The recent revelation that Secretary of Health and Human Services nominee Kathleen Sebelius has paid back taxes for mistakes in her past three years of tax returns was a painful reminder to many that Obama’s transition to power has been less than smooth, as he continues to seek appointments for many top-level cabinet positions. To look back on the two most controversial Obama appointments:

Former Senate Majority Leader Tom Daschle, nominated for the same position as Sebelius, withdrew his name from consideration after it was revealed that he owed over $100,000 in back taxes from consulting work and the use of a driving service. The incident prompted Obama to proclaim that he “screwed up”, saying “I’ve got to own up to my mistake. Ultimately, it’s important for this administration to send a message that there aren’t two sets of rules — you know, one for prominent people and one for ordinary folks who have to pay their taxes.”

U.S. Treasury Secretary Timothy Geithner faced allegations of tax fraud during the confirmation process over failure to pay Medicare and Social Security taxes while working at the IMF. He paid over $25,000 in back taxes and was confirmed by a vote of 60-34. Compare this to some past Treasury Secretaries:

• W. Michael Blumenthal (1977): confirmed by voice vote
• G. William Miller (1979): 97-1
• Donald Regan (1981): 98-0
• James Baker III (1985): 95-0
• Nicholas F. Brady (1988): 92-2
• Lloyd Bentsen (1993): confirmed by voice vote
• Robert Rubin (1995): 99-0
• Lawrence Summers (1999): 97-2
• Paul O’Neill (2001): confirmed by voice vote
• John W. Snow (2003): confirmed unanimously
• Henry Paulson (2006): confirmed by voice vote
• Timothy Geithner (2009): 60-34

Given the calls for Geithner’s resignation (albeit from fringe sources), I decided to take a look back in history to see how a rocky confirmation process impacts a cabinet member’s incumbency - namely, the rate of resignation as a function of the vote count in the confirmation process. To do this, I first collected a spreadsheet of data for most of the major cabinet appointments since the Carter administration using this table. A few initial notes about my methodology:

• The “vote count” is listed as a percentage of yea votes of total votes.
• Confirmation viva voce counts as unanimous confirmation. They are not entirely the same - arguably a voice vote indicates greater consensus then a vote at all, and this may have slightly impacted my findings for the most controversial cabinet positions. However, the general idea is to look at opposition to nominees, not the process by which they were confirmed.
• Resignation means leaving office before the end of the incumbent president’s presidency - so leaving in between terms still counts as resignation. (This Washington Profile article explains why this may be a political tactic)
• Nominees that were withdrawn before or during confirmation hearings are listed as zeros for vote count and missing variables for resignation.
• I counted John Tower, who was the only nominee during this time period to be disapproved by the Senate, as having withdrawn as well. This is to attempt to correct for the fact that his rejection represented a period of controversy at the Department of Defense greater than that of a simple rejected.
• I’ll preface all of this by saying that even the concept of using statistics to predict resignations is highly imperfect, because obviously much depends on the person rather than the trends.

On to the data… I first created a graph that looked something like this.

That is a little hard to interpret, so I abandoned the idea of plotting vote counts for all departments at once and instead broke it down into average vote count by year and by department. I’ll start with average vote count by year. This first chart lists the vote percentage described earlier, with the first column indicating the actual average percentage, the second column listing the first column to the second power, the third column listing the first column to the third power, and the fourth column listing the first column to the fourth power. The rationale behind doing this is that because the percentages are so high since so many nominees are confirmed unanimously, a higher power operation highlights difference. I’ll call the fourth power column the “standard quartic” for each set of data.

Still difficult to interpret? It’s because some rows have much less data than others. To simplify matters, I redid the chart using only rows that represented the beginning of new terms, when most cabinet appointments take place, and then graphed that data with a trendline. (Note that it is not necessary to graph the standard quartic - it visually appears the same as the linear data.)

The trend should become clear - over time, cabinet confirmation votes have become more and more split. However, the vote counts are still quite high in the first column. Bush was around 90% for both of his terms despite some very controversial appointments, and Obama is only down to around 80%. What this table doesn’t take into account is that some cabinet positions tend to have more contested votes than others. For example, as you’ll see in the chart below, no Senator has voted against the confirmation of a Secretary of Housing & Urban Development or Transportation during this time period. I’ll demonstrate this data using first a chart with the same column scheme as described earlier and then a graph of the standard quartic.

It should be clear that certain departments are much more controversial than others - notably, the Attorney General appointment. Some data points are anomalies - for example, the Veterans’ Affairs data suffers from one withdrawal in a small field since the department is new. Looking back to Geithner, notice that this data confirms the idea that Secretary of Treasury appointments aren’t that controversial - notice how the department ranks about par for the course, and that includes Geithner’s vote count.

Moving on to resignation rates, I’ll first present a bunch of data at once and then explain.

The first two rows represent average vote counts based on whether or not the cabinet member resigned in office, with a “1″ indicating resignation and a “0″ indicating no resignation. The columns are presented in standard linear, square, cubic, and quartic form. This data indicates that vote counts for cabinet members who later resigned were slightly lower, but this data is probably not very significant since 1) it includes a lot of cabinet members who resigned for reasons other than controversy - were promoted, sought other positions, etc, and 2) the difference is marginal - even the standard quartic values are close.

This chart represents the average rate of resignation and the number of resignations per term for each listed year, which corresponds to the inauguration of a new President. This makes more sense than simply taking resignations per year as described above.

This data, perhaps most significant, shows two things: the rate of resignation for cabinet members using a threshold for initial vote counts. I chose .96 as an arbitrary value because it roughly represents more than around 2 Senators voting against confirmation. G represents Geithner’s vote count. You can clearly see that the rate of resignation for cabinet members with initial vote counts below Geithner’s skyrockets from 53% to 75%.

However, it’s important to look at this in context. Two major things:

• 1) This should not be interpreted to mean that Geithner has a 75% chance of resigning. The resignation rate is obviously subject to a host of many different factors (see again the Washington Profile article I linked to before for more explanation)
• 2) There aren’t that many appointees who had lower initial vote counts than Geithner.

I’m going to follow up this article with one describing the most controversial confirmation processes as ranked by this data. I won’t spoil things too much, but Geithner makes the top ten - so the sample size for below G is small.

Conclusions

1. Obama’s cabinet members faced a historically controversial confirmation process. Not only was Obama’s average lower, he significantly lowered the averages for several departments that usually didn’t face tough confirmation processes, most notable the Treasury and Commerce departments. While some of this can be explained to increased politicization of cabinet confirmation votes over time (see the trendline graph above), Obama’s data point is still far below the trend.

2. Geithner is unlikely to survive all of Obama’s presidency. This is especially true if Obama serves two terms, and if you look at the generally higher rate of turnover for Treasury secretaries.

3. Resignation is only barely correlated to initial vote count, except in extreme cases (like Geithner’s). Examples of anomalies include controversial appointments Jim Nussle and Michael Mukasey, who both served out their terms.

You can download the whole spreadsheet here.

Stay tuned for the top 10 most controversial cabinet appointments.