Just weeks before Wisconsin's June 5 recall election, the banner headline for the Sunday edition of the Wisconsin State Journal declared "Campaign donations: Despite rhetoric, the parties' mountains of money are about even [3]," a puzzling title because all evidence showed Governor Scott Walker with a significant financial advantage over challenger Tom Barrett. Former University of Wisconsin Professor Kathy Barton looked at the numbers used in the analysis and found numerous errors that caused donations to be overstated by an estimated $13 million.

Major Differences in Fundraising, Spending

According to initial estimates, an astonishing $63.5 million was spent on the recall election, and $45 million of that sum -- more than 70 percent -- came from Walker's campaign and supporters. Walker raised seven-and-a-half times as much money as Barrett ($30.5 million to $4 million at last count). In light of these facts, claiming "the parties' mountains of money are about even" in an article published just two weeks before the gubernatorial election left readers with serious misconceptions, the Center for Media and Democracy (CMD) first reported [4].

CMD noted on May 24 that the State Journal analysis was incomplete because it only accounted for contributions disclosed to Wisconsin's Government Accountability Board (GAB), despite many groups refusing to make those disclosures because they claim to be running so-called "issue ads." The State Journal analysis entirely ignored the role of secretly-funded special groups like the Koch-funded Americans for Prosperity spending tens of millions of dollars to support Walker's reelection. Most of the secret money not reported to the GAB went to benefit Republicans, and the State Journal's failure to account for this spending significantly under-counted Walker's financial support. The article also lumped contributions made for last summer's Senate recall elections with money in the gubernatorial race, which did not present an accurate picture of the relative financial support for Barrett and Walker.

Dr. Barton, a scientist, identified more fundamental errors.

Erroneous Data and Analysis

The figures the State Journal had used in its analysis were not published, and Dr. Barton's review of the only data available -- a list of top 50 donors -- showed instances of the same person being counted twice. She asked the State Journal for the authors' analysis of political contribution data so she could replicate it. "As a scientist, I expect that when anyone does an analysis, their methods, spreadsheets, and data will be available to check on their accuracy and methodology," Dr. Barton told CMD. "If you've analyzed something, others should be able to reproduce your analysis."

After days of emails back-and-forth between Dr. Barton and the State Journal, the newspaper released a Microsoft Excel file containing raw data -- a list of political contributions, sorted by date -- but not their analysis, as requested. She asked for a description of their procedure, so she could replicate the analysis with the raw data, but was told in an email "I'm not going to walk you through how to do all that."

"I finally got tired of waiting for the analysis and dug into the data on my own," said Ms. Barton, who is a Staff Member at the Carnegie Institution for Science's Department of Plant Biology. "I quickly found that the file contained potential duplicates and alerted them to this. They ignored me until I emailed them showing that these duplicates caused greater than $1 million dollar discrepancies. They wrote and told me they would issue a correction the day after I told them I calculated there were likely to be over 60,000 duplicates with an estimated effect of $13 million dollars, that these duplicates were not present in the official GAB reports, and that there was evidence that these were not randomly distributed."

Dr. Barton wrote the following letter to the Wisconsin State Journal, which they have not yet published:

Numbers dominated the Barrett and Walker campaigns. Numbers on jobs, budgets and campaign donations were tossed around by both sides. It was tough to know whom to believe. That's why it's so important that voters can count on transparent and accurate analysis of numbers by news media.

Recently I found that the numbers the Wisconsin State Journal published in a campaign finance article were wrong ("Campaign donations: Despite rhetoric, the parties' mountains of money are about even"). I estimated over 60,000 duplicate entries, some in the million-dollar range, caused the article to report $13 million in donations that never existed. After I alerted them to this, the WSJ published a correction and a new set of numbers. The WSJ does not know how the duplicate errors arose.

So, are the new numbers right? That's impossible to know: WSJ has not made the revised analysis available. And quite frankly, I don't have the energy or time to investigate this further. It wasn't easy getting WSJ to send me information on the original article. My request got me three quarter million entries of raw data -- but no analysis. I ultimately had to crunch this data myself to try to confirm the article's findings. That's when I found the mistakes.

WSJ champions transparency in their quest for information from government agencies. WSJ should adopt an equally aggressive transparency policy for their work. Articles should provide links to analysis and raw data. Without this, WSJ's numbers are no more believable than the campaigns'.

As Dr. Barton noted in her letter, the State Journal issued a correction and a new set of numbers on June 1, which can be viewed here. [5] Because the revised analysis that led to these numbers is not available, Dr. Barton says she does not know whether the updated numbers are accurate.

Mountains of Money Still Not Even

The State Journal also corrected a quote from Mike McCabe, executive director of the Wisconsin Democracy Campaign, a nonpartisan group that tracks money in politics. The quote discussing spending on so-called "issue ads" and changed from:

"About $50 million has been spent on such advertising since recall fever started last summer, with spending fairly evenly split between the parties,"

to

"About $27 million has been spent on such advertising since recall fever started last summer, with about $22 million spent in support of Republicans and $5 million to help Democrats."

"I'd say that's quite a change," notes Barton. The State Journal explained the discrepancy by saying "Statements made by Mike McCabe have also been updated, after he said he misunderstood a question from a reporter."

Despite the State Journal identifying serious errors in its data analysis and correcting a quote that significantly alters the balance of money between Democrats and Republicans, the article's title has not been corrected -- despite all evidence to the contrary, the article still claims "the parties' mountains of money are about even."

The Wisconsin State Journal did not endorse a candidate in the recall.