Instruments for Measuring the Influence of Visual Persuasion: Validity and Reliability Tests

Abstract

In User Experience (UX) research, the instruments are often measured by means of rating scales such as Likert scale and semantic differential scale. The validity of the findings and conclusions rely heavily on the instruments used in the questionnaires. This paper provides the assessment of the validity and reliability of a new set of measures to evaluate the influence of visual persuasion on web users. The instruments will be used to assess web users' perceptions of credibility, engagement, informativeness, satisfaction, social influences, usability, and visual aesthetic. Firstly, 85 items are pilot tested by expert and novice users in an offline and online settings. Secondly, the exploratory factor analysis is carried out in which 44 items representing 12 latent variables are reduced to 39 items with some of the latent variables are combined into one. The results show: Kaiser-Meyer-Olkin (KMO) of 0.901, significant Bartlett’s test, communalities range between 0.470 - 0.829, nine factors (also known as the latent variables) emerged with eigenvalues greater than 1, explaining more than 60% of the total variance, factor loadings of 0.466 and above, factors correlations of less than 0.7, and Cronbach’s alphas are well above the limit of 0.70. Finally, a confirmatory factor analysis is carried out on the first-order and second-order latent variables using the PLS-SEM. The instruments exceed the minimum requirement of the assessments for the convergent validity, discriminant validity, reliability and collinearity. The findings suggest that the proposed 39 items are valid and reliable for measuring the persuasiveness of visual persuasion.