Journal ofcemnlology: PSYCHOLOGICAL SCIENCES 1998, Vol. 53B, No. 5, P329-P340 Copyright 1998 by The Cemntological Society of America Age ferences in Attitudes Toward Computers Sara J. Czaja 1-2 and Joseph Shark 1-3 'Miami Center on Human Factors and Aging Research, University of Miami. 2 University of Miami School of Medicine, department of Industrial Engineering, University of Miami. It is commonly believed that older adults hold more negative attitudes toward computer technology than younger people. This study examined age differences in attitudes toward computers as a function of experience with computers and computer task characteristics. A sample of 384 community-dwelling adults ranging in age from 20 to 75 years performed one of three real-world computer tasks (data entry, database inquiry, accounts balancing) for a 3-day period. A multidimensional computer attitude scale was used to assess attitudes toward computers pretask and posttask. Although there were no age differences in overall attitudes, there were age effects for the dimensions of comfort, efficacy, dehumanization, and control. In general, older people perceived less comfort, efficacy, and control over computers than did the other participants. The results also indicated that experience with computers resulted in more positive attitudes for all participants across most attitude dimensions. These effects were moderated by task and gender. Overall, the findings indicated that computer attitudes are modifiable for people of all age groups. However, the nature of computer experience has an impact on attitude change. COMPUTERS and other forms of advanced technology are being used increasingly in a variety of settings to perform a wide variety of tasks. It is not uncommon for someone to shop, to pay bills, or to make travel reservations at home using a personal computer. Automatic teller machines (ATMs) are frequently used to conduct banking transactions, and E-mail is a common form of communication. Furthermore, most workers interact with some form of computer technology in the routine performance of their jobs. Clearly, the successful adoption of technology is becoming increasingly important to a person's ability to live and function effectively within society. It is commonly believed that older people are uncomfortable with new forms of technology and that they are more resistant to using technology than are younger people. This belief often places older people at a disadvantage, because designers fail to consider older people as a potential user group when designing technology (e.g., Parsons, Terner, & Kersley, 1994). Parsons and colleagues recently conducted a study to improve the design of remote control units for seniors and found that manufacturers of these devices typically fail to consider product design issues for older consumers. Older people are also frequently bypassed when opportunities for technology training or retraining are available. For example, older workers generally have fewer opportunities than younger people do to participate in worker retraining programs to update needed work skills (Fossum, Arvey, Paradise, & Robbins, 1986; Rosen & Jerdee, 1976). Data examining age differences in the adoption and use of technology have yielded contradictory findings. For example, only 1% of people aged 65+ years own personal computers (Schwartz, 1988). Data also indicate that older people are less likely than younger people to use common forms of technology such as ATMs or VCRs (Rogers, Cabrera, Walker, Gilbert, & Fisk, 1996; Zeithaml & Gilly, 1987). Rogers and colleagues conducted an extensive survey of ATM use across the adult life span and found that people aged 65+ years were much less likely to own an ATM card or to use an ATM machine than younger or middle-aged adults. Zeithaml and Gilly (1987) also found that adults aged 65+ years were less likely to use ATMs than adults younger than age 65 years. However, the older adults were willing to use other forms of technology, such as grocery checkout scanners. In contrast, data from a survey conducted by the American Association of Retired Persons (AARP) of older people who had visited a technology center indicated that the majority of respondents were willing to use personal computers to perform routine tasks such as preparing taxes, budgeting, and accessing health or benefit information (Edwards & Engelhardt, 1989). In a study of E-mail (Czaja, Guerrier, Nair, & Landauer, 1993) that included women aged 50-95 years, the participants found it valuable to have a computer in their home and indicated that they would be willing to use computers for tasks such as paying bills and communication. Given the widespread dispersion of technology, it is important to understand what factors influence the likelihood that older adults will use technology so that strategies and interventions can be developed to maximize their potential interactions with these systems. It is generally accepted that a person's attitude (predisposition directed toward some object, person, or event) influences his or her willingness to accept and use technology, as attitudes tend to guide behavior (Regan & Fazio, 1977). According to a model outlined by Mackie and Wylie (1988), user acceptance of technology is affected by: (a) the user's awareness of the technology and its purpose; (b) the extent to which the features of the technology are consistent with the user's needs; (c) the user's experience with the technology; and (d) the availability of support, such as documentation and training. For example, Zeithaml and Gilly (1987) found that if older people were provided P329
P330 CZAJAANDSHARIT with an explanation of the benefits associated with technologies such as ATMs they were more likely to use the technologies than if they were unaware of the benefits (see also Czaja et al., 1993). Edwards and Engelhardt (1989) found that introducing the technology in a highly interactive and understandable manner was one factor that was likely to have influenced the receptivity of the seniors in their sample toward computers. In a study that examined factors influencing the willingness of older people to use ATMs, Smither and Braun (1994) found that nonusers had more negative attitudes toward ATM machines than users. They also found that people who had used ATMs had more positive attitudes than those who had never tried them. Generally, it is believed that experience with technology will result in people having more positive attitudes toward the technology (e.g., Jay & Willis, 1992; Krauss & Hoyer, 1984). However, the literature regarding the influence of experience with technology (e.g., with computers) on attitudes among older people yields mixed results. Danowski and Sacks (1980) examined the effects of participation with computer-mediated communication on attitudes toward computers among a sample of older people and found that positive experiences with computers result in more positive attitudes. In our study of text editing (Czaja, Hammond, Blascovich, & Swede, 1989), we found no change in attitudes following computer use. Dyck and Smither (1994) examined the relationships among computer anxiety, computer experience, gender, and level of education among a sample of younger and older adults; they also measured attitudes toward computers. Their data indicated age differences in attitudes such that the older participants had more positive attitudes toward computers than the younger participants. However, the older subjects also indicated less confidence about their ability to use computers. In addition, an inverse relationship between computer experience and computer anxiety was found; higher levels of experience were associated with less anxiety and more positive attitudes. Marquie, Thon, and Baracat (1994) surveyed office workers ranging in age from 18-70 years about their attitudes toward computers, their use of computers outside of work, and the amount of computer training they received. They found that experience with computers was the most important factor influencing attitudes; nonusers had more negative attitudes and anxiety toward computers than users. Age also influenced attitudes: older workers had attitudes that were more negative, more fears surrounding threats to employment, and less knowledge about the utility and operation of computers. Jay and Willis (1992) suggest that the disparate findings regarding experience with computers and attitude changes might be due to the nature of the experiences and the attitude measures used in the studies. For example, participants in the Danowski and Sacks (1980) study received 3 weeks of exposure to computers, whereas in the Czaja et al. (1989) study, participants were exposed to computers for only 1 day. Jay and Willis speculate that attitude change does not occur with limited amounts of exposure and that older people need more time to absorb and evaluate information. They examined the influence of computer experience on attitudes toward computers among a sample of communitydwelling older adults who participated in a 2-week training program on desktop publishing. They found that participation in the training program resulted in more positive attitudes among the study participants and that these effects were maintained for 2 weeks following training. The type of experience a person has with computers may also influence attitude change. As noted, Edwards and Engelhardt (1989) attributed their older participants' positive attitudes toward computers to the highly supportive manner in which the computers were introduced. Also, several investigators (e.g., Czaja et al., 1993) have shown that older adults are more receptive to using technologies such as computers if they perceive the technologies as being useful and the tasks that they are able to perform with the technologies as being valuable and beneficial. In addition to understanding factors (e.g., experience) that result in successful attitude change, it is important to understand the nature of the attitude change. Attitudes toward computers are typically multidimensional; therefore, in order to maximize the likelihood that older individuals will adopt positive attitudes toward computer technology, attitudes need to be understood at the multidimensional level. For example, Jay and Willis (1992) used a multidimensional scale to assess attitudes toward computers and found that experience with computers increased computer comfort and efficacy. These two attitude dimensions were targeted by their computer training program. Their results demonstrate that there is a relationship between the nature of the experience with computers and the specific nature of the attitude change. This type of information is important in ensuring effective attitude change. If the dimensions of attitudes that are less positive are understood, interventions can be designed to influence those dimensions specifically. Unfortunately, most of the prior research examining age differences in attitudes toward computers has used unidimensional attitude scales. The objective of this article is to examine further the influence of computer experience on the attitudes of older adults toward computer technology. The study reported here extends prior research on this topic by examining three different types of computer tasks a data entry task, a database inquiry task, and an accounts balancing task which are different from those used in prior research examining this issue. Moreover, these tasks are highly representative of computer-based tasks performed in actual work settings. In fact, design of the three tasks involved close collaboration with three large U.S. corporations where these tasks are performed. Each task places different demands on the person: the data entry task emphasizes speed, accuracy, and psychomotor skills; the database inquiry task involves file identification and visual search; and the accounts balancing task emphasizes problem solving and information integration and involves a graphical user interface. This research is part of a larger study examining age differences in the performance of computer-based work and the relationship between cognitive abilities, task experience, and task performance. Understanding whether computer task characteristics influence attitudes toward technology is important for the effective design of interfaces and training programs. Shackel
AGING AND ATTITUDES TOWARD COMPUTERS P331 (1986) suggests that attitudes are a critical component of usability and maintains that in order for a system to be usable, interaction with the system must provoke positive attitudes on the part of the user. For example, it may be that tasks that are characterized by highly structured interactions may result in more negative attitudes than those that allow more flexibility, because the user may feel less control over the former. Also, as noted earlier, once attitudes are understood it may be possible to modify them through training. Specifically, this article addresses three issues: (a) Will experience with computers result in a change in attitudes toward computers? (b) Will the effect of experience vary across attitude dimensions? and (c) Will these effects vary according to age and task characteristics? In addition, this study presents data regarding the relationship between attitudes toward computers and perceptions of workload and stress and the relationship between task performance and attitudes. Examination of these issues provides further insight into the nature of the experience effect. Finally, data are presented regarding gender differences in attitudes toward computers: Jay and Willis (1992) discuss the need to examine gender differences in attitudes toward computers as several investigators (e.g., Krauss & Hoyer, 1984) have indicated that older women are less receptive to computer technology than are older men. METHOD Sample A total of 384 subjects, including 163 men and 221 women ranging in age from 20-75 years, participated in the study. The mean age of the sample was 48.36 years ( = 17.21). In order to ensure adequate numbers of both younger and older subjects, participants were recruited in three age groups: younger (20-39 years), middle-aged (40-59 years), and older (60-75 years) adults. There were 138 subjects in the younger group, 117 subjects in the middle-aged group, and 129 subjects in the older group. One hundred and twenty-seven participants performed the data entry task, 123 performed the database inquiry task, and 134 performed the accounts balancing task. Participants were recruited from the local community through advertisements and were paid $125.00 for their participation (if they completed the entire protocol). They were required to have at least a high school education, familiarity using a keyboard (ability to type a 5-line paragraph), and 20/40 near and far vision (with or without correction). Near visual acuity was tested using the Rosenbaum-Yaeger Chart and far acuity was tested using the Snellen Chart. Participants were also screened to ensure that they were able to read characters on the computer screen. This was tested by asking the subjects to identify numbers, upper- and lowercase letters, and special characters (e.g., #, <, *) that appeared on the screen in random locations. The participants were also screened for cognitive impairments (a score > 24 on the Mini-Mental Status Exam; Folstein, Folstein, & McHugh, 1975) and Psychic Distress (Symptom Checklist 90-Revised [SCL 90-R] scores scaled by gender; Derogatis, 1977). Finally, participants were screened for occupational background and excluded if they were currently or previously employed in data entry, database inquiry, or accounts balancing jobs. The sample was fairly well educated; 21.2% (n = 81) had high school degrees, 39.4 % (n = 150) had some college or technical school education, and 39.4% (n = 150) had college degrees or beyond (there were three missing data points in reported education level). Approximately 60% (60.1%; n = 229) of the sample was unemployed or retired, 23.9% (n = 91) worked part-time, and 16% (n = 61) worked full-time (there were three missing data points in reported employment status). The demographic characteristics for the samples for each of the three tasks are presented in Table 1. Given that the tasks involved used computers and there were likely to be cohort differences in computer experience, participants were asked to complete a computer experience questionnaire. The questionnaire asked participants to indicate whether they had ever used a computer and, if so, to rate the duration of their experience, the frequency of use, and the breadth of their computer knowledge. Responses to the questionnaire were categorized according to four levels: no prior experience, very little experience (very little knowledge and infrequent use), some experience (knowledge of a few applications and occasional use), and considerable experience (broad knowledge and frequent, regular use). Approximately 27% (27.2%; n = 104) of the sample had no prior experience with computers, 21.7% (n = 83) had very little experience with computers, 36.1% (n - 138) had some experience with computers, and 14.9% (n = 57) had considerable experience with computers (there were two missing data points in prior computer experi- Variable Age Table 1. Sample Characteristics for All Three Tasks Gender Women Younger Middle-aged Older Men Younger Middle-aged Older Education High school Junior college/trade school College/graduate school Employment Full-time Part-time Other Data Entry 47.17 17.28 27 22 22 23 14 19 36 56 35 16 41 70 Task Database Inquiry 48.67 17.96 21 22 35 23 11 11 25 51 46 27 29 66 Accounts Balancing 49.19 16.50 21 31 20 23 17 22 20 43 69 Note: The numbers in the gender, education, and employment variables represent cell counts. 18 21 93
P332 CZAJAANDSHARIT ence). There was a significant difference in prior computer experience as a function of age, \ 2 (6) = 17.58, p <.01; the younger and middle-aged participants had more prior experience with computers than the older participants. There was no difference in prior computer experience according to gender. Materials, Setting, and Equipment Attitudes toward computers were assessed using the Attitudes Toward Computers Questionnaire (ATCQ; Jay & Willis, 1992). The ATCQ is a 35-item multidimensional scale assessing seven dimensions of attitudes toward computers: comfort (feelings of comfort with computers and their use); efficacy (feelings of competence with the computer); gender equality (the belief that computers are important to both men and women); control (the belief that people control computers); interest (the extent to which one is interested in learning about and using computers); dehumanization (the belief that computers are dehumanizing); and utility (the belief that computers are useful). Each dimension is assessed by 5 or 6 items that are scored using a 5-point Likert-type scale format. The scale has been used in prior research with elderly samples. Details concerning the scale construction can be found in Jay and Willis. The modified Stress Arousal Checklist (Cruickshank, 1984) was used to measure perceptions of stress and arousal. The checklist, originally developed by Mackay, Cox, Burrows, and Lazzerini (1978), represents a two-dimensional model of mood that uses adjectives for evaluating stress and arousal. The checklist contains 26 items; 18 items are related to feelings of stress (9 to high stress and 9 to low stress) and 8 items are related to feelings of arousal (4 to low arousal and 4 to high arousal). For each item, the subject is asked to rate how he or she is feeling at that moment in time. The National Aeronautics and Space Administration (NASA) Task Load Index (TL) Scale (Hart & Staveland, 1988) was used to measure workload. The scale requires the subject to rate a task on the basis of the following six dimensions comprising workload: mental demand, physical demand, temporal demand, performance, effort, and frustration level. The scale provides a rating of overall workload and a rating of each of the six dimensions. The tasks were performed using a PC in a laboratory designed to represent an office environment. Three workstations were set up with a wall between workstations. Experimental Tasks Working in close collaboration with three corporations based in the United States, three computer-based tasks were simulated: data entry, database inquiry, and accounts balancing. The simulations maintained the structural integrity of the tasks performed in the real world and have high ecological validity. The data entry task simulated a task performed at a large transportation company. The task involved entering trip record information into preformatted computer screens. The information is derived from trip records completed by truck drivers and is used to calculate fuel tax. These records include specific trip information, including odometer readings, dates of trip, states traveled (codes are used), and fuel purchases. The primary emphasis in the task is on speed and accuracy of data input. The database inquiry task was a simulation of a job performed by service representatives of a large health insurance corporation. The task involved understanding a large number of concepts related to health insurance plans and required the participant to respond to queries from "members" who purchased health insurance from the company. The participant received simulated requests for information and for file updates and changes either on paper or by telephone; the participant used both written reference materials and the computer to respond to these requests. Computer interaction required navigation through a set of computer files that corresponded to different categories of information (e.g., claims) or actions (e.g., documenting member requests). The accounts balancing task simulated the responsibilities of an accounts balancing operator in the banking industry, who makes sure that the transactions of customers are in balance. This task was entirely computer-based (there were no paper documents or reference materials) and involved a graphical user interface. The task utilized software that processed checks automatically by using electronic pictures of checks and deposit slips. The balancing operator worked only on those customer deposits that the software identified as being out of balance. The Windows-based system allowed the operator to examine the various pieces of information (e.g., checks and deposit slips) scanned by the computer to identify the causes of out of balance conditions and to make the necessary corrections to eliminate these conditions. There were various causes of out of balance conditions (e.g., an error on a deposit slip or an incorrectly scanned check). Procedure The same protocol was followed for each of the three tasks. Respondents participated for a period of 5 days for approximately 5 hours per day. On Day 1 respondents were screened for the inclusion/exclusion criteria, and they completed the computer experience questionnaire and the ATCQ. On Day 2, they were given an introduction to computers and trained to perform one of the three tasks. Specifically, participants were trained until they demonstrated an understanding of the task as evidenced by their ability to perform a set of practice problems on their own and to answer training criteria questions. If they had difficulty completing the practice problems, they could ask the experimenter for assistance. The training criteria questions were administered following completion of the practice problems. If the participants were unable to answer these questions, the concepts were explained to them and they were provided with additional practice problems. This process was repeated up to three times. At this point, participants who were still unable to meet the training criteria were paid $50.00 for their effort and their participation was terminated. For the accounts balancing task, participants were also provided with mouse and Windows training. On Days 3 through 5, participants performed the task on their own for 3 hours each day. They completed the Stress Arousal Checklist prior to and after task performance on each of the 3 days. The participants also completed a paper-based, multiple-choice job knowledge test and the NASA TL Scale
AGING AND ATTITUDES TOWARD COMPUTERS P333 following task performance (job knowledge tests were developed to assess conceptual and procedural knowledge of the task). On Day 5, they also completed the ATCQ questionnaire a second time. 136 (a) Data Entry Task RESULTS Age, Task, and Task Experience Effects on Attitudes Toward Computers An overall attitudes toward computers score was computed by summing responses across the 35 questionnaire items. This score was calculated so the results of this study could be discussed relative to other studies that used unidimensional scales. Scores were also calculated for each of the seven attitude dimensions by summing the responses to the questions within each dimension. Higher scores reflected attitudes that were more positive toward computers. Age group, task, and gender effects on attitudes toward computers (overall and for each dimension) were analyzed using a multivariate analysis of variance (MANOVA) procedure (Maxwell & Delaney, 1989), with Task Experience (pretask vs posttask) as a within-subjects factor at two levels; Age Group and Task as between-subjects factors, each at three levels; and Gender as a between-subjects factor at two levels. For the factors involving within-subjects effects (Task Experience, Age Group Task Experience, Task Task Experience, and Gender Task Experience) Wilk's lambda criterion was used as the basis for tests of significance. Prior computer experience was included in the analyses as a covariate as there were age differences in prior computer experience as well as significant relationships between prior computer experience and overall pretask attitudes toward computers, r (382) =.30, p <.01, and overall posttask attitudes toward computers, r (382) =.23, p <.01. There were no differences among the participants across the three tasks in pretask measures of overall attitudes toward computers or any of the attitude dimensions. Follow-up tests on all between-subjects comparisons (i.e., differences between age group, task, and gender, both within and across task experience) were performed using Scheffe's test for multiple comparisons (a =.05). Simple main effects associated with the Age Group Task and Gender Task interactions were analyzed using the univariate analysis of variance procedure and were further analyzed using Scheffe's procedure. For all within-subjects comparisons (i.e., differences across task experience both within and across age group, task conditions, and gender) the Bonferroni procedure was used. O 122 Younger Middle-aged Older Age Group (b) Database Inquiry Task Younger Middle-aged Older Age Group (c) Accounts Balancing Task Overall Attitudes. The results indicated a significant effect of Task Experience, F( 1,363) = 21.54, p <.001, and a significant Task Task Experience interaction, F(2,363) = 5.92, p <.01, for overall attitudes toward computers. Generally, computer task experience resulted in attitudes that were more positive attitudes toward computers. However, as shown in Figure 1, this effect was not uniform across all three tasks. Participants who performed the data entry and database inquiry tasks reported more positive attitudes toward computers following task experience, whereas there was no change in overall attitudes among participants who performed the ac- 122 Younger Middle-aged Age Group D task task Older Figure 1. Age differences in pretask versus posttask scores of overall attitudes toward computers for: (a) the data entry task; (b) the database inquiry task; and (c) the accounts balancing task.
P334 CZAJAANDSHARIT counts balancing task. There were no Age Group or Gender effects for overall attitudes toward computers. Attitude dimensions. There were significant main effects of Task Experience, F( 1,363) = 59.29, p <.001; Age Group, F(2,363) = 3.58, p <.05; and Gender, F (1,363) = 5.51, p <.05, and significant Task Task Experience, F(2,363) = 6.23, p <.01, and Gender Task Experience, F(l,363) = 6.69, p <.01, interactions for ratings of comfort with computers. As shown in Tables 2-4, ratings of comfort with computers increased among the participants performing the data entry and database inquiry tasks but not among those performing the accounts balancing task. Also, as Tables 2-4 show, the older participants indicated less comfort with computers than the other participants. Women experienced a greater increase in comfort with computers than men did following task experience (Table 5). There was also a significant main effect of Age Group, F(2,363) = 5.05, p <.01, for perceptions of control over computers. In addition, the Task Task Experience, F(2,363) = 5.37, p <.01, and Age Group Task, F(4,363) = 2.50, p <.05, interactions were significant for this attitude dimension. Generally, the older and middle-aged participants perceived themselves as having less control over computers than did the younger participants. With respect to the Task Task Experience interaction, participants who performed the data entry and database inquiry tasks indicated an increase in their perception of control over computers following task experience. There were no changes in ratings of this attitude dimension among those participants who performed the accounts balancing task (Tables 2-4). These effects, however, varied according to age group. Specifically, the younger people indicated more control over computers for the data entry task. There were no age differences in ratings of control for the other two tasks. Participants also rated computers as less dehumanizing following task experience, F( 1,363) = 12.51, p <.001. As shown in Tables 2-4, there was less of a change in ratings of dehumanization following task experience for the data entry task compared with the database inquiry and accounts balancing tasks. There were also significant effects of Gender F( 1,363) = 7.47, p <.01, and Age Group, F(2,363) = 4.60, p <.05, and a significant Gender Task Experience interaction, F(2,363) = 4.50, p <.05, for this dimension. Women rated computers as more dehumanizing than men, and the younger subjects rated them as more dehumanizing than the older subjects did. However, with task experience, ratings of dehumanization became more positive for women than men (Table 5). There were significant main effects of Task Experience, F(l,363) = 4.05, p <.05, and Age Group, F(2,363) = 6.22, Table 2. Attitude Ratings as a Function of Age for the Data Entry Task Age Group Attitude Dimension Younger Middle-Aged Older Overall 129.48 9.15 131.80 9.65 2.32 7.15 130.17 9.79 133.25 10.96 3.08 5.51 128.00 10.70 131.17 12.06 3.17 6.99 Comfort 18.12 3.47 20.32 3.04 1.50 2.71 17.42 4.12 19.11 3.83 1.69 2.57 16.73 4.05 18.56 4.04 1.83 2.77 Control 16.60 3.11 17.38 3.50 0.78 2.51 18.39 2.63 19.19 2.86 0.81 2.54 19.05 2.81 19.17 2.58 0.12 2.24 Dehumanization' 15.00 3.88 13.80 3.56-1.20 2.90 13.44 3.70 13.11 3.24-0.33 2.18 12.76 4.05 12.32 3.08-0.44 3.09 Efficacy 21.20 2.44 22.14 2.57 0.94 1.90 21.20 3.00 21.56 2.93 0.36 1.64 20.34 2.69 20.63 2.98 0.29 2.03 Gender Equality 20.12 2.50 20.26 2.93 0.14 1.90 21.47 2.85 22.00 2.93 0.53 1.81 20.27 3.12 21.29 2.93 1.02 2.03 Interest 20.86 2.48 20.92 2.48 0.06 1.87 20.97 2.50 20.94 2.63-0.03 1.21 21.10 2.51 21.15 2.71 0.05 1.69 Utility 22.80 2.56 22.96 2.70 0.16 2.67 23.33 2.72 23.25 2.66-0.08 2.29 24.02 2.57 24.19 2.94 0.17 2.32 Notes: = pretask score; = posttask score; = -. A higher score indicates that respondent believes that computers are more dehumanizing.
AGING AND ATTITUDES TOWARD COMPUTERS P335 Table 3. Attitude Ratings as a Function of Age for Database Inquiry Task Age Group Attitude Dimension Younger Middle-Aged Older Overall 130.23 9.28 134.25 9.37 4.02 6.41 130.45 6.43 133.27 10.96 2.82 7.62 127.22 8.99 132.46 9.44 5.24 7.26 Comfort 18.43 3.47 20.14 2.71 1.70 2.62 18.82 3.00 19.64 3.56 0.82 1.81 16.00 3.97 18.28 3.83 2.28 2.51 Control 17.57 3.55 18.57 3.18 1.00 1.84 17.94 2.47 19.06 2.86 1.12 2.22 18.85 1.98 19.57 2.41 0.72 2.07 Dehumanization" 14.95 3.62 14.25 4.11-0.70 2.42 14.91 3.39 14.97 4.53 0.06 2.51 14.35 3.75 13.76 3.93-0.59 2.06 Efficacy 21.25 2.51 21.73 2.44 0.48 1.64 21.33 2.34 21.70 3.40 0.36 2.41 19.94 2.34 21.20 2.65 1.26 2.12 Gender Equality 20.32 2.61 20.80 2.63 0.48 2.15 20.45 2.45 21.24 2.82 0.79 2.38 20.52 2.84 21.09 2.76 0.57 2.07 Interest 21.16 2.21 21.07 2.19-0.01 1.07 20.94 2.41 20.27 3.55-0.67 2.26 21.17 2.32 21.00 1.92-0.17 2.04 Utility 22.95 2.46 23.93 2.94 0.98 2.85 22.55 2.82 22.85 3.10 0.30 1.91 22.96 2.64 23.74 2.51 0.78 2.52 Notes: = pretask score; = posttask score; =. 'A higher score indicates that respondent believes that computers are more dehumanizing. p <.01, and a significant Task Task Experience interaction, F(2,363) = 3.87, p <.O5, for ratings of efficacy. In general, ratings of efficacy increased with computer task experience. However, this effect was moderated by task. As shown in Tables 2-4, ratings of efficacy increased among participants performing the data entry and database inquiry tasks but not among participants performing the accounts balancing task. The older subjects indicated significantly less efficacy than the other participants did. There were significant effects of Task Experience, F(l,363) = 13.86, p <.001, and Gender, F(l,363) = 6.81, p <.05, for perceptions of gender equality. In general, ratings of this dimension increased with experience, and women had higher ratings on this dimension than men did. An increased rating on this dimension indicates an increased belief that computers are important to both men and women. Finally, there was a significant effect of Task Experience, F{ 1,363) = 4.47, p <.05, and a significant Task Task Experience interaction, F(2,363) = 3.19, p <.05, for ratings of utility of computers. Generally, perceptions of the usefulness of computers increased with task experience. However, this change in attitude dimension varied across the three tasks. The change in ratings of utility were the greatest for the database inquiry task. There was a significant effect of Age Group for this dimension, F(2,363) = 3.27, p <.05. In general, the older people perceived computers as being more useful than the other subjects did. There were no effects for ratings of interest in computers. Relationships Among Computer Attitudes and Perceptions of Stress, Arousal, and Workload Correlational analyses were used to examine the relationships among attitudes toward computers and perceptions of stress, arousal, and workload. Specifically, the difference scores for overall attitude and for each of the attitude dimensions were correlated with ratings of these measures. Day 5 ratings of stress, arousal, and workload were used in this analysis because performance on this day was the most stable and the participants completed posttask attitude ratings on Day 5. As shown in Table 6, computer attitudes were related to ratings of workload but not to perceptions of stress or arousal. Specifically, the data indicated that overall attitudes toward computers and attitude dimensions of control, comfort, and efficacy were related to ratings of task frustration. As levels of frustration with the tasks increased, attitudes toward computers became less positive. Interestingly, ratings of overall attitudes, comfort, interest, and utility were also related to perceptions of performance: People who had more positive ratings of their performance also had more positive attitudes toward computers.
P336 CZAJAANDSHARIT Table 4. Attitude Ratings as a Function of Age for Accounts Balancing Task Age Group Attitude Dimension Younger Middle-Aged Older Overall 133.80 9.06 134.11 10.89 0.32 7.48 130.21 10.24 131.31 11.45 1.10 8.49 127.57 8.66 129.95 10.79 2.38 8.28 Comfort 18.91 3.99 19.61 4.01 0.70 2.35 17.81 4.58 18.65 3.70 0.83 3.14 17.55 3.58 18.02 3.60 0.48 2.90 Control Dehumanization" 19.16 2.81 14.00 4.09 18.75 2.97 13.75 4.27-0.41 2.15-0.25 2.69 18.69 2.92 14.58 4.50 18.46 3.16 13.83 4.94-0.23 2.34-0.75 2.35 18.24 2.62 13.62 3.58 18.93 2.74 13.24 3.63 0.69 2.26-0.38 2.43 Efficacy 21.93 2.14 21.73 2.82-0.20 2.19 21.33 2.36 21.54 2.87 0.21 2.29 20.05 1.97 20.12 2.96 0.07 2.72 Gender Equality 20.66 2.62 21.39 2.89 0.73 2.63 20.21 3.42 20.77 3.50 0.56 2.47 20.31 2.91 21.26 2.49 0.95 2.45 Interest 21.75 1.94 21.72 2.12-0.48 1.69 21.06 2.52 21.08 3.04 0.02 2.15 21.07 1.94 21.12 2.30 0.05 2.06 Utility 23.52 2.42 23.61 2.97 0.09 2.30 22.90 3.04 23.3 3.22 0.40 2.38 23.29 2.64 23.90 2.81 0.62 2.76 Notes: = pretask score; = posttask score; = -. A higher score indicates that respondent believes that computers are more dehumanizing. Relationships Between Computer Attitudes and Performance Level A series of analyses were performed to determine whether changes in attitudes toward computers were related to performance levels on the three tasks. In these analyses, one performance measure from each of the three computer tasks was used to represent task performance; the measures selected were considered to be the most important and reliable measures for each of the tasks. For the data entry task, the measure used was the number of trip records entered into the computer; for the database inquiry task, the measure used was the accuracy of responses to the telephone inquiries; and for the accounts balancing task, the measure used was the number of transactions balanced. Given that the emphasis in these analyses was on performance level, only the data from subjects representing the upper and lower quartiles of performance for each task were used. Two sets of analyses were performed, with prior computer experience included in each analysis as a covariate. In both sets of analyses, the effects of performance level on attitude change were analyzed using a MANOVA procedure with Attitude Change (pretask computer attitude scores vs posttask computer attitude scores) as a within-subjects factor at two levels and Performance Level (upper quartile of performance vs lower quartile of performance) as a between-subjects factor at two levels. The first set of analyses focused on the first day of performance (Day 3) and was performed to determine whether changes in computer attitudes were affected by initial levels of task performance. The results indicated a significant effect of Attitude Change for the data entry task, F(l,48) = 4.87, p <.05, and for the database inquiry task, F(l,57) = 9.88, p <.01, with attitudes toward computers becoming more positive with experience on each of these tasks (Figures 2a and 2b). For the accounts balancing task, a trend toward a significant effect for Performance Level was found, F(l,49) = 3.49, p <.10, with those subjects performing in the upper quartile having more positive attitudes toward computers than the subjects in the lower quartile of performance (Figure 2c). The second set of analyses followed the same procedure as the first, except the upper and lower quartiles of performance on each task were computed for the change in performance between Day 3 and Day 5. The purpose of the second set of analyses was to determine whether changes in computer attitudes were influenced by changes in performance. For the data entry task, the results indicated a significant effect of Attitude Change, F(l,49) = 10.60, p <.01, in the direction of more positive attitudes toward computers as task experience increased. For the database inquiry task, significant effects were found for Attitude Change, F(l,58)
AGING AND ATTITUDES TOWARD COMPUTERS P337 = 9.04, p <.01, and the Attitude Change Performance Level interaction, F(l,58) = 5.53, p <.05. Follow-up analysis of the interaction effect indicated no significant differences in pretask computer attitude scores between the subjects in the upper and lower quartiles of performance. However, the subjects associated with the lower quartile of performance had significantly higher posttask computer attitude scores (p <.05) than their counterparts in the upper quartile. Finally, there were no significant effects found for the accounts balancing task. Attitude Dimension Overall Comfort Control Dehumanization' Efficacy Gender Equality Interest Utility Table 5. Attitude Ratings as a Function of Gender 129.88 9.43 17.39 4.11 18.18 2.87 14.75 3.81 20.94 2.53 20.78 2.80 21.15 2.31 23.10 2.67 Women 132.84 10.54 19.01 3.84 18.70 2.88 14.00 4.03 21.53 2.96 21.40 2.85 20.98 2.62 23.54 2.83 2.96 7.41 1.62 2.70 0.53 2.39-0.75 2.54 0.59 2.25 0.61 2.21-0.18 1.91 0.44 2.42 129.38 9.36 18.39 3.62 18.39 2.95 13.48 3.94 20.99 2.42 19.99 2.81 21.07 2.32 23.20 2.65 Men 131.84 10.65 19.33 3.41 18.83 3.14 13.20 3.91 21.22 2.71 20.70 2.89 21.04 2.45 23.56 2.93 2.46 7.32 0.94 2.63 0.44 2.17-0.28 2.52 0.23 1.99 0.70 2.17 -.03 1.68 0.36 2.55 Notes: = pretask score; = posttask score; = -. "A higher score indicates that respondent believes that computers are less dehumanizing. DISCUSSION The intent of this study was to examine whether attitudes toward computers are influenced by direct computer experience and whether these attitudes vary as a function of age, gender, and computer task characteristics. Specifically, the study examined three tasks that are inherently computerbased and are performed across a wide variety of work settings. The study also explored the relationships between attitudes toward computers and ratings of workload, stress and arousal, and between attitudes toward computers and task performance. Overall, the findings indicated that attitudes toward computers are modifiable and that, irrespective of age or gender, direct experience with computers resulted in more positive attitudes. These results parallel the findings of other investigators (e.g., Danowski & Sacks, 1980; Dyck & Smither, 1994; Jay & Willis, 1992; Marquie et al., 1994) and underscore the importance of providing people, especially those who have had little or no experience with computers, with opportunities to interact with computer technology. Generally, the literature suggests that user attitudes have important implications with respect to the acceptance and use of innovations such as computer technology (Grudin & Markus, 1997). The results of this study also provide insight into the dimensions of attitudes that are influenced by computer experience. The data indicated that experience with computers increased participants' feelings of comfort with technology, competence with computers, and feelings that computers are useful. Furthermore, direct experience increased the perception that computers are important to both men and women. In an earlier study, Jay and Willis (1992) also found that direct experience with computers increased feelings of competence and comfort with technology. Similarly, Marquie and colleagues (1994) found that experienced computer users perceived computers as more interesting, more useful, and less threatening than did nonexperienced users; experienced computer users were also more likely to use computers. Understanding the dimensions of attitudes that are influenced by computers provides information regarding specific impact of experience on users and highlights dimensions of attitudes that are not influenced by experience. Information of this type is important to the design of inter- Table 6. Correlations Between the ference Scores in Computer Attitude and Workload, Stress and Arousal Measures Workload (n = 371) Overall Effort Level Frustration Level Performance Level Mental Demand Physical Demand Temporal Demand Stress (n = 369) Arousal (n = 369) Computer Attitude Overall Comfort Control Dehumanization Efficacy Gender Interest Utility -.069 -.065 -.080 -.067 -.081.007.023.076 -.034 -.023 -.018 -.060 -.010 -.059 -.040 -.016 -.246** -.210** -.124*.011 -.212** -.085 -.078 -.047.169**.174**.054.008.095.002.124*.118* -.053.026 -.075.013 -.106* -.103 -.088 -.010 -.044 -.082.003 -.062.002.076 -.033 -.046.064 -.057 -.008.086.052.141**.094.145**.051.076 -.039 -.019.045.079 -.016.041 -.049 -.088.060 -.069 -.010.005 -.016 -.048 *Correlation is significant at the 0.05 level (2-tailed). **Correlation is significant at the 0.01 level (2-tailed).
P338 CZAJAANDSHARIT 120 120 task (a) Data Entry Task task (b) Database Inquiry Task task task (c) Accounts Balancing Task task task Lower Quartile Upper Quartile Figure 2. ferences in pretask versus posttask scores of overall attitudes toward computers between participants in the upper and lower quartiles of performance on Day 3 for: (a) the data entry task; (b) the database inquiry task, and (c) the accounts balancing task. ventions such as training programs. For example, in this study task experience had no effect on the participants' ratings of interest in computers, which suggests that potential real-world applications of computers need to be stressed during training. As noted, interest in technology is a strong predictor of willingness to use technology in the future. Marquie and colleagues (1994) found that resistance to change largely depends on the lack of knowledge of the innovative effects of a new technology. In the present study, the lack of change in interest in computers with increased task experience may have reflected the nature of the experimental tasks. The tasks investigated in this study were limited to actual work tasks, and the participants were not exposed to other, more versatile computer applications such as spreadsheets, word processing, and E-mail. The data also indicated that changes in attitudes were moderated by task characteristics. For example, there were no changes in overall attitude or in ratings of comfort, control over computers, or competence with computers for people performing the accounts balancing task. This task was more cognitively demanding than the other two tasks and involved a graphical user interface that required manipulating a mouse and using Windows. In fact, the participants rated this task as more mentally demanding than the other tasks and found it to be more frustrating (Sharit et al., in press). Our data indicated that as levels of frustration increased, overall attitudes toward computers became less positive as did feelings of comfort, control, and competence. In addition, attitudes toward computers were related to ratings of performance; specifically, people who rated their performance higher had more positive attitudes toward computers (Table 6). Overall, these findings point to the importance of providing users with adequate training so they have the skills needed to operate computers successfully. The findings also underscore the importance of usability with respect to interface design. Other important issues relate to how one's initial experience in the performance of a computer task influences attitude change and how attitude change is influenced by changes in performance with task experience. The results of this study indicate that the nature of the computer task can affect both of these relationships. With respect to initial levels of task performance, change in attitude toward computers was not a function of performance level for the relatively less cognitively demanding data entry and database inquiry tasks. However, for the accounts balancing task, the better performers had more positive attitudes toward computers than the subjects who had more difficulty initially grasping the task. These results suggest that extra attention should be given to training and design strategies that can minimize mismatches between the cognitive demands of the computer task and the cognitive skills of the user. Otherwise, there is a risk that users may feel that they are not capable of handling the task and may adopt negative attitudes toward computers. This, in turn, may influence their willingness to use computers in the future. The results of this study also indicated age effects for computer attitudes. Although there were no age differences in overall attitudes, which is consistent with findings of other investigators (e.g., Czaja et al., 1989), there were age
AGING AND ATTITUDES TOWARD COMPUTERS P339 effects for several of the attitude dimensions. Specifically, the older people reported less comfort and less competence with computers, and felt they had less control over computers. They also perceived computers to be more dehumanizing than did the other participants. These data support the findings of Marquie and colleagues (1994) who also found that the older workers in their sample were less comfortable with computers and were more sensitive to the lack of flexibility in operating procedures when using computers to perform tasks. In our study, the older participants had less prior computer experience than the other participants, and prior experience with computers was positively related to the ratings of comfort, competence, and efficacy. However, the age effects were found even after controlling for differences in prior computer experience. This suggests that other age-related factors were important to attitude ratings, and these factors need to be investigated in future research. The data also demonstrate the importance of evaluating the various dimensions of the attitude scale. As we have shown, there are multiple dimensions along which attitudes vary, and reliance on a unidimensional measure of attitudes may mask these effects. Finally, the data indicated relatively few gender differences in attitudes. Specifically, the women experienced a greater increase in comfort with computers following task experience than did the men. However, the women also found computers to be more dehumanizing following task experience. These findings largely refute the suggestion that women are less receptive and have more negative attitudes toward computers than men. Moreover, there were no age group by gender interaction effects indicating that older women are as receptive to computer technology as younger women are. Overall, the results of this study support the belief that attitudes toward computers are modifiable and that providing users with an opportunity to interact with new technologies, such as computers, is an effective means of attitude change. Our findings also demonstrate that factors including level of frustration and level of performance during initial interaction with a technology have an influence on attitude change. In this regard it is important to ensure that users are provided with adequate support during their interactions with technologies. The data also highlight the importance of understanding how individual characteristics (e.g., age) influence attitudes toward computers. This type of knowledge can be used to develop more effective methods for introducing computers to various user groups. Given that older people typically have had less experience and exposure to technology, it is critical for them to be introduced to technologies such as computers in a manner that allows them to feel comfortable with the technology, experience some success in the performance of computer tasks, and understand the utility and benefits associated with using technology. ACKNOWLEDGMENTS This research was supported in part by National Institute on Aging Grant AG11748-05 and was conducted in association with the Miami Center on Human Factors and Aging Research, one of the Edward R. Roybal Centers for Research on Applied Gerontology. The authors thank K. Ercan Dilsen, Chin Chin Lee, and Sankaran Nair for their invaluable assistance. Address correspondence to Dr. Sara J. Czaja, Miami Center on Human Factors and Aging Research, University of Miami School of Medicine, 1425 NW 10th Avenue, Miami, Florida 33136. E-mail: sczaja@eng.miami.edu REFERENCES Cruickshank, P. J. (1984). A stress and arousal mood scale for low vocabulary subjects: A reworking of Mackay et al. (1978). British Journal of Psychology, 75, 89-94. Czaja, S. J., Guerrier, J. H., Nair, S. N., & Landauer, T. (1993). Computer communication as an aid to independence for older adults. Behavior and Information Technology, 12, 197-207. Czaja, S. J., Hammond, K., Blascovich, J. J., & Swede, H. (1989). Agerelated differences in learning to use a text-editing system. Behavior and Information Technology, 8, 309-319. Danowski, J. A., & Sacks, W. (1980). Computer communication and the elderly. Experimental Aging Research, 6, 125-135. Derogatis, L. R. (1977). SCL-90R (Rev. ed.) [Manual]. Baltimore, MD: John Hopkins University School of Medicine, Clinical Pychometrics Unit. Dyck, J. L., & Smither, J. A. (1994). Age differences in computer anxiety: The role of computer experience, gender and education. Journal of Educational Computing Research, 10, 239-247. Edwards, R., & Engelhardt, K. G. (1989). Microprocessor-based innovations and older individuals: AARP survey results and their implications for service robotics. International Journal of Technology and Aging, 2, 56-76. Folstein, M. F., Folstein, S. A., & McHugh, P. R. (1975). Mini-Mental State: A practical method for grading the cognitive state of patients for the clinician. Journal of Psychiatric Reseach, 12, 189-198. Fossum, J. A., Arvey, R. D., Paradise, C. A., & Robbins, N. E., (1986). Modeling the skill obsolescence process: A psychological/economic integration. Academy of Management Review, 11, 362-374. Grudin, J., & Markus, M. L. (1997). Organizational issues in development and implementation of interactive systems. In M. G. Helander, T. K. Landauer, & P. V. Prabhu (Eds.), Handbook of human-computer interaction (2nd ed., pp. 1457-1475). Amsterdam: Elsevier Publishing. Hart, S. G., & Staveland, L. E. (1988). Development of NASA-TL (Task Load Index): Results of empirical and theoretical research. In P. A. Hancock & N. Meshkati (Eds.), Human mental workload (pp. 139-183). Amsterdam: North Holland. Jay, G. M., & Willis, S. L. (1992). Influence of direct computer experience on older adults' attitudes toward computers. Journal of Gerontology: Psychological Sciences, 47, 250-257. Krauss, I. K., & Hoyer, W. J. (1984). Technology and the older person: Age, sex, and experience as moderators of attitudes towards computers. In P. K. Robinson, J. Livingston, & J. E. Birren (Eds.), Aging and technological advances (pp. 349-350). New York: Plenum ss. Mackay, C, Cox, T., Burrows, G., & Lazzerini, T. (1978). An inventory for the measurement of self-reported stress and arousal. British Journal of Social and Clinical Psychology, 17, 283-284. Mackie, R. R., & Wylie, C. D. (1988). Factors influencing acceptance of computer-based innovations. In M. Helander (Ed.) Handbook of human-computer interaction (pp. 1081-1106). New York: Elsevier Publishing Co. Marquie, J. C, Thon, B., & Baracat, B. (1994). Age influence on attitudes of office workers faced with new computerized technology. Applied Ergonomics, 25, 130-142. Maxwell, S. E., & Delaney, H. D. (1989). Designing experiments and analyzing data. Belmont, CA: Wadsworth Publishing Company. Parsons, H. M., Terner, J., & Kearsley, G. (1994). Design of remote control units for seniors. Experimental Aging Research, 20, 211-218. Regan, D. T, &. Fazio, R. (1977). On the consistency between attitudes and behavior: Look to the method of attitude formation. Journal of Experimental and Social Psychology, 13, 28-45. Rogers, W. A., Cabrera, E. F., Walker, N., Gilbert, D. K., & Fisk, A. D. (1996). A survey of automatic teller machine usage across the adult life span. Human Factors, 38, 156-166. Rosen, B., & Jerdee, T. H. (1976). The nature of job-related stereotypes. Journal of Applied Psychology, 61, 180-183. Schwartz, J. (1988). The computer market. American Demographics, 10, 38-41.
P340 CZAJA AND SHARIT Shackel, B. (1986). Ergonomics in design for usability. In M. D. Harrison tors affecting the adoption of automatic teller machines. The Journal of & A. F. Monk (Eds.), People and computers: Designing for usability. General Psychology, 121, 381-389. Proceedings of the second conference of the BCS HC1 Specialist Zeithaml, V. A., & Gilly, M. C. (1987). Characteristics affecting the accep- Group. Cambridge: Cambridge ss. tance of retailing technologies: A comparison of elderly and nonelderly Sharit, I, Czaja, S. J., Nair, S. N., Hoag, D. W., Leonard, D. C, & Dilsen, consumers. Journal of Retailing, 63, 49-68. K. E. (in press). Subjective experiences of stress, workload, and bodily discomfort as a function of age and type of computer work. Work & Stress.. Received October 13, 1997 Smither, A. J., & Braun, C. C. (1994). Technology and older adults: Fac- Accepted March 30, 1998 We've Moved! On September 1, 1998, we moved our offices to a new location: The Gerontological Society of America National Academy On An Aging Society 1030 15th Street, NW Suite 250 Washington, DC 20005-1503 Our telephone and fax numbers remain the same: GSA: (202) 842-1275 telephone NAAS: (202) 408-3375 telephone (202) 842-1150 fax (202) 842-1150 fax