3. DATA DESCRIPTION AND EXPLORATORY ANALYSIS

3.1 Analysis Outline

3.2 Survey Structure

Because there are an unbalanced number of respondents in each group, it is important to remember that the overall sample results will be heavily influenced by the disproportionate number of prospective telecenter users. However, this effect only applies to certain questions where the study group means are significantly different. In these cases, the overall mean is biased towards the average response of telecenter users. In general, the small sample sizes for all three groups make the results susceptible to sampling error. Thus, the average response of the home-based telecommuters surveyed, for example, is not likely to provide a good estimate of the true mean response of all home-based telecommuters. Of the three study groups, the center-based telecommuters are best represented due to the high number of respondents in the sample and the low number in the population. Additionally, self-selection is another likely source of error. The control group members, all of whom volunteered for the study, are likely to have more experience with telecommuting than an average employee. Familiarity with the topic may bias the results in favor of telecommuting since those with previous telecommuting experience may be prevented from telecommuting in their current position. Although there are a number of likely sources of error, this likelihood of error is considered acceptable in an exploratory study.

For clarity to the respondent and convenience of the designer, the survey is divided into seven parts: job characteristics, job satisfaction, perception of different work environments, the importance of work environment characteristics, the amount of telecommuting, travel characteristics, and sociodemographic information. The important results from each part are described below in corresponding sections. The surveys for the different groups are nearly identical with only slight modifications to a few questions in order to accommodate the different sample groups.

3.3 Part A: Job Characteristics

The respondents worked under a variety of work schedules. Although a plurality worked a conventional work week (37.1%), many worked under fixed (21.6%) and variable (17.5%) flextime arrangements. Others worked a compressed work week (11.3%) or were employed part-time (9.3%). Three respondents had schedules that did not fit these common definitions. The only noteworthy differences across groups are that the non-telecommuters had the smallest percentage of conventional workers (33.3%) and that most of the part-time workers are prospective telecenter users (see Table 3-2).

Each respondent was asked to classify the time spent during the week into five work activity categories (see Table 3-3). Respondents said that they spent approximately 48% of their time working independently. They split the time working with others evenly into face-to-face (e.g., in-person meetings) and remote (e.g., phone calls) communication. Work at a specific location and work-related travel each accounted for about 7% of the work day. Not surprisingly, the home-based telecommuters said they spent more time working independently (56%); however, prospective telecenter users spent the least time of all groups in this category (44%). Those planning to use the telecenter spent almost three times as much time on location-dependent work and on work-related travel than the other two groups.

3.4 Part B: Job Satisfaction

The final section of part B asks the respondent to give an opinion on various job satisfaction components (for the averages for each question by study group, see Appendix B, Charts B3A to B3N). An analysis of variance (ANOVA) was conducted to determine if there were significant variations among the study groups in the mean responses to each question. Because this is an exploratory study, the level of significance is set relatively high at 0.1. Despite the high level of significance, the group averages were significantly different for only three questions (see Table 3-4). Looking at all the questions in order, the following observations may be made. On average, respondents communicate well with their supervisor, with non-telecommuters communicating most effectively. Most respondents feel that they have a good opportunity for promotion: members of the telecommuting groups feel they have better chances overall than non-telecommuters (significant at a p-value of 0.049). The members of the sample, on average, have enough resources to do the job, with prospective telecenter users most satisfied by resource availability. In general, respondents feel they are part of an effective work team and work well with their supervisor.

Overall, the sample members who had clients (83.5%) are not particularly bothered by their demands, but home-based telecommuters feel significantly (at level of confidence of 0.1) that clients' demands do take too much of their time. For overall job satisfaction, the expected telecenter users (4.0) are slightly more satisfied than the control groups (3.8). Prospective telecenter users also feel a greater sense of accomplishment than the other groups. The respondents from different groups feel significantly different levels of supervisor appreciation (p = 0.0301). Again, those planning to use the telecenter (4.1) feel more appreciated by their supervisor than either home-based (3.4) or non- telecommuters (3.5). However, these perceived differences in supervisor appreciation may be due to the gratification of the study group at having just been selected to telecommute.

All groups have similar, strong positive opinions about their relations with co- workers. Although all groups are very confident in their abilities, the non-telecommuters are less confident than the other two groups. The non-telecommuters also stand out as relatively more likely to look for a new job (3.1). Of the 35 (36%) respondents who supervised other employees, home-based telecommuters agree most strongly (4.5) that they worked well with them, and prospective telecenter users agree least strongly (4.0). Finally, most respondents do not find their job boring or tedious; home-based telecommuters more strongly expressed this sentiment than the other two groups.

3.5 Part C: Work Environment

Unlike the part B questions, the analysis of variance for part C has numerous significant findings. Differences among study groups are significant for half of the thirty questions. The most significant p-value is for the question on not having equipment at the workplace, in which the non-telecommuters have the mean closest to agreement with the question. Other significant questions (p-values less than 0.002) include the commute hassle and beneficial to the environment, where center-based telecommuters have different averages compared to the other two groups. Differences among workplaces are significant for almost every question. Only the opinions on household conflicts and having work judged by the results show similar responses across all workplaces. Two-way interactions are also significant for slightly more than half of the questions (16 out of 30). Seven questions have significant two-way interactions with p-values less than 0.0005. They include the three questions mentioned previously for study groups, as well as motivation, distractions, and both cost opinions.

3.6 Part D: Importance of Work Characteristics

The lowest average importance scores are not as extreme as the highest scores since the lowest score is still above 2.0, which corresponds to a rating of slightly important. Except for home-based telecommuters (who found dependent care at least relatively important), all groups considered the ability to care for dependents to be unimportant. This result is probably influenced by the respondents in each group who do not have dependents. Also, home-based telecommuters rated working in a professional environment as the least important factor. This group also stated that interacting socially at work and dressing "the way I like" were unimportant. The latter result is surprising since the ability to dress comfortably is often touted as an advantage, albeit one of minor importance, of telecommuting from home (USDOT 1993). Telecenter users also felt that social interaction at work was unimportant.

3.7 Part E: Amount of Telecommuting

When asked from which location they prefer to work, the respondents, on average, allocated only 50% of their time to the regular workplace (see Table 3-8). The telecommuting center was the next most preferred site at 26.5%, and home was given 19%. The remaining 4.5% was allocated to site visits and other various work locations. The variation in these averages is quite high; the percentages were significantly different for each study group. Telecenter users allocated a nearly identical amount of time to the regular workplace and the telecenter with home and other locations making up only about 15% of the total. Home-based telecommuters, on the other hand, split their time evenly between the regular workplace and home. Finally, the non-telecommuters prefer to spend the majority of their time at the regular workplace, 25% at home, and a meager 4% of the work time at a telecenter. Thus, it appears that each group prefers the arrangement it is currently (or soon will be) using.

The last five questions in part E compare telecommuting from a center and from home. Almost 92% of the sample think that the nature of the job allows them to work from the telecommuting center for some amount of time which compares to only 81% for working from home. In the prospective telecenter users group, approximately 20% are unable to telecommute from home because of the nature of the job. Respondents also felt that their supervisor was more willing to let them work at the telecenter (90%) than at home (76%). Even more of the anticipated center-based telecommuters (26%) have supervisors who are unwilling to allow them to telecommute from home. So, for a sizeable portion of the prospective center users, the telecenter is the only telecommuting option available.

Although the survey was to have been administered prior to the start of telecenter use, it was not always possible to reach a respondent or to have a survey ready before telecommuting actually began. Thirty-seven percent of the respondents stated that they currently telecommute from a center. This percentage is composed of 34 telecenter users (56%) and 2 home-based telecommuters. As a result of the confusion in the survey administration, this question is discarded from further consideration. Almost half of the sample currently telecommutes from home: 15 home-based telecommuters (100%), 24 telecenter users (40%), and 6 non-telecommuters (40%) (who do it infrequently). In contrast, approximately 80% of those surveyed would like to telecommute from the telecenter or from home for some time assuming there were no work-related constraints. Many home-based telecommuters were uninterested in working at the center (33%), and likewise, some telecenter users did not want to work at home (20%). Finally, two-thirds of the respondents expected to be telecommuting at least some of the time six months from completing the survey. Again, some telecenter users did not expect to be telecommuting from home (33%), and many home-based telecommuters did not expect to be telecommuting from a center (66%). However, two non-telecommuters and an additional two home-based telecommuters were expecting to be telecommuting from a center. Overall, the results from the choice, preference, and expectation questions indicate that simultaneous home and center telecommuting appears to be a popular option.

3.8 Part F: Travel Characteristics

3.9 Part G: Sociodemographic Characteristics

The average household size is 2.7 persons. On average, non-telecommuters have a smaller household at 2.3 persons, and home-based telecommuters have a larger household at 3.3 persons (see Table 3-11). Only 7.2% of those sampled have a household member, other than preschoolers, who needed special care. Thirty-two percent have children under the age of six, and 46.4% have children under the age of 16. Corresponding to the household size, fewer non-telecommuters and more home-based telecommuters have children than the center-based telecommuters. Finally, each household has, on average, 1.4 full-time and 0.3 part-time workers. However, there are slightly more part-time workers (0.5) in home-based telecommuter households and slightly fewer (0.1) in non-telecommuter households than on average.

In the total sample, only one respondent (a non-telecommuter) did not have a driver's license. On average, there were 1.9 licensed drivers per household. Again reflecting average household size, non-telecommuters had fewer and home-based telecommuters had more drivers per household than center-based telecommuters. There were 1.9 vehicles available to the household for trips. Non-telecommuters had fewer vehicles (1.6) than the other two groups. Finally, there are 1.1 vehicles per licensed driver for the study group. Prospective telecenter users had more vehicles per driver and non-telecommuters had fewer.

The respondents are asked to give the highest level of education they have attained (see Table 3-12). Overall, forty percent of the respondents have college degrees, an additional 9.3% have taken some graduate school, and another 26.8% have completed at least one graduate degree. Of the remaining respondents, most have some college education (23.9%), and only three percent have simply a high school diploma. The telecenter users have the largest proportion of respondents with graduate schooling.

The distribution of annual household income before taxes is skewed to the higher categories (see Table 3-13). The two largest income brackets are $35,000 to $54,999 (25.8%) and $55,000 to $74,999 (24.7%). The highest categories are also well represented with 19.6% who make $75,000 to $94,999 a year and 16.5% who make more than $95,000 a year. Approximately twelve percent earn less than $35,000 a year. The members of the extreme income groups, both less than $35,000 and greater than $95,000, are nearly all member of the telecenter group. It should be noted that the income is measured at the household level, so the number of workers in the household will affect the income level.

3.10 Implications for Preference Modeling

Return to the Abstracts of Non-Refereed Publications page.
Return to the Main page.