Loneliness and poor social connections significantly affect health, making reliable measurement essential. This study assessed a 13-item survey, combining the UCLA Loneliness Scale, ONS Direct Measure, social capital questions, COVID-19 isolation items and demographic items, in a cross-sectional online sample of 134,164 valid responses of NHS patients in England. Rasch analysis evaluated item calibration, model fit, reliability and differential item functioning. Findings confirmed unidimensionality and a clear hierarchy of items measures, with some redundancy in the scale among loneliness items. The survey demonstrated strong psychometric properties, and Wright maps highlighted clusters and gaps in item distribution. These results validate the questionnaire as a reliable tool for measuring loneliness and social connection and offer a framework for refining public health interventions and guiding future research across diverse populations.