Explanatory Power Of Regression Models Research Articles

Digitalisation dynamics in SMEs: An approach from systems dynamics and artificial intelligence

This paper addresses the study of digitalisation dynamics in SMEs. Improving on existing research and its methodological limitations, we provide an understanding of the digital transformation in SMEs by approaching the research from a non-linear and complex perspective. We empirically test our hypotheses using the Eurostat Flash Eurobarometer No. 486 data set, with a final sample of 16,365 SMEs. Our first contribution shows that an adequate understanding of digital transformation not only implies the identification of drivers of digitalisation but also a grasp of how these drivers act, highlighting the differential effect that internal capabilities and external support of the company in interaction have on digital transformation. Moreover, the results show that the effect of interactions between variables is transferred to the output variable in a non-linear process, which may contain an optimum produced by a differential combination of input variables. Second, the paper extends the research methodology, emphasising the importance of combining classic regression analysis with machine-learning techniques. Thus, using a systemic approach, we conclude that the combination of the explanatory power of regression models and machine learning allows us to quantify and explain how variables act, solving complex and non-linear problems.

Second-Hand Clothing Shoppers’ Motivations. An Exploratory Study among Ethnic Hungarians from the Szeklerland Region of Romania

Abstract We conducted a non-representative online questionnaire survey among inhabitants of the Szeklerland region in Romania with a view to looking into their second-hand shopping habits. Based on an adapted version of an international scale, the present analysis aims to explore the motivational background of these shopping activities. The exploratory factor analysis indicated five motivational dimensions: economic/austerity, critical/environmental, originality, social interaction, nostalgic and self-expression. The three latter motivational dimensions resemble the hedonic motivational dimensions indicated by the literature. The motivational dimensions can be explained by a series of independent variables; however, the explanatory power of regression models is marginal. As a second step, we adopted a cluster analysis in order to model second-hand shoppers’ typical consumer segments. We found three clusters: austerity social interactionist, originality seekers, and nostalgic critics.

Scale‐dependent patterns and drivers of plant diversity in steppe grasslands of the Central Alborz Mts., Iran

AbstractQuestionsWhich are the main environmental drivers of plot scale alpha‐diversity and beta‐diversity in arid steppes? Do they vary between functional groups and across grain sizes?LocationCentral Alborz Mts., N Iran.MethodsWe sampled vascular plants in 23 nested‐plot series with nine grain sizes from 0.0001 m2 to 1,000 m2 and 334 25‐m2 vegetation plots in different vegetation types of natural dry steppe grasslands. To assess beta‐diversity, we calculated overall and local z‐values of species–area relationships modelled with the power function. As potential drivers of species richness and z‐values, we used topographic, edaphic and climatic variables as well as management types. Generalised linear models (GLMs), and generalised linear mixed‐effect models (GLMMs) if spatial autocorrelation occurred, were used in a multi‐model inference framework to build statistical models.ResultsMean annual temperature was the most important predictor for total species richness and richness of functional groups across grain sizes, with a unimodal relationship for grains of 25–100 m2, but mostly increasing for finer grain sizes. Precipitation of the driest month and cover of gravel were influential drivers at the smallest grains. The explanatory power of regression models increased towards larger grain sizes. The overall z‐values showed a high positive relationship with precipitation of the driest month, mean annual temperature and mean soil depth.ConclusionsRelated to our more than 3,000‐m elevational gradient, mean annual temperature (highly negatively correlated with elevation) was the most influential and consistent driver across functional groups and grain sizes with mostly unimodal relationships for alpha‐diversity and a positive effect on beta‐diversity. Findings for other drivers were less consistent, and overall the explained variance of our models was relatively low, calling for additional studies to determine whether in the arid grasslands of Iran stochasticity is just higher or there are additional important variables.

Effects of physical and psychological factors on users’ attitudes, use patterns, and perceived benefits toward urban parks

This paper studied perceived physical and psychological factors that influence relations between people and urban parks. Current literature showed that these factors influence attitudes, behavior, and perceived benefits toward urban parks. While most research focused on either physical factors or psychological determinants, much less grouping them and simultaneously examining their effects on people-environment relations. To study the issue, we collected data by interviewing five hundred park visitors from ten urban parks in Hong Kong. Results showed that both categories of influence are significantly associated with relations between people and urban parks; facilities and management in physical dimension and perceived accessibility in psychological factors are variables most strongly associated with these relations. Besides, psychological factors added explanatory power of regression models. Nevertheless, the inclusion of psychological factors crowded out physical factors as significant variables, and the mediation test suggested that psychological factors play a potential mediating role in the associations between physical factors and the people-environment relations. The findings highlighted the salient factors of urban parks and the effects on health-related benefits.

Невербальные предикторы оценок достоверности/недостоверности сообщаемой информации

A microstructural analysis of perception of a partner in communication was carried out. Mute video recordings of 15 clips of a structured conversation in which communicants expressed true and false judgments, were subjected to complex coding. In each 40 ms frame 51 nonverbal signs/102 binary markers indicating the state of the facial zones, the nature of the movements of the head, hands and body were considered. Based on expert estimates, the proportion of frames was calculated, in which each of the markers is present at selected time intervals. Looking at the video clips, 35 observers intuitively, by external features, determined the fragments when the communicator is telling truth and when lying. The frequency and time of occurrence of markers were analyzed. Frequency regression models of “true” and “false” response of communicants were built. It is shown that the required estimates are performed by the observer 1.5—2 seconds before the answer. High-frequency features form a stable core of the impression of the reliability of the communicator’s judgments, complemented with changeable low-frequency features explaining the growth of explanatory power of regression models. Markers have been identified that contribute to adequate perception of the reliability of information reported. The style of non-verbal behavior of people implementing alternative communication strategies is described.

Social Capital, the Quality of Social Networks and Redistribution Preferences: Evidence from the International Social Survey Programme

Social capital can be a useful asset in the labor market. Individuals who have more or better-quality connections in their social networks have an easier time finding employment. Since they face a lower probability of unemployment, one might expect them to express less support for redistributive tax policies. Using survey data from 20 advanced industrialized countries included in the 2006 Role of Government module of the International Social Survey Programme (ISSP), I find some evidence that individuals with better-quality social networks tend to oppose increasing taxes on the rich and lowering taxes on the poor, and that they believe the government ought to be responsible for providing jobs to those who want them, and for ensuring a decent living standard for the unemployed. At the same time, I find that higher social network quality does not appear to correlate with an individual’s belief that the government ought to reduce income differences, or increase spending on unemployment benefits. The social network quality measures do not, however, appear to increase the explanatory power of regression models significantly, indicating that any implied effects are fairly tenuous.

Changing towards electric vehicle use in Greater Stockholm

This paper studies electric vehicle (EV) adoption in Greater Stockholm in Sweden using the Transtheoretical Model of Change (TTM) and the Protection Motivation Theory as a framework and considers socio-cognitive, behavioural and socio-economic attributes that may influence the process towards electric vehicle use. TTM considers behavioural change as a process consisting of five stages-of-change rather than as an event. Some key findings were made: (1) from the earlier to the later stages-of-change, the attitude towards EVs becomes more positive, the knowledge about EVs increases and the self-efficacy is consistently increasing. (2) The threat appraisal and response efficacy of EVs increase from stage to stage in the stages prior to the actual change but have a lower level for the stages after the change. (3) The explanatory power of regression models modelling both pre-contemplation and all stages-of-change increases significantly when incorporating socio-cognitive variables such as self-efficacy, threat-appraisal, response efficacy and attitudes towards EVs. (4) The modal share of the car is consistently increasing throughout the stages-of-change. The results indicate that policy measures aiming at increasing knowledge and self-efficacy of car drivers related to EV use can stimulate electric vehicle adoption. Also, the relative advantages of EVs for car drivers should get more attention rather than only emphasizing the environmental advantages.

Using satellite-based measurements to explore spatiotemporal scales and variability of drivers of new particle formation.

New particle formation (NPF) can potentially alter regional climate by increasing aerosol particle (hereafter particle) number concentrations and ultimately cloud condensation nuclei. The large scales on which NPF is manifest indicate potential to use satellite-based (inherently spatially averaged) measurements of atmospheric conditions to diagnose the occurrence of NPF and NPF characteristics. We demonstrate the potential for using satellite-based measurements of insolation (UV), trace gas concentrations (sulfur dioxide (SO2), nitrogen dioxide (NO2), ammonia (NH3), formaldehyde (HCHO), and ozone (O3)), aerosol optical properties (aerosol optical depth (AOD) and Ångström exponent (AE)), and a proxy of biogenic volatile organic compound emissions (leaf area index (LAI) and temperature (T)) as predictors for NPF characteristics: formation rates, growth rates, survival probabilities, and ultrafine particle (UFP) concentrations at five locations across North America. NPF at all sites is most frequent in spring, exhibits a one-day autocorrelation, and is associated with low condensational sink (AOD × AE) and HCHO concentrations, and high UV. However, there are important site-to-site variations in NPF frequency and characteristics, and in which of the predictor variables (particularly gas concentrations) significantly contribute to the explanatory power of regression models built to predict those characteristics. This finding may provide a partial explanation for the reported spatial variability in skill of simple generalized nucleation schemes in reproducing observed NPF. In contrast to more simple proxies developed in prior studies (e.g., based on AOD, AE, SO2, and UV), use of additional predictors (NO2, NH3, HCHO, LAI, T, and O3) increases the explained temporal variance of UFP concentrations at all sites.

Predicting the similarity of parasite communities in freshwater fishes using the phylogeny, ecology and proximity of hosts

Parasite communities tend to be dissimilar in hosts that are geographically, phylogenetically, ecologically and developmentally distant from one another. The decay of community similarity is a powerful and increasingly common method of studying parasite beta diversity, but most studies have examined only a single type of distance. Here, we evaluate distances based on the phylogeny, ecology, spatial proximity and size of hosts, as predictors of the similarity of parasite communities in individual hosts, host populations and host species. We surveyed parasites in six species of fish collected simultaneously from six localities in the St. Lawrence River, Canada, and species in a common group of larval parasites were discriminated using DNA sequences from barcode region of cytochrome c oxidase I. Distances based on the habitat use patterns of host species were good predictors of short‐term, ecological similarity of parasite communities, such as that operating at the scale of the individual host. The genetic distance between host species was associated with almost all types of similarity at all scales, particularly qualitative and phylogenetic similarity of parasite communities at the level of populations and meta‐populations of hosts. The trophic level, diet, spatial proximity and size of hosts were poor predictors of parasite community similarity. The increased taxonomic resolution provided by molecular data increased the explanatory power of regression models, and different factors were implicated when parasite species were distinguished with DNA barcodes than when larval parasites were lumped into morphospecies, as is commonly practiced.

The Inflation-Hedging Ability of Canadian Direct Real Estate Return and its Components, Income, and Appreciation

Executive Summary.This paper examines the inflation hedging characteristics of Canadian private real estate between 1973 and 2007. In the short term, total and appreciation returns are a complete hedge against inflation, expected and unexpected inflation, and remain a complete hedge after the introduction of macroeconomic variables increasing materially the explanatory power of regression models. Cointegration tests show a positive long-term relationship between inflation and total return, as well as appreciation return. Appreciation return is thus identified as the true inflation-hedging source for Canadian real estate. Granger causality tests reveal causality from total and appreciation returns to inflation. Therefore, total and appreciation returns may help forecast future inflation rates.

Value Relevance of Earnings and Cash Flows from Operations: Effects of Earnings Permanence, Growth and Firm Size in Tunisian Stock Exchange

The purpose of this study is to examine the information content of accounting earnings and CFO to explain security returns in Tunisian Stock Exchange through the study of their response coefficients and the explanatory power of regression models in the period of 1997 - 2001 (an application of a New Accounting System of Companies). We hypothesis that (i) earnings (CFO) have an incremental information content beyond CFO (earnings) when the two variables are jointly used in return regression, (ii) the response of returns to CFO (earnings) increases (decreases) when earnings is transitory, (iii) security returns response to both earnings and (CFO) will be higher (lower) for high (low) growth firms and (iv) the security returns response to accounting earnings (CFO) are weaker (higher) for large (small) firms. We have introduced market-to-book ratio and market value of equity as additional variables in regression models and as proxies of growth and firm size. Results can not affirm that CFO plays a more important role in marketplace when earnings are transitory. Results in relation to earnings growth and firm size hypothesis show that CFO can join Market-to-Book ratio and market value of equity as proxy of risk.

Length of comorbidity lookback period affected regression model performance of administrative health data

Background and Objective The impact of different comorbidity ascertainment lookback periods on modeling posthospitalization mortality and readmission was examined. Methods Index cases comprised medical ( n = 326,456) and procedural ( n = 349,686) patients with a hospital admission from 1990–1996. Administrative hospital data were extracted for 102 comorbidities, ascertained at index admission and for 1-, 2-, 3-, and 5-year lookback periods. Deaths and readmissions were identified within 12 months and 30 days of separation, respectively. Hierarchically nested and nonnested Cox regressions as well as Receiver Operator Characteristic Area Under the Curve (ROC-AUC) were used to determine model-fit and predictive ability of lookback period models. Results The 1-year lookback period provided the best model-fit for both patient groups when modeling mortality. A similar model-fit was seen at index admission for procedural but not medical patients. The superior readmission model employed 5 years of lookback for both patient groups. With one exception, all lookback period models were superior to those abstracting comorbidity from index admission only. Similar results were evident from ROC-AUC, although greater predictive ability was seen with modeling of mortality (0.847–0.923) compared with readmission (0.593–0.681). Conclusion The explanatory power of regression models, when adjusting for comorbidity, is influenced by length of lookback, outcome investigated and clinical subgroup. Shorter periods (∼1 year) appear appropriate for modeling posthospitalization mortality, whereas longer lookback periods are superior for readmission outcomes.

Hedge Fund Risk Factors With Option-like Structures

The authors start with the observation that hedge funds are not obliged to disclose sources of earnings and trade strategies. For this reason little is known on how returns are achieved. One approach to shed some light on this matter is based on Sharpe [1992], which introduced a model for the explanation of returns of traditional mixed funds (e.g., composed of equities and bonds). Mixed funds usually pursue a strategy of “long only” whereas hedge funds also use short selling and leveraging, and act in a much more flexible way. These trade strategies of hedge funds lead to option-like structures, which are not covered by the basic Sharpe model. They discuss and describe the possibilities to include such structures into a regression model. Offering data on the various such strategies, they present the empirical results of an analysis of the risk factors of the Standard & Poor9s hedge fund indices. They conclude with the view that their study confirms the findings documented in the literature and that further research could be directed toward capturing further risk factors in order to increase the explanatory power of regression models.

The Binary Regression Quantile Plot: Assessing the Importance of Predictors in Binary Regression Visually

We present a graphical measure of assessing the explanatory power of regression models with a binary response. The binary regression quantile plot and an area defined by it are used for the visual comparison and ordering of nested binary response regression models. The plot shows how well various models explain the data. Two data sets are analyzed and the area representing the fit of a model is shown to agree with the usual likelihood ratio test.