Removing Biases in Communication of Severity Assessments of Intimate Partner Violence: Model Development and Evaluation.

BACKGROUND: To support a victim of violence and establish the correct penalty for the perpetrator, it is crucial to correctly evaluate and communicate the severity of the violence. Recent data have shown these communications to be biased. However, computational language models provide opportunities for automated evaluation of the severity to mitigate the biases. OBJECTIVE: We investigated whether these biases can be removed with computational algorithms trained to measure the severity of violence described. METHODS: In phase 1 (P1), participants (N=71) were instructed to write some text and type 5 keywords describing an event where they experienced physical violence and 1 keyword describing an event where they experienced psychological violence in an intimate partner relationship. They were also asked to rate the severity. In phase 2 (P2), another set of participants (N=40) read the texts and rated them for severity of violence on the same scale as in P1. We also quantified the text data to word embeddings. Machine learning was used to train a model to predict the severity ratings. RESULTS: For physical violence, there was a greater accuracy bias for humans (r2=0.22) compared to the computational model (r2=0.31; t38=-2.37, P=.023). For psychological violence, the accuracy bias was greater for humans (r2=0.058) than for the computational model (r2=0.35; t38=-14.58, P<.001). Participants in P1 experienced psychological violence as more severe (mean 6.46, SD 1.69) than participants rating the same events in P2 (mean 5.84, SD 2.80; t86=-2.22, P=.029<.05), whereas no calibration bias was found for the computational model (t134=1.30, P=.195). However, no calibration bias was found for physical violence for humans between P1 (mean 6.59, SD 1.81) and P2 (mean 7.54, SD 2.62; t86=1.32, P=.19) or for the computational model (t134=0.62, P=.534). There was no difference in the severity ratings between psychological and physical violence in P1. However, the bias (ie, the ratings in P2 minus the ratings in P1) was highly negatively correlated with the severity ratings in P1 (r2=0.29) and in P2 (r2=0.37), whereas the ratings in P1 and P2 were somewhat less correlated (r2=0.11) using the psychological and physical data combined. CONCLUSIONS: The results show that the computational model mitigates accuracy bias and removes calibration biases. These results suggest that computational models can be used for debiasing the severity evaluations of violence. These findings may have application in a legal context, prioritizing resources in society and how violent events are presented in the media.

Latent Semantic Analysis Discriminates Children with Developmental Language Disorder (DLD) from Children with Typical Language Development.

Computer based analyses offer a possibility for objective methods to assess semantic-linguistic quality of narratives at the text level. The aim of the present study is to investigate whether a semantic language impairment index (SELIMI) based on latent semantic analysis (LSA) can discriminate between children with developmental language disorder (DLD) and children with typical language development. Spoken narratives from 54 children with DLD and 54 age matched controls with typical language development were summarized in a semantic representation generated using LSA. A statistical model was trained to discriminate between children with DLD and children with typical language development, given the semantic vector representing each individual child's narrative. The results show that SELIMI could distinguish between children with DLD and children with typical language development significantly better than chance and thus has a potential to complement traditional analyses focussed on form or on the word level.

Transcranial direct current stimulation based on quantitative electroencephalogram combining positive psychotherapy for major depression.

Frontal cortex activity is reduced in the left hemisphere during depression. Transcranial direct current stimulation is a noninvasive neuromodulation technique that can increase frontal cortex activity. Therapy based on transcranial Direct Current Stimulation and positive psychology therapy was applied for improving patients' quality of life. The present study compared three conditions of subjects with clinical depression; (a) transcranial Direct Current Stimulation therapy, (b) positive psychotherapy, and(c) combined treatment. Hamilton Depression Rating Scale, Adult State Hope Scale and Optimism/Pessimism Scale was used at baseline, 2-week, 4-week, and 3-month follow-up. Combined condition participants showed greater reduction in depressed mood, improved hope and optimism after 4-weeks as well as during 3-month follow-up than the other conditions. Results are discussed in terms of additive or synergistic relation between transcranial direct current stimulation and positive psychology treatment.

The role of sparsely distributed representations in familiarity recognition of verbal and olfactory materials.

We present the generalized signal detection theory (GSDT), where familiarity is described by a sparse binomial distribution of binary node activity rather than by normal distribution of familiarity. Items are presented in a distributed representation, where each node receives either noise only, or signal and noise. An old response (i.e., a "yes" response) is made if at least one node receives signal plus noise that is larger than the activation threshold, and item variability is determined by the distribution of activated nodes as the threshold is varied. A distinct representation leads to better performance and a lower ratio of new to old item variability, than a more distributed and less distinct representations. Here we apply the GSDT to empirical data on verbal and olfactory memory and suggest that verbal memory relies on a distinct neural item representation, whereas olfactory memory has a fuzzy neural representation leading to poorer memory and inducing a larger ratio of new to old item variability.

Self-Rated Attentiveness Interacts with Transcranial Direct Current Stimulation and Noise Stimulation in Reaction Time in a Go/No-Go Task.

Previous research has found that stimulating inattentive people with auditory white noise induces enhancement in cognitive performance. This enhancement is believed to occur due to a statistical phenomenon called stochastic resonance, where noise increases the probability of a signal passing the firing threshold in the neural cells. Here we investigate whether people with low attentiveness benefit to a larger extent than attentive people from stimulation by auditory white noise and transcranial direct current stimulation (tDCS). The results show, for both auditory noise and tDCS stimulation, that the changes in performance relative to nonstimulation correlate with the degree of attentiveness in a Go/No-Go task, but not in a N-back task. These results suggest that the benefit of tDCS may interact with inattentiveness.

Quantifying Semantic Linguistic Maturity in Children.

We propose a method to quantify semantic linguistic maturity (SELMA) based on a high dimensional semantic representation of words created from the co-occurrence of words in a large text corpus. The method was applied to oral narratives from 108 children aged 4;0-12;10. By comparing the SELMA measure with maturity ratings made by human raters we found that SELMA predicted the rating of semantic maturity made by human raters over and above the prediction made using a child's age and number of words produced. We conclude that the semantic content of narratives changes in a predictable pattern with children's age and argue that SELMA is a measure quantifying semantic linguistic maturity. The study opens up the possibility of using quantitative measures for studying the development of semantic representation in children's narratives, and emphasizes the importance of word co-occurrences for understanding the development of meaning.

Long-term memory for odors: influences of familiarity and identification across 64 days.

Few studies have investigated long-term odor recognition memory, although some early observations suggested that the forgetting rate of olfactory representations is slower than for other sensory modalities. This study investigated recognition memory across 64 days for high and low familiar odors and faces. Memory was assessed in 83 young participants at 4 occasions; immediate, 4, 16, and 64 days after encoding. The results indicated significant forgetting for odors and faces across the 64 days. The forgetting functions for the 2 modalities were not fundamentally different. Moreover, high familiar odors and faces were better remembered than low familiar ones, indicating an important role of semantic knowledge on recognition proficiency for both modalities. Although odor recognition was significantly better than chance at the 64 days testing, memory for the low familiar odors was relatively poor. Also, the results indicated that odor identification consistency across sessions, irrespective of accuracy, was positively related to successful recognition.

The semantic structure of accuracy in eyewitness testimony.

In two studies, we examined if correct and incorrect statements in eyewitness testimony differed in semantic content. Testimony statements were obtained from participants who watched staged crime films and were interviewed as eyewitnesses. We analyzed the latent semantic representations of these statements using LSA and BERT. Study 1 showed that the semantic space of correct statements differed from incorrect statements; correct statements were more closely related to a dominance semantic representation, whereas incorrect statements were more closely related to a communion semantic representation. Study 2 only partially replicated these findings, but a mega-analysis of the two datasets showed different semantic representations for correct and incorrect statements, with incorrect statements more closely related to representations of communion and abstractness. Given the critical role of eyewitness testimony in the legal context, and the generally low ability of fact-finders to estimate the accuracy of witness statements, our results strongly call for further research on semantic content in correct and incorrect testimony statements.

The ABC of happiness: Validation of the tridimensional model of subjective well-being (affect, cognition, and behavior) using Bifactor Polytomous Multidimensional Item Response Theory.

Background: Happiness is often conceptualized as subjective well-being, which comprises people's evaluations of emotional experiences (i.e., the affective dimension: positive and negative feelings and emotions) and judgements of a self-imposed ideal (i.e., the cognitive dimension: life satisfaction). Recent research has established these two dimensions as primary parts of a higher order factor. However, theoretical, conceptual, and empirical work suggest that people's evaluations of harmony in their life (i.e., the sense of balance and capacity to behave and adapt with both acceptance and flexibility to inter- and intrapersonal circumstances) constitutes a third dimension (i.e., the behavioral dimension). This tridemensional conceptualization of subjective well-being has recently been verified using Unidimensional Item Response Theory (UIRT) and Classical Test Theory (CTT). Here, we use a recently developed and more robust approach that combines these two methods (i.e., Multidimensional Item Response Theory, MIRT) to simultaneously address the complex interactions and multidimensionality behind how people feel, think, and behave in relation to happiness in their life. Method: A total of 435 participants (197 males and 238 females) with an age mean of 44.84 (sd = 13.36) responded to the Positive Affect Negative Affect Schedule (10 positive affect items, 10 negative affect items), the Satisfaction with Life Scale (five items), and the Harmony in life Scale (five items). We used Bifactor-Graded Response MIRT for the main analyses. Result: At the general level, each of the 30 items had a strong capacity to discriminate between respondents across all three dimensions of subjective well-being. The investigation of different parameters (e.g., marginal slopes, ECV, IECV) strongly reflected the multidimensionality of subjective well-being at the item, the scale, and the model level. Indeed, subjective well-being could explain 64 % of the common variance in the whole model. Moreover, most of the items measuring positive affect (8/10) and life satisfaction (4/5) and all the items measuring harmony in life (5/5) accounted for a larger amount of variance of subjective well-being compared to that of their respective individual dimensions. The negative affect items, however, measured its own individual concept to a lager extent rather than subjective well-being. Thus, suggesting that the experience of negative affect is a more independent dimension within the whole subjective well-being model. We also found that specific items (e.g., "Alert", "Distressed", "Irritable", "I am satisfied with my life") were the recurrent exceptions in our results. Last but not the least, experiencing high levels in one dimension seems to compensate for low levels in the others and vice versa. Conclusion: As expected, the three subjective well-being dimensions do not work separately. Interestingly, the order and magnitude of the effect by each dimension on subjective well-being mirror how people define happiness in their life: first as harmony, second as satisfaction, third as positive emotions, and fourth, albeit to a much lesser degree, as negative emotions. Ergo, we argue that subjective well-being functions as a complex biopsychosocial adaptive system mirroring our attitude towards life in these three dimensions (A: affective dimension; B: behavioral dimension; C: cognitive dimension). Ergo, researchers and practitioners need to take in to account all three to fully understand, measure, and promote people's experience of the happy life. Moreover, our results also suggest that negative affect, especially regarding high activation unpleasant emotions, need considerable changes and further analyses if it is going to be included as a construct within the affective dimension of a general subjective well-being factor.

Women Are Perceived to Have More Power Than Men in Domains That Are Viewed As Important in U.S. Tweets.

Previous research of gender differences in power has largely focused on the public domain (e.g., leadership positions), and to a lesser extent power on the private domain (e.g., power in private relationships), where people may perceive these domains to be more or less important in their lives. We studied gender differences in preference weighted power (PWP) in different domains by weighting the perceived power by the perceived importance in life in a large set of Twitter messages from the United States (N = 9,286,471). The results showed that men (tweets including "he") are semantically related to general power, and women (tweets including "she") are related to importance in life, which in previous research have been connected to the public and private domains, respectively. Importantly, women had higher PWP than men. In conclusion, men are perceived to have more general power and women more PWP in U.S. tweets.

Assessment of depression and anxiety in young and old with a question-based computational language approach.

Middle aged adults experience depression and anxiety differently than younger adults. Age may affect life circumstances, depending on accessibility of social connections, jobs, physical health, etc, as these factors influence the prevalence and symptomatology. Depression and anxiety are typically measured using rating scales; however, recent research suggests that such symptoms can be assessed by open-ended questions that are analysed by question-based computational language assessments (QCLA). Here, we study middle aged and younger adults' responses about their mental health using open-ended questions and rating scales about their mental health. We then analyse their responses with computational methods based on natural language processing (NLP). The results demonstrate that: (1) middle aged adults describe their mental health differently compared to younger adults; (2) where, for example, middle aged adults emphasise depression and loneliness whereas young adults list anxiety and financial concerns; (3) different semantic models are warranted for younger and middle aged adults; (4) compared to young participants, the middle aged participants described their mental health more accurately with words; (5) middle-aged adults have better mental health than younger adults as measured by semantic measures. In conclusion, NLP combined with machine learning methods may provide new opportunities to identify, model, and describe mental health in middle aged and younger adults and could possibly be applied to the older adults in future research. These semantic measures may provide ecological validity and aid the assessment of mental health.

Precise language responses versus easy rating scales-Comparing respondents' views with clinicians' belief of the respondent's views.

BACKGROUND: Closed-ended rating scales are the most used response format for researchers and clinicians to quantify mental states, whereas in natural contexts people communicate with natural language. The reason for using such scales is that they are typically argued to be more precise in measuring mental constructs; however, the respondents' views as to what best communicates mental states are frequently ignored, which is important for making them comply with assessment. METHODS: We assessed respondents' (N = 304) degree of depression using rating scales, descriptive words, selected words, and free text responses and probed the respondents for their preferences concerning the response formats across twelve dimensions related to the precision of communicating their mental states and the ease of responding. This was compared with the clinicians' (N = 40) belief of the respondent's view. RESULTS: Respondents found free text to be more precise (e.g., precision d' = .88, elaboration d' = 2.0) than rating scales, whereas rating scales were rated as easier to respond to (e.g., easier d' = -.67, faster d' = -1.13). Respondents preferred the free text responses to a greater degree than rating scales compared to clinicians' belief of the respondents' views. CONCLUSIONS: These findings support previous studies concluding that future assessment of mental health can be aided by computational methods based on text data. Participants prefer an open response format as it allows them to elaborate, be precise, etc., with respect to their mental health issues, although rating scales are viewed as faster and easier.

Psychometric properties of persian version of escapism scale among Iranian adolescents.

The aim of this study was to assess the psychometric properties of the Persian version of the Escapism Scale among Iranian adolescents aged 14-18. Between January 2021 and August 2021, cross-sectional study was conducted using a convenience sampling method to select 566 participants (340 girls and 226 boys) to investigate the relationship between physical activity and mental health in adolescents. The participants completed several questionnaires, including the Escapism Scale, Erikson Psychosocial Stage Inventory (EPSI), Satisfaction with Life Scale (SWLS), Hope Scale (AHS), Smartphone Addiction Scale-Short Version (SAS-SV), and General Self-efficacy (GSE). Construct validity, reliability using Cronbach's alpha, and concurrent validity were used to evaluate the Escapism Scale's validity and reliability. Results of the Confirmatory Factor Analysis (CFA) indicated that a two-factor model provided a good fit for the data: sbX2 = 179.99 (p < 0.01); SRMR = 0.07; RMR = 0.56, CFI = 0.91; NFI = 0.89; IFI = 0.91; NFI = 0.89; GFI = 0.93; AGFI = 0.90, RMSEA = 0.076). The Cronbach's alpha coefficient for escapism was 0.73. The study found a significant positive relationship between escapism and smartphone addiction (r = 0.19). Additionally, a significant negative relationship was observed between escapism and hope (r=-0.31), satisfaction with life (r=-0.34), and general self-efficacy (r=-0.33). Furthermore, a significant relationship was found between escapism and gender. Lastly, the study found a significant relationship between escapism and identity confusion (r = 0.164, P < 0.01) and identity coherence (P < 0.01, r = 29). In conclusion, the Escapism Scale is a valid and reliable tool for assessing escapism and psychological evaluations in Iranian adolescents. These results may inform future research and suggest re-testing in clinical populations.

Natural language analyzed with AI-based transformers predict traditional subjective well-being measures approaching the theoretical upper limits in accuracy.

We show that using a recent break-through in artificial intelligence -transformers-, psychological assessments from text-responses can approach theoretical upper limits in accuracy, converging with standard psychological rating scales. Text-responses use people's primary form of communication -natural language- and have been suggested as a more ecologically-valid response format than closed-ended rating scales that dominate social science. However, previous language analysis techniques left a gap between how accurately they converged with standard rating scales and how well ratings scales converge with themselves - a theoretical upper-limit in accuracy. Most recently, AI-based language analysis has gone through a transformation as nearly all of its applications, from Web search to personalized assistants (e.g., Alexa and Siri), have shown unprecedented improvement by using transformers. We evaluate transformers for estimating psychological well-being from questionnaire text- and descriptive word-responses, and find accuracies converging with rating scales that approach the theoretical upper limits (Pearson r = 0.85, p < 0.001, N = 608; in line with most metrics of rating scale reliability). These findings suggest an avenue for modernizing the ubiquitous questionnaire and ultimately opening doors to a greater understanding of the human condition.

Language and gender: Computerized text analyses predict gender ratios from organizational descriptions.

Previous research has shown that language in job adverts implicitly communicates gender stereotypes, which, in turn, influence employees' perceived fit with the job. In this way, language both reflects and maintains a gender segregated job market. The aim of this study was to test whether, and how, language in organizational descriptions reflects gender segregation in the organizations by the use of computational text analyses. We analyzed large Swedish companies' organizational descriptions from LinkedIn (N = 409), testing whether the language in the organizational descriptions is associated with the organizations' employee gender ratio, and how organizational descriptions for organizations with a majority of women and men employees differ. The statistical analyses showed that language in the organizational descriptions predicted the employee gender ratio in organizations well. Word clouds depicting words that differentiate between organizations with a majority of women and men employees showed that the language of organizations with a higher percentage of women employees was characterized by a local focus and emphasis on within-organizations relations, whereas the language of organizations with a higher percentage of men employees was characterized by an international focus and emphasis on sales and customer relations. These results imply that the language in organizational descriptions reflects gender segregation and stereotypes that women are associated with local and men with global workplaces. As language communicates subtle signals in regards to what potential candidate is most sought after in recruitment situations, differences in organizational descriptions can hinder underrepresented gender groups to apply to these jobs. As a consequence, such practices may contribute to gender segregation on the job market.

Changes in self and object representations following psychotherapy measured by a theory-free, computational, semantic space method.

We propose a theory-neutral, computational and data-driven method for assessing changes in semantic content of object representations following long-term psychodynamic psychotherapy. Young adults in psychotherapy are compared with an age-matched, non-clinical sample at three time points. Verbatim transcripts of descriptions of the self and parents were quantified in a semantic space constructed by Latent Semantic Analysis. In the psychotherapy group, all representations changed from baseline to follow-up, whereas no comparable changes could be observed in the comparison group. The semantic space method supports the hypothesis that long-term psychodynamic psychotherapy contributes to sustained change of affective-cognitive schemas of self and others.

Reevaluating the Influence of Leaders Under Proportional Representation: Quantitative Analysis of Text in an Electoral Experiment.

We propose that leaders play a more important role in voters' party sympathy in proportional representation systems (PR) than previous research has suggested. Voters, from the 2018 Swedish General Election, were in an experiment asked to describe leaders and parties with three indicative keywords. Statistical models were conducted on these text data to predict their vote choice. The results show that despite that the voters vote for a party, the descriptions of leaders predicted vote choice to a similar extent as descriptions of parties. However, the order of the questions mattered, so that the first questions were more predictive than the second question. These analyses indicate that voters tend to conflate characteristics of leaders with their parties during election campaigns, and that leaders are a more important aspect of voting under PR than previous literature has suggested. Overall, this suggests that statistical analysis of words sheds new light of underlying sympathies related to voting.

Voters' view of leaders during the Covid-19 crisis: Quantitative analysis of keyword descriptions provides strength and direction of evaluations.

Objectives: Previous research suggests that governments usually gain support during crises such as the Covid-19. However, these findings are based on rating scales that only allow us to measure the strength of this support. This article proposes a new measure of how voters evaluate Prime Ministers (PM) by asking for descriptive keywords that are analyzed by natural language processing. Methods: By collecting a representative sample of citizens' own key words describing their PM in 15 countries in Europe during the outbreak of Covid-19, and analyzing these by latent semantic analysis and a multiple OLS regression, we could quantify the strength and direction of voters' view. Results: The strength analysis supported previous studies that describing the PM with positive words was strongly associated with vote intention. Furthermore, a change in the direction of the attitudes from "good" to "honest" was found. A new finding was that the pandemic was associated with an increase in polarization. Conclusions: The keyword evaluation analysis provides opportunities of evaluating both strength and direction of voters' view of their PM, where we show new results related to increased polarization and shift in the direction of attitudes.

Freely Generated Word Responses Analyzed With Artificial Intelligence Predict Self-Reported Symptoms of Depression, Anxiety, and Worry.

BACKGROUND: Question-based computational language assessments (QCLA) of mental health, based on self-reported and freely generated word responses and analyzed with artificial intelligence, is a potential complement to rating scales for identifying mental health issues. This study aimed to examine to what extent this method captures items related to the primary and secondary symptoms associated with Major Depressive Disorder (MDD) and Generalized Anxiety Disorder (GAD) described in the Diagnostic and Statistical Manual of Mental Disorders (DSM-5). We investigated whether the word responses that participants generated contained information of all, or some, of the criteria that define MDD and GAD using symptom-based rating scales that are commonly used in clinical research and practices. METHOD: Participants (N = 411) described their mental health with freely generated words and rating scales relating to depression and worry/anxiety. Word responses were quantified and analyzed using natural language processing and machine learning. RESULTS: The QCLA correlated significantly with the individual items connected to the DSM-5 diagnostic criteria of MDD (PHQ-9; Pearson's r = 0.30-0.60, p < 0.001) and GAD (GAD-7; Pearson's r = 0.41-0.52, p < 0.001; PSWQ-8; Spearman's r = 0.52-0.63, p < 0.001) for respective rating scales. Items measuring primary criteria (cognitive and emotional aspects) yielded higher predictability than secondary criteria (behavioral aspects). CONCLUSION: Together these results suggest that QCLA may be able to complement rating scales in measuring mental health in clinical settings. The approach carries the potential to personalize assessments and contributes to the ongoing discussion regarding the diagnostic heterogeneity of depression.

Computational Language Assessments of Harmony in Life - Not Satisfaction With Life or Rating Scales - Correlate With Cooperative Behaviors.

Different types of well-being are likely to be associated with different kinds of behaviors. The first objective of this study was, from a subjective well-being perspective, to examine whether harmony in life and satisfaction with life are related differently to cooperative behaviors depending on individuals' social value orientation. The second objective was, from a methodological perspective, to examine whether language-based assessments called computational language assessments (CLA), which enable respondents to answer with words that are analyzed using natural language processing, demonstrate stronger correlations with cooperation than traditional rating scales. Participants reported their harmony in life, satisfaction with life, and social value orientation before taking part in an online cooperative task. The results show that the CLA of overall harmony in life correlated with cooperation (all participants: r = 0.18, p < 0.05, n = 181) and that this was particularly true for prosocial participants (r = 0.35, p < 0.001, n = 96), whereas rating scales were not correlated (p > 0.05). No significant correlations (measured by the CLA or traditional rating scales) were found between satisfaction with life and cooperation. In conclusion, our study reveals an important behavioral difference between different types of subjective well-being. To our knowledge, this is the first study supporting the validity of self-reported CLA over traditional rating scales in relation to actual behaviors.