تفاوت سنی در درک قبول واقعیت PDA و عملکرد

Abstract The present study addresses two basic determinants of technology utilization: the attitude towards a certain technology and the performance when using it. According to the technology acceptance model (TAM), perceived ease of use and usefulness are assumed to be strong determinants of the actual and successful utilization of technology. However, the relationship between the acceptance of technical devices and their successful utilization (i.e. performance) is not completely understood. In this study, users’ attitudes towards technology and their performance when interacting with a computer simulated PDA device were examined. Moreover, the moderating role of individual variables like age, gender, subjective technical confidence, and computer expertise in the relationship between technical performance and acceptance was analyzed. The results showed significant associations between performance and TAM factors. However, this interrelation was much stronger for the older group, especially between performance and the ease of use. The factors computer expertise and technical self-confidence played a minor role. Gender effects on technical self-confidence and TAM factors were identified, although they did not affect performance. Future research should focus on training formats for the older age group, which facilitate a successful interaction with technical devices.

Introduction Information technology (IT) has proliferated into most professional and private areas in the last decades. There are only a few areas left that are not permeated by IT. A prominent role in this context play mobile communication technologies, e.g. mobile and smart phones, communicators and electronic organizers, which show continuously increasing rates of growth each year (Shiffler, Smulders, Correia, Hale, & Hahn, 2005). Contrary to former times, when the utilization of IT was restricted to a sophisticated user group, nowadays broader user groups have access to IT. Moreover, the utilization of IT is no longer voluntary – its effective use has become an essential requirement in today’s working and private life. The organization of professional and private activities, events and transactions heavily depends on the utilization of technical devices and demands the acceptance and application of IT from our society. However, the purchase of technical devices does not guarantee their effective utilization. Users show a broad variety of behaviors when they deal with new devices. They may completely reject them, may only partially use selected functions or they may completely adopt the technology and all functionalities offered. The central question is: What are the underlying factors causing different levels of acceptance and different reactions towards the utilization of technology? From a sociopolitical, psychological and educational point of view, the investigation of the nature and key factors affecting the utilization behavior of IT is important for both, research and practice. 1.1. Technology acceptance and utilization Technology acceptance has become a key concept in a broad field of research areas, such as marketing, ergonomics, pedagogic and psychology. It can be described as the approval, favorable reception and ongoing use of newly introduced devices and systems. Regarding the concept of technology acceptance a further distinction between “attitude-acceptance” and “behavior-acceptance” was established. Attitude-acceptance includes an affective (motivational-emotional) and a cognitive component, which implies a cost-benefit-analysis of system usage. The attitude-acceptance of a user is not directly observable. “Behavior-acceptance” refers to the observable part of technology acceptance and describes the adoption of innovations by using them. In other words: acceptance contains an attitude towards a certain behavior and the behavior itself. The issues of technology acceptance, actual utilization and their mutual relationship have been researched from multiple theoretical perspectives, e.g. diffusion of innovation or social psychology. Melenhorst, Rogers, and Caylor (2001) explain the acceptance and decision to use communication technologies in terms of a cost-benefit analysis. Not all users perceive technology as advantageous and helpful for them. Therefore, users weigh the individually expected benefits and costs (e.g. investment of money and energy, frustration while learning to use the system) before adopting a new technology. Another approach to explain technology acceptance is the task-technology-fit (TTF) model (Goodhue & Thompson, 1995), which postulates that the acceptance of a system depends on the individual estimation of system performance. In turn, the estimation of system performance is influenced by characteristics of the task (complexity), the technology (functions) and the individual (skills and abilities). More specifically, the TTF model suggests that technology adoption depends on how well the new technology fits into the requirements of a particular task. The TTF model provides an informative basis about context- and individual variables, that might influence the issue of attitude acceptance, but it does not explain the behavior acceptance, i.e. the actual use of the system. The most influential theoretical approach in studying the determinants of IT utilization is the technology acceptance model (TAM, Davis, 1989). Based on the theoretical assumptions of the theory of reasoned action (Ajzen & Fishbein, 1975) the TAM offers a link between technology acceptance and utilization behavior. Following the theory of reasoned action, a person’s behavior is determined by the intention to perform a certain behavior. This intention is assumed to be a function of one’s own attitude towards the behavior and individual norms. According to the TAM, a users’ decision to use a new technical device or software package is determined by the behavioral intention to use the system. This behavioral intention is in turn determined by the perceived ease of use of the system and its perceived usefulness. The ease of use describes “the degree to which a person believes that using a particular system would be free from effort”, the perceived ease of use is “the degree to which a person believes that using a particular system would enhance his or her job performance” (Davis, 1989). Empirical findings showed that the perceived usefulness is the main predictor of the behavioral intention to use the system, whereas the ease of use strongly influences the perceived usefulness. Furthermore, the TAM assumes “some external variables” to influence the perceived ease of use as well as the usefulness, but neither the nature of these external variables nor their modes of functioning are described in the model. Furthermore, a rather critical assumption of this model is, that perceived ease of use and perceived usefulness fully mediate the influence of external variables on IT usage behavior. As a consequence, this assumption allows to exclude the effects of external variables, i.e. those variables related to a specific task, to system characteristics and user differences. In the extended version of the technology-acceptance-model (Venkatesh, 2000 and Venkatesh and Davis, 2000) a number of external variables was added, which were assumed to influence the behavioral intention to use a system, e.g. social and cognitive processes (subjective norm, system image and relevance, quality of output). However, even in the revised TAM the inclusion of influential external variables is not completely satisfying. Up to now, the impact of user characteristics like demographic variables (age, gender, expertise) or cognitive (abilities) and attitudinal (self-efficacy) concepts on perceived ease of use and usefulness as the main factors that influence the behavioral intention to use the system, is not adequately regarded yet. Therefore, the following section deals with the role of individual differences in technology acceptance and user behavior. 1.2. Individual differences in technology acceptance and user behavior Individual differences such as demographic variables, computer experience, cognitive abilities and personality factors have long been important in end user computing research (e.g. Zmud, 1979). It was found that individual differences are significant factors in explaining both, technology acceptance and user behavior (Arning and Ziefle, submitted, Chua et al., 1999, Gefen and Straub, 1997, Harrison and Rainer, 1992 and Ziefle et al., 2004). In the context of this study we refer to a key set of variables like age, computer expertise, gender and subjective technical confidence. The factor user age plays an important role in the explanation of variability in system acceptance and performance. Zajicek and Hall (2000) state that perceived usefulness of a technology is lower in older adults, because they weigh the perceived usefulness against the time to learn how to operate the system. Related to this balancing procedure is the fear of failure as additional cost, which is much more pronounced in older adults. Argarwal and Prasad (1999) conceptualized the factor user age as tenure in workforce, but found no significant associations to attitude variables or behavioral intentions. The question arises, if tenure is an adequate operationalization of user age. First, it does not convey information about age-related chances in physical and psychic/mental functioning and second, it does not explain individual differences in different age groups. The findings presented so far refer to older users’ attitudes towards technology. Regarding performance when using a device, previous studies congruently showed that older users usually have greater difficulties in handling a computer device or in the acquisition of computer skills (e.g. Goodman, Gray, Khammampad, & Brewster, 2004). However, the knowledge about the influence of age on the estimation of perceived ease of use and usefulness as well as its relation to performance is limited. Prior computer experience and its influence on attitude and performance when using technical devices has been investigated in a huge number of studies (e.g. Argarwal and Prasad, 1999, Downing et al., 2005 and Ziefle, 2002). Generally it was found that experts show a superior performance with respect to the utilization of technology. This finding is rather trivial as long as the relationship between computer experience, technology acceptance and performance is not completely clarified. Argarwal and Prasad (1999) found that prior experience was associated with ease of use, but did not directly affect the behavioral intention. Since Argarwal and Prasad (1999) did neither report associations between individual variables and expertise nor assessed performance measures, the relationship between expertise and individual variables, acceptance and performance is not fully understood. Although gender differences have been missing a long time from IT research, they are now widely discussed as an important factor in the explanation of computer attitudes and performance. In the TAM no references to the impact of gender are found. Gefen and Straub (1997) examined gender differences in the perception and utilization of an email-system according to the TAM. Gender effects were confirmed for the TAM factors, however in a differential way: While women reported higher values of perceived usefulness, men were found to report a higher ease of use with computers. Gender differences though were not found with respect to the reported frequency of email usage, thus attitudes towards technology were not found to affect the actual use. On the other hand research has shown that women usually report lower levels of computer-related self-efficacy and a higher computer anxiety (e.g. Busch, 1995) as well as a lower subjective technical confidence when using technical devices ( Arning and Ziefle, submitted and Ziefle et al., 2004). Even though women’s attitudes towards technology are more negative and their self-confidence when using technical devices is significantly lower, gender effects seem to be limited to attitudes and subjective measures, and do not involve lower performance outcomes for female users (e.g. Arning and Ziefle, submitted and Ziefle and Bay, 2006). This shows – once more – that the relationship between individual variables, attitudes and performance requires further examination. The subjective technical confidence (STC) is also discussed as an influential factor of technology acceptance and system usage. The STC is an individual belief in one’s own ability to solve technical problems ( Beier, 1999). The construct of STC has its source in the concept of locus of control, developed by Rotter (1955). The locus of control refers to individuals’ generalized expectations concerning where control over subsequent events resides. The original locus of control formulation classified generalized beliefs concerning who or what influences things along a bipolar dimension from internal to external control. Internal locus of control means that individuals usually attribute success to their own competency, while individuals with an external locus of control ascribe their action outcomes not to themselves but rather to chance or others. The concept of subjective technical confidence also assumes one dimension, ranging from low subjective technical confidence (external control) to high subjective technical confidence (internal control). In comparison to other related concepts, which evaluate users’ beliefs about their capacity to handle technical devices (e.g. domain-specific self-efficacy), the STC refers to a broader range of technical interactions or situations. Studies have shown that high scores in computer self-efficacy are related to navigational performance and reported ease of use (e.g. Brosnan, 1998 and Liu and Grandon, 2003). However, the selection, treatment and conceptualization of individual variables, technology acceptance and user behavior varies considerably in these studies. For example, the often-cited model of Zmud (1979) about individual variables and technology acceptance is based on a literature review without an empirical confirmation of the influence of individual variables. Moreover, the operationalization of “user behavior” was also realized in different ways. In some studies, user behavior was assessed by the rated intention to use the system or the reported frequency to use technical systems (e.g. Gefen & Straub, 1997). In other studies, measurements of performance when interacting with a system were carried out (e.g. Busch, 1995). In the context of performance measurements, effectiveness and efficiency are usually under study, evaluating different designs or measuring the influence of user characteristics on performance. However, it was found that “perceived usefulness” as the main predictor of system utilization in the TAM is poorly correlated with actual use (e.g. Straub, Limayem, & Karahanna-Evaristo, 1995), self-reported utilization or intent to use. Therefore, the TAM-variables “ease of use” or “usefulness” may not be an appropriate operationalization for the actual use, because users are poor estimators of aspects of their own behaviors (Szajna, 1996). Therefore, the actual utilization and performance when using a system should be considered in research activities. 1.3. Research model Summarizing the knowledge so far, the perceived ease of use and the perceived usefulness of a technical device are regarded as important determinants for the acceptance and utilization of technology. Both are assumed to influence behavior intentions to use technical devices. However, prior research has reported inconsistent and incomplete results about the relationship between individual variables, attitudes and performance. The knowledge about which and how external or individual factors determine the perceived ease of use and the perceived usefulness is rather limited. This particularly applies for the factors age and gender, but also for the level of computer experience and the subjective technical confidence when using technical devices. Moreover, quantifying performance when interacting with a device does not allow to draw conclusions regarding attitudes that may have influenced performance outcomes. Comparably few studies were concerned with both aspects simultaneously, i.e. the determination of the impact of attitudes towards technology on performance when using technology (e.g. Liu & Grandon, 2003). A second limitation – which applies to all mentioned research streams – is concerned with the fact that mainly young adults were examined in the studies. Therefore it is unknown, if the outcomes can be generalized to the older adult group which has a very different upbringing, expertise, learning history and culture regarding the utilization of technology. Accordingly, the present study focuses on the co-acting and interdependence of participants’ age, gender, expertise and subjective technical confidence, their performance when using a technical device, the reported usefulness and the perceived ease of using the device. Due to the significance of perceived ease of use for technology acceptance, it was examined, which variables contribute to the ease of use of a system and therefore, to the improvement of technology acceptance. As a technical device, an electronic organizer (personal digital assistant, PDA) was selected, a typical representative of a mobile device. Two main research goals were pursued in this study: (1) We wanted to analyze the contributing factors to the perceived ease of use. Therefore the goal was to determine whether task performance or subjective technical confidence are the main predictors or if both variables jointly explain the ease of use. (2) In addition, by considering gender, age and the level of computer experience, we wanted to examine whether individual variables have a moderating impact on the relationship between subjective technical confidence, task performance and acceptance. In order to structure the number of variables (individual, performance and acceptance variables), a working model was developed, which serves as an underlying framework (Fig. 1). It comprises the different variables and their hypothesized relationships. On the left hand side, the individual variables are given, which are assumed to influence subjective technical confidence, performance outcomes and perceived ease of use as well as perceived usefulness. Research model. Fig. 1. Research model. Figure options According to the model, the following hypotheses were specified. H1: Individual variables such as age, gender and expertise are related to the subjective technical confidence (STC), perceived ease of use (PEU) and perceived usefulness (PUF). Young adults, male users and experts report a higher subjective technical confidence, a higher ease of use and a higher usefulness. H2: Younger adults and experts perform better in navigational tasks. H3: STC, PEU and PUF are positively associated with navigational performance. H4: Young age, high levels of STC and performance a positively related. H5: High degrees of STC and good navigation performance are positively related to PEU. H6: PEU and PUF are positively related.

. Results 3.1. Reliability and validity of scales The reliability of the STC and the TAM measures (PEU and PUF) was assessed by scale reliability analysis. The Cronbach alpha values were 0.91 for the STC and 0.96 for the TAM. Compared to the acceptance level of 0.7 for empirical research, the observed scale reliabilities were extraordinarily high and indicated that the items for each scale were internally consistent and reliable. A principal factor analysis with varimax rotation (with Kaiser Normalization procedure) was conducted to ensure that items for the same construct measure the particular trait, while items for another construct measure a different trait. Three factors that explained 79.1% of variance in all items were extracted. The rotated factor matrix in Table 1 shows that all items loaded on the correct latent constructs (indicated by bold values). Therefore, the factorial validity of the scales can be taken for granted. Table 1. Factor analysis of the measurement scales PEU, PUF and STC Factor loadings PEU PUF STC TAM1 (PEU) .910 .302 .183 TAM2 (PEU) .845 .401 .141 TAM3 (PEU) .878 .294 .123 TAM4 (PEU) .902 .291 .198 TAM5 (PEU) .797 .331 .255 TAM6 (PEU) .770 .453 .193 TAM7 (PUF) .306 .848 .046 TAM8 (PUF) .404 .753 .005 TAM9 (PUF) .362 .863 .075 TAM10 (PUF) .384 .815 .080 TAM11 (PUF) .270 .899 .129 TAM12 (PUF) .183 .890 .211 STC1 .338 .251 .777 STC2 .176 .090 .785 STC3 .047 .408 .788 STC4 .526 .161 .675 STC5 −.059 .238 .724 STC6 .106 .025 .770 STC7 .106 −.166 .789 SCT8 .406 −.319 .641 Bold values indicate items’ loadings on correct latent constructs. Table options 3.2. Individual variables and STC, PEU and PUF H1 states differences in the degree of one’s STC and TAM (PEU and PUF) depending on individual variables like age, gender and expertise. To confirm that, a MANOVA with the between-factors age, gender and expertise on the dependent variables STC, PEU and PUF was conducted. Multivariate tests revealed a significant main factor of age: older adults score significantly lower in STC, PEU and PUF than younger adults (F(3, 22) = 4.7; p < 0.05, Table 2). Table 2. Means (and SD) for both age groups for STC, PEU and PUF (n = 32) Factor “age” STC PEU PUF Young 75.3 (14.8) 23.7 (4.5) 18.6 (7.4) Old 62.0 (14.8) 17.0 (8.2) 14.1 (6.8) Table options The factors expertise and gender missed statistical significance in the overall F-test (F(3, 22) = 3.3, p > 0,05), but the interaction of age and gender reached significance (F(3, 22) = 3.3, p < 0.05). Regarding the STC ( Fig. 2 (left)), young male users show a higher technical confidence (M = 85.3) compared to young female users (M = 66.9) and older adults (Mmale = 61.6; Mfemale = 61.7). The examination of Fig. 2 (left) shows, that the interaction of age and gender is ordinal, that means, that differences in the STC only occur for young male adults. For the PEU ( Fig. 2 (center)), a hybrid interaction occurs, i.e. only one – the factor age – of the two factors gender and age can be globally interpreted: younger adults (Mmale = 25.8, Mfemale = 21.1) report a higher ease of use than older adults (Mmale = 13.7, Mfemale = 20.9). A disordinal type of interaction exists for the PUF, i.e. none of the main factors must be interpreted ( Fig. 2 (right)). The degree of perceived usefulness differs depending on age and gender: young male adults score the highest on the PUF-scale (M = 22.8), in contrast to older male adults, who score the lowest (M = 12.4). For female participants, the relationship is inverted (Myoung = 14.9; Mold = 15.9) (see Table 3). Interactions of age and gender for STC (left), ease of use (center) and ... Fig. 2. Interactions of age and gender for STC (left), ease of use (center) and usefulness (right). Figure options Table 3. Means (and SD) for both age groups for number of tasks solved and Time on task (n = 32) Factor “age” Number of tasks solved [max = 2] Time on task Young 1.8 (0.5) 178.9 (78.5) Old 0.6 (0.7) 321.2 (61.9) Table options Concluding the results so far, H1 is supported partially. Individual variables like age and gender influence the extent of technical confidence, ease of use and usefulness. Younger and especially male adults report the highest ratings regarding STC, PEU and PUF. Gender differences only play a role in relation to age: younger male adults reach the highest values on the STC, PEU and PUF scale. Contrary to expectations, expertise was not found to be an influential factor, neither regarding subjective technical confidence, nor for ease of use and usefulness. 3.3. Individual variables and performance The influence of individual variables on navigation performance was formulated in H2, stating that performance is higher for young participants and participants with higher computer expertise. In order to rule out gender differences, the factor gender was included into the data analysis. A 2 × 2 × 2 analysis of variance with the factors age, gender and expertise as between-subject variables and effectiveness (number of tasks solved) and time on task as dependent variables was carried out. A highly significant main effect of age was revealed in the omnibus F-test (F (2, 23) = 18.5, p < 0.01): young participants solved more tasks more quickly than older participants. The performance of male and female participants as well as novices and experts did not differ significantly. Neither for the factor gender nor the factor expertise significant effects were observed. Summarizing the results, H2 is partially confirmed: only the factor age has a considerable impact on performance outcomes. 3.4. Relationship between STC, PEU and PUF and navigational performance To examine the positive relationship between the STC, PEU and PUF on the one hand and performance on the other hand (as stated in H3), correlational analyses were conducted (Table 4). Results reveal that STC, PEU and PUF show significant associations with time on task, indicating that a high technical confidence (r = −.64; p < 0.05), high ease of use (r = −.69; p < 0.05) and usefulness (r = −.41; p < 0.05) were related to a shorter time on task. The same pattern is obtained for task effectiveness: participants reporting a high technical confidence (r = .58; p < 0.05) and a high ease of use (r = .72; p < 0.05) are more successful in solving the experimental tasks. Perceived usefulness, in contrast, was not correlated with the number of tasks solved. Table 4. Bivariate correlations between STC, PEU, PUF and performance variables (n = 32) STC PEU PUF Number of tasks solved Time on task STC 1 .46⁎⁎ .30 .58⁎⁎ −.64⁎⁎ PEU 1 .67⁎⁎ .72⁎⁎ −.69⁎⁎ PUF 1 .35 −.41⁎ Number of tasks solved 1 −.86⁎⁎ Time on task 1 ⁎ p < 0.05. ⁎⁎ p < 0.01. Table options On the basis of these correlations, H2, which states a relation between attitude variables like STC, PEU and PUF and navigational performance, is confirmed. Participants, who report a highly pronounced technical confidence, ease of use and usefulness, tend to perform better in the experimental PDA-tasks. In order to determine if the correlation outcomes are present in both age groups, further correlation analyses were carried out for both age groups separately (Table 5). Table 5. Bivariate correlations between STC, PEU, PUF and performance variables for younger (top) and older adults (bottom) STC PEU PUF Number of tasks solved Time on task Younger adults STC 1 .57⁎ .42 .35 −.55⁎ PEU 1 .57⁎ .52⁎ −.66⁎⁎ PUF 1 .091 −.27 Number of tasks solved 1 −.61⁎ Time on task 1 Older adults STC 1 .23 −.05 .51⁎ −.51⁎ PEU 1 .70⁎⁎ .67⁎⁎ −.61⁎ PUF 1 .26 −.28 Number of tasks solved 1 −.87⁎⁎ Time on task 1 ⁎ p < 0.05. ⁎⁎ p < 0.01. Table options The comparison of Table 4 and Table 5 shows, that the relationship between STC and the number of tasks solved is only significant for the older group, whereas this relationship misses significance for the younger adult group (rold = .51, p < 0.05 vs. ryoung = .35, n.s.). Furthermore, a stronger association between PEU and task effectiveness for older adults is revealed: The age-group-specific correlation analyses show, that the linear relationship between individual beliefs and task effectiveness is stronger for older adults (rold = .67, p < 0.01 vs. ryoung = .52, p < 0.05). As the multivariate and correlational analyses did not reveal any systematic relationship between expertise, individual variables, attitude variables and performance, the variable “expertise” was excluded from further analyses. 3.5. Regression analysis So far, results show that individual variables like age and gender, subjective beliefs like STC, PEU and PUF and task performance are closely interrelated. To get a deeper insight into the relationship of those variables and to examine the research model, individual sets of multivariate regression analyses were conducted. It was of central interest to find out what contributes more to the acceptance of mobile devices like PDAs. Is objective task performance the main predictor of ease of use? Or do subjective perceptions of capabilities (STC) predict the degree of acceptance? Additionally, the – probably moderating – effects of age were of interest. In order to detect effects of multicollinearity, i.e. when predictors are correlated to each other, leading to biases in the regression model, collinearity diagnosis were carried out. As the condition index ⩽10 showed, no collinearity between the predictors was found. 3.5.1. Prediction of subjective technical confidence (STC) As addressed before, H1 declares an association between age and gender for the subjective technical confidence. The stepwise regression analysis on STC with the variables age, gender, task effectiveness and efficiency, PEU and PUF was highly significant (at α = 0.000): Time on task (as performance component) and gender (as individual variable) are significant predictors of the subjective technical confidence ( Table 6). Interestingly, instead of age – what would have been expected – the predictor task efficiency (time on task) was included in the regression model, indicating that age alone does not have much predictive power in explaining age-related differences in the STC. Table 6. Regression analysis on STC (n = 32) (TOT, time on task) Dependent variable Adj. R2 Condition index Predictor β p t-Value STC 0.48 6.1 TOT −0.63 0.00 −4.81 Gender −0.33 0.02 −2.50 Table options As shown in the interaction of age and gender for the STC in Fig. 2, the significant influence of gender in the regression model is caused by extremely high STC-scores of young male adults. This was proven again in the following analysis. Testwise regressions for both age groups show, that the factor gender has predictive power only for the STC of younger adults (Table 7). For older adults, time on task is the only predictor for the STC. The negative β-coefficient expresses that a high STC is associated with short processing time and vice versa. Table 7. Testwise regression analysis on STC for young adults (n = 16) and older adults (n = 16) Dependent variable Age group Adj. R2 Condition index Predictor β p t-Value STC Young 0.62 5.7 Gender −0.64 0.00 −3.84 TOT −0.36 0.05 −2.20 STC Old 0.21 10.8 TOT −0.51 0.04 −2.21 Table options 3.5.2. Prediction of performance H4 claims a positive relationship between young age, high levels of STC and performance. Because performance was assessed by two variables – task effectiveness and efficiency – the regression analyses were carried out for each performance measure separately. Due to high correlations between the performance variables, one of each performance variables was excluded from the regression analyses on the other performance variable, respectively. According to the Condition Indices ⩽10, multicollinearity was ruled out for the regression models. 3.5.2.1. Task effectiveness (number of tasks solved) For the whole group of participants, the perceived ease of use and the factor age explain 68% of variance in task effectiveness (α = 0.000, Table 8). Participants, who report a high ease of use after solving the tasks, also reached high task effectiveness. Table 8. Regression analysis on number of tasks solved (n = 32) Dependent variable Adj. R2 Condition index Predictor β p t-Value Number of tasks solved 0.68 10.1 PEU 0.50 0.00 4.38 Age group −0.48 0.00 −4.23 Table options In testwise regressions for both age groups, the predictive power of perceived ease of use is lower for younger adults (R2 = 22%) than for older adults (R2 = 41%). This indicates that the relationship between PEU and task effectiveness is weaker for younger adults than for older adults (see Table 9). Table 9. Testwise regression analysis on performance (number of tasks solved) for young adults (n = 16) and older adults (n = 16) Dependent variable Age group Adj. R2 Condition index Predictor β p t-Value Number of tasks solved Young 0.22 10.9 PEU 0.52 0.04 2.27 Number of tasks solved Old 0.41 4.5 PEU 0.67 0.00 3.38 Table options 3.5.2.2. Efficiency (time on task) Focusing on time on task as predictor variable, the regression model explains 71% of variance (α = 0.000, Table 10). The strongest predictor is age, indicating that older participants need considerably more time to work on the experimental tasks. Again, ease of use is a highly significant predictor of the performance variable. Participants with high scores on the PEU-scale are faster in the completion of the experimental tasks. For the whole group, the STC also has explanatory power, which means that participants with a high subjective technical confidence need less time to process the tasks. The integration of the STC into the regression model causes a rather high Condition Index (17.3), which indicates moderate collinearity. Without including the SCT into the regression model, the adjusted R2 is slightly reduced to 65% with an Condition Index of 9.6. Therefore, the results of this regression model (including the STC) have to be interpreted cautiously. Table 10. Regression analysis on time on task (n = 32) Dependent variable Adj. R2 Condition index Predictor β p t-Value TOT 0.71 17.3 Age group 0.44 0.00 3.88 PEU −0.35 0.00 −3.07 STC −0.29 0.02 −2.57 Table options The age group-specific testwise regression analysis did not contribute additional information to the prediction of the variable time on task. As age is the main predictor in the regression analysis, the exclusion of age reduces the overall adjusted R2 to R2 = 39% and R2 = 32%, respectively. For both age groups, the perceived ease of use is the main predictor of time on task ( Table 11). Table 11. Testwise regression analysis on performance (time on task) for young adults (n = 16) and older adults (n = 16) Dependent variable Age group Adj. R2 Condition index Predictor β p t-Value TOT Young 0.39 10.8 PEU −0.66 0.01 −3.28 TOT Old 0.32 4.5 PEU −0.61 0.01 −2.86 Table options Concluding the results regarding H4, the STC does not act as a predictor of performance. Thus, performance outcomes are not influenced by participants’ subjective technical confidence. In contrary, age and the perceived ease of use were found to be the main predictors of performance. 3.5.3. Prediction of ease of use (PEU) In order to examine H5, which declares the positive influence of technical confidence and performance on the perceived ease of use, a stepwise regression with the variables age, gender, performance (number of tasks solved, time on task), STC and perceived usefulness was conducted. The regression model was significant at α = 0.000. Adjusted R2 values suggest that task effectiveness and perceived usefulness can predict 70% of the variance in PEU. Obviously, participants correspond sensitively to their performance outcomes and the perceived usefulness when they define the ease of use of handling the device. Contrary to expectations, subjective beliefs like the STC do not affect the ease of use ( Table 12). Table 12. Regression analysis on PEU (n = 32) Dependent variable Adj. R2 Condition index Predictor β p t-Value PEU 0.70 5.6 Number of tasks solved 0.56 0.00 5.28 PUF 0.47 0.00 4.45 Table options Testwise regressions show age-specific differences in the prediction of the perceived ease of use (Table 13). For younger adults, the task effectiveness does not predict ease of use, but rather task efficiency (time on task). Apparently, younger adults do not include their success in solving the PDA tasks (effectiveness) into their judgment of ease of use. Possibly, due to their overall good performance when working with the PDA, the effectiveness may not be relevant as an indicator. They rather rely on time on task when assessing the ease of use. In contrast, for older adults, task effectiveness plays a major role in the prediction of ease of use. Apart from task effectiveness, the main factor predicting ease of use in older adults is the perceived usefulness of the device. Probably, older adults integrate general attitudes towards the device into their formulation of ease of use. This can also be observed in the regression model for younger adults, but the perceived usefulness of the device is a much weaker predictor of the perceived ease of use for younger than for older adults. Table 13. Testwise regression analysis on perceived ease of use (PEU) for young adults (n = 16) and older adults (n = 16) Dependent variable Age group Adj. R2 Condition index Predictor β p t-Value PEU Young 0.54 8.7 TOT −0.55 0.01 −3.00 PUF 0.42 0.04 2.33 PEU Old 0.72 5.1 PUF 0.56 0.00 3.84 Number of tasks solved 0.52 0.00 3.58 Table options As the STC was excluded in the stepwise regression model, which predicted the perceived ease of use, we wanted to find out how much variance of the perceived ease of use is explained by the STC alone. Therefore a regression with the STC as the sole predictor of perceived ease of use was run. Only 18.9% of variance in the perceived ease of use is explained by the STC. The picture changes however, when testwise regressions are carried out for both age groups, separately. In testwise regressions the STC was found to have a different effect on perceived ease of use depending on age. For younger adults, the STC explains 27.7% of variance, the regression model is significant at α = 0.021. For older adults the overall R value (R2 = −.16) does not even suggest a linear relationship between STC and perceived ease of use. 3.5.4. Prediction of perceived usefulness (PUF) In order to confirm H6, which states an association between ease of use and usefulness, a multiple stepwise regression with the independent variables age, gender, STC, performance measures (number of tasks solved, time on task) and the perceived ease of use was carried out. According to expectations, the perceived ease of use was the main predictor of perceived usefulness, with an adjusted R2 = 42% ( Table 14). The same results were found in age-specific testwise regressions ( Table 15). In the group of young adults and – to a stronger extent – in the group of the older adults, perceived usefulness was a significant predictor of perceived usefulness. The comparably low amount of explained variance for the group of younger adults indicates, that they might integrate further variables in the formulation of the perceived usefulness of a PDA, beyond the ease of using the device. Table 14. Regression analysis on PUF (n = 32) Dependent variable Adj. R2 Condition index Predictor β p t-Value PUF 0.42 5.8 PEU 0.66 0.00 4.88 Table options Table 15. Testwise regression analysis on perceived usefulness (PUF) for young adults (n = 16) and older adults (n = 16) Dependent variable Age group Adj. R2 Condition index Predictor β p t-Value PUF Young 0.28 10.9 PEU 0.57 0.02 2.60 PUF Old 0.45 4.5 PEU 0.70 0.00 3.65