To get the percentage we computed seats lost / initial seats for both chambers to get individual seats lost percentages. We then average the percentages together to get a single number, essentially weighting the seats (Senate vs. House) by their relative individual significance.
Congressional elections are often counter-cyclical, and a recent question that has arisen is how out of line the Republican gain in the legislature and at the state level with historical norms.
The theory is that while a presidential election may have coattails, they are more than offset but subsequent mid-terms, and higher mobilization of the opposition in the end of the president's term. So we will be measuring the difference between the number of seats held in January 2001 vs. number of seats held in January 2009. It is worth noting that this does not give us the greatest difference, since early vacancies generally swing towards the President's
Starting in 1953 the breakdown was 47-48, favoring Republicans: Senator Morse, elected as a Republican, declared independence and did not caucus with either party. Additionally Hawaii and Alaska were not yet states. On opening day in 1961 the split was 65-35 (favoring Democrats). This represents a 27% loss of seats.
In 1953 the total was 214 Democrats to 221 Republicans (Rep. Reams caucused with Democrats). By 1961 the Democrats dominated 264 to 173, for a loss of 48 seats, or 22%.
Data -- every year of the Senate
Wikipedia