In this study, all video content was converted to MPEG-4 format and integrated into an online experiment. Each video clip had a five-second countdown timer and was accompanied by emotion sliders ranging from 0 to 100% intensity, with the question, "How does the person in the video make you feel?" The videos were divided into two blocks: the first block contained ambiguous or neutral facial expressions to establish a baseline, while the second block featured smiles from politicians associated with reward and affiliation. Participants in the study were exclusively from England and intended to vote for one of the three major political parties. To ensure a representative sample, participants came from all 124 postal code zones in England and were recruited through a Qualtrics survey panel, receiving £5 for their participation. A total of 344 participants took part, with an average age of 57 and a gender distribution of 219 males and 125 females. Most participants (58%) intended to vote for the Conservative and Unionist Party, followed by 22% for The Labour Party and 20% for The Liberal Democrat Party. Emotional responses were assessed using eight sliders, combined into four 100-point variables: anger, distress, happiness, and affinity. The Cronbach's α values for these scales ranged from 0.78 to 0.97. Data from participants who responded both before and after the UK General Election were analysed in this study. Key to dataset The variable names refer to affective rating that each voter reported when presented with one of three facial displays that were engendered by each of the political leaders from one of the two data collection timepoints i.e., 1 week before (T1) or 1 one week after (T2) the UK 2019 General election. For example, T1_JSNeutHap is the column header for data that was collected before the election and here participants were shown the video excerpt of the, then, leader of the Liberal Democrats, Joanna Swinson, engendering a neutral or ambiguous facial display. The data in this specific column is the collapsed data for participant rated measures of excitement and happiness which were collapsed to from the single variable of happiness.