Using correspondence analysis to map the portrayal of Dutch political parties in newspaper

In this thesis, the portrayal of Dutch political parties by Dutch national newspapers is analyzed. Relevant terms related to politics are found and associated with newspapers, political parties and over time. This is done by finding the frequencies of occurrences in each newspaper, co-occurrences with political parties and occurrences within each time period. Using correspondence analysis (CA) these frequencies are visualized. In addition, using Procrustes rotation the CA solutions are aligned to each other to examine the movements over time. With bootstrapping techniques, stability measures are calculated and confidence ellipses found to be included in the visualization. The extracted relevant terms are found to be associated differently with each political party. The portrayal shows that a division exists between coalition and opposition parties, both for co-occurrences of terms and within newspapers. Moreover, a division is found between financial and green, humanistic themes among which movements of parties are shown.