Keywords

1 Introduction and Theoretical Background

It is well-known that personal characteristics like experience, expertise, but also self-confidence and attitudes towards computer technology play an important role in how people interact with computers. In this regard, cognitive theories are promising to better understand user behavior and to design computer technology that better fits the user needs.

This paper is devoted to two well-known cognitive theories, namely self-concept and attribution theory, and investigates their connections in the field of human-computer interaction (HCI).

The self-concept is defined as the sum of an individual’s self-related beliefs, attitudes, and expectations [1]. These perceptions refer to a number of areas, for example, intellectual status, physical and physical appearance, personality traits, and emotional tendencies. Theorists consider the self-concept as generated from lifelong experiences, which are developed by individuals’ interactions with their surroundings [2].

Attribution theory, in turn, strives to explain how causal explanations influence individuals’ motivation, emotions, and behavior [3]. Attribution theorists assume that people try to understand themselves and the world around them in order to achieve cognitive mastery of their environment’s causal structure [4].

In prior research, these two theories have been independently investigated in the field of computer-related self-perceptions and both provided valuable insights into the understanding of learning and behavior of computer users. However, the direct link between these two theories has not been investigated yet, especially in the field of HCI, which is the contribution of the current study. The goal is to shed more light on computer-related attitudes, behaviors, and cognitive characteristics of computer users.

This paper is structured as follows: Sect. 1 introduces the two constructs of self-concept and attribution theory in general as well as with regard to HCI, followed by a brief description of the connections between these theories. Section 2 provides a description of our research questions and methodology. The description of the analysis and the results obtained in this study are outlined in Sect. 3. Finally, we conclude with a discussion of implications for practice and research, limitations, as well as an outlook for future research.

1.1 Self-concept

The self-concept, in general, denotes the total of all cognitive representations which an individual has stored in his/her memory [5,6,7]. These self-referred representations relate to different areas, for example, one’s own intellectual abilities, aspects of one’s body, or one’s experiences and behavior. Hence, the self-concept is multifaceted and can be subdivided, for example, into an academic, a social, an emotional and a physical self-concept, which in turn can be differentiated into further sub-areas [8].

In the field of social sciences, the self-concept is one of the best-explored constructs (Hattie 1992) and many theories about the “self” have been proposed in the last 100 years. In this regard, an early distinction has been made between two global aspects of the self: the self-knowing self (self as knower, I, pure ego) and the recognized self (self as known, me, empirical ego). Moreover, in describing the self, a distinction can be made between the material self (body, family, and possessions), the social self (views others hold of the individual), and the spiritual self (emotions) [9]. Another example is the looking-glass self, which describes the self-concept as an developing result of the perceived impressions and evaluations in social interaction [10].

There are two main theoretical approaches in self-concept research: the unifactorial construct and the multifaceted hierarchical construct.

Representatives of the unifactorial approach hold the opinion that the self-concept is dominated by a single general or global self-concept and that individual or specific factors can not be adequately differentiated [e.g., 11, 12]. For example, they claim that children are too immature to make distinctions between different facets of self-concept.

The proponents of the multifaceted hierarchical construct claim that the global self-concept is built up from a number of different facets. The global self-concept is at the top of the hierarchical tree, which then splits into smaller components such as academic, social, emotional and physical self-concepts [e.g., 8, 13].

Hansford and Hattie [14] examined the correlations between different self-concept variables and performance measures. They found that the correlation became significantly larger when they merely considered studies in which only the academic self-concept was measured. Other researchers discovered that school performance is highly correlated with the domain-specific self-concept, but is only moderately related to the general self-concept [ 15 ]. These findings illustrate that the self-concept is domain-specific and thus today’s self-concept research assumes that self-related cognitions mainly refer to specific sub-areas. For this reason, recent research mostly does not refer to a general self-concept but is narrowed to domain-specific self-concepts. This study is conducted within the multi-faceted hierarchical paradigm and focuses on assessing individuals’ levels of specific computer-related self-concept.

Computer-Related Self-concept.

The computer-related self-concept is a quite new psychological construct that is especially related to computer-related experiences, interests, motivations, attitudes, and competencies [16]. The construct is based on the Three-Components-Model of Attitudes [17] and comprises:

  • The conative component representing the concrete actions, behaviors, or specific experiences with regard to dealing with computers (experiences in childhood, youth, and adulthood);

  • The motivational component describing emotional and content-specific motifs for using computers (fun, joy, fascination, computer anxiety, and individual motives in using computers);

  • The cognitive component including the subjectively perceived competence and self-efficacy regarding the handling of computers, attribution processes as well as strategies for dealing with information technology (computer-related self-perception of competencies, self-efficacy, attribution processes, and strategies).

1.2 Attribution Theory

Attribution theory originates in the 1950s and was developed when psychologists began to be more interested in studying the human’s perception of causes for their behavior rather than their perception of the behavior itself. Attributions are subjective causal explanations for successful and unsuccessful outcomes that are known to influence individuals’ behavior, motivation, and emotions [18].

Heider [4] was one of the first who investigated how people try to understand the causes of their own actions. He argued that individuals are trying to analyze, as a kind of ‘naive scientist’, the causes of an action. According to Heider, the cause of an action is attributed either to internal or external factors. For example, a person may either feel responsible for a positive or negative outcome (internal) or relate it to external circumstances [4]. This first dimension is called the locus (or locus of control) dimension and served as the basis for subsequent attribution theories as well as further attributional dimensions [19].

The subsequent theory that we built upon in this paper is called the achievement-related attribution theory. It states that individuals generally attribute their failure and success in performance-related situations to one of four attributional patterns. These attributional patterns are ability, effort, task difficulty, and luck [18]. Ability is defined as the knowledge and abilities a person believes to have to perform a task. Effort is defined as the physical and mental energy a person exercises in performing a task. Task difficulty is defined as how easy or difficult a task is perceived. Finally, luck is defined as that role that opportunity plays in performing a task [18]. These attributions, in turn, can be determined by means of the four attributional dimensions of locus, stability, controllability, and globality [20]. Table 1 illustrates the relationships.

Table 1. Attributional classification scheme: relations between attributional dimensions and attributional patterns for failure and success situations [20].
Full size table

Ability and effort are considered as internal attributions because they relate directly to the person who makes the attribution. On the contrary, task difficulty and luck are regarded as external attributions because the person ascribes the cause to external circumstances.

The second dimension to classify attributions is the stability dimension. In this regard, causes are considered as stable over time (recurring) or as unstable (singular event) [18]. Ability and task difficulty are regarded as stable attributions. Conversely, effort and luck are considered as unstable attributions.

Two more dimensions are used to describe individuals’ attributions: The controllability dimension describes whether a cause is perceived as controllable (easy to change) or as uncontrollable (hard to change or even unchangeable) [18].

Finally, the globality dimension distinguishes between causes perceived as global (generally valid) or only valid for specific (certain) situations [21].

Computer-Related Attribution Theory.

The investigation of attributions is fairly young in the field of HCI. However, distinct attributional patterns with favorable as well as unfavorable attribution styles were observed by Niels and Janneck [22] in the information systems context. Moreover, demographic factors (e.g., age, gender, self-assessed computer skills) [23], as well as the impact of different attribution patterns on the assessment of computer systems were examined [24]. However, to our knowledge, relations between computer-related self-concept and causal attributions have not been investigated yet.

1.3 Connection Between Self-concept and Attribution Theory

In other research domains, particularly in the educational context, research on attribution theory and self-concept has already spawned important implications for research and practice. Attribution theory allowed researchers to gain insights into what causes students identify as the reason for their performance. They also found that these perceived causes are crucial because they influence students’ future expectations, motivation, and subsequent behavior [25]. Furthermore, the self-concept also affects learning behavior: Those with a positive self-concept performed better than those with a negative self-concept [2].

First connections between self-concept and attributions were discovered rather by chance when spontaneous remarks by participants were additionally recorded in a self-concept study [26]. It turned out that those participants with a positive self-concept externalized causes for insoluble tasks while participants with negative self-concept rather attributed their failure to internal causes. Subsequent research was able to replicate these results [e.g., 18, 27,28,29,30]. This connection was then further investigated by other researchers [e.g., 31,32,33] and especially by Weiner [18]. Weiner found that individuals with a positive self-concept tend to attribute success to their own ability and individuals with a negative self-concept tend to attribute failure to unstable causes such as effort or luck. This was also found in an educational context. For example, students with a positive self-concept were more likely to attribute failures to external-unstable factors (luck), while students with a negative self-concept tend to attribute failures to internal-stable factors (ability). In situations of success, the opposite is the case. In summary, more correlations in situations of failure than in situations of success were found [34,35,36,37,38].

In this regard, a number of studies focused on the attributional dimensions (i.e., locus, stability, controllability, globality), but there is also a series of studies that considered the four causal attributional patterns (ability, effort, task difficulty, and luck). However, both methods yield very similar results.

In contrast to the other components of the self-concept, attributions are considered as influenceable, for example, through measures such as the so-called reattribution training. The aim of the training is to transform unfavorable attribution patterns into favorable attribution patterns. This transformation, in turn, has a positive effect on the self-concept [39,40,41].

2 Research Questions and Methodology

So far, research has shown a fairly consistent pattern: individuals with a positive self-concept mainly attribute success to internal and failures to external causes. For individuals with a negative self-concept, the opposite is the case. However, most of these prior studies were conducted with children or students in educational context. Thus, in order to draw conclusions in the field of HCI, it is important to examine these relations in this specific context and with “average” computer users since the self-concept [42] as well as attributions [43] are deemed to be domain-specific.

The present study differs from previous studies by exploring relationships between self-concept and causal attributions by focusing on computer-related situations; in particular, regarding computer-related failure and success situations experienced in “real-life”. The aim is to shed more light on computer-related attitudes, behaviors, and cognitive characteristics of computer users. For this purpose, we investigated causal attributions as well as the computer-related self-concept in a laboratory setting by conducting usability tests. We investigated correlations between the attributional dimensions (locus, stability, controllability, and globality) as well as the attributional patterns (ability, effort, task difficulty, and luck) and the computer-related self-concept. This is one of the first studies combining these two cognitive theories with regard to computer use, thus making a unique contribution to the body of knowledge in the field of HCI.

2.1 Measurements

Next to the assessment of general demographic aspects (age, gender, education, general computer use and skills), two validated questionnaires were used to cover the aspects of computer-related self-concept and computer-related causal attributions.

Computer-Related Self-concept Questionnaire (CSC).

The standardized Computer-Related Self-concept Questionnaire developed by Janneck, Vincent-Höper and Ehrhardt [16] was used to measure the participants’ level of computer-related self-concept. The measurement focuses on the individual’s self-perceptions rather than attempting to infer their self-concept by observing the behavior or the attributions of others. The questionnaire consists of 11 subscales with a total of 27 items, col-lecting conative, motivational, and cognitive computer-related self-ratings using a five-point Likert scale (“1 = strongly disagree” to “5 = strongly agree”). The Computer-Related Self-Concept Questionnaire has proven to have a satisfactory reliability and validity [16]. Table 2 shows the English version of the questionnaire. This questionnaire was chosen because, to the best of our knowledge, it is the only questionnaire that relates to computer-related situations. Moreover, the questionnaire is simple to evaluate and provides quantitative values that allow easy comparison.

Table 2. The Computer-Related Self-Concept Questionnaire [16].
Full size table

Computer-Related Attribution Questionnaire.

The standardized Attribution Questionnaire developed by Guczka and Janneck [44] was used to measure computer-related attributions. The questionnaire distinguishes between failure and success situations, as usually done in attribution research, and contains four items to measure the attributional dimensions of locus, stability, controllability and globality. The questionnaire is based on the Sport Attributional Style Scale, SASS [45]. Table 3 shows the English version of the questionnaire relating to situations of failure. Items measuring attributions of success are worded analogously. This questionnaire was chosen because it allows examining the attributional dimensions separately, but also to determine the attributional patterns.

Table 3. Excerpt from the Attribution Questionnaire for failure situations [44].
Full size table

Overall, evaluative statements made by individuals about themselves are considered to be valid and reliable data sources and self-reports provide valuable information about the individual [e.g., 46,47,48].

2.2 Procedure

Data was collected in a laboratory setting, conducting usability tests. First, the participants filled out a questionnaire containing the demographic questions as mentioned above, as well as the Computer-Related Self-Concept Questionnaire (Table 2).

Afterwards, the participants were asked to edit three task pairings on three different applications or devices, whereby one task of each pairing was easy to solve (situation of success – e.g.: Search for the district office opening hours on a municipal home page) and one task was hard or even unsolvable (situation of failure – e.g.: Search for the building regulations of a certain district, which did not exist on the homepage). The tests included two tasks with a web application (website), two tasks with a desktop application (spreadsheet program), and two tasks with a mobile application (nutrition application).

To slightly increase the pressure, the participants were told that they have five minutes only to solve each task. In fact, the timeframe was not strictly adhered to. If necessary the participants had little more time to solve the success tasks and less time to solve the failure tasks. However, the processing time varied one minute at maximum.

Before the participants started with the first task, they were explained with the aid of an example how to answer the questions that would follow the tasks. Then the participants started working on the tasks. After each task, the participants completed the Computer-Related Attribution Questionnaire (Table 3).

In all, a total of 340 situations (163 failure and 177 success situations) were recorded. This imbalance was due to individual perceptions of the outcome of the task: For example, some participants also succeeded in the hard task condition, while others were not successful in the easy task condition. Moreover, the participants decided by themselves whether they successfully solved a task or not. For example, some participants failed objectively but nevertheless believed they had been successful. On average, the participants needed one hour to complete the test. They were not paid for their participation.

2.3 Sample

In order to obtain a balanced sample, the participants were selected according to the following criteria: Approximately the same number of female and male persons, about one-third of them aged between 14 and 25 years, one-third between 26 and 45 years and one-third aged 46 years and above.

In all, 64 persons participated in the study (46,9% female, 53,1% male). Mean age was 38.20 years (Median = 31, SD = 16.97, range: 17–75 years). The general level of education was quite high (75% with a high school or university degree). Participants were rather experienced computer users. On average they had 13.88 years (Median = 14, SD = 7.17, range: 3–32 years) of experience in computer use and they used computers on average 5.77 h a day (Median = 6, range: 3–14 h). Participants self-rated their computer skills on a Likert scale ranging from 1 (low) to 7 (expert) on average at 4.27 (Median = 4.33, SD = 1.61, range: 1–7).

3 Data Analysis and Results

In order to examine the relations between computer-related self-concept and computer-related causal attributions, we first analyzed the computer-related self-concept mean values for each subscale as well as the mean value of the total scale. Secondly, we analyzed the attribution questionnaire and calculated the means for each attributional dimension (locus, stability, controllability, and globality), separately for situations of failure and success. Furthermore, the attributional patterns (ability, effort, task difficulty, and luck) were determined based on Weiner’s attribution matrix [3].

Finally, correlations between self-concept scales and attributional dimensions as well as correlations between self-concept scales and attributional patterns were calculated. In the following sections, methods and results are explained.

3.1 Computer-Related Self-concept

In a first step, inverse coded item values were transformed, as the order of positive and negative terms is randomized in the questionnaire. Secondly, means (M), standard deviations (SD), and internal consistencies (Cronbach’s α) for the subscales as well as for the overall self-concept (total scale) were calculated (Table 4). Higher values indicate a more positive computer-related self-concept.

Table 4. Results Computer-Related Self-Concept Questionnaire subscales and total scale.
Full size table

In general, participants reported a quite positive computer-related self-concept (M = 3.12). The subscales with the high-est means were computer anxiety (M = 4.11), strategies (M = 3.77), and external control beliefs (M = 3.30). The scales indicating a negative or unfavorable self-concept were tool perspective (M = 2.41), internal attribution (M = 2.67), and creating (M = 2.77). The subscales show acceptable to good reliability coefficients.

3.2 Computer-Related Causal Attributions

In our analysis, we distinguished between situations of failure and success, as it is usually done in attribution research. First, the mean value for each attributional dimension was calculated (Table 5). Results show that the locus dimension is nearly equally distributed. The participants see internal as well as external reasons for their failure and success. Furthermore, the causes are perceived to be stable over time, persist in different situations (global), and are perceived as controllable. Overall, this represents more positive attributional patterns.

Table 5. Results Attribution Questionnaire dimensions for success and failure situations.
Full size table

Secondly, the attributional patterns were determined for each participant separately for failure and success situations, based on Weiner’s attribution matrix [3]. This was done by combining the attributional dimensions of locus and stability and clustering the combinations into the four attributional patterns of ability, effort, task difficulty, and luck. Participants who attributed the reasons for success or failure, respectively, to internal/stable causes were assigned to the attributional pattern ability. Participants who attributed the causes to external/stable reasons were assigned to the task difficulty pattern, participants who attributed internal/unstable to the effort pattern, and participants who attributed the reasons to external/unstable causes to the attributional pattern of luck. The attributional pattern matrix, as well as the number of participants in each pattern, separately for situations of failure and success, is shown in Table 6.

Table 6. Attributional pattern classification matrix based on Weiner’s [3] attribution theory and number of participants for each pattern in situations of failure and success.
Full size table

3.3 Relationship Between Self-concept and Attributional Dimensions

Correlations (Spearmen’s Rho) between the computer-related self-concept total scale as well as the subscales and the attributional dimensions (locus, stability, controllability, and globality) were calculated separately for failure and success (Table 7).

Table 7. Correlations (Spearmen’s Rho) between computer-related self-concept subscales (CSC) and computer-related attributional dimensions in situations of failure and success.
Full size table

Regarding the self-concept total scale, analyses revealed significant correlations for the locus dimension: Users with a positive computer-related self-concept attributed causes for failure (r = 0.550, p = 0.000) and success (r = 0.266, p = 0.034) predominantly to external reasons. The latter is interesting, as this does not correspond to the findings from other domains where individuals with positive self-concept usually attribute success to internal reasons [34,35,36,37,38]. For the other attributional dimensions, fewer correlations were found.

With regard to the self-concept subscales in failure situations, significant positive correlations between the attributional dimension of locus and all self-concept subscales were found. Furthermore, a positive correlation between the self-concept subscale internal attributions and the attributional dimension of controllability was found (r = 0.265, p = 0.034). Individuals who generally tend to attribute failures to external causes are more likely to perceive the cause to be uncontrollable. Moreover, a negative correlation between the self-concept subscale of internal attributions and globality could be established (r = −0.304, p = 0.015). Individuals who attribute failures to external causes believe that the cause only affects a specific situation and is not valid in other computer-related situations.

In success situations, fewer significant correlations were found. First, significant positive correlations between the attitudinal dimension of locus and the self-concept subscales practical experiences (r = 0.328, p = 0.008), positive emotions (r = 0.262, p = 0.036), competencies (r = 0.256, p = 0.041), self-efficacy (r = 0.301, p = 0.016), internal attributions (r = 0.256, p = 0.041) and strategies (r = 0.358, p = 0.004), were found. Hence, individuals who have a lot of practical experience in dealing with computer technology, have fun with computer technology, consider themselves to be competent in handling computer devices, have a high degree of computer self-efficacy, and have positive computer-related learning strategies, rather attribute success to the system (external) than to their own abilities (internal). Furthermore, there is a positive correlation between the self-concept scale creating and the attributional dimension of controllability (r = 0.274, p = 0.029). Accordingly, people who like to use computers as a design tool tend to perceive the cause of success as uncontrollable. In addition, a positive correlation between the self-concept scale strategies and globality emerged (r = 0.273, p = 0.029). People who believe that the success refers to a global cause tend to explore systems more intuitively than persons attributing the cause to a specific situation.

3.4 Relationship Between Self-concept and Attributional Patterns

Attributional patterns and self-concept scores were tested globally for differences followed by post-hoc tests for pairwise comparison. Because of partially non-normally distributed data Kruskal-Wallis test was used instead of analyses of variance.

Regarding situations of failure, tests yielded significant differences for the total self-concept score (p = 0.001) as well as for the subscales practical experiences (p = 0.033), positive emotions (p = 0.024), computer anxiety (p = 0.001), competencies (p = 0.000), self-efficacy (p = 0.001), internal attributions (p = 0.001), external control beliefs (p = 0.043), and strategies (p = 0.005). Merely for understanding, creating, and tool perspective no differences were found. In situations of success, tests showed no significant differences for any of the self-concept scales (Table 8 and Fig. 1).

Table 8. Relations between self-concept total scale as well as subscales (CSC) and attributional patterns in situations of failure and success – results of Kruskal-Wallis tests.
Full size table
Fig. 1.
figure 1

Mean values of computer-related self-concept scale and subscales for each attributional pattern in situations of failure. Note: PX = practical experiences, PE = positive emotions, CA = computer anxiety, UD = understanding, CR = creating, TP = tool perspective, CT = competencies, SE = self-efficacy, IA = internal attributions, EC = external control beliefs, ST = strategies, TS = total scale.

Full size image

Post-hoc tests were calculated to identify the relations between the individual attribution patterns and self-concept scales. In situations of failure, the analysis showed significant differences between the attributional patterns and the self-concept scales (Table 9). Results for situations of success are not reported since prior Kruskal-Wallis tests showed no significant differences.

Table 9. Relations between computer-related self-concept total scale as well as subscales (CSC) and computer-related attributional patterns in situations of failure – Post-hoc test (LSD).
Full size table

In all, individuals that attributed failures to the ability pattern showed the lowest total self-concept score (M = 2.195) while individuals that attributed failures to task difficulty showed the highest score (M = 3.569).

Regarding the self-concept subscales in situations of failure, the following significant differences were found: Individuals who attributed their performance to ability (M = 2.152) scored their practical experience significantly lower than those who attributed their performance to task difficulty (M = 3.361) or luck (M = 3.067). They also showed fewer positive emotions towards computer technology than those who attributed their performance to task difficulty (M = 1.955 vs. M = 3.354).

Moreover, individuals who attributed their performance to ability (M = 2.849) showed higher values of computer anxiety than those who attributed their performance to effort (M = 4.246), task difficulty (M = 4.431) or luck (M = 4.500).

Individuals who attributed their performance to ability (M = 1.667) also considered themselves to have fewer computer competencies than those who attributed their performance to effort (M = 2.526), task difficulty (M = 3.431) or luck (M = 3.200). Furthermore, individuals who attributed their performance to effort (M = 2.526) consider themselves to have fewer computer competencies than those who attributed their performance to task difficulty (M = 3.431).

Individuals who attributed their performance to ability (M = 2.136) showed lower values of self-efficacy than those who attributed their performance to effort (M = 3.118), task difficulty (M = 3.813) or luck (M = 3.100). Moreover, individuals who attributed their performance to effort (M = 3.118) showed lower values than those who attributed their performance to task difficulty (M = 3.813) and those, in turn, showed lower values than individuals who attributed their performance to luck (M = 3.100).

Individuals who attributed their performance to ability (M = 1.864) or effort (M = 2.263) showed lower values of internal attributions than those who attributed their performance to task difficulty (M = 3.125) or luck (M = 3.250). This is theoretically plausible since lower values on this scale refer to internal attributions and this, in turn, corresponds to the attributional pattern of ability and effort. Conversely, high values refer to external attributions, which correspond to the patterns of task difficulty and luck. These findings also support the construct validity of our measurement.

Similarly, individuals who attributed their performance to ability (M = 2.409) showed lower values of external control believe than those who attributed their performance to effort (M = 3.290), task difficulty (M = 3.604) or luck (M = 3.550). They believe to have little control over computer problems that occur due to their abilities.

Finally, individuals who attributed their performance to ability (M = 2.500) scored their computer-related strategies significantly lower than those who attributed their performance to effort (M = 3.658), task difficulty (M = 4.438) or luck (M = 3.750). Moreover, individuals who attributed their performance to effort (M = 3.658) scored significantly lower than those who attributed their performance to task difficulty (M = 4.438).

4 Discussion

This study aimed to examine the relationship between computer-related self-concept and computer-related attributions in computer users. This section discusses the findings of the present study, its limitations, and offers suggestions for future research and practice.

4.1 Relationship Between Self-concept and Causal Attributions

The present study shows that users with more positive or negative computer-related self-concept differ in attributing their computer performance. Results yield significant differences especially with respect to the attributional dimension of locus. In situations of failure, similar results to other research domains were found. Users with a positive self-concept attributed failures to external reasons and tend to blame the system when something goes wrong. However, they also attribute successful outcomes to external factors. This is an interesting finding since this is contrary to the findings of a number of previous studies in other domains [34,35,36,37,38]. Other researchers even found the opposite in a study conducted with students in a classroom environment: Students with a positive self-concept attributed success and failure more to internal causes, while students with negative self-concept attributed success and failure more to external causes [49]. It is likely that the assigned tasks were perceived as too easy and the participants with a positive self-concept had ample experience with the respective applications or devices. On the other hand, participants with a negative self-concept may have less experience and were, therefore, proud to be able to solve the task at all. Therefore, they rather attributed the success to themselves.

This was also reflected in regard to the attributional patterns (ability, effort, task difficulty, and luck) but merely in situations of failure. Users that attribute causes of failures to their own abilities (internal/stable) showed a significantly more negative computer-related self-concept than users that attribute causes of failures to factors such as effort, task difficulty, or luck.

In summary, the present study found that users with positive and negative self-concept partly differ in respect to their attributions for computer-related outcomes. The findings also suggest that the relations regarding computer technology differ from the relations in other research domains.

4.2 Implications and Recommendations

The findings can be used in human-computer research and practice to understand better why users think, feel, or behave in a certain way. Thus, design principles could be developed to support different types of users in a specific way. To our knowledge, this is the first study that combines these two cognitive theories and directly examines the impact of computer-related causal attributions on users’ computer-related self-concept, and vice versa.

Therefore, this study contributes to a more complete and detailed knowledge of users’ computer-behavior. The results encourage further research on the relations between different cognitive theories in the field of HCI.

There are also implications for practitioners who develop and design computer systems. This study sheds light on different types of computer users who show characteristics of self-concept. In order to assist people to develop a more positive self-concept, several measures might be explored. For example, attributional retraining [50], which suggests that individuals’ performance will increase when they learn to ascribe causes to more favorable attributions, could be a promising approach. Thus, our results are valuable for developing and improving existing computer learning training strategies and methods, as well as support and assistance mechanisms for users. Practitioners should attempt to adapt these findings and design specified systems by, for example, including attributional retraining strategies. This could be done, for example, by providing feedback that changes the beliefs of the users about the cause of computer-related outcomes (e.g., comments that contain the desired attributions). System developer and designers should bear this in mind and future research should take this into consideration.

4.3 Limitations and Future Research

The present study also faces some limitations. First, the relatively small sample size, especially regarding some subgroups, limits the generalizability of this study. This is particularly true for the attributional patterns (ability, effort, task difficulty, and luck). Future studies with larger samples would enable to delineate the relations more clearly, especially regarding group comparisons.

Despite the careful selection of the participants, the sample is relatively homogeneous. It consists predominantly of educated and experienced computer users who self-assessed their computer expertise as high. This limits the generalizability of our findings since there is some evidence that socio-demographic factors, like e.g. the educational level, have an impact on attribution processes [23]. Maybe this is also an explanation regarding the few significant differences in situations of success. To investigate the possible influence of the sample characteristics we re-ran our analyses with these factors (level of education, years of computer experience, self-assessed computer skills) as a covariate. However, the analysis showed no differences, as the variance of the values turned out to be too low. Therefore, future research will need to involve a more heterogeneous or representative sample.

The research design of this study also carried certain limitations. Standardized use situations were chosen in this study to create a similar experience for all participants. However, a drawback of this method is that the situations were somewhat artificial and unrelated to the participants’ normal use habits, which might result in reduced intensity and significance of the experience (see [22] for a comparison of different data collection methods).

Furthermore, participants are from Germany only. In regard to possible intercultural differences, future studies should investigate cultural differences by expanding into a more international context.

The results presented here give first insights regarding the relation of computer-related self-concept and computer-related attributions. More research is needed to provide a rich understanding of how and to what extent these factors play a role in HCI research and practice. In this regard, it should be noted that this study has an explorative character. Therefore the results should be interpreted with caution. Nevertheless, this calls for more research to corroborate the findings.

Our next step is to investigate the relations in more detail as well as the effects of reattribution training methods on the computer-related self-concept.