Specification Curve Analysis Research Articles

Meta-analyzing intelligence and religiosity associations: Evidence from the multiverse.

Over the past century, a remarkable body of research about the relationship of intelligence and religiosity has accumulated. So far, the majority of studies that investigated this relationship showed a negative correlation, indicating lower cognitive abilities of individuals reporting stronger religious beliefs. Although the effect direction has been observed to be largely consistent across studies, the reported effect strength varied substantially across studies. Several potentially moderating variables such as different intelligence and religiosity assessment methods, educational status of samples, and participant sex have been proposed as likely candidates for explaining systematic differences in effect strengths. However, the effects of these moderators are to date unclear. Consequently, we focused in investigating effects of these moderating variables on the intelligence and religiosity link in an update of prior meta-analytical investigations in n = 89 (k = 105; N = 201,457) studies. Random-effects analyses showed a small but robust negative association between intelligence and religiosity r = -.14 (p < .001; 95% CI [-.17, -.12]). Effects were stronger for (i) psychometric intelligence tests than for proxy measures such as grade point averages and (ii) general population and college samples than pre-college samples. Moreover, we provide evidence from combinatorial, multiverse, and specification curve analyses that further corroborates the robustness of the investigated association. Out of 192 reasonable specifications all 135 (70.4%) significant summary effects were negative. In all, our results show small but robust negative associations between religiosity and intelligence that are differentiated in strength but generalize in terms of direction over moderating variables.

Specification curve analysis shows that social media use is linked to poor mental health, especially among girls

An important 2019 paper applied a novel analytic technique called Specification Curve Analysis (SCA) to data from three large-scale community samples to investigate the association between adolescent technology use and mental health/well-being. The paper concluded that an association exists but is tiny, with median betas between −0.01 and −0.04. This association was reported to be smaller than links between mental health and various innocuous variables in the datasets such as eating potatoes, and therefore to be of no practical significance. The current paper re-ran SCA on the same datasets while applying alternative analytic constraints on the model specification space, including: 1) examining specific digital media activities (e.g., social media) separately rather than lumping all “screen time” including TV together; 2) examining boys and girls separately, rather than examining them together; 3) excluding potential mediators from the list of controls; and 4) treating scales equally (rather than allowing one scale with many subscales to dominate all others). We were able to reproduce the original results with the original configurations. When we used the revised constraints, we found several much larger relationships than previously reported. In particular: among girls, there is a consistent and substantial association between mental health and social media use (median betas from −0.11 to −0.24). These associations were stronger than links between mental health and binge drinking, sexual assault, obesity, and hard drug use, suggesting that these associations may have substantial practical significance as many countries are experiencing rising rates of depression, anxiety, and suicide among teenagers and young adults.

探索日间功能障碍与认知功能之间相关性的稳健性：使用标准曲线分析的方法

探索日间功能障碍与认知功能之间相关性的稳健性：使用标准曲线分析的方法

Multimethod assessment of pubertal timing and associations with internalizing psychopathology in early adolescent girls.

Early pubertal timing has consistently been associated with internalizing psychopathology in adolescent girls. Here, we aimed to examine whether the association between timing and mental health outcomes varies by measurement of pubertal timing and internalizing psychopathology, differs between adrenarcheal and gonadarcheal processes, and is stronger concurrently or prospectively. We assessed 174 female adolescents (age 10.0-13.0 at Time 1) twice, with an 18-month interval. Participants provided self-reported assessments of depression/anxiety symptoms and pubertal development, subjective pubertal timing, and date of menarche. Their parents/guardians also reported on the adolescent's pubertal development and subjective pubertal timing. We assessed salivary dehydroepiandrosterone (DHEA), testosterone, and estradiol levels and conducted clinical interviews to determine the presence of case level internalizing disorders. From these data, we computed 11 measures of pubertal timing at both time points, as well as seven measures of internalizing psychopathology, and entered these in a Specification Curve Analysis. Overall, earlier pubertal timing was associated with increased internalizing psychopathology. Associations were stronger prospectively than concurrently, suggesting that timing of early pubertal processes might be especially important for later risk of mental illness. Associations were strongest when pubertal timing was based on the Tanner Stage Line Drawings and when the outcome was case-level Diagnostic and Statistical Manual of Mental Disorders, 4th edition (DSM-IV) depression or Hierarchical Taxonomy of Psychopathology (HiTOP) distress disorders. Timing based on hormone levels was not associated with internalizing psychopathology, suggesting that psychosocial mechanisms, captured by timing measures of visible physical characteristics might be more meaningful determinants of internalizing psychopathology than biological ones in adolescent girls. Future research should precisely examine these psychosocial mechanisms. (PsycInfo Database Record (c) 2022 APA, all rights reserved).

Does compulsory schooling affect health? Evidence from ambulatory claims data

Using claims data on more than 23 million statutorily insured, we investigate the causal effect of schooling on health in the largest and most comprehensive analysis for Germany to date. In a regression discontinuity approach, we exploit changes in compulsory schooling in West Germany to estimate the reduced form effect of the reforms on health, measured by doctor diagnoses in ICD-10 format covering physical as well as mental health conditions. To mitigate the problem that empirical results depend on subjective decisions made by the researcher, we perform specification curve analyses to assess the robustness of findings across various model specifications. We find that the reforms have, at best, very small impacts on the examined doctor diagnoses. In most of the specifications we estimate insignificant effects that are close to zero and often of the “wrong” sign. Therefore, our study questions the presence of the large positive effects of education on health that are found in the previous literature.

Clinicopathologic significance of DNA replication licensing factors in head and neck diffuse large B-cell lymphoma

Diffuse large B-cell lymphoma (DLBCL) harbors defects in the proliferation pathway. We performed multiparameter analysis of proteins expressed during different cell cycle phases and correlated them with clinical parameters of head and neck DLBCLs. Thirty-nine DLBCLs were staged and immunohistochemically stained with MCM2, Ki67, and geminin. The receiver operating characteristic curve and its area under the curve were calculated, and sensitivity vs specificity curve analysis was performed. The highest labeling index was in MCM2, followed by Ki67 and geminin (P < .001). All pairs showed significant differences (P < .001). The best cutoff points to differentiate limited from advanced disease were 68% and 45% for MCM2 and Ki67, respectively. There was no acceptable cutoff for geminin (area under the curve=0.667, P=.134). MCM2/Ki67 (P=.293) and geminin/Ki67 (P=.233) ratios did not differ between the stages. The median (interquartile range) of the geminin/Ki67 ratio was 0.57 (0.68), translating to a reduced G1. We suggest a role for cell cycle-related proteins in the biology and behavior of DLBCLs. MCM2 and Ki67 cutoffs can be a potential option to differentiate limited from advanced disease, where imaging and laboratory techniques are unavailable. The G1 decrease and the significantly higher MCM2 expression compared to Ki67 indicate replication disturbances, making factors involved in the G1 phase targets for treatment.

Chronic stress and executive functioning: A specification-curve analysis

To investigate the association between chronic stress and executive functioning (EF), we assessed 514 young to middle-aged adults in three EF tasks (i.e., Number-Letter, 2-Back, Go/Nogo) that assessed shifting, updating, and inhibition. Chronic stress was assessed by various self-report measures and hair cortisol concentrations as indicators of subjective and objective chronic stress, respectively. In order to test the association between chronic stress and EF, we fit a structural equation model with a latent common EF factor predicted by subjective and objective chronic stress on Kaplan-Meier estimates of response times. Controlling for participants' sex, age household income and the delay between cognitive testing and hair sample collection, neither subjective nor objective chronic stress showed a meaningful association with common EF. Exploratory analyses suggested a moderation effect of income on the association between subjective chronic stress and common EF, with a smaller association for high-income participants. Additionally, we conducted a specification-curve analysis on the association between chronic stress and EF to assess the influence of different analysis choices on results in our dataset. This analysis confirmed the absence of a coherent association between chronic stress and EF by showing that the majority of analytical choices produced null effects and only a small number of analytical choices produced meaningful associations (negative or positive). Taken together, our findings suggest that common EF likely remains preserved under the influence of chronic stress. Our specification-curve analysis, however, also shows that chronic stress may also have either a positive or a negative effect on EF, depending on the choice of covariates and measures of chronic stress and EF. Consequently, more research on the role of these factors for the association between chronic stress and EF is needed to avoid the interpretation of non-replicable stress-EF associations caused by analytical choices or selection bias.

Hyperparameter Tuning and Pipeline Optimization via Grid Search Method and Tree-Based AutoML in Breast Cancer Prediction.

Automated machine learning (AutoML) has been recognized as a powerful tool to build a system that automates the design and optimizes the model selection machine learning (ML) pipelines. In this study, we present a tree-based pipeline optimization tool (TPOT) as a method for determining ML models with significant performance and less complex breast cancer diagnostic pipelines. Some features of pre-processors and ML models are defined as expression trees and optimal gene programming (GP) pipelines, a stochastic search system. Features of radiomics have been presented as a guide for the ML pipeline selection from the breast cancer data set based on TPOT. Breast cancer data were used in a comparative analysis of the TPOT-generated ML pipelines with the selected ML classifiers, optimized by a grid search approach. The principal component analysis (PCA) random forest (RF) classification was proven to be the most reliable pipeline with the lowest complexity. The TPOT model selection technique exceeded the performance of grid search (GS) optimization. The RF classifier showed an outstanding outcome amongst the models in combination with only two pre-processors, with a precision of 0.83. The grid search optimized for support vector machine (SVM) classifiers generated a difference of 12% in comparison, while the other two classifiers, naïve Bayes (NB) and artificial neural network—multilayer perceptron (ANN-MLP), generated a difference of almost 39%. The method’s performance was based on sensitivity, specificity, accuracy, precision, and receiver operating curve (ROC) analysis.

Be Yourself and Behave Appropriately: Exploring Associations Between Incongruent Personality States and Positive Affect, Tiredness, and Cognitive Performance

Personality traits describe how people typically think, feel, and behave. Personality states describe how people think, feel, and behave in a given moment. In their daily lives, people often behave the way they typically do (they enact trait-congruent personality states), but occasionally behave differently from how they typically do (trait-incongruent personality states). Several theories propose that such incongruent personality states should be associated with undesirable outcomes such as less positive affect or more tiredness, but the current state of evidence is inconclusive and mostly based on one personality dimension: extraversion. In this study, we contribute to filling important gaps in the literature by examining congruence of personality dimensions other than extraversion, considering characteristics of the situation, and modeling congruence with state-of-the-art response surface analyses. We aimed to manipulate state honesty-humility and state agreeableness as well as perceived adversity and deception of the situation in a prisoner’s dilemma paradigm. The manipulations mostly had the intended effects but they also had additional unspecific effects on other personality states and situation characteristics. The study thus emphasized the difficulty of manipulating personality states, situation characteristics, and trait–state and state–situation congruence. In pre-registered analyses of variance, response surface analyses, and specification curve analyses, we then examined how trait–state congruence and state–situation congruence were associated with positive affect, tiredness, and performance in a numerical Stroop task. Neither trait–state congruence nor state–situation congruence were associated with positive affect, tiredness, or cognitive performance. However, in light of this study’s limitations, more studies that are carefully designed, carefully operationalized, and well-powered are needed to examine trait–state and state–situation congruence. Because experimental research can advance the understanding of personality dynamics substantially, future research should additionally further aim to develop valid and reliable manipulations of personality states and situation characteristics.

Comparing clinical performance of current implantable cardioverter-defibrillator implantation recommendations in arrhythmogenic right ventricular cardiomyopathy.

AimsArrhythmogenic right ventricular cardiomyopathy (ARVC) patients have an increased risk of ventricular arrhythmias (VA). Four implantable cardioverter-defibrillator (ICD) recommendation algorithms are available The International Task Force Consensus (‘ITFC’), an ITFC modification by Orgeron et al. (‘mITFC’), the AHA/HRS/ACC guideline for VA management (‘AHA’), and the HRS expert consensus statement (‘HRS’). This study aims to validate and compare the performance of these algorithms in ARVC.Methods and resultsWe classified 617 definite ARVC patients (38.5 ± 15.1 years, 52.4% male, 39.2% prior sustained VA) according to four algorithms. Clinical performance was evaluated by sensitivity, specificity, ROC-analysis, and decision curve analysis for any sustained VA and for fast VA (>250 b.p.m.). During 6.4 [2.8–11.5] years follow-up, 282 (45.7%) patients experienced any sustained VA, and 63 (10.2%) fast VA. For any sustained VA, ITFC and mITFC provide higher sensitivity than AHA and HRS (94.0–97.8% vs. 76.7–83.5%), but lower specificity (15.9–32.0% vs. 42.7%-60.1%). Similarly, for fast VA, ITFC and mITFC provide higher sensitivity than AHA and HRS (95.2–97.1% vs. 76.7–78.4%) but lower specificity (42.7–43.1 vs. 76.7–78.4%). Decision curve analysis showed ITFC and mITFC to be superior for a 5-year sustained VA risk ICD indication threshold between 5–25% or 2–9% for fast VA.ConclusionThe ITFC and mITFC provide the highest protection rates, whereas AHA and HRS decrease unnecessary ICD placements. ITFC or mITFC should be used if we consider the 5-year threshold for ICD indication to lie within 5–25% for sustained VA or 2–9% for fast VA. These data will inform decision-making for ICD placement in ARVC.

Testosterone fluctuations in response to a democratic election predict partisan attitudes toward the elected leader

Intergroup competitions such as democratic elections can intensify intergroup polarization and conflict. Partisan attitudes toward the elected leader can also shift from before to after an election, but the biology underlying these attitudinal shifts remains largely unknown. An important factor could be the hormone testosterone, which is theorized to fluctuate during competition and to influence status seeking. In a naturalistic study of 113 registered voters, we measured changes in testosterone levels and attitudes toward the winner of the 2012 US Presidential Election. We found that supporters of the losing candidate (Mitt Romney) showed acute increases in testosterone levels compared to supporters of the winner (Barack Obama) on the evening of Election Day. Supporters of the losing candidate also demonstrated flatter diurnal testosterone slopes on Election Day that persisted up to two days after the election. Furthermore, greater increases in acute testosterone levels and flatter diurnal slopes among supporters of the losing candidate were associated with less positive evaluations of the winning candidate. These testosterone-moderated attitudinal shifts observed in the days after the election showed a directionally similar pattern with a weaker effect size six months later. Finally, we confirmed that the main results were robust to alternative data analytic choices using multiverse specification curve analysis. The findings from this paper suggest that hormonal responses to large-scale intergroup competitions may shape how we perceive our elected leaders, shedding light on the biology of intergroup relations.

Understanding the effects of conceptual and analytical choices on ‘finding’ the privacy paradox: A specification curve analysis of large-scale survey data

ABSTRACT The privacy paradox suggests that privacy concerns do not relate to privacy-related behavior. Although it has inspired numerous studies, findings remain inconclusive. Some of the inconsistencies in published findings may be explained by a strong heterogeneity in the conceptual and analytical choices that researchers implement when investigating the privacy paradox. Based on representative survey data of the 27 EU member states (2011: n = 8,962; 2015: n = 10,526; 2019: n = 11,428), I investigated the effect of conceptual and analytical decisions on ‘finding’ the privacy paradox. Specification curve analyses revealed that the magnitude and statistical significance of the relationship between privacy concerns and information disclosure is contingent on the operationalization of the independent variable, the inclusion of covariates, and the age of the studied population. The relationship between online privacy concerns and using social media privacy settings, in contrast, was less influenced by analytical decisions. Yet, the relationship was stronger in younger people and increased over time. The findings call for more transparency in analyzing research data. Evaluating the implications of analytical choices will help to establish best practices and advance cumulative knowledge creation in privacy research.

Association between adolescent oral contraceptive use and future major depressive disorder: aprospective cohort study.

BackgroundBecause of the widespread use of oral contraceptives (OCs) and the devastating effects of depression both on an individual and a societal level, it is crucial to understand the nature of the previously reported relationship between OC use and depression risk. Insight into the impact of analytical choices on the association is important when interpreting available evidence. Hence, we examined the association between adolescent OC use and subsequent depression risk in early adulthood analyzing all theoretically justifiable models.MethodsData from the prospective cohort study TRacking Adolescents’ Individual Lives Survey, among women aged 13–25 years were used. Adolescent OC use (ages 16–19 years) was used as a predictor and major depressive disorder (MDD) in early adulthood (ages 20–25 years), as assessed by the Diagnostic and Statistical Manual of Mental Disorders‐IV oriented Lifetime Depression Assessment Self‐Report and the Composite International Diagnostic Interview, was used as an outcome. A total of 818 analytical models were analyzed using Specification Curve Analysis in 534 adolescent OC users and 191 nonusers.ResultsOverall, there was an association of adolescent OC use and an episode of MDD in early adulthood [median odds ratio (OR)median = 1.41; ORmin = 1.08; ORmax = 2.18, p < .001], which was driven by the group of young women with no history of MDD (ORmedian = 1.72; ORmin = 1.21; ORmax = 2.18, p < .001).ConclusionsIn summary, adolescent OC use was associated with a small but robust increased risk for experiencing an episode of MDD, especially among women with no history of MDD in adolescence. Understanding the potential side effects of OCs will help women and their doctors to make informed choices when deciding among possible methods of birth control.

Reconsidering evidence of moral contagion in online social networks.

The ubiquity of social media use and the digital data traces it produces has triggered a potential methodological shift in the psychological sciences away from traditional, laboratory-based experimentation. The hope is that, by using computational social science methods to analyse large-scale observational data from social media, human behaviour can be studied with greater statistical power and ecological validity. However, current standards of null hypothesis significance testing and correlational statistics seem ill-suited to markedly noisy, high-dimensional social media datasets. We explore this point by probing the moral contagion phenomenon, whereby the use of moral-emotional language increases the probability of message spread. Through out-of-sample prediction, model comparisons and specification curve analyses, we find that the moral contagion model performs no better than an implausible XYZ contagion model. This highlights the risks of using purely correlational evidence from large observational datasets and sounds a cautionary note for psychology's merge with big data.

The relationship between mental well-being and dysregulated gaming: a specification curve analysis of core and peripheral criteria in five gaming disorder scales.

Gaming disorder (also known as dysregulated gaming) has received significant research and policy attention based on concerns that certain patterns of play are associated with decreased mental well-being and/or functional impairment. In this study, we use specification curve analysis to examine analytical flexibility and the strength of the relationship between dysregulated gaming and well-being in the form of general mental health, depressive mood and life satisfaction. Dutch and Flemish gamers (n = 424) completed an online survey containing five unique dysregulated gaming measures (covering nine scale variants) and three well-being measures. We find a consistent negative relationship; across 972 justifiable regression models, the median standardized regression coefficient was −0.39 (min: −0.54, max: −0.19). Data show that the majority of dysregulated gaming operationalizations converge upon highly similar estimates of well-being. However, variance is introduced by the choice of well-being measure; results indicate that dysregulated gaming is more strongly associated with depressive mood than with life satisfaction. Weekly game time accounted for little to no unique variance in well-being in the sample. We argue that research on this topic should compare a broad range of psychosocial well-being outcomes and explore possible simplifications of the DSM-5 gaming disorder criteria. Given somewhat minute differences between dysregulated gaming scales when used in survey-based studies and largely equivalent relationships with mental health indicators, harmonization of measurement should be a priority.

Differences in outcomes following an intensive upper-limb rehabilitation program for patients with common central nervous system-acting drug prescriptions.

Background: Difficulty using the upper-limb is a major barrier to independence for many patients post-stroke or brain injury. High dose rehabilitation can result in clinically significant improvements in function even years after the incident; however, there is still high variability in patient responsiveness to such interventions that cannot be explained by age, sex, or time since stroke. Methods: This retrospective study investigated whether patients prescribed certain classes of central nervous system-acting drugs—γ-aminobutyric acid (GABA) agonists, antiepileptics, and antidepressants—differed in their outcomes on the three-week intensive Queen Square Upper-Limb program. For 277 stroke or brain injury patients (167 male, median age 52 years (IQR: 21), median time since incident 20 months (IQR: 26)) upper-limb impairment and activity was assessed at admission to the program and at six months post-discharge, using the upper limb component of the Fugl-Meyer, Action Research Arm Test, and Chedoke Arm and Hand Activity Inventory. Drug prescriptions were obtained from primary care physicians at referral. Specification curve analysis was used to protect against selective reporting results and add robustness to the conclusions of this retrospective study. Results: Patients with GABA agonist prescriptions had significantly worse upper-limb scores at admission but no evidence for a significant difference in program-induced improvements was found. Additionally, no evidence of significant differences in patients with or without antiepileptic drug prescriptions on either admission to, or improvement on, the program was found in this study. Although no evidence was found for differences in admission scores, patients with antidepressant prescriptions experienced reduced improvement in upper-limb function, even when accounting for anxiety and depression scores. Conclusions: These results demonstrate that, when prescribed typically, there was no evidence that patients prescribed GABA agonists performed worse on this high-intensity rehabilitation program. Patients prescribed antidepressants, however, performed poorer than expected on the Queen Square Upper-Limb rehabilitation program. While the reasons for these differences are unclear, identifying these patients prior to admission may allow for better accommodation of differences in their rehabilitation needs.

Community-acquired pneumonia severity assessment tools in patients hospitalized with COVID-19: a validation and clinical applicability study

ObjectiveTo externally validate community-acquired pneumonia (CAP) tools on patients hospitalized with coronavirus disease 2019 (COVID-19) pneumonia from two distinct countries, and compare their performance with recently developed COVID-19 mortality risk stratification tools. MethodsWe evaluated 11 risk stratification scores in a binational retrospective cohort of patients hospitalized with COVID-19 pneumonia in São Paulo and Barcelona: Pneumonia Severity Index (PSI), CURB, CURB-65, qSOFA, Infectious Disease Society of America and American Thoracic Society Minor Criteria, REA-ICU, SCAP, SMART-COP, CALL, COVID GRAM and 4C. The primary and secondary outcomes were 30-day in-hospital mortality and 7-day intensive care unit (ICU) admission, respectively. We compared their predictive performance using the area under the receiver operating characteristics curve (AUC), sensitivity, specificity, likelihood ratios, calibration plots and decision curve analysis. ResultsOf 1363 patients, the mean (SD) age was 61 (16) years. The 30-day in-hospital mortality rate was 24.6% (228/925) in São Paulo and 21.0% (92/438) in Barcelona. For in-hospital mortality, we found higher AUCs for PSI (0.79, 95% CI 0.77–0.82), 4C (0.78, 95% CI 0.75–0.81), COVID GRAM (0.77, 95% CI 0.75–0.80) and CURB-65 (0.74, 95% CI 0.72–0.77). Results were similar for both countries. For the 1%–20% threshold range in decision curve analysis, PSI would avoid a higher number of unnecessary interventions, followed by the 4C score. All scores had poor performance (AUC <0.65) for 7-day ICU admission. ConclusionsRecent clinical COVID-19 assessment scores had comparable performance to standard pneumonia prognostic tools. Because it is expected that new scores outperform older ones during development, external validation studies are needed before recommending their use.

Trust in government regarding COVID-19 and its associations with preventive health behaviour and prosocial behaviour during the pandemic: a cross-sectional and longitudinal study.

The effective implementation of government policies and measures for controlling the coronavirus disease 2019 (COVID-19) pandemic requires compliance from the public. This study aimed to examine cross-sectional and longitudinal associations of trust in government regarding COVID-19 control with the adoption of recommended health behaviours and prosocial behaviours, and potential determinants of trust in government during the pandemic. This study analysed data from the PsyCorona Survey, an international project on COVID-19 that included 23 733 participants from 23 countries (representative in age and gender distributions by country) at baseline survey and 7785 participants who also completed follow-up surveys. Specification curve analysis was used to examine concurrent associations between trust in government and self-reported behaviours. We further used structural equation model to explore potential determinants of trust in government. Multilevel linear regressions were used to examine associations between baseline trust and longitudinal behavioural changes. Higher trust in government regarding COVID-19 control was significantly associated with higher adoption of health behaviours (handwashing, avoiding crowded space, self-quarantine) and prosocial behaviours in specification curve analyses (median standardised β = 0.173 and 0.229, p < 0.001). Government perceived as well organised, disseminating clear messages and knowledge on COVID-19, and perceived fairness were positively associated with trust in government (standardised β = 0.358, 0.230, 0.056, and 0.249, p < 0.01). Higher trust at baseline survey was significantly associated with lower rate of decline in health behaviours over time (p for interaction = 0.001). These results highlighted the importance of trust in government in the control of COVID-19.

Hospitality innovation strategies: Robustness analysis of paths to firm performance

Empirical knowledge on the relationship between innovation and firm performance in the hospitality industry is important for theoretical progress and for decision making. Martin-Rios and Ciobanu’s (2019) recent contribution identifies eight different innovation strategies being significantly associated with firm performance including the ambitious strategy of simultaneously innovating products, processes, organizations, and marketing. By replicating the analysis and testing the robustness of the results with a specification curve analysis using the same dataset, we find only three strategies to be robustly associated with firm performance. All three include process innovation as a necessary component: i) a pure backstage-oriented strategy of process and organizational innovation that can be ii) extended with product innovation and iii) a frontstage-oriented strategy, where marketing innovations complement new products and processes. This substantiates the relevance of the process-based view of innovation in services and call for efforts in theorizing about how different configurations of innovation strategies can represent paths to firm performance.

Investigating Associations Between Screen Time and Symptomatology in Individuals With Serious Mental Illness: Longitudinal Observational Study.

BackgroundIncreasing screen time exposure from digital devices like smartphones has shown a variety of mixed associations with cognition, behavior, and well-being in adults and children but little is known about its associations with symptomatology in individuals with serious mental illness.ObjectiveTo determine the range of associations between screen time and symptoms of individuals with mental illness, we utilized a method called specification curve analysis.MethodsIn this observational study, we recruited smartphone-owning adults (≥18 years old) with schizophrenia and healthy controls. We installed 2 research-source smartphone apps, mindLAMP and Beiwe, to collect survey results, cognitive test results, and screen time metrics over a period of 3 months. Surveys were scheduled for twice a week, but participants were instructed to take the surveys naturally as much or as little as they wanted. Screen time was collected continuously in the background. A total of 140 participants was recruited from the outpatient clinic population as well as through general public advertising. Age-matched, smartphone-owning healthy controls were also part of the recruitment pool. A specification curve analysis was a priori designed to explore the relationship between every combination of independent variable and dependent variable in order to demonstrate to what degree screen time relates to symptoms in individuals with serious mental illness.ResultsThe sample consisted of 88 participants (54 with schizophrenia and 34 healthy controls) who completed both the initial and follow-up visits, completed at least one self-reported survey, and had a minimum passive data cutoff of 5 consecutive days. While we found an association between smartphone screen time metrics and cognition (adjusted R2=0.107, P<.001), specification curve analysis revealed a wide range of heterogenous associations with screen time from very negative to very positive. The effects differed based on diagnostic group, age bracket, type of regression model used, and the specific independent and dependent variables selected for analysis.ConclusionsThe associations between screen time and mental health in patients with schizophrenia are heterogenous when examined with methods that reduce analytical bias. The heterogeneity in associations suggests that complex and personalized potential effects must be understood in the greater context of an individual. This analysis of longitudinally collected screen time data shows potential for future research that could benefit from high resolution metrics on smartphone use.