THE PERCEPTION OF TWITTER CREDIBILITY AND YOUNG ADULTS’ LOYALTY IN SHAH ALAM

Social media has become an important channel for political information and discussion among young adults in Malaysia. This study examines perceptions of Twitter credibility and their relationship with loyalty among young adults in Shah Alam, Selangor. Drawing on the Source Credibility Model, the study considers perceived expertise, trustworthiness, reputation and promotional incentives as dimensions associated with Twitter credibility, alongside perceived loyalty and information sharing. A quantitative, cross-sectional questionnaire was distributed online to young adults aged 18–25 in Shah Alam. The study reports a sample of 382 respondents drawn from a population estimate of 74,366 young adults. Responses were collected using a five-point Likert-scale instrument organised around seven constructs. The analysis examined respondents’ demographic characteristics, descriptive measures and relationships between the constructs. The reported correlations between credibility and its proposed antecedents were positive: perceived expertise (0.56), trustworthiness (0.65), reputation (0.60) and promotional incentives (0.61). Twitter credibility was also positively associated with perceived loyalty (0.59) and verified information sharing (0.61). These findings suggest that trustworthiness and the reputation of information sources are relevant to how young adults assess political content on Twitter, while perceived credibility is related to continued engagement and information-sharing practices. The study contributes to discussions of social-media credibility and political information use in a Malaysian urban setting. It also highlights the importance of helping young users evaluate source reliability, distinguish account verification from factual accuracy and approach political news critically. Given the cross-sectional self-report design and the geographic and age restrictions, the findings should be interpreted as associations rather than causal effects, and their generalisability beyond the sampled respondents remains limited.