Dec 11, 2012

Polling on taxes – response to a critic


On November 29, the CCPA released an opinion research study led by myself and Randy Galawan called Beyond the 1%: What British Columbians think about taxes, inequality and public services. The study involved an extensive online survey (poll) of a broad sample of British Columbians, conducted by Environics Research, and nine group interviews held in Metro Vancouver, Nanaimo and Kamloops. Our findings suggest that public attitudes towards taxes are more complex than pundits and politicians tend to assume; that high levels of public concern about income inequality have translated into majority support for tax increase on upper-income earners and corporations; and that people are surprisingly open to paying a little more themselves if they know tax increases will lead to concrete improvements in public services and quality of life.

A couple days after the release, political pundit and former NDP MLA David Schreck heard me talking about the study on CKNW’s Sean Leslie show, and took issue with what he heard. Following a string of unfounded criticisms on Twitter, he posted a more measured but nevertheless problematic critique on his website. Most people are unlikely to be interested in this level of back-and-forth about polling methodology, but it’s worth putting a response on the record.

Willingness to pay more income tax

Schreck’s main criticism of our survey relates to a question about people’s own willingness to pay more income tax. The question asked respondents if they would consider paying a slightly higher share of their own income in provincial income tax, to achieve a series of 11 different policy changes. Respondents could indicate if they were willing to pay 0.5%, 1%, 2% or 3% more of their income to support each objective — and were also given an option for “I am not willing to pay any more tax at all for this.” Along with the question, respondents were presented with a table showing approximately what these % increases would mean in dollar terms for someone with $40,000/year of income, and someone with $80,000/year of income. Each policy objective (such as “invest in an affordable housing strategy” and “provide more access to home and community based health care services for seniors”) was presented one at a time, in randomized order. You can see a screen shot of this question from the actual survey here.

When we analyzed the results, we grouped people’s responses for each of the 11 policy objectives into two categories — willing to pay (whether someone chose 0.5%, 1%, 2% or 3% of income) and not willing to pay. We also looked at how many of the 11 objectives people were willing to pay for — and treated this as an overall indication of their willingness to pay higher tax. For example, we calculated how many people were willing to pay for between 1 and 3 of the objectives, between 4 and 7, between 8 and 10, for all 11, and for none of them. We also calculated the average number of policies people were willing to pay for.

Schreck takes issue with both the structure of the question and our interpretation of the results, using some problematic logic.

On the structure of the question, he argues that we should not have asked whether people are willing to pay a higher share of their income to taxes, but rather whether they’d be willing to pay the equivalent increase in their provincial income tax rate. He notes,

A $400 per year tax increase for a person with $80,000 in gross income [the 0.5% increase in share of income example from the poll question] is an 8% tax increase for a person with no dependents, and a 9% increase for a person with one or two dependents. A spread of 8% to 14% in tax hikes sounds a lot different than 0.5% of income. The CCPA example of paying 3.0% more of income translates to a 60% tax hike for a single person earning $40,000 per year.

Well, yes — that’s exactly the point actually. Debates about income taxes in BC get completely skewed when people toss around these huge-sounding %-change figures. The reality is that the vast majority of us pay less than 4% of our income in provincial income tax. A 25% increase in the provincial tax rate sounds huge – but would amount to 1% of income or less for most people. And 1% of income for most people isn’t a huge dollar figure. That is why we opted to ask people whether they’d pay a slightly higher share of income to taxes, and gave them estimates for just how much money that means at different income levels. We deliberately sought to make these potential increases concrete and meaningful for the poll respondents.

I’m not sure how much more honest and transparent you can get. What Schreck is advocating — putting huge % increases in the provincial income tax rate in front of people without explaining how that computes in actual dollar terms — is misleading. Unless it’s for a poll of tax accountants — who probably carry around a little more information about effective tax rates in their heads than the rest of us do.

On the interpretation side, Schreck goes on to argue that because a 0.5% increase in income amounts to a relatively small dollar figure for most people, it’s not legitimate to ‘count’ someone who is willing to pay that much as “willing to pay” more taxes. In fact, he thinks they should be lumped in with the people who said they were not willing to pay any more tax, and recalculates the results (using the detailed results table we made available on our website) using a new category he calls “willing to pay little or nothing.”

Setting aside the fact that this is an irresponsible interpretation of the results, it is based on contradictory logic. First Schreck argues that a 0.5% increase in income represents a huge tax increase and therefore our question somehow tricks people into saying they’d pay a higher tax rate. Then he says 0.5% of income is so negligible it is effectively zero. It can’t be both.

Indeed, the point coming out of all this is a pretty simple one: Effective provincial income tax rates for most people are very low, less than 4%. A relatively small increase in the share of our incomes going to income tax doesn’t add up to a huge amount of money for most people — a few hundred dollars per year, less than a cup of coffee per day (Timmies, not Starbucks). But pooled together, this represents a lot of money. Indeed, a 20% increase in provincial income tax rates — again, sounds huge, but for most people, we’re still talking a few hundred dollars per year or less — would raise almost $1.8 billion in new provincial revenues. That’s enough money to fund, for example:

  • 2,000 new units of social housing per year; and,
  • an increase to welfare rates to help people meet basic food and shelter needs; and,
  • reduced class sizes in K-12 and more specialist teachers; AND
  • the first phase of a $10/day child care program.

A little can go a long way, when we pool our resources.

Transparency about online polling methodology

In his initial reactions to the poll on Twitter, Schreck claimed our survey methodology was weak, citing a ‘qualification’ we included in our publication of the results, which indicated that the survey did not use a random probability sample. He pointed to Vision Critical’s online surveys as a better approach. This reflects a lack of knowledge about how online polls work (and a failure to look closely at Vision Critical’s own reporting of results).

No online poll can be conducted with a truly random probability sample for the entire population. It is literally impossible — there is no central directory listing every British Columbians’ name and their various email addresses (never mind that some people don’t use email), from which one could extract a random selection of people to participate in the survey. Instead, polling firms construct a “panel” of people who agree to participate in online surveys (typically these panels are very large, with tens of thousands of people), and then invite a random sample of that panel to participate in any given poll (for example, I signed up to be in Angus Reid’s online panel, and I get periodic emails inviting me to fill out surveys about all manner of things).

As a result, while it is possible to calculate a margin of error (meaning, the measure of confidence that if the poll were conducted again, we’d get exactly the same results) in relation to the polling firm’s online panel members, it is not possible to do so for the broader population the survey aims to learn about (in this case, British Columbians). This is different from the traditional telephone poll, where random-digit dialing allows for a probability sample, and thus reporting of a margin of error. Go on the website of any major polling firm that uses online surveys, and you’ll find they all use some variation of the ‘qualification’ statement we included with our results (that goes for Vision Critical too, incidentally).

There’s been lots of debate in recent years about the relative merits of online vs telephone polls, and the limitations of relying on phone polls in a world of cell phones and caller ID, etc. My sense at this point is that most polling firms would likely say that online polls are at least as accurate as phone polls.

We chose to use an online poll for three key reasons: first, you can ask people a lot more questions in a shorter amount of time; second, people can digest more complex information and ideas when they can see the words in front of them on the screen, compared to hearing them read out over the phone; and third, online polls allow you to use interactive tools, such as the income pie charts we used to ask about estimated and ideal income distribution.

We set demographic quotas to ensure the respondents matched the broader demographic profile of BC in terms of age, gender, and region.

What is politically possible

Schreck also claims that we don’t understand politics, and haven’t learned from the HST experience. It’s not entirely clear what he thinks the HST lesson is (unless perhaps it’s simply that everyone should shut up about tax policy, period). Do our results mean that any politician or political party who sets out to increase taxes, no matter what the proposal, would get an easy ride. Um, no. That’s not what we’re saying.

What we are saying is that most British Columbians want upper income earners and corporations to pay a fair share — 57% of survey respondents want to see people with incomes over $100,000/year pay more, and 67% say corporations aren’t paying enough. We’re also saying that when tax proposals are presented in a clear and transparent way, and are linked to concrete policy outcomes, British Columbians are remarkably open to the possibility of paying a little more tax themselves.

Too many politicians and pundits talk about taxes in a polarizing way, reducing any public discussion about tax policy to a “taxes good vs taxes bad” dichotomy. This is unhelpful. Our research suggests that British Columbians are open to a more thoughtful, democratic conversation. I for one would like us to focus on more useful questions. Like what revenues do we need to tackle challenges such as inequality, the affordability crisis in housing, child care, and seniors care, and climate change. And how do we ensure everyone — rich, poor and in between — has a role in dealing with these challenges, by paying a fair share of the costs.

Elevate the debate

On that note, I must say that what I found distressing about David Schreck’s reaction to our work has nothing to do with the substantive critiques dealt with above, but rather how he conducted himself in delivering them online. Unfortunately, a lot of people seem to bring their worst selves to social media, which in the context of policy and politics can be corrosive to respectful democratic debate. David’s decision to cast insults on Twitter, such as repeatedly calling the research “pathetic” and “propaganda,” is exactly the kind of inflammatory, hyper-partisan behaviour that turns so many people off politics, and by extension, from participating in democratic processes. Let’s just not go there.

Topics: ,