Controlling response style in life satisfaction evaluation

• Grant number: 17/13144-1
• Support Opportunities: Scholarships in Brazil - Master
• Start date: September 01, 2017
• End date: December 31, 2017
• Field of knowledge: Humanities - Psychology - Fundamentals and Assessments in Psychology
• Principal Investigator: Cristian Zanon
• Grantee: João Paulo Araújo Lessa
• Host Institution: Universidade São Francisco (USF). Campus Campinas. Campinas , SP, Brazil
• Associated research grant: 15/05741-4 - Local independence evaluation and the use of anchoring-vignettes on the satisfaction with life scale, AP.JP

Abstract

Response style is an old phenomenon described in the literature (Lentz, 1938). While some tend to endorse items, others tend to disagree; Others still respond to Likert scales using predominantly the extremes or central points (He, Bartram, Inceoglu, & Van de Vijver, 2014). All those ways of responding to items introduce a measurement error in the evaluation of the construct of interest and reduce its reliability and validity (Johanson & Osborn, 2004). The situation is no different with the scale with life satisfaction that is composed of positive items which they present considerable content of social desirability. Thus, adequating the evaluation of life satisfaction is essential for the understanding about the construct and also for its implications in public policy interventions, as well as the existence of methods to refine its evaluation are important contributions to psychometry and subjective well-being. No local study has investigated the relevance of response style in assessing life satisfaction yet. Therefore, this project proposes the creation of a procedure to control the style of response to the Satisfaction with Life Scale (SWLS).This method consists in subtracting the scores of each item from the Dissatisfaction with Life Scale (proposed in the scientific initiation work) of the Satisfaction with Life Scale items. The mean of these differences in total items could be considered as a coherence index in the responses. For example, if someone marks 7 (totally agree) on an ESV item that says "If I could live my life again, I would not change almost anything", but mark 7 (I totally agree) on the item to be created from Dissatisfaction with Life Scale, which says "If I could live my life again, I would change many things", it is possible to perceive total incoherence in the answers. This degree of inconsistency could be considered an indicator of response style. Thus, we could subtract the value of each participant's incoherence index by the group mean and divide this difference by the standard deviation of the group's incoherence index. This would result in standardized life satisfaction scores controlling response style. Similar procedure was used in Soto et al. (2008) to control response style in a Big Five Inventory. The usefulness of this procedure could be tested by comparing correlations of life satisfaction, with and without control of response style, with variables that the literature indicates as being associated, for example: personality and health. If response style is a relevant factor in the process of responding to life satisfaction items, it is expected that the control of these styles will reduce the measurement error related to this factor and, therefore, increase the reliability of the test and, consequently, With related variables (eg, personality, health, among others).Method: participants will be 300 adults (18+) who will respond to the Satisfaction with Life Satisfaction, Dissastifaction with Life Scale, Big Five Inventory and the Goldberg General Health Questionnaire. Participants will respond to online scales and will be invited through the dissemination of research on social networks.Data analysis: initially, descriptive statistics and reliability statistics of all scales will be produced. Subsequently, the life satisfaction scores will be calculated with and without acquiescence control. Finally, correlations of life satisfaction scores with and without acquiescence control will be performed with the personality and health factors. Correlation difference (Fisher's z) tests will be used to assess whether correlations with and without response style control differ significantly. (AU)