3. ATTITUDINAL SURVEY ANALYSIS

3.1 Quasi-Experimental Design Structure

The data collection process was designed to make comparisons along three dimensions. First of all, both employees and their managers were polled on their attitudes about telecommuting. Both surveys focus on job performance and satisfaction of the employee, perceptions of telecommuting, and feasibility of telecommuting for the employee. Identical or parallel wording was used between employee and manager surveys to the extent possible. The manager surveys also include a section on the organization's opinions about telecommuting.

Three study groups make up the second survey dimension: center-based telecommuters, home-based telecommuters, and non-telecommuters. The center-based telecommuters surveyed in this study come both from the RABO telecenters and from two southern California non-RABO telecenters which the university was contracted to evaluate. Members of the other two groups (including both employees and their managers) were recruited from the same organizations through the telecenter users or their employers and were chosen, where possible, to have a position similar to that of the telecenter user or the user's manager, respectively. The control groups were included to see how center-based telecommuting differed from more familiar work arrangements. Where changes in attitude and behavior over time are noted in the telecenter group, the control groups help to determine whether those changes are a result of telecommuting from a center, common to telecommuting in general, or a consequence of background changes affecting all workers. Thus, the control group surveys were designed to parallel the telecenter group surveys as closely as possible, with the obvious exception that the latter surveys contained some questions about the telecenter experience that were not applicable to the other groups.

Finally, the surveys were administered in two waves. In the first wave, the prospective telecenter user and the associated study group employees and managers were to be given surveys before telecommuting from the center began. Although this was the ideal situation, in some cases the surveys did not reach the respondents until soon after center use had already begun. Starting in April 1995, the second wave of surveys was distributed (5 to 18 months after the before surveys, an average of 9.0 months) in order to obtain opinions after familiarity with center-based telecommuting had been achieved. The two survey waves allow for an analysis of how perceptions of center-based telecommuting change with actual experience since the before wave benchmarks job satisfaction and performance levels. Although some modifications were made between the before and after versions, the surveys are primarily the same containing many identically worded questions.

The data tabulated in this interim report comes from a subset of the total surveys returned. To allow sufficient time for after data collection, analysis, and reporting, only those respondents who returned their initial surveys before December 5, 1994, are included in this analysis. The second wave surveys were mailed to telecenter users who had spent at least five months as project participants and to the associated study group members. For this report, the cutoff date for after survey data entry was July 21, 1995. Thus, the before surveys were completed during the period from July 1993 to December 1994, and the after surveys from April through July 1995. For both waves, approximately one-fourth of the data comes from non-RABO sites.

The number of surveys received in each of the categories is shown in Table 3-1. Additional before and after surveys have been received since the cutoff dates by both existing and new participants associated with both existing and newly opened telecenters. This additional data will be analyzed in conjunction with the final report.

Type	Wave	Study Group			Total
		Center-based Telecommuters	Home-based Telecommuters	Non-telecommuters
Employees	Before	61	15	21	97
	After	39	11	17	67
Managers	Before	33	10	6	49
	After	28	4	2	34

This interim report focuses on the attitudinal surveys for telecenter users and their managers, only (see Appendices E and F). Comparisons to the other study groups will be conducted in the future as resources permit. Key findings for the employee telecenter user surveys are discussed in Section 3.2, which describes the characteristics of the telecenter users and compares certain characteristics before and after the beginning of telecommuting. Later on in the report, Section 5.2 presents a comparison of the traits of those that continued to telecommute from a center with those who dropped out of the program. Section 3.3 covers the surveys for managers of telecenter users. Finally, the responses of employees and managers are compared and contrasted in Section 3.4.

3.2 Employee Survey Results

Unless otherwise specified, the findings reported below come from the 39 after surveys (see Appendix E) returned by the center-based telecommuters (see Table 3-1). Where changes over time are of particular interest, before and after responses are compared for the 27 respondents who have completed both versions of the telecenter user survey.

Due to overlapping survey distribution and data entry cutoff dates, the total response rate for either the before or after surveys is too complex to determine easily at this point in the evaluation. However, for the pre-12/5/94 sample, 27 of the 39 respondents to the after survey also completed the before survey (the remaining twelve surveys came from non-RABO respondents who were telecommuting from a center prior to the start of the RABO project). Thus, one can calculate a response rate that measures the number of surveys received from those who were eligible to receive the after surveys. Of the 61 before surveys received, 30 respondents had quit before the after surveys were distributed in April 1995. As a result, the after survey response rate for participants who also submitted a before survey is equal to 27 divided by 31 (61 - 30), or 87.1%.

The employee surveys contain six sections: job characteristics, job performance and satisfaction, work environment characteristics, the amount of telecommuting, travel characteristics, and demographic information. Findings from each of these sections are discussed below, although in a different order from their appearance in the survey. In particular, demographic information is presented first in order to characterize the nature of the sample.

3.2.1 Demographic Characteristics

Since the survey was designed to be administered to the same people in both waves, demographic information that was not expected to change was obtained only on the initial survey. As a result, the characterization of the after data set in terms of gender and age is restricted to the 27 respondents who returned both surveys. (Demographic data are now being collected for those who have only completed the after surveys, thus augmenting the information that will be analyzed in the final report). The sample population has more females, 16 (59.3%), than males, 11 (40.7%). Nearly half the sample falls into the median age category of 35 to 44 years of age (48.1%). The adjacent categories, from 25 to 34 and from 45 to 54 years of age, are the next largest at 22.2% each. The remainder, 7.4%, goes to the next highest category, 55 to 64 years of age (see Table 3-2).

Characteristic	Number (Proportion)	Characteristic	Mean (Std. Dev.)
Female1	16 (59.3%)	Household size	2.97 (1.33)
Age 25 to 34 years1	6 (22.2%)	Full-time workers	1.54 (0.64)
Age 35 to 44 years1	13 (48.1%)	Part-time workers	0.15 (0.37)
Age 45 to 54 years1	6 (22.2%)	Vehicles per household	2.29 (1.13)
Age 55 to 64 years1	2 (7.4%)	Vehicles per worker	1.38 (0.51)
Dependent care	2 (5.1%)	Vehicles per licensed driver	1.13 (0.33)
Children less than 6	13 (33.3%)	Licensed drivers per household	2.00 (0.65)
Children less than 16	21 (53.8%)

¹ N = 27

In the full data set (39 respondents), the average household size was 3.0 persons. Only two persons (5.1%) of those sampled had a household member who needed special care. One-third have children under the age of six, and more than half (53.8%) have children under the age of 16. Each household had, on average, 1.5 full-time and 0.2 part-time workers. Between the survey administrations (based on the reduced sample size of 27), the number of full-time workers increased slightly (from 1.4 to 1.5), and the number of part-time workers fell slightly (from 0.3 to 0.2). No one in the sample is without a driver's license. There are 2.0 licensed drivers per household and 2.3 vehicles available to the household for trips. Finally, there are 1.1 vehicles per licensed driver.

The respondents on the whole were very well educated. Again using the reduced data set, forty-four percent held only college degrees, an additional 7.4% had taken some graduate school, and another 22.2% had completed at least one graduate degree (see Table 3-3). Of the remaining respondents, most had some college education (18.5%), and only 7.4% had simply a high school diploma.

The data from the after surveys show that annual household income before taxes is skewed towards the higher categories (see Table 3-3). The most frequent annual household income bracket is $75,000 or more (43.6%). The next two categories, $35,000 to $54,999 and $55,000 to $74,999, each have 25.6% of the sample. The remaining 5.1% earned between $25,000 and $34,999 per year.

Education Category (N = 27)	Number (Proportion)	Household Income (N = 39)	Number (Proportion)
High school graduate	2 (7.4%)	Less than $25,000	0
Some college	5 (18.5%)	$25,000 to $34,999	2 (5.1%)
College degree	12 (44.4%)	$35,000 to $54,999	10 (25.6%)
Some graduate school	2 (7.4%)	$55,000 to $74,999	10 (25.6%)
Graduate degree	6 (22.2%)	$75,000 or more	17 (43.6%)

3.2.2 Job Characteristics

The first section of the survey asks questions about the type of work the respondent does (see Table 3-4). Returning to the full after data set, the majority of the sample (53.8%) classified their current position as professional/technical. The other categories are manager/administrator (23.1%), sales/marketing (12.8%), and administrative support (10.3%). Although there is a wide range in responses, respondents had worked for their immediate supervisor for 3.1 years on average, for their present employer for 8.2 years, and in their present occupation for 10.2 years. The respondents reported working an average of 83.2 hours in a two-week period.

The respondents worked under a variety of work schedules (see Table 3-5). The two top categories are variable flextime (38.5%) and fixed flextime (25.6%), with the conventional schedule (15.4%) coming in at a distant third. Additionally, there were some workers on a compressed work week (15.4%) and some working part-time (5.1%).

Job Type	Number (Proportion)	Work Duration	Mean (Std. Dev.)
Manager / Administrator	9 (23.1%)	Years worked for supervisor1	3.12 (3.02)
Professional / Technical	21 (53.8%)	Years worked for employer2	8.21 (6.53)
Administrative support	4 (10.3%)	Years worked in occupation	10.23 (6.80)
Sales / Marketing	5 (12.8%)	Hours worked in two weeks	83.23 (10.97)

¹ N = 36    ² N = 37

Work Schedule	Reduced (N = 27)		Full (N = 39)
	Before	After	After
Part-time	4 (14.8%)	2 (7.4%)	2 (5.1%)
Conventional	3 (11.1%)	6 (22.2%)	6 (15.4%)
Fixed flextime	9 (33.3%)	7 (25.9%)	10 (25.6%)
Variable flextime	8 (29.6%)	10 (37.0%)	15 (38.5%)
Compressed work week	3 (11.1%)	2 (7.4%)	6 (15.4%)

It is of interest to analyze the impact of telecommuting on the type of work schedule. One might expect more flexible schedules to become more popular with the use of telecommuting centers. However, in some programs, employees have been required to choose between telecommuting and other forms of flexible work which suggests that an increase in telecommuting may result in a decrease in other flexible schedules. The reduced data set shows that respondents did change their work schedules between survey measurements. The number of part-time and fixed flextime workers dropped slightly (by 2 respondents each), while the number of conventional and variable flextime workers went up (by 3 and 2 respondents, respectively). The data show that both less flexible and more flexible schedules increased, thus supporting the hypothesis that either effect could be expected.

The survey asked each respondent to classify the time spent on the job into five activity categories (see Table 3-6). Respondents reported that they spent approximately 47% of their time working independently. They split the time working with others evenly (about 18% each) into face-to-face and remote communication. Work at a specific location (9.2%) and work-related travel (6.8%) accounted for the rest of the work day. Although there may be variations at the individual level, the two sets of responses for the 27 who answered both survey versions shows that the aggregate percentages remain essentially the same between the two survey waves.

Work Activity	Reduced (N = 27)		Full (N = 39)
	Before	After	After
Independent	45.5% (27.2)	48.7% (26.2)	47.4% (26.4)
Face-to-face	17.5% (8.7)	16.5% (9.6)	19.1% (12.1)
Remote	18.1% (15.6)	17.8% (19.6)	17.5% (17.7)
At a specific location	7.8% (12.3)	9.3% (9.6)	9.2% (9.6)
Travel	10.7% (14.6)	7.8% (10.2)	6.8% (9.0)
Other	0.4% (1.9)	0%	0%

3.2.3 Job Performance and Satisfaction

Respondents were asked a number of questions on their perceptions regarding their job, the first of which asks the respondent to rate his/her job performance according to four aspects. The averages for the four aspects all fall near 4.0 (that is, "good") on a five point scale. The telecenter users rated their amount of work done as 4.26, work quality as 4.44, ability to meet deadlines as 4.18, and productivity as 4.36 (see Table A-1, Appendix A). The balanced data set shows minimal change between before and after measurements, indicating that telecommuting did not significantly affect the respondents' opinions of their work performance (see Table B-2, Appendix B). However, all questions except the ability to meet deadlines were rated slightly higher on the later survey.

The second question asks respondents how their supervisor would rate them on the same four work aspects. The average scores are very similar to the first question with only the quality of work being slightly lower at 4.32 (see Table A-1, Appendix A). The later measurement for the reduced data set shows that the respondents thought their supervisor would rate them slightly lower on the four aspects after using the telecenter. However, only for the ability to meet deadlines is the decrease significant according to a t-test of the means (see Table B-2, Appendix B).

The final section asks the respondent to give an opinion on various job satisfaction components, again on a five-point scale. Looking at all the questions in order, the following observations may be made (although there were both positively- and negatively-oriented questions, the average scores reported below have all been changed to positive orientation, see Table A-1, Appendix A). On average, respondents feel they communicate well with their supervisor (4.2), have a good opportunity for promotion (3.6), have sufficient resources to do the job (3.5), are part of an effective work team (3.9), and work well with their supervisor (4.1). Most respondents found their job to be not tedious or boring (4.3), working gave them a sense of accomplishment (4.0), and they had appreciative supervisors (4.1). The telecenter users also were very confident of their work ability (4.5), got along very well with their co-workers (4.5), and were unlikely to look for a new job (3.8). Overall, the sample members who had clients (38%) are not particularly bothered by their demands (3.5). Of the 24 respondents who supervised other employees (62%), most worked well with them (3.9). The average response for overall satisfaction was 4.1. The average responses from the before and after surveys for the balanced data set show much variation, but none of the changes are statistically significant (see Table B-2, Appendix B).

3.2.4 Work Environment Characteristics

The questions in the third section of the survey ask about the respondent's attitudes toward characteristics of the work environment at three different work locations (see Table A-2, Appendix A). Similarly to the job satisfaction questions, t-tests were conducted for the responses to the work characteristic questions to look for significant differences between the before and after surveys on the reduced data set (see Table B-3, Appendix B). Very few of the averages changed significantly after telecommuting from the center began. However, distractions from others were more of a problem at the telecenter than originally envisioned (from a mean of 1.81 to 2.37, at a p-value of 0.029).

For the full after data set, the means for each question are graphed according to each of the three workplaces (regular workplace, telecommuting center, and home) in Figure A-1 of Appendix A. A one-way analysis of variance was conducted for each question to test for significantly differing means among workplaces. All questions except two show significant differences among workplaces (see Table A-3, Appendix A). Having work judged by the results and communication with the supervisor were unaffected by a change in workplace location.

On the before survey, a section was included where respondents were asked to rate the importance of the same work environment characteristics on a four-point scale (see Table A-4, Appendix A). Using the balanced data set, the characteristics that have the highest and lowest average importance rankings overall are shown in Tables 3-7 and 3-8 respectively. The prospective telecenter users feel that working effectively, having the needed equipment, and having their work judged by its results were very important. They also rate supervisor communication and a hassle-free commute as important 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. The respondents rate the ability to care for dependents the least important; however, this result is probably influenced by the number of respondents who have dependents.

3.2.5 Amount of Telecommuting

At the time of the after survey wave, the center-based telecommuters had, on average, about one year's experience with telecommuting from the center. (This estimate includes both the RABO and non-RABO telecenter users, the latter of whom had been telecommuting much longer, on average). Less than half of the sample (43.6%) reported that they had also telecommuted from home for an average of 2.84 years. In the initial survey, 19 out of 27 (70.4%) had not telecommuted before participating in the study.

When asked about the ideal distribution of work time among various locations (see Table 3-9), the respondents, on average, allocated nearly equal amounts of their time to both the regular workplace (44.6%) and to the telecommuting center (41.5%). Home was a distant third at 8.6%, with other locations, such as site visits and sales meetings, making up the remaining 5.4%. Although the variation within the data is quite high, the averages themselves remained basically the same for both versions of the survey (for those who answered both). The respondent was also asked to distribute his/her work hours if the telecenter were unavailable. In this case, the members of the sample would spend two-thirds of their time at the main office (67.8%), 23.7% at home, and 8.5% at other locations, which evenly divides the time allocated to the center in the previous question between the regular workplace and home.

The ideal distribution of time at the various workplaces can also show which combinations of work locations are preferred. Table 3-10 shows the frequency of selecting the possible workplace combinations from the after survey data. Not surprisingly, the most preferred alternative was to work at both the regular workplace and the telecenter (41.0%) and, on average, to split the work time evenly between the two locations. However, the next most preferred alternative was to work at home in addition to the previous two locations (23.1%). So, even though the sample is composed of telecenter users, a sizeable percentage are still interested in doing some telecommuting from home. If the other work location is included with the regular workplace (both of which can be considered essential to the job), then nearly all participants fall into two work arrangement categories: regular workplace/telecenter (53.8%) and regular work-place/telecenter/home (30.8%).

The survey included a question designed to find out what the telecommuter does with the time saved by telecommuting. The most frequently checked responses (more than one response could be marked) were spending time with family or friends (66.7%), getting more sleep (51.3%), and relaxing by myself (43.6%). Other important activities included housework or yardwork (38.5%) and exercise (35.9%). Equal numbers of respondents marked working, shopping, and participating in an outdoor activity (25.6%). Only 4 respondents (10%) said there was no significant time saved. When asked which one of the previously-checked activities they did most frequently during the time saved by telecommuting, a sizeable portion of the respondents cited spending time with family or friends (38.5%). Other "most frequent" activities included working (12.8%) and housework or yardwork (10.3%).

Workplace Combination	Number (Proportion)	Mean Ideal Distribution of Time1
		RW	TC	HM	OL
RW/TC	16 (41.0%)	50.2%	49.8%	---	---
RW/TC/HM	9 (23.1%)	55.6%	28.7%	15.8%	---
RW/TC/OL	5 (12.8%)	44.0%	29.0%	---	27.0%
ALL	2 (5.1%)	22.5%	27.5%	22.5%	27.5%
TC/HM	2 (5.1%)	---	70.0%	30.0%	---
RW/HM	1 (2.6%)	80.0%	---	20.0%	---
TC	1 (2.6%)	---	100.0%	---	---
TC/HM/OL	1 (2.6%)	---	40.0%	50.0%	10%
Total	37 (94.9%)	44.6%	41.5%	8.6%	5.4%
Missing	2 (5.1%)

¹ Workplace locations are the regular workplace (RW), the telecommuting center (TC), home (HM), and other location (OL).

A set of five questions compares telecommuting from a center and from home. The first two questions ask about the possibility of telecommuting based on job suitability and manager support, respectively, and the last three ask for the current, preferred, and predicted amount of telecommuting (see Table 3-11). The mean telecommuting frequencies given in Table 3-11 were calculated by averaging the responses weighted by the average for each frequency category as shown in the lower section of the table.

For the full data set, telecenter users thought their jobs were suitable for telecommuting from a center about 40% of the time, on average. However, they only currently telecommuted from a center 30% of the time even though they wanted to use the center 50% of the time. Average frequencies of telecommuting from home were lower than from a telecenter for all five questions. Respondents felt that their jobs were suitable for telecommuting from home only about 27% of the time on average, suggesting that some parts of their jobs could be done remotely but required the context of a typical office environment. Accordingly, the telecenter users chose to telecommute from home much less frequently (9.6% of the time, on average) than from a center and preferred to telecommute from home much less frequently (20%). Over the next six months, the respondents predicted a slight increase in the frequency of both forms of telecommuting. (The levels of actual telecenter use based on the attendance logs are described in Section 4.4.2).

It is of interest to analyze how the responses to these five questions change with telecenter experience. For the reduced data set, the frequency of telecommuting was significantly different between the two waves for three of the questions according to a pairwise t-test of the means (see Table B-1, Appendix B). After gaining experience with telecommuting from the center, respondents thought that their supervisors were likely to allow them to telecommute from home more frequently than before (p = 0.047). Thus, familiarity with one form of telecommuting makes the adoption of other forms more likely. Not surprisingly, the frequency of telecommuting from the center increased significantly (p = 0.021). However, the change between the survey waves is not as great as might be expected due to a number of respondents who had already begun to telecommute from the center. 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. In fact, fifty-two percent of the respondents (from the reduced data set) stated that they currently telecommuted from a center on the before survey, although most of those had probably been telecommuting less than a month. Additionally, the expected frequency of center-based telecommuting declined between surveys (p = 0.013). Whereas respondents initially expected on average to be telecommuting from a center 50% of the time within six months, their predicted future telecommuting frequency was only 38% after having telecommuted for some time, which is essentially equal to their current actual frequency of 39%. Thus, the later estimate appears to be a more realistic alignment of expectations with reality

¹ N = 38     ² N = 37     ³ N = 26     ⁴ N = 26
⁵ Bolded after means are significantly different from their before counterparts at a .LE. 0.05. See Table B-1, Appendix B for the test statistics.

Survey Question Definitions

Job	Considering the requirements of your current job, how much do you think the nature of your job would allow you to telecommute . . . ?
Supervisor	Considering the characteristics of your current supervisor, how much do you think your supervisor would let you telecommute . . . ?
Choice	How much do you currently telecommute . . . ?
Prefer	Assuming that there are no work-related constraints, how much would you like to telecommute . . ?
Expect	Six months from now, how much do you expect to be telecommuting . . . ?

c. Values Used in Frequency Calculation

Frequency Category	Definition	Assigned Value
Not at all	0 days per month                         0/22 = 0	0
Less than once a month	0.5 day per month                       0.5/22 = 0.0227	0.0225
About 1 - 3 days a month	2 days per month                         2/22 = 0.091	0.09
1 - 2 days a week	1.5 days per week                       1.5/5 = 0.30	0.30
3 - 4 days a week	3.5 days per week                       3.5/5 = 0.70	0.70
5 days a week	5 days per week                          5/5 = 1.0	1.00
Occasional partial days	Equated with "Less than once per month"	0.0225

Comparing the distribution of the frequency categories for the five questions shows some important differences between telecommuting locations (see Table 3-12). Although all respondents felt that the nature of the job allows them to work from the telecommuting center for some amount of time, approximately 20% of the respondents considered themselves unable to telecommute from home for the same reason. Even more respondents (about 30%) felt that their supervisor was unwilling to let them work at home at all. Cross-tabulating these two results shows that the option of telecommuting from home is not available to almost 40% of the telecenter users which indicates that centers may help spread the transportation and other benefits of telecommuting to a larger segment of the workforce than home-based telecommuting alone would. Setting an arbitrary telecommuting frequency criterion at one day per month or more, the full after survey results show that 79.5% of the respondents meet this level of telecommuting at the center. Using the same marker, 20.6% currently telecommute from home. In fact, 12.5% spend one day per month or more working at both telecommuting locations.

From a Center1	Not at all	Less than once a month	About 1-3 days a month	1-2 days a week	3-4 days a week	5 days a week	Occasional partial days
Job	0	0	9 (23.1%)	18 (46.2%)	8 (20.5%)	4 (10.3%)	0
Manager2	0	0	8 (20.5%)	20 (51.3%)	4 (10.3%)	6 (15.4%)	0
Choice	1 (2.6%)	6 (15.4%)	10 (25.4%)	11 (28.2%)	9 (23.1%)	1 (2.6%)	1 (2.6%)
Prefer	0	1 (2.6%)	2 (5.1%)	15 (38.5%)	17 (43.6%)	3 (7.7%)	1 (2.6%)
Expect	4 (10.3%)	0	9 (23.1%)	16 (41.0%)	9 (23.1%)	1 (2.6%)	0
From Home1
Job	7 (17.9%)	3 (7.7%)	5 (12.8%)	18 (46.2%)	2 (5.1%)	3 (7.7%)	1 (2.6%)
Manager2	11 (28.2%)	1 (2.6%)	3 (7.7%)	15 (38.5%)	2 (5.1%)	4 (10.3%)	2 (5.1%)
Choice2	25 (64.1%)	4 (10.3%)	0	6 (15.4%)	1 (2.6%)	1 (2.6%)	1 (2.6%)
Prefer2	12 (30.8%)	3 (7.7%)	5 (12.8%)	12 (30.8%)	2 (5.1%)	2 (5.1%)	2 (5.1%)
Expect3	20 (51.3%)	4 (10.3%)	3 (7.7%)	5 (12.8%)	2 (5.1%)	1 (2.6%)	2 (5.1%)

¹ See Table 3-11 for the text of the survey questions.
² N = 38     ³ N = 37

It appears that the respondents would like to work from the telecommuting center more often than they are currently doing so since the three highest frequency categories are much larger than for the question on current amount of telecommuting. Also, a good portion of the telecenter users do not want to work at home (30%). Finally, although most respondents expected to be telecommuting more often six months after completing the survey, four of the respondents (10.3%) planned to stop using the telecenter. Again, some telecenter users did not expect to be telecommuting from home (51.3%). Overall, the results from the choice, preference, and expectation questions indicate that combined home and center telecommuting appears to be a popular option.

3.2.6 Travel Characteristics

On average, the commute to the regular workplace took 62.2 minutes to travel 44.2 miles. The average commute to the telecommuting center took about 10 minutes to travel 7.3 miles. Thus, on days that the participants used the center instead of going to the regular workplace, their commute travel was reduced, on average, by 36.9 miles. (This estimate includes two participants who traveled 5 and 12.5 miles, respectively, farther to the center than to the regular workplace. For the first case, no first-hand explanation is available, but it is supposed that the employee used the center to work with her supervisor, a telecenter user, and/or work at a nearby field site. In the other case, the employee was actually using the center temporarily before obtaining permanent space for a branch office near the center.)

On the initial survey, respondents were asked to divide their total commute between miles on freeways and miles on other roads. For the commute to the regular workplace, 35.9 miles were traveled on freeways compared to 5.9 miles on other roads, on average. The trip to the telecenter was composed of 4.6 freeway miles and 2.4 non-freeway miles. Interestingly, a larger portion of the travel to the telecenter was done on freeways than on surface streets.

Examining the difference between the RABO sites, which were established near residential areas for the most part, and the non-RABO sites, at least one of which is located in a light-industrial park, provides some insights into the effects of telecenter location (see Table 3-13). Although RABO participants had shorter commutes to the telecenter (6.5 vs. 9.1 miles), non-RABO respondents had a greater reduction in travel since their regular workplace was farther away, on average (53.1 vs. 39.1 miles). Non-RABO participants traveled more on freeways (6.8 vs. 4.1 miles) when commuting to the telecenter than RABO participants and approximately the same amount on other roads (2.3 vs. 2.4 miles).

Location	Roadway Type	One-way Distance (miles)
		RABO (N=21)	Non-RABO (N=5)	Total (N=26)
Regular workplace	Freeway	32.5	50.5	35.9
	Surface streets	6.6	2.6	5.9
	Total	39.1.	53.1	41.8
Telecenter	Freeway	4.1	6.8	4.6
	Surface streets	2.4	2.3	2.4
	Total	6.5	9.1	7.0

The section of the survey covering travel characteristics also includes a series of questions about residential relocation. Of the 39 respondents, only 2 (5.1%) had moved in the past year: one moved closer to work and the other moved farther away. Contrary to expectation, the former respondent stated that telecenter use was a somewhat important factor in the relocation decision, while the latter respondent stated that telecenter use was not a factor. A larger number of respondents were considering residential relocation. Of these 7 (17.9%), 5 planned to move closer to work and the other two would move either closer to or farther from work. However, a majority of those planning to move did not believe that telecommuting (5 of 7) or telecenter location (4 of 7) were important factors in the relocation decision. So, although telecommuting is anticipated to allow residents to locate farther from work since the total weekly commute distance decreases, there is little measurable impact for this small sample at this early point in the telecommuting experience. For a more complete analysis of the travel impacts of the telecenters, including a comparison of modes of travel to the regular workplace and to the telecommuting center, see Chapter 6.

3.2.7 Summary of Employee Survey Results

In this interim report, the description of the survey data is restricted to the telecenter users only. Primarily, the data from the after-wave surveys is used to characterize center-based telecommuters (sample size of 39). However, where appropriate, data from both before and after waves are utilized to highlight changes related to the use of telecenters (a reduced sample size of 27). A summary of the results from the six parts of the survey is provided below.

The section on demographics asked for general characteristics, such as age, income, and education. There are slightly more female telecenter users than male ones, and nearly half the sample is between the ages of 35 to 44 years. The average household size of three persons is consistent with the fact that more than half of the respondents have children under 16 years of age. Vehicle availability is high among the respondents with 2.3 vehicles per household and 1.4 vehicles per worker. The telecenter users are highly educated: about 30% have had additional schooling after college. Additionally, many of the participants have high incomes (about 70% have annual household incomes greater than $55,000).

Job characteristics varied among the center-based telecommuters. Slightly over half of the sample hold professional/technical positions which are usually easily adapted to telecommuting, and as a whole, they are experienced in their field with an average length of time in the profession of 10.2 years. Flextime schedules are popular among the telecenter users (used by nearly 65%). Finally, the respondents spend a good portion of their workday working independently (47%) or remotely (18%), both of which are good indicators of positions with telecommutable tasks.

The responses from the attitudinal sections on job performance and satisfaction and work environment characteristics show primarily positive results. There is little change in performance or satisfaction characteristics between survey waves suggesting that working from a telecommuting center does not drastically change these factors. The only significant change was a slight drop in the perceived opinion of the supervisor on the telecommuter's ability to meet deadlines. The ratings on the statements about work characteristics also remained primarily the same between survey measurements. Distractions at the telecenter were slightly more of a problem than originally envisioned, but the average response on the after survey is still to disagree with the statement that distractions were a problem. Finally, the most important work characteristics to the respondents are working effectively, having needed equipment, and having work judged by the results.

The survey also measured the amount of telecommuting the telecenter users had done, are currently doing, and plan to do in the future. The average experience with telecommuting from a center was about one year at the time of the after survey, and about half also had experience with home-based telecommuting. On the other hand, about 40% of the respondents did not have the option to telecommute from home which indicates that centers may help spread the transportation and other benefits of telecommuting to a larger segment of the workforce. With the time saved by telecommuting, the respondents most often spend time with family or friends, get more sleep, and/or relax by themselves.

When distributing their work time for the ideal situation, the respondents preferred to work from the regular workplace and the telecenter about equal amounts, 40% to 45% of their time (each) on average. However, they actually reported telecommuting only about 30% of the time even though their jobs were suitable for telecommuting for about 40% of the time, on average (see Section 4.4.2 for actual telecommuting frequency based on attendance log data). The respondents predicted greater frequencies of future center-based telecommuting than current levels (38%), but that expected frequency was substantially lower than was reported on the before survey (50%). In addition, the results from the choice, preference, and expectation of telecommuting indicate that combined home and center telecommuting appears to be a popular option.

In the section on travel, the commute to the regular workplace was reported as 44.2 miles in length, while the commute length to the telecommuting center was given as 7.3 miles, on average. The resulting average commute travel savings by using the center instead of going to the main office for the after survey respondents is 36.9 miles. Despite the reduction in travel, the majority of travel to the telecenter is on freeways, suggesting that the centers are far from the average participant's residence. This is especially true of the respondents from non-RABO centers who have longer commutes than RABO telecommuters, on average, to both the regular workplace (53.1 vs. 39.1 miles) and the telecommuting center (9.1 vs. 6.5 miles). Additionally, telecenter use was not found to have much effect on residential relocation decisions in this short time frame.

The tabulations of the attitudinal surveys provide a good characterization of the telecommuting center users. However, the survey data will also be used to model the decision to adopt telecommuting. Further analysis will be presented in the final report and/or in subsequent studies.

3.3 Manager Survey Results

Attitudinal surveys for managers targeted the supervisors of the employees who participated in the project. The number of returned manager surveys for the two waves and three study groups is shown in Table 3-1. There were a total of 49 respondents for the before survey (33 managers of center-based telecommuters; 16 managers of home-based and non-telecommuters), and 34 for the after survey (28 center-based; 6 home-based and non-telecommuters). Attrition in the second wave is due both to employees dropping out of the study (in which case an "after" survey from the manager is unnecessary) and to failure on the part of the manager to return the questionnaire.

Similarly to the previous section, this section focuses on analyzing the after survey for telecenter participants, that is the perceptions of the supervisors after their employees had used the telecenter for several months. The findings are based mainly on the information of the 28 managers of telecenter users who returned the after survey. These 28 surveys represent a 60.9% response rate for 46 managers of the telecenter users.

Of the 28 respondents, 13 were from RABO sites and 15 were from non-RABO sites. In addition, only 14 supervisors responded to both before and after surveys. When comparisons of responses between the two survey waves are of interest, the analysis is based on the information collected from these 14 respondents. Of the remaining 14 supervisors who returned the after survey but not the before survey, two respondents were new managers for continuing telecommuters and the other 12 respondents were managers of new participants in the study. These new participants, who were from the non-RABO site at Ontario, had already been telecommuting for some time before joining the study, and hence neither they nor their managers could complete the before survey.

Although the surveys of the participating employees and their managers were conducted in parallel, there is not necessarily a manager survey for each employee and vice versa. Among the telecenter participants, 39 employee surveys and 28 manager surveys were returned for the after wave. However, there are only 19 employee-manager pairs in the sample. Therefore, the average responses of the 28 managers may not completely reflect the behavior of the 39 employees whose responses are presented in Section 3.2. Comparisons of the responses between managers and employees are discussed in detail in Section 3.4.

The questionnaire design was similar for both survey waves, with the after survey obtaining additional information about telecommuting. The survey contains six parts:

Part A:	Job characteristics of the employee.
Part B:	Assessments of the employee's job performance and satisfaction.
Part C:	Attitudes toward the employee working at three different work environments: regular workplace, telecenter and home.
Part D:	Perceptions of the advantages of telecommuting.
Part E:	Perception of telecommuting as a work option, including general attitudes toward telecommuting within various management levels of the organization, information on the amount of time for the employee to telecommute, and intentions regarding continuing the telecommuting program.
Part F:	Demographic data, including gender, age, job tenure, education level, and frequency of computer usage.

The sections that follow describe the responses to each part of the survey in the order in which each part appears, except that the demographic information is presented first in order to describe the nature of the sample. The after survey of managers of telecenter users is attached to this report as Appendix F.

3.3.1 Demographic Characteristics

The demographics given below describe the 28 managers who returned the after survey. The gender of these respondents was equally distributed, with 14 females and 14 males. Half of the supervisors were between the ages of 45 and 54 years old. The next largest age groups, which each accounted for 17.9% of the sample, were 25 to 34 and 35 to 44 years of age. The remainder (14.3%) were in the category of 55 to 64 years old. On average, respondents had worked for their current organizations for 11.9 years and had worked as supervisors for more than 8 years.

The majority (71.4%) used a computer for several hours per day on the job. Only 2 persons did not use a computer at all while working. Approximately 79% of the respondents held at least a four-year college degree: 25.1% had obtained a bachelor's degree and another 53.5% had completed one or more graduate degrees. Half of the 28 supervisors (50%) came from organizations with more than 500 employees. Eight persons (28.6%) worked for an organization with 10 to 49 employees. Of the remainder, 1, 2, and 3 respondents came from organizations with 1 to 9, 50 to 99, and 100 to 499 employees, respectively. Thus, in this small sample, large organizations are the dominant participants in center-based telecommuting. This is in contrast to other, larger-sample studies of home-based telecommuting, where adoption has been found to be more prevalent among small-to-medium size employers.

3.3.2 Job Characteristics of Employees

In the first part of the survey, respondents were asked to give some background information on their employees' jobs. Supervisors of the 28 telecenter users reported that their participating employees worked for 83.9 hours in two weeks on average. Table 3-14 shows the items or services which the supervisors thought that the employees would need to work as effectively at the telecenter as they did at the regular workplace. The most frequently-selected items were personal computer and copier. Other frequent responses include fax machine, printer, software and modem. More than half of the supervisors thought that voice mail, conference calling, and electronic mail were essential to employees working effectively at the telecenter. Thus, the ability of center-based telecommuters to keep in touch with their supervisors, clients, and co-workers is important to most of the managers.

Table 3-14: Items or Services Thought to Help Employees
Work Effectively at the Telecenter (N=28)

3.3.3 Assessment of Employee Performance

In Part B of the survey, supervisors were asked to evaluate their employees' performance in various ways. The average responses are shown in Table 3-15. The first question asked the supervisors to evaluate their employees (on a five-point scale from "terrible" to "excellent") regarding the amount of work completed, quality of work, ability to meet deadlines, and overall productivity. Most of the employees were rated as "good" (4) or "excellent" (5).

Table 3-15: Supervisors' Ratings of Their Employees in the After Survey (N=28)

¹ Statements were negatively worded on the survey, but changed to a positive orientation for comparison with other responses.
² N = 24     ³ N = 13

To see if telecommuting had an impact on supervisors' assessments, responses in the before and after waves are compared for the 14 managers completing both surveys. The results of conducting paired-t tests on each statement are given in Table 3-16. None of the statements shows a significant change under even a 0.10 level of significance. Thus, for this small sample, it appears that telecommuting did not alter the managers' perception of their employees' work abilities either for better or for worse.

The second question asks the respondents to express an opinion (a five-point scale from "strongly disagree" to "strongly agree") regarding the work effectiveness of their employees along several dimensions. On the whole, the supervisors were satisfied with their employees' performance. Employees were believed to perform well on a team and to work well with the supervisor as well as with their own subordinates (where applicable). Also, they were still considered to communicate well with supervisors and their co-workers after the start of telecommuting.

Statements	Before Mean (S. D.)	After Mean (S. D.)	2-tail p-value
Amount of work completed	4.36 (0.75)	4.57 (0.51)	0.189
Quality of work	4.43 (0.65)	4.50 (0.52)	0.189
Ability to meet deadlines	4.29 (0.73)	4.50 (0.52)	0.435
Overall productivity	4.29 (0.83)	4.43 (0.65)	0.327
Employee communicates effectively1	4.07 (0.83)	4.29 (0.47)	0.336
Adequate resources are available1	4.07 (0.48)	3.93 (0.73)	0.435
Employee's work team is effective	3.85 (0.90)	4.23 (0.60)	0.175
Employee works well with supervisor	4.07 (0.83)	4.21 (0.70)	0.612
Clients demand a reasonable amount of time1, 2	3.10 (1.10)	3.20 (0.92)	0.343
Supervisor is satisfied with employee's performance	4.43 (0.85)	4.50 (0.52)	0.720
Supervisor expresses enough appreciation1	3.57 (0.94)	3.21 (1.05)	0.315
Employee gets along well with co- workers1	4.14 (0.66)	4.21 (0.70)	0.720
Employee has ability to do the job	4.38 (0.93)	4.50 (0.52)	0.500
Employee is likely to stay at current job1	3.36 (1.08)	3.71 (0.83)	0.208
Employee works well with subordinates3	4.67 (0.58)	4.00 (1.00)	0.184

¹ Statements were negatively worded on the survey, but changed to a positive orientation for comparison with other responses.

² N = 10     ³ N = 3

3.3.4 Attitudes Toward Different Work Environments

In Part C of the survey, supervisors' attitudes toward their employees working at three different work environments (regular workplace, telecenter, and home) were investigated using a series of attitudinal statements with a five-point response scale (from "strongly disagree" to "strongly agree"). The average ratings on each statement for each of the three workplaces are shown in Figure C-1, Appendix C. A three-level one-way analysis of variance (ANOVA) test was conducted for each statement to determine if supervisors' attitudes differ significantly across the three workplaces. Table 3-17 shows the results of the F-test for equality of means for each ANOVA, as represented by the p-values. Under a 0.10 level of significance, the effect of the workplace factor was significant for 16 of the 22 statements. The effect was insignificant for the employee indulging (C5), relative independence (C6), having the option to work when sick or disabled (C12), having the freedom to adjust the work schedule (C13), having work judged by the results (C15), and having high worker's compensation liability (C21).

In general, the supervisors favor working at home the least. The mean rating for the telecenter generally falls between the average rating for the regular workplace and home. Despite this, the respondents rated the telecenter similarly to or better than the regular workplace on the following aspects: employee being easily motivated (C1), supervisor's level of comfort (C2), workplace having a professional appearance (C3), having fewer distractions from others (C4), being good for the environment (C7), supervisor feeling confident (C10), costing too much (C11), and employee working effectively (C18).

Meanwhile, center-based telecommuting had some relative disadvantages compared to the regular workplace, with the telecenter being rated less positively on statements such as the employee not being visible to management (C8), the employee not having needed equipment (C9), the employee being unavailable (C14), not having enough professional interaction (C16), poor communication with the employee (C17), telecommuting being an administrative burden (C19), security of confidential information (C20), and lower perceived security for the organization's property (C22). However, it is important to realize that the mean ratings on all of these statements were neutral or favorable, just less favorable than the ratings for the regular workplace. Hence, these attitudes toward the telecenter are not negative in the absolute sense. Further, these attitudes seem to apply to telecommuting in general since home-based telecommuting had even more unfavorable average responses than the telecenter in each of the above statements except the last two (C20 and C22).

To examine how the supervisor's perceptions of the various workplaces may have changed with the introduction of center-based telecommuting, two-way ANOVA tests were performed on the same attitudinal statements for the 14 respondents common to both survey waves. The results are shown in Table 3-18. The effect of the workplace factor is significant for most of the same statements as in the previous ANOVA. As to the effect of the wave factor, the data indicates that the mean ratings are statistically different (at a 0.10 level of significance) for the following four statements: costing too much (C11), having the option available to work when sick or disabled (C12), employee having the freedom to adjust the work schedule (C13), and employee not having enough professional interaction (C16). In other words, the supervisors' attitudes on these statements changed significantly after their employees began telecommuting from the center.

Table 3-17: One-way ANOVA Results for Supervisors' Attitudes Toward Different Work
Environments in the After Survey (N=28)

¹ The numbers given are the p-values of the F-test for equality of means across workplaces. Values in boldface type are significant at a .LE. 0.10.

Table 3-18: Two-way ANOVA Results for Supervisors' Attitudes Toward Different
Work Environments in Both Survey Waves (N=14)

Statements1	Workplace Factor2	Wave Factor3	Interaction4
C1. Employee is easily motivated	0.002	0.881	0.437
C2. Supervisor feels uncomfortable	0.004	0.561	0.838
C3. Workplace has professional appearance	0.000	0.496	0.815
C4. Employee is distracted by others	0.002	0.524	0.237
C5. Employee indulges	0.542	0.797	0.806
C6. Employee is relatively independent	0.614	0.134	0.505
C7. Beneficial to the environment	0.000	0.641	0.605
C8. Employee is not visible to management	0.000	0.561	0.804
C9. Employee does not have needed equipment	0.000	0.173	0.854
C10. Supervisor feels confident in employee	0.026	0.516	0.983
C11. Costs too much	0.001	0.092	0.337
C12. Offer option when sick or disabled	0.016	0.000	0.415
C13. Employee can adjust work schedule	0.272	0.003	0.537
C14. Employee is unavailable	0.001	0.350	0.516
C15. Employee's work is judged by results	0.283	0.171	0.971
C16. Employee does not have enough prof. interaction	0.000	0.083	0.855
C17. Poor communication with employee	0.018	0.782	0.874
C18. Employee works effectively	0.006	0.357	0.752
C19. Administrative burden	0.002	1.000	0.589
C20. Security of confidential information	0.393	0.140	0.955

¹ Statements C21 and C22 appeared only in the after survey, and hence cannot be tested across waves.
² The numbers given are the p-values of the F-test for equality of means across workplaces. Values in boldface type are significant at a .LE. 0.10.
³ The numbers given are the p-values of the F-test for equality of means across survey waves. Values in boldface type are significant at a .LE. 0.10.
⁴ The numbers given are the p-values of the F-test for the interaction between the workplace and wave factors.

The average ratings for the four statements with a significant wave effect are shown in Figure C-2 of Appendix C. We see that first, the perception that center-based telecommuting costs too much has changed. On average, respondents disagreed more strongly on the after survey with the statement that telecommuting from a center would cost the organization too much. However, managers similarly changed their opinion about the costs of working from the regular workplace, suggesting that the change may not be simply caused by telecenter use. Second, it was found in the after wave that the supervisors were more likely to offer employees the option to work from any of the three workplaces when sick or disabled. It may be that telecommuting has raised supervisors' awareness of the potential for accommodating such needs from a variety of locations. Third, in the after survey these supervisors reported that at all three workplaces their employees had even more freedom to adjust their work schedules than they had indicated prospectively in the before survey, with the two telecommuting options having the highest scores. Perhaps the trust gained in telecommuting has spread to other parts of the job. Finally, although telecommuting is still perceived to offer less professional interaction than the regular workplace, the managers were less concerned about interaction after several months' experience with telecommuting. For the other 16 statements, there were no significant differences between survey waves.

3.3.5 Potential Advantages of Telecommuting

In Part D of the survey, the perceptions of the supervisors themselves and their perceptions of their organizations' official views on the advantages of implementing telecommuting were obtained through a series of attitudinal statements using a four-point scale ("no opinion", "not significant", "moderately significant", and "extremely significant"). The survey instructions specifically requested the respondent to distinguish between "no opinion" and "not significant", where the latter in fact represents a considered opinion. While acknowledging that the distinction may sometimes be difficult to make in practice, the belief is that a high proportion of "no opinion" ratings on a given characteristic may indicate an undervalued advantage of telecommuting. This suggests that marketing strategies emphasizing those characteristics may help to raise the perceived value of telecommuting to management.

Table 3-19 lists the proportion of supervisors and organizations having no opinion on each of the 11 advantages named in Part D. From one-sixth to one-third of the organizations did not have opinions on certain potential advantages of telecommuting. In contrast, most of the supervisors had their own opinions on these statements. Assuming that supervisors' direct contact with the telecommuters affords them greater experience with the benefits of telecommuting than upper management wouldhave, and assuming that supervisors were offering their own opinions with the good of the organization as a whole in mind, it would appear that a full awareness of the business advantages of telecommuting have not filtered up to the top levels in the organization. Obviously, such awareness on the part of organization decision-makers will be important if telecommuting is to expand.

However, even supervisors had a relatively high "no opinion" rate on four advantages of telecommuting: an improved ability to recruit employees (11%), reduced health costs (14%), compliance with environmental regulations (14%), and an improved disaster response capability (21%). This suggests the need to raise awareness of the potential benefits of telecommuting in these areas.

Table 3-19: Percentage of Supervisors and Organizations With "No Opinion"
on Advantages of Telecommuting (N=28)

¹ N = 24     ² N = 25

The remaining three points on the scale ("not significant" to "extremely significant") have an ordinal relationship to each other and were assigned values of 1, 2, and 3, respectively. The mean ratings on each advantage for those who had an opinion are shown in Table 3-20. According to the supervisors' average rating, six potential advantages are viewed as being at least moderately significant following the introduction of center-based telecommuting: improved employee retention (D1), improved ability to recruit employees (D4), increased productivity (D5), compliance with environmental regulations (D10), improved employee relations (D11) and (marginally) reduced absenteeism (D8). Five potential advantages are viewed as being insignificant: savings on office space costs (D2), improved customer service (D3), savings on parking costs (D6), improved disaster response capability (D7), and reduced health costs (D9). The mean ratings for the organizations were lower than the managers' ratings on every statement except compliance with environmental regulations. However, none of these differences were statistically significant, except for increased productivity (D5). Supervisors view increases in employee productivity as a more important advantage of telecommuting than they think their organizations do.

3.3.6 Telecommuting as a Work Option

In Part E of the survey, respondents were asked about their experience with telecommuting, the extent to which telecommuting may be adopted by the organization, and how desirable it may be for the employee. To assess the general level of support for telecommuting within the organization, the survey asked about the existence of formal policies and general attitudes toward telecommuting on the part of responding managers themselves, their immediate supervisors, and the organization as a whole. Among the 28 responses, 71.4% of the telecommuter managers themselves, 60.7% of their supervisors, and 60.7% of the organizations had formal policies supporting telecommuting. Twenty-five percent of the respondents, 32% of their supervisors, and 21.4% of the organizations did not have formal policies on telecommuting. One respondent thought that the organization had a formal policy against telecommuting. Four respondents (14.3%) did not know if there was a formal policy on telecommuting in the organization, and two did not know if their immediate supervisor had a formal policy.

Table 3-20: Average Ratings and ANOVA Results for Potential Advantages of
Telecommuting in the After Survey (N=28)

Statements	Supervisor	Organization	Opinion Factor11
D1. Improve employee retention	2.14	1.851	0.136
D2. Save office space costs	1.36	1.201	0.329
D3. Offer better customer service	1.442	1.373	0.702
D4. Improve ability to recruit employees	1.884	1.655	0.193
D5. Increase productivity	2.112	1.726	0.032
D6. Save parking costs	1.192	1.176	0.877
D7. Improve disaster response capability	1.367	1.318	0.808
D8. Reduce absenteeism	1.64	1.423	0.204
D9. Reduce health costs	1.299	1.116	0.166
D10. Comply with environmental regulations	1.969	2.1010	0.532
D11. Improve employee relations	2.29	2.0510	0.119

¹ N=20     ² N=27     ³ N=19     ⁴ N=25     ⁵ N=17     ⁶ N=18
⁷ N=22     ⁸ N=16     ⁹ N=24     ¹⁰ N=21
¹¹ The numbers given are the p-values for the t-test for equality of means across the two management levels. Values in boldface type are significant at a .LE. 0.10.

When asked about the general attitude toward telecommuting, 93% of the respondents had a positive response. However, that percentage dropped to 61% for their supervisors and to 50% at the organizational level. Although none of the respondents themselves had negative attitudes toward telecommuting, they reported that nearly 11% of their supervisors and 18% of their organizations did. Seven percent of the respondents, 21% of the respondents' supervisors, and 32% of the organizations held neutral opinions on telecommuting. Thus, the higher the management level, the less positive was the attitude toward telecommuting. Indeed, it appears that some managers are supporting telecommuting for their staff in the face of actively negative attitudes on the part of upper management.

To ascertain the respondents' experience in managing telecommuters, they were asked how long they had supervised telecommuting employees. On average, respondents had supervised center-based telecommuters for 15.6 months and home-based telecommuters for 14.2 months. When asked about the proportion of the organization's workforce allowed to telecommute from the telecenter and from home within the next two years, around half of the respondents thought that it would increase for each form of telecommuting. However, approximately 40% of respondents expected no change in the proportion of telecommuters at their organization.

Overall, 82% of the supervisors had a high or very high level of satisfaction with center-based telecommuting. The other eighteen percent were neutral; none had low satisfaction with center-based telecommuting. With the center in its current state, half of the respondents believed that the organization would be likely to offer center-based telecommuting to other employees. Although only two respondents thought that other employees were unlikely to be offered the opportunity to telecommute from a center, 39% of the sample were unsure whether it would be made available to others.

When asked what, if anything, would need to change to make the organization likely to offer center-based telecommuting, the most frequently selected response (39.3%) was that "nothing more is needed, we are already likely to offer it". Other common responses included lowering the cost (29%), obtaining manager acceptance (27%), and having the ability to quantify the benefits to the organization (21%). Collectively, these three responses were cited as most important by 39% of the respondents. At least two of these three responses relate to justifying the cost of the center. Even though rental costs for using the center were partially or completely waived for many participants at this point in the demonstration, managers were likely responding both to existing costs of other kinds (monetary and non-monetary) and to expected future rental costs in a post-demonstration environment. Thus, it appears as though cost-justification is a critical issue in increasing the acceptance of center-based telecommuting. Manager acceptance may still be an issue apart from cost-benefit considerations (that is, even if telecenters were demonstrably cost-effective, managers may have other considerations); however, acceptance is likely to increase as proof of the benefits of telecommuting (relative to costs) becomes stronger.

Supervisors were asked about the ideal distribution of their employees' work time among alternative work locations. On average, they wanted their employees to spend 64% of the time at the regular workplace, 29% at the telecenter, 5% at home, and 4% at other locations such as client offices. Table 3-21 shows the distribution of responses among various workplace combinations and, for each combination, the average ideal distribution of work time at each location. The most favorable workplace combination is regular workplace and telecenter, which accounts for 57% of the responses. On average, under this workplace combination, the managers prefer their employees to spend 73.8% of work time at the regular workplace and 26.3% at the center. The second most common response (25%) is regular workplace, telecenter and home, where the average work time is distributed 60.7%, 30.0% and 9.3%, respectively. Collectively, the results of Table 3-21 suggest that (1) the manager still feels that the regular workplace is the primary work location, to be used three or more days out of the work week, (2) the manager is willing for the employee to telecommute nearly three times as much from the center as from home on average, and (3) for nearly one-third of the managers, some mixture of center and home-based telecommuting is considered ideal.

Table 3-21: Supervisors' Distribution of Workplace Combinations and
Ideal Distribution of Work Time (N=28)

¹ Workplace locations are the regular workplace (RW), the telecommuting center (TC), home (HM), and other location (OL).

A series of questions asks the supervisor about several aspects of the employee's telecommuting frequency from a center and from home. In contrast to the above question in which respondents filled in blanks for the "percent of time" an employee would ideally spend at each location, responses to this series of questions consisted of categories ranging from "not at all" to "5 days a week". The first question relates to the nature of the employee's job and its suitability for telecommuting. The other questions ask for the current amount, the allowable amount, and the predicted amount of telecommuting of the employee. Table 3-22 shows the distribution of responses to the four questions. More than half of the respondents reported that their employees telecommuted from the telecenter one to two days per week. The distributions of the responses are similar for the four questions, suggesting that the responses are consistent across all four indicators of possible and actual telecommuting frequency. Looking six months into the future, the supervisors also expect that the amount of center-based telecommuting will remain approximately equivalent to current levels except that one manager thought the employee would not be telecommuting from the center at all. It is noteworthy that few managers expected their employees to be telecommuting from the center nearly full time. This is consistent with the result for the question regarding the ideal distribution of time across work locations.

Table 3-23 shows the mean telecommuting frequencies for the four categories based on the managers' responses. The calculation follows the similar method described in Section 3.2. The nature of the job allows employees to telecommute from a center about 1.5 days per week (32%), on average. Although the permitted frequency is 37%, the current rate and the expected future rate for center-based telecommuting is about 30% of the work time. It appears that the supervisors feel that employees are restricted in the amount of time that they can telecommute due to the nature of the job rather than due to the restrictions imposed by management.

¹ N = 27

Job	Considering the requirements of your employee's current job, how much do you think the nature of your job would allow him/her to telecommute . . . ?
Choice	How much does your employee currently telecommute . . . ?
Permit	How much would you allow your employee to telecommute . . . ?
Expect	Six months from now, how much do you expect your employee to be telecommuting . . . ?

Table 3-23: Relative Telecommuting Frequency ­ Supervisor Responses¹

After	Proportion of Work Week - Mean and Standard Deviation (N=28)
	Job	Choice	Permit	Expect
From a center	32.3% (21.2)	28.3% (21.7)2	36.7% (22.3)	29.7% (22.9)
From home	18.3% (22.0)	5.6% (10.5)	18.0% (19.5)	7.6% (11.2)

¹ See Table 3-11 for the values used in the frequency calculation and Table 3-22 for the definition of the categories.
² N = 27

Home-based telecommuting was not perceived as positively as center-based telecommuting with respect to job suitability and permitted frequency: only 28.6% of the managers thought that the nature of the job allowed their employees to telecommute from home one to two days per week; 35.7% were willing for the employee to telecommute from home that often. About 61% of the managers reported that their employees did not telecommute from home currently, and 46.3% of the managers did not expect their employees to be telecommuting from home within the next six months. The results are consistent with prior hypotheses that center-based telecommuting could be superior to home-based telecommuting in terms of job suitability and manager acceptance (e.g., Bagley, et al., 1994) but are at odds with other researchers' findings that telecenters are not viewed favorably by managers (Bernardino and Ben-Akiva, 1996). Clearly, the self-selection bias of this sample should be kept in mind when interpreting these results; that is, managers who are willing to participate in this project are more likely to have a positive view of telecenters than the general population of managers.

3.3.7 Summary of Manager Responses

This section summarizes the survey results from 28 supervisors of center-based telecommuters. On the whole, the respondents reported an optimistic and positive attitude toward telecommuting. The analysis showed clearly that supervisors' opinions of the performance of their employees did not diminish with the introduction of telecommuting.

Characteristics dealing with the workplace atmosphere (such as motivation, professional appearance, and distractions) were considered to be similar at both the regular workplace and the telecenter. Areas in which the center was perceived less positively than the regular workplace concerned the supervisor-employee relationship (such as communication, availability, professional interaction, and administrative burden) as well as security of information and property. However, mean ratings for the telecenter on these characteristics were all neutral or better, indicating that the disadvantage is relative, not absolute. These attitudes seem to be generic to telecommuting in general since they tended to be even less favorable for home-based telecommuting.

Nearly all of the managers (93%) indicated having a positive attitude toward telecommuting in general, and 82% rated their level of satisfaction with center-based telecommuting as high or very high. (However, a selection bias must be noted, as managers who were dissatisfied with telecommuting would be less likely to have lasted long enough to complete an after survey. As indicated in Section 5.3, supervisor-related concerns were cited by 5 of the 20 employees from whom reasons for quitting could be obtained as important reasons for quitting telecommuting). Six potential advantages were viewed by managers to be at least moderately significant following the introduction of center-based telecommuting: improved employee retention, improved ability to recruit employees, increased productivity, compliance with environmental regulation, improved employee relations, and (marginally) reduced absenteeism. However, from 11% to 21% of the managers reported "no opinion" on the four following potential advantages of telecommuting: improved ability to recruit employees, reduced health costs, compliance with environmental regulations, and improved disaster response capability. This suggests the need to raise awareness of the potential benefits of telecommuting in these areas.

It is an important result that the perceived advantages of telecommuting are those for which the benefit is difficult to quantify (customer service and productivity), while telecommuting is not perceived to offer advantages on "hard" money items such as office space and parking costs. This will continue to make center-based telecommuting difficult to justify in purely economic terms. Indeed, while 39% of the respondents indicated that the organization was likely to (continue to) offer center-based telecommuting, an equal proportion cited reduced costs, the ability to quantify the benefits, and increased manager acceptance as factors that needed to change before the organization would be likely to offer center-based telecommuting.

About half of the managers expected that more of the organization's workforce would be telecommuting from a center in the future. However, from one-sixth to one-third of the organizations themselves did not have official opinions on various potential advantages of telecommuting according to the respondents. When opinions by the management levels above the supervisors were expressed, they tended to be less positive than those of the supervisors. Indeed, it appears that some managers are supporting telecommuting for their staff in the face of actively negative attitudes on the part of upper management. This suggests the need for upper-level management to have increased exposure to the benefits of telecommuting.

Although the employees performed well at the telecenter or even better than they did at the regular workplace in some aspects, managers still preferred telecommuting to be a part-time alternative for their employees. Very few managers expected their employees to be telecommuting from the center full-time. The managers' average ideal distribution of work time for their employees included nearly 64% at the regular workplace and 29% for center-based telecommuting. The current and the expected future telecommuting frequencies of 30% (which is equivalent to 1.5 days per week) are consistent with the managers' ideal work time distribution. However, in the managers' perception, the appropriate telecommuting frequency for their employees was more constrained by job suitability (32% of work time on average) than by the managers' willingness (37%). In any case, the managers still feel that the regular workplace is the primary work location, to be used three or more days out of the work week.

Home-based telecommuting was not perceived as positively as center-based telecommuting with respect to job suitability and permitted frequency, although the self-selection bias of the sample must be taken into account in interpreting this result. The managers were willing for the employees to telecommute nearly three times as often from the center as from home. Also, some mixture of center and home-based telecommuting was considered ideal by nearly one-third of the managers.

This expectation of part-time telecommuting may act to inhibit the adoption of telecommuting centers. If employees are only using the center one or two days per week, there may be little opportunity for their space at the regular workplace to be used for other purposes. If an organization must continue to offer the same amount of space at the regular workplace as before, plus pay rent on space at the telecenter, other telecommuting advantages will have to be that much stronger to compensate for the added cost.

3.4 Comparison of Employee and Manager Results

As can be seen from the preceding sections, the employee and manager surveys were designed such that the responses to certain questions could be compared between the two groups. These questions include job performance and satisfaction, attitudes about work environment characteristics, and the amount of telecommuting. However, the results of the comparison of telecenter employee and manager survey results must be viewed with some caution. Although future analysis may match up each telecenter user with his/her supervisor, this has not yet been done since the sample size is small (there are only 19 employee-manager pairs in the after telecenter data set analyzed for this interim report). So the comparison presented here shows only the group averages for those who returned surveys. In addition, the comparison group sizes vary considerably: 39 telecenter users and 28 managers (see Table 3-1). Therefore, variations in survey responses between the two groups may be due in part to not having completely matched responses, as well as to the differing perspectives of employees and their specific managers.

3.4.1 Job Performance and Satisfaction

Similar questions were asked of both the employees and the managers about employee job performance and job satisfaction. For the characteristics of job performance, two comparisons can be made between the employee and manager results (see Table 3-24). A direct comparison of the average employee's assessment of his/her performance and the average manager's assessment shows only slight differences. A somewhat unusual result is that for two of the four factors, the managers gave a higher average rating than the employees. In fact, the greatest average difference is for the ability to meet deadlines where employees rated themselves lower than the managers did. An additional comparison can be made between the employee's opinion of the supervisor's assessment and the supervisor's assessment. Again, there are only slight variations between the two averages for each factor; however, the values are generally closer together than in the first comparison. The exception is the "ability to meet deadlines" factor mentioned previously. It appears that employees perceive meeting deadlines to be more of a problem after the start of telecenter use than their managers do.

Table 3-24: Comparison of Job Performance

The majority of the average ratings on the factors of job satisfaction are also similar between both telecenter users and managers of telecenter users (see Table 3-25). However, three factors show notable differences. First, managers were more likely to be satisfied with the availability of resources for completing tasks than employees were. Second, employees were more bothered by client demands on their time than managers were. These two results are not particularly startling given that employees must deal with these problems more often than their supervisors have to. Surprisingly, employees had a higher average rating for the supervisor showing appreciation than the supervisors had. Perhaps supervisors preceive that they need to work at this part of their job more than they actually need to.

Table 3-25: Comparison of Job Satisfaction

Question	Rating (Mean and Standard Deviation)
	Employees (N = 39)	Managers (N = 28)
Supervisor/employee communication3	4.16 (1.03)	4.32 (0.48)
No lack of resources for employee1	3.54 (1.21)	4.11 (0.63)
Employee's work team is effective	3.85 (0.84)	4.04 (0.66)
Work well with supervisor/employee	4.08 (0.94)	4.21 (0.69)
Supervisor shows appreciation for employee1	4.05 (0.83)	3.14 (1.04)
Employee works well with co-workers1	4.49 (0.88)	4.32 (0.67)
Confidence in employee's ability to do the job	4.51 (0.56)