February 18, 2011

I love doing taxes.  This season, I will have spent every Thursday evening (when I am not travelling for IPA workshops and meetings) volunteering as a preparer at a tax assistance site in Washington DC.  What does this have to do with IPA? Well, as I was taking the certification course it struck me how similar it was to surveyor training.  We sit down with respondents and conduct interviews.  Based on their answers, we can skip some sections of a two page survey and fill out additional forms in other sections.  A reviewer comes by at the end, scrutinizes my work, and makes sure I did not skip any zeros when copying down DC's incredibly long Employer Identification Numbers from W2 forms.  We have auditors re-survey respondents to make sure that the surveyors did not make mistakes or make up responses.  Aside from simply copying a number from a W2 into the "income" section, we make the same judgement calls that we ask of surveyors:

What exactly constitutes a member of a household?  What if they live in college dorm rooms 9 months of the year?  What if they moved out in August 2010?  What if they are foster children?  If your spouse left six months ago, but you aren't officially divorced, do you still have to file as married?

We all have to worry about precision and accuracy in responses.  Dean Karlan's research has found that clients of microfinance tell their loan officers one thing about how they spend loan proceeds, when in fact they might spend the funds on consumption rather than business investments (see his presentation from last October's Microfinance Impact conference here).  Taxpayers may inflate the value of their charitable donations or business deductions.  The cost of auditing tax returns, or of getting the most precise and accurate responses, is sometimes more than the value of this accuracy.
But IPA surveys always have a sort of safety valve, in that we have codes for "don't know" and "refuses to answer" and "Other, specify: _________".  The IRS, on the other hand, has to have a rule that accounts for every single respondent.  Even if their income is from bartering, gambling, working outside the US, babysitting, investments, or hobbies.  This means that tax preparers were handed instructions, a.k.a. "surveyor manuals," that were two feet tall and weighed about as much as a small child. 
Like IPA, the IRS is beginning to transition to electronic surveying.  In fact, there are about a hundred firms competing to produce electronic versions of the interview, and charging respondents anywhere from $10 to $75 to use the web-based version.  But, I don't think the IRS does double-data entry for their paper surveys.
Their results are more profitable.  Ours are more interesting.