Which of the “big five” personality dimensions refers to being talkative and energetic?

Big Five Personality Traits

As stated earlier, MBTI has attracted criticism in recent years. The proposed alternative by most critics is the big five personality traits. The five-factor model (FFM), which the big five personality traits is based on, was developed by several researchers throughout the past decades, including Norman (1967), Smith (1967), Goldberg (1981), and McCrae and Costa (1987).11 The key strength claimed by the FFM model is that it is based on empirical research that shows consistency across time, culture, and age groups. It is also considered more structured because the five traits do not overlap. At a high level, the traits are (Figure 8.2):

Openness to experience: People with a strong tendency in this trait are considered to be imaginative and creative. They are willing to try new things and are open to ideas.

Conscientiousness: People with a strong tendency in this trait are considered to be goal focused and organized and have self-discipline. They follow rules and plan their actions.

Extraversion: People with a strong tendency in this trait are considered to be outgoing and energetic. They obtain their energy from the company of other people and are defined as being assertive and enthusiastic.

Agreeableness: People with a strong tendency in this trait are considered to be compassionate, kind, and trustworthy. They value getting along with other people and are tolerant.

Neuroticism: People with a strong tendency in this trait are considered to be anxious, self-conscious, impulsive, and pessimistic. They experience negative emotions relatively easily.

There also have been studies that investigate applying the FFM to software engineering teams.12,13 These studies focus on multiple roles within a software team and not particularly on software architects. However, they do look at key traits for software designers. Based on their analysis, the key trait that has to be strong for software architects is agreeableness. This result supports the Continuous Architecture view that at least 50% of the role of an architect is to focus on communication. (We discuss this in detail in Chapter 9.) The other traits that we think are important for architects are openness to experience and conscientiousness.

It is interesting to note that most of these studies do not explicitly define the role of an architect but refer to the role of a software designer. We are not particularly concerned about the difference between what we would call a solution architect and a software designer. The main differences are in the scale they are operating in. The role of a software designer can be said to happen at a lower level of granularity than a solution architect. Regardless, both are involved in making architecture or design decisions related to a software product. Because we have already stated that Continuous Architecture also applies to the dimension of scale, then both roles should reflect similar capabilities and responsibilities.

Narcissism and Twitter

While narcissism is associated with social media usage in various genres as noted above, it is especially associated with Twitter usage. This section will review various findings. It is too bad that we cannot tweet these results as we are writing this for more narcissistic coverage. McKinney, Kelley, and Duran (2012) argue that Twitter is a good venue for narcissists because it allows them to answer the question, “What are you doing?” in terms of 140 characters or less. Followers are supposedly interested in one’s moment-to-moment postings, which suggests egocentrism, self-aggrandizement, and self-importance—the very characteristics of narcissistic individuals. Their study revealed that being open about sharing information about oneself was significantly related to the frequency of using Facebook and Twitter to provide self-focused updates, while high scores on narcissism were associated with a larger number of Facebook friends and with the number of self-focused “tweets” that people send. In addition, posting selfies on social media is another reflection of narcissism (Murray, 2015).

The Big Five personality traits are stable, primordial personality traits that consist of neuroticism, openness, conscientiousness, extraversion, and agreeableness (Cobb-Clark & Schurer, 2012; Honeycutt et al., 2013; McCrae & Terracciano, 2005). Openness reflects the degree of intellectual curiosity, creativity, and preference for novelty and variety. Conscientiousness is the predisposition to show self-discipline and refers to planning, organization, and dependability. Extraversion reflects the need to seek stimulation in the company of others, sociability, and talkativeness. Agreeableness is the tendency to be compassionate and cooperative towards others. Finally, neuroticism reveals the tendency to experience negative emotions such as anger, anxiety, depression, or vulnerability. Neuroticism reflects emotional stability and control of impulses.

McCain and Campbell (2016) summarize a few findings on the Big Five traits and social media usage. They indicate how the traits associated with narcissism reflect a trait model of narcissism as opposed to a state or situational model in which people are narcissistic in some platforms and less narcissistic in others. In Big Five terms, grandiose narcissism is associated with high levels of extraversion and openness and low levels of agreeableness (Miller et al., 2011). Extraverts have larger social networks in general (Pollet, Roberts, & Dunbar, 2011; Roberts, Wilson, Fedurek, & Dunbar, 2008) and spend more time and generate more content on social media sites (Gosling, Augustine, Vazire, Holtzman, & Gaddis, 2011). Thus narcissists’ tendency to have more friends and generate more content on social media may be associated with their extraversion. Conversely, vulnerable narcissism is associated with low agreeableness and neuroticism. These findings suggest that anxiety is associated with increased social media usage.

Qiu, Lin, Ramsay, and Yang (2012) measured the “Big Five” personality traits of openness, conscientiousness, extraversion, agreeableness, and neuroticism among 142 Twitter users. They analyzed their participants’ tweets over a month-long period and used a software program called Linguistic Inquiry and Word Count to look for patterns in the language they used. They found that extraverts used more assent words, fewer functional words, and fewer impersonal pronouns. Openness was negatively related to the use of adverbs, swear words, affect words, and nonfluent words, but positively related to prepositions. When Qiu and his colleagues asked those who had never met the Twitter users to judge their personalities based only on their Twitter feeds, they found that people could accurately judge two of the Big Five dimensions—neuroticism and agreeableness.

Path Analysis for Twin Research

Behavior genetics uses quantitative genetic analysis as a tool. Quantitative genetic analysis is the theoretical basis for the statistical analysis of variation in populations. The statistical tool is more commonly known under the names “structural equation modeling” or “path analysis.” Model estimation has many advantages over the classic approach. The graphical reproduction of structural models is very helpful for making the assumptions of the twin design more explicit (see Fig. 1). Using path models makes it possible to test different models against each other and to opt for the model that best fits the data. The fit of the model can be expressed in goodness-of-fit statistics and the estimates for genetic and environmental effects are given together with their standard errors.

Many different complex and multivariate models are already commonly used, but here quantitative genetic analysis using the most basic univariate model for genetic analysis is illustrated. This model can easily be extended to multivariate and longitudinal designs as well as designs that go beyond the twin design and investigate the more complex genetic and environmental relationships between different relatives.

Figure 1 represents the simplest path model for MZ and DZ twins reared together. The genetic theory as outlined above is reflected in the model. By convention, the observed or dependent variables are drawn as rectangles and the latent, independent variables are shown as circles. Single-headed arrows are used to define causal relations or paths and double-headed arrows are used to define covariances. Also by convention, uppercase letters are used to define the latent variables and lowercase letters are used to represent the paths and double-headed arrows. An example from research on extraversion is used to illustrate the model. Extraversion is one of the Big Five personality traits. It is associated with active, impulsive, and social behavior, where people who exhibit high levels of these behaviors are called extraverts and people who exhibit low levels of these behaviors are called introverts. The variables in the squares are the observed levels of extraversion for twin 1 and co-twin 2. The latent variables (circles) come from behavior genetic theory. E stands for the unique environment and by definition random error is incorporated in E; A stands for the additive genetic effects, D stands for the effects of dominant genes and epistasis, and C stands for the effects of the common environment. The covariation between both genetic effects is defined for MZ as well as for DZ twins; where MZ twins show a correlation of 1, they are of the same genotype and DZ twins show a correlation of 0.5 for additive genetic effects and 0.25 for dominance effects.

Since quantitative genetic analysis has a strong regression component, the model can be also defined from its underlying regression structures

P1=eE1+aA1+dD1+cC1,

and

P2=eE2+a A2+dD2+cC2,

where P1 and P2 are the phenotypes of two co-twins. As can be derived from the regression equation, the phenotype is assumed to be a linear function of the underlying genetic and environmental effects. The total variance of the observed measure is composed from the factor loadings as:

Vp=a2+d2+c2+e2.

In the classical twin study that uses twins that are reared together, C and D cannot be modeled in one analysis, because then they become confounded and the model cannot be identified. Using this path diagram, different models can be tested. First, the simplest model that takes only the unique environmental and additive genetic effects into account is examined. In the second step, this model can be extended with common environmental or dominance effects, depending on the difference between the intraclass correlations of the MZ and DZ twins in the sample. When the intraclass correlation of the MZ twins is less than twice the intraclass correlation of the DZ twins, a model that allows for common environmental effects is chosen, because the DZ twins resemble each other more than could be expected on the basis of their genotype alone. When the intraclass correlation of the MZ twins is larger than twice the intraclass correlation of the DZ twins, a model that allows for dominance effects is chosen, because the DZ twins differ more than could be expected from theory. Finally, a model that excludes every genetic effect, the CE model that states that all individual differences are attributable to environmental effects, can be chosen.

Using these three models, the way that models are compared and tested is discussed here. In Table I, the path estimates, as well as the goodness-of-fit measure (here, the normed fit index), are given. In the study here examined, the intraclass correlation for MZ twin pairs was 0.5 and for DZ twin pairs it was 0.39, which could be an indication of common environmental effects.

Table I. Results of a Univariate Quantitative Genetic Analysis on Extraversion

Note: AE, model includes additive genetic and unique environmental effects. ACE, model includes additive genetic and unique and common environmental effects. CE, model includes only environmental effects. VA, phenotypic variance explained by genetic effects. VE, phenotypic variance explained by unique environmental effects. VC, phenotypic variance explained by common environmental effects. NFI, normed fit index for large samples. AIC, Akiaki's information criterion.

These results show that a model that incorporates only additive genetic and unique environmental effects has a nice fit (χ2/df < 1; NFI = 0.98). From the data, it can be seen that 47% of the variance in extraversion levels in the population can be attributed to additive genetic effects and 53% of the total variance can be attributed to unique environmental effects. But is this the best-fitting model or has the shared environment also had a significant effect on the development of the trait? To address this question, a model that includes additive genetic effects, unique environmental effects, and common environmental effects needs to be tested. To accomplish this, one degree of freedom must be sacrificed. This model also fits (χ2/df < 1; NFI = 0.98), but does it fit significantly better compared to the more parsimonious AE model? When models are nested, the likelihood ratio test is used to solve this question. As goodness-of-fit measure, the χ2 values are obtained. The difference between χ2 values can be interpreted as a measure for the significance of path C, χ2AE − χ2ACE = 0.681, which is not large enough to compensate for the loss of one degree of freedom. Thus, the model including common environmental effects does not explain the data significantly better than the first model with only additive and unique effects.

The next model tested does not assume any genetic effects on the trait, only environmental effects. This model also seems to fit the data (χ2/df < 1). Because this model (CE) is not nested in the first model (AE), it is not possible to compare both models using a likelihood ratio test. When models are not nested, the model with the smallest Akaiki's information criterion (AIC) is considered the best model, because AIC is a measure of the parsimoniousness of a model. Therefore, it can be concluded that a model with only additive genetic and unique environmental effects suffices to explain the data.

3.3 Associations between Problematic Facebook Use, personality and self-esteem

Results about the Big Five personality traits showed that the traits more clearly correlated with problematic Facebook use were neuroticism and conscientiousness with opposite signs: a mean positive correlation of r = .22, 95% CI [.19, .26], k = 16, Z = 10.96, p < .001 for neuroticism, and a mean negative correlation of r = −.16, 95% CI [-.21, −.09], k = 15, Z = −4.82, p < .001 for conscientiousness. The other three traits were negatively, but only mildly associated with problematic Facebook use (for detailed results see Table 2). Moderator analyses showed that the mean age of the sample significantly moderated the link between problematic Facebook use and openness (β = .017, p < .001), indicating that this effect was larger for older samples.

Regarding self-esteem, as expected the effect was negative, r = −.23, 95% CI [-.28, −.18], k = 8, Z = −8.39, p < .001. Significant heterogeneity emerged also for this variable (Q(7) = 15.65, p < .05) but none of the moderators were significant.

3.1 Big-Five personality traits recognition

Modern psychologists and computer scientists are in a constant pursuit of automatic personality recognition and classification and to this end, they utilise all available information sources. For example, researchers in [107] analysed audio from meetings in order to classify participants according to the Big Five personality traits model. Quite recently, mobile sensing technology has been also employed in order to investigate similar phenomena. According to a study for the relationships between personality and mobile phone use [108], it was found that personality traits can explain patterns of mobile phone usage, for instance extraverts and perhaps disagreeable individuals were less likely to value incoming calls, while disagreeable extraverts also reported using mobile phones more and spent more time adjusting ringtones and wallpapers. Vice versa, the process of inferring personality traits from patterns of smartphone usage is the purpose of several research works in mobile affective computing.

One such work is described in [109] where the authors analyse the relationship between smartphone usage and personality traits. The use of web, music, video, maps and other applications together with the traditional call and SMS usage, the proximity information derived from Bluetooth and the use of camera, are employed to demonstrate that aggregated features obtained from smartphone usage data can be indicators of the Big-Five personality traits. Additionally, TIPI questionnaire [81] is used to measure self-perceived personality and assess inference results which were found rather promising. The research team extended its work in [103] both in terms of experiment’s population and duration as well as in terms of the experimental framework which was enhanced in order to anonymise sensitive information extracted from usage logs and phone sensors.

Similarly, in [110] the use of anonymised mobile phone call usage data to automatically infer users’ personality (as characterised by the Big Five model) in a privacy-preserving manner is described. The necessary information is extracted from call detail records, including variables obtained from social network analysis of the calls, and a model selection applied to come up with the selected features for predicting users’ personality. The chosen features included the duration of received phone calls, the number of received and placed phone calls as well as the number of SMSs and MMSs sent or received at different times of the day. Participants filled in a 50-item version of the IPIP public domain Big Five questionnaire [84] to collect ground truth about participants’ personality profile. Staiano et al. [111] carried on the work further by extending the social network structural properties for the purpose of inferring and predicting personality and other psychological variables through contextual data collected from mobile phones. The data used in the study are proximity data derived from Bluetooth sensor, calls logs from which a social network is built and data coming from surveys where participants enter self-reported information about personality (Big Five) and relationships among subjects.

To meet a similar goal for predicting users Big-Five personalities, and thus emotional stability, in [112] researchers demonstrated that user personality can be reliably inferred from basic information accessible from all commodity smartphones. The features used in the research fall under 5 broad categories: basic phone use (e.g., number of calls, number of texts), active user behaviours (e.g., number of call initiated, time to answer a text), location (radius of gyration, number of places from which calls have been made), regularity (e.g., temporal calling routine, call and text inter-time) and diversity (call entropy, number of interactions by number of contacts ratio). Participants completed the Big Five Inventory (BFI-44) [75] for measuring their personality and the classification reached up to 61% accuracy on a three-class problem on each of the five personality dimensions.

4.3 Results and discussion

The results will be evaluated using accuracy metric as the key performance indicator, which is the official metric used for big five personality trait prediction in the literature as a binary classification task. Accuracy is the classification percentage that is right in all data classification.

The formula for quantifying binary accuracy is:

(4)Accuracy=TP+TF TP+TF+FP+FN

where: TP = True positive; FP = False positive; TN = True negative; FN = False negative

5.2 What are the major factors to be considered when developing a serious game?

Game Design factors are incorporated into the games by the developers, and game play factors are related to the feelings and experience felt by the players when playing the game [81]. Among the 25 studies included in this contribution, 20 studies provided at least one design factor to be considered when designing a game, 2 studies [36,62] suggested research guidelines for future work, and 6 studies reported neither any design factors nor any research guidelines. From the reviewed studies, we can find following basic design factors: