Enhancing estimation methods for integrating probability and nonprobability survey samples with machine-learning techniques. An application to a Survey on the impact of the COVID-19 pandemic in Spain.

Web surveys have replaced Face-to-Face and computer assisted telephone interviewing (CATI) as the main mode of data collection in most countries. This trend was reinforced as a consequence of COVID-19 pandemic-related restrictions. However, this mode still faces significant limitations in obtaining probability-based samples of the general population. For this reason, most web surveys rely on nonprobability survey designs. Whereas probability-based designs continue to be the gold standard in survey sampling, nonprobability web surveys may still prove useful in some situations. For instance, when small subpopulations are the group under study and probability sampling is unlikely to meet sample size requirements, complementing a small probability sample with a larger nonprobability one may improve the efficiency of the estimates. Nonprobability samples may also be designed as a mean for compensating for known biases in probability-based web survey samples by purposely targeting respondent profiles that tend to be underrepresented in these surveys. This is the case in the Survey on the impact of the COVID-19 pandemic in Spain (ESPACOV) that motivates this paper. In this paper, we propose a methodology for combining probability and nonprobability web-based survey samples with the help of machine-learning techniques. We then assess the efficiency of the resulting estimates by comparing them with other strategies that have been used before. Our simulation study and the application of the proposed estimation method to the second wave of the ESPACOV Survey allow us to conclude that this is the best option for reducing the biases observed in our data.

Development and validation of a tool to measure collaborative practice between community pharmacists and physicians from the perspective of community pharmacists: the professional collaborative practice tool.

BACKGROUND: Collaborative practice between community pharmacists and physicians is becoming increasingly common. Although tools and models to explore collaborative practice between both health care professionals have been developed, very few have been validated for their use in clinical practice. The objective of this study was to develop and validate a tool for measuring collaborative practice between community pharmacists and physicians from the perspective of community pharmacists. METHODS: The DeVellis method was used to develop and validate the Professional Collaborative Practice Tool. A pool of 40 items with Likert frequency scales was generated based on previous literature and expert opinion. This study was undertaken in Spain. A sample of community pharmacists providing medication reviews with follow-up and a random sample of pharmacists providing usual care were invited to participate. Exploratory and confirmatory factor analysis was used to assess the tool's reliability and content validity. RESULTS: Three hundred thirty-six pharmacists were invited with an overall response rate of 84.8%. The initial 40 items selected were reduced to 14 items. Exploratory Factor Analysis provided a 3-factor solution explaining 62% of the variance. Confirmatory Factor Analysis confirmed the three factors "Activation for collaborative professional practice," the "Integration in collaborative professional practice," and the "Professional acceptance in collaborative professional practice." The tool demonstrated an adequate fit (X2/df = 1.657, GFI = 0.889 and RMSEA = 0.069) and good internal consistency (Cronbach's alpha = 0.924). CONCLUSIONS: The Professional Collaborative Practice Tool has shown good internal reliability and criterion validity. The tool could be used to measure the perceived level of collaborative practice between community pharmacists and physicians and monitor changes over time. Its applicability and transferability to other settings should be evaluated.

Propensity score adjustment using machine learning classification algorithms to control selection bias in online surveys.

Modern survey methods may be subject to non-observable bias, from various sources. Among online surveys, for example, selection bias is prevalent, due to the sampling mechanism commonly used, whereby participants self-select from a subgroup whose characteristics differ from those of the target population. Several techniques have been proposed to tackle this issue. One such is Propensity Score Adjustment (PSA), which is widely used and has been analysed in various studies. The usual method of estimating the propensity score is logistic regression, which requires a reference probability sample in addition to the online nonprobability sample. The predicted propensities can be used for reweighting using various estimators. However, in the online survey context, there are alternatives that might outperform logistic regression regarding propensity estimation. The aim of the present study is to determine the efficiency of some of these alternatives, involving Machine Learning (ML) classification algorithms. PSA is applied in two simulation scenarios, representing situations commonly found in online surveys, using logistic regression and ML models for propensity estimation. The results obtained show that ML algorithms remove selection bias more effectively than logistic regression when used for PSA, but that their efficacy depends largely on the selection mechanism employed and the dimensionality of the data.

Temperament and characteristics related to nomophobia.

Nomophobia is defined as the fear of being out of mobile phone contact and is considered to be a phobia of the modern age. The current study set out to establish the relationship between temperament and personality and the development of nomophobia. The sample was composed of 968 participants selected from the Andalusian population, of which there were 182 males and 785 females aged from 23.19 years. The instruments used were the Questionnaire to Assess Nomophobia (QANIP; Olivencia-Carrión et al., 2018) and the Temperament and Character Inventory Revised (TCI-R; Cloninger et al., 1993). We found that cooperation is a characteristic that significantly reduces nomophobic levels, particularly for the two factors of Mobile Phone Addiction and Negative Consequences. Furthermore, Reward Dependence appears to be positively related to two of the factors involved in nomophobia, namely Mobile Phone Addiction and Loss of Control, suggesting a relationship between Nomophobia and personality. These findings are discussed in terms of their usefulness for identifying the personality predictors of nomophobia in order to develop preventive and intervention strategies.

[Comparability of the editions 2006/07 and 2011/12 of the Spanish Health Survey.] / Comparabilidad de las ediciones 2006/07 y 2011/12 de la Encuesta Nacional de Salud de España.

OBJECTIVE: The Spanish Health Survey (SHS) is one of the main sources on social determinants of health. The objectives were to describe the data merging process of the 2006/07 and 2011/12 SHS editions, as well as to provide the merged databases and the directory of the variables. METHODS: The comparable variables were identified from the questionnaires. Other new variables were created to achieve comparability or to provide new indicators of interest. We obtained two completed merged databases, as well as two variable directories with their dictionaries (each one for the adults and children samples). RESULTS: 188 variables were identified as potentially comparable ones. 36,7% of them had difficulties for being comparable and 8% had to be removed. The topic with the worst consequences due to the lack of comparability was the working conditions and employment from the adult sample. CONCLUSIONS: The merging of both SHS editions had a lot of problems, especially in the adult's sample. That affects to its comparability. This work may be useful to assess, design and merge these and other editions of the SHS, as well as to serve as a model to be applied in other cross-sectional population-based surveys.

OBJETIVO: La Encuesta Nacional de Salud de España (ENSE) es una de las fuentes principales sobre determinantes sociales de la salud. Los objetivos de este trabajo fueron describir el procedimiento de fusión de las ediciones 2006/07 y 2011/12 de la ENSE, y proporcionar acceso libre a las bases de datos y directorio de variables. METODOS: Identificamos las variables comparables a partir de los cuestionarios, construyendo otras nuevas para conseguir la comparabilidad o para proporcionar nuevos indicadores de interés. Obtuvimos dos bases de datos fusionadas y dos directorios de variables con sus diccionarios (muestra de adultos y de menores). RESULTADOS: Identificamos 188 variables potencialmente comparables. Un 36,7% tuvieron algún problema para su comparabilidad y un 8% tuvieron que ser eliminadas, fundamentalmente en la muestra de adultos. El ámbito con peores consecuencias de la falta de comparabilidad fue el de condiciones de trabajo y empleo en la muestra de adultos. CONCLUSIONES: La fusión de ambas ediciones de la ENSE tuvo muchas incidencias en la muestra de adultos, afectando de manera importante a su comparabilidad. Este trabajo es útil para evaluar, diseñar y fusionar estas y otras ediciones de la ENSE, e incluso sirve como modelo para la fusión de otras encuestas transversales de base poblacional.

Comparabilidad de las ediciones 2006/07 y 2011/12 de la Encuesta Nacional de Salud en España / Comparability of the editions 2006/07 and 2011/12 of the Spanish health survey

Fundamentos: La Encuesta Nacional de Salud de España (ENSE) es una de las fuentes principales sobre determinantes sociales de la salud. Los objetivos de este trabajo fueron describir el procedimiento de fusión de las ediciones 2006/07 y 2011/12 de la ENSE, y proporcionar acceso libre a las bases de datos y directorio de variables. Métodos: Identificamos las variables comparables a partir de los cuestionarios, construyendo otras nuevas para conseguir la comparabilidad o para proporcionar nuevos indicadores de interés. Obtuvimos dos bases de datos fusionadas y dos directorios de variables con sus diccionarios (muestra de adultos y de menores). Resultados: Identificamos 188 variables potencialmente comparables. Un 36,7% tuvieron algún problema para su comparabilidad y un 8% tuvieron que ser eliminadas, fundamentalmente en la muestra de adultos. El ámbito con peores consecuencias de la falta de comparabilidad fue el de condiciones de trabajo y empleo en la muestra de adultos. Conclusión: La fusión de ambas ediciones de la ENSE tuvo muchas incidencias en la muestra de adultos, afectando de manera importante a su comparabilidad. Este trabajo es útil para evaluar, diseñar y fusionar estas y otras ediciones de la ENSE, e incluso sirve como modelo para la fusión de otras encuestas transversales de base poblacional (AU)

Background: The Spanish Health Survey (SHS) is one of the main sources on social determinants of health. The objectives were to describe the data merging process of the 2006/07 and 2011/12 SHS editions, as well as to provide the merged databases and the directory of the variables. Methods: The comparable variables were identified from the questionnaires. Other new variables were created to achieve comparability or to provide new indicators of interest. We obtained two completed merged databases, as well as two variable directories with their dictionaries (each one for the adults and children samples). Results: 188 variables were identified as potentially comparable ones. 36,7% of them had difficulties for being comparable and 8% had to be removed. The topic with the worst consequences due to the lack of comparability was the working conditions and employment from the adult sample. Conclusion: The merging of both SHS editions had a lot of problems, especially in the adult’s sample. That affects to its comparability. This work may be useful to assess, design and merge these and other editions of the SHS, as well as to serve as a model to be applied in other cross-sectional population-based surveys (AU)