Methods and apparatus to generate corrected online audience measurement data are disclosed. An example apparatus includes an impressions collector to receive a first audience count indicative of first impressions attributed to a first media category for a demographic group, a second audience count indicative of second impressions attributed to a second media category accessed by the demographic group, and a third audience count indicative of third impressions attributed to the first media category and the second media category accessed by the demographic group, the first audience count including a duplicated audience count; a factor generator to calculate a deduplication factor using the first audience count, the second audience count, and the third audience count; an audience calculator to determine a corrected audience count of the first media category to correct the duplicated audience count; and a ratings data generator to generate ratings information for the first media category.