individuals in a region or even the nation, and they often are accurate to within a few percentage points. Assuming you have a reasonable sample—let’s say a few hundred—and that the sample was randomly chosen and you had a good response rate so the sample wasn’t biased in some way, you can apply statistical tests to the data.

Notice that one possibility already was mentioned. When the first sample didn’t seem to yield useful results, we did another sample. Theoretically, we could conduct 100 surveys, each a random sample of, say, a few hundred librarians. We then could calculate the average salary for women and for men in each sample, and arrange these averages in a table. If, in 95 of the 100 samples, women had a higher salary than men, we would feel comfortable generalizing this result and asserting that the average salary of female librarians in the U.S. was higher than for male librarians. On the other hand, if, the average salary of women was higher than that of men in only 55 of the 100 cases, we probably would not feel comfortable asserting generally that the average salary for women was higher than for men.

While this approach would yield better results and would let us decide our comfort level for reaching a conclusion—if the salaries were higher 80% of the time, or 99% of the time—this clearly is a laborious way to do research. In this scenario, where we conducted 100 surveys and compared the results, we really were comparing the averages of each survey against the others. Over the centuries, statistical techniques have been created which will use the results of one survey to do the same thing.

In this example, we could do one survey by taking a random sample of U.S. librarians, list the salaries of males and females, and compare the averages using a statistical procedure (in this case, a t-test). Tables have been created which are based on the notion of taking many samples of one size or another from a large population. By using these tables (actually, statistical software has the computations built in) with just one sample, the t-test would look at the actual averages of the male and female salaries and would look at the dispersion of the salaries (how spread out or how narrow the salary ranges are). By comparing both the averages and the dispersions for the particular sample the statistical test yields a number which indicates how likely we would be to achieve similar results if we were to conduct 100 such samplings of the entire population. In social sciences research the threshold for "statistically significant" usually is at 95%. That is, if the statistical test indicates the same result is likely at least 95% of the time, they hypothesis is asserted as being true. In statistical parlance, such a result would be described as "significant at the .05 level". If a result is not replicable at least 95% of the time, the hypothesis usually is described as "not statistically significant". (Actually, hypothesis testing is more subtle than that, but this description gives a general impression of the process and reasoning.)

It is important to note that statistical tests do take sample size into account, and there are techniques for estimating an appropriate sample size for a given hypothesis and set of data. Because of the power of statistical tests, samples often are preferable to surveying the entire population if the samples are chosen properly. (In many cases this involves selecting randomly from a population so the sample will not have a bias.) Surveying an entire population in most cases is impossible because of the size of the population (for example, all U.S. librarians). Also, selecting a reasonably sized sample allows for the possibility of following up on non-respondents and ensuring that most individuals selected actually respond. Even though the sample itself may not be biased, if the response rate is poor, the results still may be biased because certain subgroups might have been more likely than others to respond.

Statistical tests also have other benefits. They allow for the testing across several variables and may be able to demonstrate connections which would not be apparent just by looking at the raw data or simple averages. In order to make these connections, however, the survey instrument, usually a questionnaire, has to be constructed so the right type of data is collected. Designing a questionnaire is not merely a process of asking good questions; it also involves having an understanding of what desirable results might be and the type of data needed in order to obtain those results. If, in this hypothetical survey of librarians’ salaries, respondents were asked to indicate a salary range (such as less than $25,000, $25,000-50,000, and so on) we would not have been able to test whether or not women’s salaries were higher than men’s because the averages are so close that our categories would have obliterated this small difference.

Most of the statistical tests used in analyzing the data were quite straightforward, as were many of the issues we were interested in examining. Knowing the details of the tests is not necessary for understanding the conclusions. We believed it was important to point out that we did use standard statistical procedures, and, that while these are by no means foolproof, they do give a solidity and assurance to our conclusions that looking at the raw data would not. Of course, statistical tests are used to verify (or deny) hypothesis. Much of the information gleaned from the survey is descriptive, and this information often is useful in its own right.

The working group, comprised of individuals representing diverse MLA interests, focused on four major areas in designing the questionnaire: demographics, education and professional achievement, work, and MLA participation. The questionnaire was revised many times, and was pre-tested on a small group of individuals at the 1997 annual meeting. In the late spring of 1997, 380 questionnaires were sent out: 300 to MLA members selected at random from the master MLA member mailing list, and to 80 non-members who had attended a recent conference or were chapter members but not national members. 246 replies were received (65%); of those, 245 were useable. 213 of the 300 MLA members replied (70%). (There were three follow-ups to the questionnaire in order to elicit the maximum number of respondents. These follow-ups included letters, email, and phone calls to non-respondents.) These response rates are far higher than is customary for this type of survey (one researcher remarked that this response rate was "phenomenal"). The surveys were returned from respondents to Pat Fisken, who recoded the original questionnaire identifying numbers according to a random number scheme so it was impossible to tell who was associated with any particular questionnaire. The questionnaires then were returned to David Lesniaski, who hired students to enter the responses into an Access database. After all the data was entered, extensive tests were run to ensure the data was clean: that all responses were entered in the appropriate fields, and the numbers and information were entered as correctly and as consistently as possible. (For example, when asked what percentage FTE a person’s job was, some respondents replied ‘1’ rather than ‘100%’, so the data was adjusted to make all responses consistent.) This second database was then converted to other formats and then imported to SPSS for running statistical analyses.

Three significant trends seemed to emerge from the data. First, the MLA is a very diverse organization of highly-educated professionals; second, for the most part there is equality of opportunity and of achievement within the organization; third, among MLA national members, there is remarkable equality of professional accomplishment in work-related measures. Although the MLA consists of a majority of academic librarians, it is not an overwhelming majority. There are many other types of librarians in the organization, and many MLA members are not librarians (or have primary responsibilities outside librarianship). Despite this diversity of interests, and despite the diversity of subgroups studied in the analyses (e.g. academic librarians vs. public librarians, various categories of primary job responsibility, gender, etc.) most measures of participation in the organization indicated equality in MLA participation and in professional and educational achievements. Men and women, for example, served on the board in equal proportion to their numbers in the membership, have the same degrees, and same average salary. This consistency holds for many other subgroups as well. There are some differences, of course, and those will be detailed in the report below. These are anomalies, however, when compared to the larger picture.

In the four major areas considered by the working group questionnaire (demographics, education and professional achievement, work, including job responsibilities, job satisfaction, and salary, and participation in MLA activities), there were many points of contact and overlap. The subgroups studied consistently included gender, type of library, and type of job. The latter two categories are discussed more fully in the "work" section.

The average age of MLA members is slightly over 49 years with a standard deviation of ~ 12.5 years. (This means that roughly two-thirds of MLA members are between 37 and 61.) In graphic terms:

Of the 213 MLA members responding to the survey, 115, (54%) are female, while 95 (44%) are male. (The remaining 2% did not respond to this question.)

Not surprisingly, 187 (88%) indicated European/Anglo American heritage. 166 (78%) indicated their sexual preference/orientation as heterosexual, 22 (10%) as lesbian/gay, and 4 (2%) as bisexual. However, please note that 21(10%) did not respond to that question. If we extrapolate from the percentages for those responding to this question, the figures become 86% heterosexual, 11% lesbian/gay, and 2% bisexual (numbers do not add to 100% because of rounding).

93% of MLA members responding to the survey indicated they work in the US, followed by Canada (slightly over 1%); other countries accounted for an additional 2%, and 4% did not respond to this question. We are rather geographically diverse. Of those who listed the state in which they worked, 32 (17%) were from New York, 24 (12%) from California, followed by Ohio (n=13, 6%). All other states had fewer than 10 respondents (or less than 5% each of the total pool). Given the relatively small sample size (of 213) and the small number of individuals in each state, it probably is not reasonable to make any conclusions on state-by-state comparisons. Grouping by region makes more sense for the size of the sample. If we group states into commonly accepted regions, we have [for example]

"Northeast" (ME, MA, RI, CT, VT, NH, NY, NJ, PA, MD, DE, DC): 70, or 37% of respondents,

"Midwest" (OH, IN, MI, IL, MO, IA, MN, WI, SD, ND, NE, KS): 46, or 24%,

"West/Southwest" (WA, OR, CA, ID, MT, WO, UT, CO, OK, TX, NM, AZ, AK, HI): 42, or 22%

"South" (VA, NC, SC, GA, FL, AL, MS, TN, KY): 32, or 17%

This list is intended to be suggestive, not conclusive! Clearly, one could look at other regional definitions. Some analyses (under the "work" section, below) did use the above geographic divisions. Though the specific numbers would change with changes to regional groupings, it is unlikely the conclusions would be affected. Most members were content with their geographic location (86% were "satisfied" or "very satisfied") and there were no significant differences in these figures by region.

One point to come from this survey is that MLA is not dominated by members from any one region of the country, but is truly a national organization. However, it is overwhelmingly a U.S. organization.

One of the survey questions related to city size. The largest single group (31%) works in a city of over one million people, and nearly 40% of MLA members work in a region of over one million people. Not surprisingly, "urban" is the characterization of the type of area most MLA members work in (less than 25% work in an area of less than 100,000 people).

MLA members are a highly educated group. Only 1 respondent (<1%) lacked a college degree.

Most MLA members have undergraduate education in music, but with a fairly wide spread of actual majors. However, those MLA members who pursue graduate work (in addition to the MLS) have a much narrower focus, and tend to specialize in musicology. 206 MLA members (97%) had at least one graduate degree. 134 (63%) had either the MA or MM, and 43 (20%) had a doctorate in music. 147 (69%) had an accredited MLS, and 7 (3%) had a doctorate in library science. It is interesting to note that nearly one-third of MLA national members do not have the MLS.

One concern some members voiced is that there may be differences in achievement between men and women in MLA. This is not the case in terms of graduate education. If we look at the percentage of MLA members with a graduate degree by gender, we see that 98% of the men but 94% of the women responding have a graduate degree (MA, MM, MLS, PhD, etc.). While there is a small difference between these two numbers, that difference is not statistically significant. Neither are there gender differences in scholarly and musical pursuits. Similarly, there are no significant differences among administrators, catalogers, and public services librarians in educational achievement. While this might seem counterintuitive at first glance, the fact that so many MLA members have advanced degrees means that the subset of those lacking advanced degrees is quite small, so it may not show up in already small statistical cells. However, one of the few educational differences we did discover is that academic librarians are more likely to have advanced degrees in music than public librarians. Interestingly, this does not translate into difference in professional or scholarly work: there are no differences among any of these groups with regard to scholarly activity.

Academic librarians form a large subgroup of MLA. This table provides a brief summary of some of their educational and professional characteristics.

Of the 213 MLA members responding to the survey (71% return rate), 124 identified themselves as working in an academic library. 93 of those have the MLS, and that’s the group we’re considering as professional music librarians working in an academic library.

Continuing education is important to MLA members. 85 (42%) of the 204 respondents to this question indicated they attend at least 1 nationally sponsored workshop annually, while 48% (of 198 responding) attended at least one regionally sponsored workshop annually and 33% (of 184 responding) take at least one class annually. 118 (of 197 responding), or 60%, indicated that continuing education activities were "important" or "extremely important". Fortunately, 157 (74%) indicated free or discounted tuition, convention/meeting reimbursement, or paid sabbatical leave as a work benefit, and 71% indicated they were satisfied with the continuing education opportunities available through their work. Probably because so many MLA members were satisfied with continuing education opportunities available to them, statistical analyses did not show any unusual dissatisfaction regarding continuing education by any particular subgroup, type of library, or type of job held.

Most MLA members indicated at least basic knowledge of one or more languages other than English. The most popular languages were, not surprisingly, German (200), French (190), Italian (92), Spanish (80), Latin (49), and Russian (24). No other single language had more than a 10% response. The typical proficiency was "bibliographic knowledge..." for these 6 languages. Again, no particular subgroup of MLA as tested by job category (catalogers, reference, etc.) showed any significant differences in the number of languages known or language proficiency.

We are a very active group of people:

For purely musical activities, over half (54%) play an instrument,49% participate in choral singing, and 48% in ensemble playing.

It is interesting to compare the two tables above: for academic music librarians and for MLA members generally. We can see that many of the numbers are quite close. The statistical tests which show that for many variables there is no significant difference between academic librarians and other MLA members confirm this impression.

This is the institutional setting for our work:

About 18% of members indicated multiple categories. During the course of the analysis, it was decided not to eliminate these duplications. In many cases it would have distorted the data to force individuals into one category or another—many do work in more than one setting, or have a position which crosses traditional boundaries.

The following chart provides a visual representation:

Although academic librarians are the largest group by far, there are many other groups of both librarians and non-librarians in the organization. Interestingly, in the responses from non-MLA members, the percentage of public librarians was much higher (30%). This suggests that there may be many librarians working in public libraries with responsibilities for music who are chapter members, or attended MLA conferences in their region, but who are not national members. This possibility should be explored further

Academic librarians form quite a diverse group. The largest subgroups are those who work in institutions with a significant graduate component (ca. 50% of academic librarians), and those who work at state-supported institutions (60%) as opposed to private institutions. Nevertheless, substantial numbers work in undergraduate institutions, or in conservatories. (See the chart in the appendix under question 9 for further detail.)

One concern arose from follow-ups to non-respondents. A disproportionately large number of non-respondents were retired, and saw no point in returning the survey; they clearly felt marginalized professionally (though this had no reflection on MLA; indeed, they were pleased that the organization cared what they thought). It is likely the number of retired or semi-retired MLA members is much greater than this survey indicates.

21% of us are unionized. Union members are overwhelmingly in public libraries, at least by percentage (24/125, or 20% of academic librarians are unionized; 15/26, or 60% of public librarians are unionized). There is a significant difference in salary between those belonging to unions and those who do not. This is true in both academic and public libraries (salaries of union members are higher). 33% of MLA national members have faculty status. 82% have a permanent, continuing, or multi-year renewable contracts.

Most of us are pleased with many aspects of our work:

Salary received a disproportionate number of negative responses. In fact, it was the only category we analyzed in detail, since the responses to so many other variables were so overwhelmingly positive.

For many analyses, we split librarians into 4 categories: administrators, catalogers, public service librarians, and generalists. For the first three, the criterion was that the individuals each had to spend half time or more on the appropriate activities: administrators: administration and supervision; catalogers: cataloging; public service: a combination of reference, BI, circulation, ILL, and liaison. Generalists could not spend more than 25% of their time on any one activity. Interestingly, there was almost no overlap among these categories. In the few (4) cases where an individual could fall into more than one category, we looked at their total job responsibilities and assigned them to whatever category seemed to best fit the profile. Of course, many people don’t fit into one of these categories, but there weren’t any other breakdowns which made sense and which were mutually exclusive of each other or these four categories.

The numbers in these categories are not large—none contains over 15% of the membership. Nevertheless, some noticeable trends emerged from these groupings.

Salary issues received much exploration. The average MLA member salary in summer 1997 was about $41,000,with a standard deviation of about $12,500 (this means about 2/3 of MLA members earn between $29,000 and $53,000—not a terribly large spread) . The distribution was approximately normal, but there was a slight "bump" in the upper 40s and lower 50s. As it turned out, this bump was due in large part to the salaries received by those categorized as administrators. Public service librarians, generalists, and catalogers showed no significant salary differences. Administrators, on the other hand, did have a salary average significantly higher than that of other job categories. There is no difference in salary by gender—men and women earn (statistically speaking) the same amount. Public librarians (as a group) showed no significant difference in salary from academic librarians (as a group).

In summary: the following factors seem to have a statistically significant effect on salary:

The following factors to not seem to have a significant effect on salary:

There was the hope of devising a formula to answer the question: how much should I be making? Unfortunately, the data do not support the establishment of such a formula. It appears that salary setting is largely a local phenomenon, and that factors within an individual’s control (education, experience) may not have as significant an immediate effect as other characteristics (the specific job responsibilities, and geographic location). However, these factors may have a long-term effect on salary. The largest difficulty in looking at salary is the presence of confounding variables. If, for example, people who wrote 10 or more articles were paid more than those who wrote fewer than 10, the difference may be due to writing articles or may be caused by other factors (such as experience: perhaps people who wrote 10 or more articles have significantly more professional experience). The data set is small enough and salaries are uniform enough overall as to make this type of analysis very difficult and open to a variety of interpretations.

Perhaps not surprisingly, those who were dissatisfied with their salaries had significantly lower salaries than those who were satisfied with their salaries.

One of the few instances where there are gender differences is the issue of making changes job or education due to family responsibilities. The only significant differences are in ‘resigned position to relocate with spouse or family,’ which happened to women significantly more often than men. Most MLA members, however, did not indicate that family responsibilities caused them to make changes in job or education.

Since academic librarians form the largest subgroup of MLA, the data summary for academic librarians is fairly comprehensive.

This table points to one of the more disturbing aspects of faculty status:

The column "faculty status" indicates how many librarians in that category of library have faculty status. "Faculty feel general" indicates how many librarians in each type of institution feel they have equality with teaching faculty. "Faculty feel of faculty status" indicates how many of those with faculty status feel they have equality with teaching faculty. While about half of all MLA academic librarians have faculty status, that status varies widely from one type of institution to another. More distressing is that overall most MLA academic librarians do not believe they have equal status with other faculty; this holds true even among those with faculty status. Why this is the case is beyond the scope of the questionnaire, but is worthy of study in MLA.

There is no statistically significant difference between the salaries of those with faculty status and those without.

The possession of an advanced degree (additional masters or doctorate beyond the MLS) has no effect on salary.

This table shows more than it might seem at first glance:

One response which was consistent for MLA national and chapter members across several questions was the importance of contact with other members. The importance of this factor cannot be overestimated.

Different groups of people have different value on some of these items. For example, those who thought Notes was "very important" (category 4) are significantly older than those who selected responses 1-3. Just the opposite obtains for the placement service: older members are significantly less interested in this than younger members, as is to be expected. There also is a significant difference in the length of time one has been an MLA member between those who rated the placement service "1" and those who rated it higher. Additionally, there is a significant difference in the length of time one has been a professional librarian between those who rated it 3 or 4 and those who rated it 1 or 2. In short, there is a small, but significant and changing group of people who value the placement service. Hence even though the number of individuals rating it "important" or "very important" is less than for other items, the service still performs an important function for a significant subset of MLA members.

The overall point is that different subgroups of MLA value different offerings of the organization. These are the two most obvious examples from the data and the questionnaire. The remaining variables seem to have a much more even spread and grouping of responses, and have overwhelmingly positive reactions among the vast majority of members.

Many individuals are very active in the organization:

In the few particular categories we tested, there are no significant differences in participation among any of the subgroups: gender, academic / public librarians, or catalogers vs. public service librarians vs. administrators vs. generalists. Unfortunately, the data is somewhat ambiguous here. In looking at participation in MLA, it appears that while on specific measures (e.g. board membership) there are no significant differences between public and academic library participation, there may be differences overall. This may be due to a cumulative effect not noticeable in a particular variable. This, however, is very hard to measure. A chi-square, separating MLA members into "academic", "public", and "other", and participation into different categories of activity (individuals participating in 0, 1-2, and 3+ categories in the table above) does yield significant differences between academic librarians and others. Limiting this to just academic vs. public librarians also yields a (barely) significant difference on this measure of participation.

Whether or not these differences should be of concern is beyond the scope of this report. The differences are not huge, and it took some statistical manipulation to find them. On the other hand, the issue does deserve further study, and it might be prudent to research just how effectively subgroups other than academic librarians are provided for in convention and other programming.

Sorting out the reasons for such differences is difficult. Public librarians attend fewer meetings, and it is possible that meeting attendance has a connection with MLA participation.. Because of the difficulty of measuring each of these variables accurately, it is not advisable to try to use the current data. This is an area for further investigation. However, the reasons for lower meeting attendance are not clear, since there were no significant differences found on variables strongly related to meeting attendance (e.g. institutional support). This is an area which should be more carefully explored. It is also clear that "other" MLA members (not academic or public librarians) which form a significant part of the membership (>25%) have lower meeting attendance and participation than academic librarians especially.

Issues pertaining to convention attendance were particularly difficult to address. It was very difficult to find a useable measure of convention attendance. Simply counting meetings attended is far too simplistic (for this doesn’t take into account how many years someone might be a member). We considered only recent meeting attendance (the past 3 years), selected only those individuals who were MLA national members for the past 3 years, and split this group into the categories of 0,1,2 or 3 meetings attended, then collapsed categories (e.g. 0 vs. 1-3 meetings attended).

Interestingly enough, there seems to be little connection between attendance at MLA national meetings and the actual amount of institutional support for attending meetings. There is a significant difference between the attendance of those who receive no support and those who receive some support, but there does not seem to be a clear relationship between meeting attendance and the amount of institutional support. Something, clearly, is a lot better than nothing. (Again, this applies only to the last 3 meetings. Also, giving people an option of reporting $ or % support just diffused the data, making it harder to regroup it and find meaningful results.)

There may be significant differences in national conference attendance by the types of libraries people work in. One test indicates that academic librarians are significantly more likely to have attended recent conferences (past 3) than public librarians. Whether or not the same is true for academic librarians vs. other categories of MLA members cannot be determined by the present data, since cell sizes are too small to achieve significant results.

The entire issue of conference attendance is somewhat troubling and the data are insufficient to draw unambiguous conclusions. It is especially troubling that the major reason for non-attendance is financial (54% of respondents), yet the data do not show a clear connection between the amount of support and attendance, only that there is a connection between attendance and some financial support vs. no institutional support. This issue deserves further research.

Several questions for MLA national members were also asked of chapter members. Many of the results are similar, not surprisingly. The percentage of academic librarians is even greater for local chapters than for the national organization. Since the focus of the questionnaire was not on chapters, there is no way to tell if the concerns noted above about participation of academic librarians as opposed to others also applied to chapters. Just as with the national organization, chapter members noted contact with other members as the most important aspect of chapter membership and meeting attendance.

We had hoped to investigate the characteristics of non-members who might be potential members. Unfortunately, it was not possible to obtain a random sample of such individuals, so results presented in the data summary should be taken with some caution. In general, it seems that on many variables non-members parallel members but the data are a little more diffuse. For example, many non-members have graduate degrees in music, but not as high a percentage as members. Nonmember salaries are significantly lower. It seems that a lower percentage are professional librarians. Unfortunately the reasons for not joining the national organization are quite diverse. Although this is not a large group of people, it might be worthwhile to follow up on these individuals to see if the national organization can be of use and service to them.

We believe this project has illuminated many characteristics of MLA members and their relationship to the organization. We hope it confirms much of what we know to be the best features of MLA members: their diversity, their educational and professional achievements, and their involvement with MLA. We also hope it confirms what we hope are the best feature of the organization: equality of opportunity and for participation in the work of the organization, and programs which speak to a diverse but committed membership.

No one survey can hope to accomplish everything, answer all questions it set out to answer, or provide data to speak fully to concerns not part of the design of the survey. Nevertheless, we believe we have considered many critical questions, and, even in cases where we were unable to have answers as complete as we would like, we believed we have framed the questions and helped to identify the data necessary to answer those questions

This report suggests several areas which bear further investigation, and readers of the report likely will find yet other questions to explore. Topics which come readily to mind include salary (which had the largest negative response of all the work-related issues), meeting attendance and support for meeting attendance, the relationship of meeting attendance to participation in other MLA activities, the possible variability of participation in MLA activities by different subgroups within the organization, and the relevance of MLA to potential members (such as chapter members or meeting attendees). There are many points of contact with the earlier self-study and with Plan 2001. We hope some of the points of contact will be explored.

The Working Group would like to thank the members of the various boards involved with the project for their support, and especially would like to thank the MLA members who participated in the survey.

July 1999; rev Jan 2000