Confounding and collinearity in regression analysis: a cautionary tale and an alternative procedure, illustrated by studies of British voting behaviour

Johnston, Ron; Jones, Kelvyn; Manley, David

doi:10.1007/s11135-017-0584-6

Cited by 764 publications

(436 citation statements)

References 26 publications

(23 reference statements)

Supporting

Mentioning

425

Contrasting

Unclassified

Order By: Relevance

“…We employed a linear regression to check for the existence of potential confounders and to verify whether they were not confusing the association between dependent and independent variables (Johnston et al 2018). First, because the independent variables included in the model could highly correlate with each other, we conducted an index of tolerance and the measure of variance inflation factors (VIF) to detect collinearity.…”

Section: Resultsmentioning

confidence: 99%

The Mediating Effect of Coping Strategies on Religious/Spiritual Struggles and Life Satisfaction

Szczęśniak¹,

Kroplewski²,

Szałachowski³

2020

Religions

View full text Add to dashboard Cite

A growing number of researchers are testing potential problematic forms of religiousness that denote anxieties regarding sacred matters. However, only a few studies have assessed how religious/spiritual struggle is associated with positive outcomes. Because people’s coping responses to stressors are key determinants of their well-being, we expected that different coping strategies could be potential mediators between religious problems/tensions and life satisfaction. The research was conducted on a group of 744 Roman Catholics. We used the Religious Comfort and Strain Scale, Satisfaction with Life Scale, and Brief Coping Orientation to Problems Experienced (COPE) questionnaire. The outcomes show that religious comfort correlated positively with life satisfaction, while fear/guilt, negative emotions toward God, and negative social interactions surrounding religion correlated negatively with life satisfaction. Our research amplifies the understanding of the religious/spiritual struggles and life satisfaction relationship, mediated by “secular” coping strategies. It confirms that both religious and secular methods of experiencing different strains seem to coexist with multiple other strategies in the context of broadly understood life satisfaction.

show abstract

Section: Resultsmentioning

confidence: 99%

The Mediating Effect of Coping Strategies on Religious/Spiritual Struggles and Life Satisfaction

Szczęśniak¹,

Kroplewski²,

Szałachowski³

2020

Religions

View full text Add to dashboard Cite

show abstract

“…Equation 1 showed the linear model specification with dummy variables for the demand function. Equation 2 showed the double log model specification derived from the Cobb-Douglas function (exponential) (Gujarati and Porter 2009) with a logarithm transformation, including dummy variables. !…”

Section: Model Specificationmentioning

confidence: 99%

“…The correlation was then conducted to determine which explanatory variables are correlated. A correction was made by omitting one of the closely correlated variables (Gujarati and Porter 2009).…”

Section: Specification Testingmentioning

confidence: 99%

Demand Analysis of Termite Control Service in Jakarta

Rahman

Nandika

Simangunsong

2020

JSL

View full text Add to dashboard Cite

Termite is an economically important pest species in the pest control industry and considered as one of the urban pests. Although it had caused a great loss, only a few studies on termite control demand were found. This study attempted to identify determinants and build the econometric model of termite control demand in Jakarta. The findings are expected to give the pest control industry a better understanding of the pest control market. Two ad-hoc models, linear and double log models, were investigated using the Least Square Dummy Variable (LSDV) technique. The results showed that the double log model was found better than the linear model based on sign expectation and significance. The price of termite control service, building permits, price of structural metals and dummy variables for the large company were statistically significant determinants, whereas dummy variables for risk class were not. Termite control demand for medium and small companies was not significantly different, but both of them were significantly lower than demand for the large company. This study also found that the demand for termite control was elastic.Keywords: demand analysis, econometric modeling, pest control, termite control, urban pest

show abstract

“…Multicollinearity test for selected individual-level predictors was done to ensure in ated standard errors due to many predictors measuring the same characteristics are controlled. In this research, the parameters for variance in ation factor (VIF) ≤ 2.5 and tolerance ≥ 0.4 were set as recommended by Johnston et al [41]for the logistic regression model to identify potentially redundant variables due to collinearity.…”

Section: Multilevel Regression Modelmentioning

confidence: 99%

Assessing the Contextual Effect of Community in the Utilization of Postnatal Care Services in Ghana

Dankwah

Zeng

Feng

et al. 2020

Preprint

View full text Add to dashboard Cite

Background Inequalities in the use of postnatal care services (PNC) in Ghana have been linked to poor maternal and neonatal health outcomes. This has ignited a genuine concern that PNC interventions with a focus on influencing solely individual-level risk factors do not achieve the desired results. This study aimed to examine the community-level effect on the utilization of postnatal care services. Specifically, the research explored clusters of non-utilization of PNC services as well as the effect of community-level factors on the utilization of PNC services, with the aim of informing equity-oriented policies and initiatives. Methods The 2014 Ghana Demographic and Health Survey GDHS dataset was used in this study. Two statistical methods were used to analyze the data; spatial scan statistics were used to identify hotspots of non-use of PNC services and second two-level mixed logistic regression modeling was used to determine community-level factors associated with PNC services usage. Results This study found non-use of PNC services to be especially concentrated among communities in the Northern region of Ghana. Also, the analyses revealed that community poverty level, as well as community secondary or higher education level, were significantly associated with the utilization of PNC services, independent of individual-level factors. In fact, this study identified that a woman dwelling in a community with a higher concentration of poor women is less likely to utilize of PNC services than those living in communities with a lower concentration of poor women (Adjusted odds ratio (AOR) = 0.60, 95%CI: 0.44–0.81). Finally, 24.0% of the heterogeneity in PNC services utilization was attributable to unobserved community variability. Conclusion The findings of this study indicate that community-level factors have an influence on women’s health-seeking behavior. Community-level factors should be taken into consideration for planning and resource allocation purposes to reduce maternal health inequities. Also, high-risk communities of non-use of obstetric services were identified in this study which highlights the need to formulate community-specific strategies that can substantially shift post-natal use in a direction leading to universal coverage.

show abstract

Confounding and collinearity in regression analysis: a cautionary tale and an alternative procedure, illustrated by studies of British voting behaviour

Cited by 764 publications

References 26 publications

The Mediating Effect of Coping Strategies on Religious/Spiritual Struggles and Life Satisfaction

The Mediating Effect of Coping Strategies on Religious/Spiritual Struggles and Life Satisfaction

Demand Analysis of Termite Control Service in Jakarta

Assessing the Contextual Effect of Community in the Utilization of Postnatal Care Services in Ghana

Contact Info

Product

Resources

About