Content uploaded by John William D'Attoma
Author content
All content in this area was uploaded by John William D'Attoma on Dec 18, 2018
Content may be subject to copyright.
Measuring Budgetary Preferences Using Interactive
Budget Simulations: A Holistic Approach∗
John D’Attoma†
Kim-Lee Tuxhorn‡
Sven Steinmo §
November 10, 2018
Abstract
The literature on budget preferences has become increasingly bifurcated with schol-
ars studying preferences over spending and revenue as separate outcomes of interest.
Certainly, this substantive divide has led to important advancements in explaining spe-
cific questions related to each side of the ledger, but the distinction within the field has
come at a theoretical and empirical cost. In this paper, we argue that a holistic ap-
proach, accounting for spending and revenue preferences, can overcome several existing
issues found in the literature. Yet, a major barrier to studying budget preferences in
this way lies with traditional methods, such as survey questionnaires. We introduce
online interactive budget tools as a flexible, multi-purpose method for scholars to col-
lect attitudinal measures on spending and revenue items. We test whether the holistic
approach elicits different behaviors than the more singular approaches using data from
a 2018 Mechanical Turk survey with an embedded budget tool, and a randomized ex-
periment. We make two important contributions. First, we uncover that participants
reduce spending less when given both size of the budget compared to when they are
given spending separately. On the tax side, there is little variation between treatments,
which aligns well with the theory that Americans have quite stable preferences for tax-
ation. Secondly, we uncover that our budget tool decreases polarization considerably.
Indeed, when given both sides of the budget there are very little differences on taxing
and spending between conservatives and democrats.
∗Paper Prepared for the Annual Conference at Northeastern Political Science Association, Montreal, Nov
8-10 2018. Please request permission from one of the authors before citing.
†University of Exeter Business School, Tax Administration Research Centre (TARC),
j.w.dattoma@exeter.ac.uk
‡University of Calgary, Department of Political Science, kimlee.tuxhorn@ucalgary.ca
§University of Colorado, Boulder, steinmo@colorado.edu
1
A Holistic Approach to Studying Budget Preferences
Introduction
Decisions over government budgets engender political battles. Accordingly, the literature on
mass attitudes toward government budgets is nothing less than immense, spanning a large
disciplinary range, including political science, economics, and sociology. Over the years,
the research on budget preferences has increasingly drifted further apart with some scholars
only focusing on the tax/revenue side of the ledger while others seek to explain spending
preferences. In fact, it may be argued that the scholarship on spending and tax preferences
are so large that they now constitute separate research programs. Certainly, this substantive
divide has led to tremendous advancements in explaining specific questions related to each
side of the ledger, but the distinction within the field has come at a theoretical and empirical
cost.
Specifically, we raise three issues related to what may be called the “singular approach” –
that is focusing on only taxes or spending – to explaining budget preferences (Beck, Rainey
and Traut 1990). Most importantly, few studies have forced respondents to consider the real-
world fiscal trade-offs between taxes and spending. This omission of budgetary constraints
may lead to extreme positions on the issue of taxation and spending items. As (Citrin
1979: p. 114) pointed out early on in the literature “[S]ome surveys elicit opinions without
forcing people to confront, albeit hypothetically, the trade-off between taxing and spending.
When the survey questions properly point to the existence of “price effects” by warning that
reduced taxes mean fewer services or that increased spending would result in higher taxes, the
tendency to demand “something for nothing” declines.” Second, many studies narrowly focus
on specific items within the federal budget such as spending on education or corporate tax
on the revenue side. While such item-specific approaches can be useful to explain preferences
on a particular budgetary line, it is difficult to generalize these preferences to the budget at-
large, and it is unclear what the composition of the budget would be if these preferences were
translated into actual budget items. Third, existing studies on preferences over the budget
and budget items often lack context, giving respondents insufficient information about how
their decisions impact other parts of the budget.
In this paper, we argue that a “holistic approach” – studying spending and revenue prefer-
ences in tandem – offers a solution to these issues. The inherent trade-offs between spending
and revenue are more transparent when respondents are given both sides of the ledger. A
holistic approach accommodates multiple spending and revenue items and gives a broader
context to how individual items are situated in the budget-at-large. Certainly, the call for
a holistic approach is not entirely new, but its adoption has been met with resistance be-
2
A Holistic Approach to Studying Budget Preferences
cause of existing methodological challenges. We present online interactive budget tool as a
solution to these limitations and enable researchers to more effectively match empirics with
theory.
To demonstrate the connection between a holistic approach and our method, we partnered up
with Balancing Act, an organization focusing on public engagement in government budgets,
to build an interactive budget tool based on the general line items of the 2018 federal budget.1
Interest in interactive budget tools has grown in recent years with policymakers and non-
profits using them to educate the public about local and state budgets. We believe that
this is a powerful but underutilized tool for political scientists studying attitudes toward
taxation and spending policies. The tool allows researchers to obtain micro-level data on
individual line items and macro-level data on general trends, such as do conservatives support
increasing taxes on the rich or do liberals support cutting defense spending. Moreover,
future research can utilize budget tools to measure issue salience, generate dichotomous and
continuous measures for specific budgetary items, and determine whether and how people
balance budgets.2
We make two contributions to the literature. We use the data collected from this study to
illustrate that a holistic approach to studying taxes and spending elicits different preferences
than studying taxes and spending separately by leveraging a randomized controlled exper-
iment. Secondly, we examine whether ideological polarization over important budgetary
positions is indeed an facet of the American citizenry or an artefact of the “singular” ap-
proach. To preview our results: We uncover that participants reduce spending less when
given both size of the budget compared to when they are given spending separately. On the
tax side, there is little variation between treatments, which aligns well with the theory that
Americans have quite stable preferences for taxation.
Budget Preferences: A Tale of Two Literatures
There is a long scholarly history of studying the public’s support for budgets. Early scholars
began studying budgets by employing what we call a holistic approach. Much of this early
literature suggests that the public holds unrealistic preferences, which we believe has facil-
itated a methodological disconnect between taxes and spending. If the public has a hard
1Balancing Act creates interactive budget models for voters to learn about public budgets and the
choices involved in creating them. More information about Balancing Act can be found on their web-
sitehttp://www.abalancingact.com
2Our data will be made publicly available on the authors’ website.
3
A Holistic Approach to Studying Budget Preferences
time making realistic decisions when presented with both sides of the budget, then it makes
some sense to disentangle spending and taxes as to improve our measurements, as well as to
focus on more specific questions. Today, the literature has mainly splintered so that scholars
either study taxes or spending, but, rarely, consider both sides of the budget ledger together.
This divorce, we argue, has led to theoretical assumptions that may not align with actual
budget preference. Scholars such as Beck, Rainey and Traut (1990), Hansen (1998), and
Bonica (2015) increasingly seem to share this sentiment.
Early pioneers who examined public support for budgets usually accounted for attitudes
toward taxes and spending together. Mueller (1963), for example, fielded a survey in 1960-
61 which asked respondents about preferences toward certain public programs, followed by a
series of questions about raising taxes to support those same programs. She concludes that
there is a lack of congruence between the levels of spending that people want and the extent
of taxes that they are willing to pay to support those programs. Similarly, Citrin (1979: 113)
referring to the public’s incoherent budget preferences states, “the public’s readiness to
demand and consume government programs is understandably greater than their willingness
to pay for them.” Sears and Citrin (1982) and Ladd Jr et al. (1979) also indicate that people
enjoy high levels of spending without the required taxes to fund that spending. In other
words, Ladd Jr et al. (1979) concludes that “the message Americans are giving on the issues
of taxing and spending is clear: Reduce taxes; maintain “big government”; end the waste
and inefficiency of big government.”
This could be due to what Freed and Cantril (1967) coined as value “schizophrenia,” meaning
that individuals generally support public spending, but at the same time, laissez-faire ideals.
While Freed and Cantril (1967) see this as a failure to adapt to changing political and
economic structures, others (Rokeach 1973;Lipset 1979;McClosky and Zaller 1984;Feldman
and Zaller 1992) perceive this as a conflict between deeply entrenched American values, such
as freedom and equality. Inconsistent, incoherent, and unrealistic preferences were thus all
considered to be important and enduring characteristics of American public opinion and
ideology.
Though there is some consensus, it’s still not exactly clear whether individuals actually
hold contradictory preferences. For instance, extending upon Citrin (1979) and Ladd Jr
et al. (1979), Susan Welch (1985) instead suggests that the idea that most people have
unrealistic preferences is not totally accurate. In her study only a minority of individuals
want something for nothing, while most people were willing to fund increased spending with
higher taxes. Welch (pg. 316) claims, however, that “we do not know whether citizens have
4
A Holistic Approach to Studying Budget Preferences
very accurate ideas of how much money might be raised by reallocating from their less desired
services nor do we know how they would balance cost against benefits in their reallocation
preferences.” This speaks more to the methodological problems associated with measuring
peoples threshold for tradeoffs without forcing them to make an actual tradeoff. Similarly,
Hansen (1998) indicates that most Americans have very little trouble determining their
views, recognizing tradeoffs, and making consistent decisions. He concludes that Americans’
budget preferences are “remarkably well structured” (pg. 526). But he also makes a similar
observation about some of the methodological limitations when stating, “[E]ven the best
survey instruments have neglected the essential aspect of the public budgeting problem:
the tradeoffs inherent in establishing public priorities. Typically, survey questions offer
respondents unrealistic choices among incomplete sets of options.” (pg. 514).
We believe that this tacit agreement among academics that Americans hold unrealistic pref-
erences and the lack of literature demonstrating otherwise, has facilitated a methodological
disconnect between taxes and spending, which over time reinforces the divorcing of these
research agendas. According to Beck, Rainey and Traut (1990: 72), “Citizen thinking about
taxes, services, and the tradeoffs between them is not well understood. Most previous re-
search has restricted its attention to only one side of the tradeoff equation and has approached
it from a single theoretical perspective.” They go on to say, “With such divergence in focus
and theoretical approach, it is little wonder that citizen fiscal thinking remains somewhat
puzzling to social scientists.
Yet, despite the ambiguous findings and this call for a more holistic approach, the vast
majority of studies still only consider one side of the budget. This is not to say, however,
that there are not good reasons for separating the two research agendas. Both Hetherington
(2005) and Rudolph and Evans (2005), for example, make significant contributions to the
spending literature as they they put forth a theory which suggests that spending preferences
are not necessarily singular, but rather are context-dependent. Their contributions have been
influential for understanding how both trust and ideology can shape individuals’ support for
public spending. Therefore, by focusing on only spending they were able to also look at
very specific areas of the budget. Still, this literature assumes that individuals can only
have relative preferences on stand-alone issues, because they cannot process something as
complex as a federal budget.
Furthermore, Wlezien (1995) makes an important theoretical contribution to the spending
literature by pinpointing support for very specific spending items. This study indicates that
preferences for public spending are not so static. Individuals form preferences in relation to
5
A Holistic Approach to Studying Budget Preferences
current policy changes, so that individuals adjust their preferences for public spending down
when appropriations increase and up when appropriations decrease. Similarly, Kolln and
Wlezien (2016), using a conjoint experiment, further demonstrate that individuals do hold
strong preferences when asked to make complex/multidimensional decisions. Both of these
studies provide both theoretical and methodological contributions by focusing on one side
of the budget. However, singular models that leverage only one feature of the budget fail
to account for dynamic preferences. The literature cited above, for example, assumes that
women’s, seniors’, and Democrats’ preferences would remain unchanged even when given
richer contextual information. Indeed, this might be so, but without examining both sides of
the ledger and providing realistic budgetary context there is no way to know whether those
preferences are real or simply an artifact of a lack of prior contextual information.
There is also an expanding amount of literature on preferences for taxation (Ballard-Rosa,
Martin and Scheve 2017;Scheve and Stasavage 2010;Barnes 2014;Ballard-Rosa, Martin
and Scheve 2017;Bartels 2005;Rudolph and Evans 2005). Although less attention has been
given to taxation, the general conclusion among most scholars is that people are averse to
tax increases generally and demonstrate somewhat unrealistic preferences. For example,
Bartels (2005) demonstrates that most Americans supported the Bush tax cuts, although
the redistributive effects were more regressive than progressive. Most recently, Ballard-Rosa,
Martin and Scheve (2017) have recently developed a conjoint model which demonstrates that
although most people are concerned for fairness and progressivity, their policy preferences
would bring in significantly less revenue than current policy. The authors acknowledge,
however, that this result could also be an artifact of a lack of context addressing the spending
side of the budget. This is an important contribution to the literature as it introduces taxes
as a multi-dimensional issue instead of singular. However, all of these literatures still only
consider the one side of the budget.
Although academics have made considerable progress since the pioneering work of Freed and
Cantril (1967), Citrin (1979), and Welch (1985), more recent research still leaves gaps for
further investigation. For example, Rudolph and Evans (2005) and Rudolph (2009) consider
both sides of the ledger, however they are studied independently. Ballard-Rosa, Martin
and Scheve (2017) use conjoint analysis experiments to examine progressive taxation, but
until now they have only considered taxation. Similarly, Kolln and Wlezien (2016) exploit
a conjoint experiment, and yet, they only consider spending. Moreover, although conjoint
experiments are extremely valuable for examining trade-offs, researchers also have to make
significant trade-offs when deciding which budgetary items should be included in the trade-off
decision.
6
A Holistic Approach to Studying Budget Preferences
An Argument for A Holistic Approach to Studying Bud-
get Preferences
There are a number of scholars that argue for a more holistic approach to studying budget
preferences, as they do not believe that the “unrealistic” preferences argument holds up
when individuals are given context about both sides of the budget ledger. Even more recent
studies that considers only one side of the budget point out the limitations to examining
either taxes or spending. Kolln and Wlezien (2016: 18) recognize that even when giving indi-
viduals real-world trade-offs on spending items through a conjoint experiment, “our analysis
still focuses on what are largely unconstrained preferences, and in future research we will
address the consequences of imposing real budget constraints and allowing for higher/lower
taxes.” Similarly, Ballard-Rosa, Martin and Scheve (2017: 15) state “although our experi-
ment explicitly encouraged respondents to consider the revenue consequences of the tax rate
plans that they chose, the most favored set of rates would raise substantially less revenue
than current policy. Further research is needed to determine whether this is because the
effects of these cuts on public services is not made explicit.” They therefore acknowledge
the benefits of research that considers spending preferences alongside tax preferences and
encourage others to take steps in this direction.
It could be argued then that this one-sided methodology could be driving the “unrealistic”
preference for high spending and low taxes (Freed and Cantril 1967;Sears and Citrin 1982).
Beck, Rainey and Traut (1990: 72) argue that this is indeed the product of a methodological
“singular” approach, stating:
Most previous research has restricted its attention to only one side of the trade-
off equation and has approached it from a single theoretical perspective. Schol-
ars favoring sociopsychological approaches typically have concentrated on service
evaluations and have employed political attitudes and beliefs as the major ex-
planatory factors. By contrast, taxes have been the special domain of scholars
adopting microeconomic perspectives with their explanatory emphasis on mate-
rial cost-benefit calculations.
Most importantly, few studies have forced respondents to consider the real-world fiscal trade-
offs between taxes and spending. The omission of budgetary constraints may lead to extreme
positions on the issue of taxation and spending items. For example, Ladd Jr et al. (1979),
using public opinion polling, demonstrate that vast majorities of Americans thought they
paid too much in taxes, but very few thought that public services should be reduced. On the
7
A Holistic Approach to Studying Budget Preferences
other hand, Blinder and Holtz-Eakin (1983) produced polling that shows 67 percent of Amer-
icans support a balanced budget amendment. Additionally, although Eismeier (1982: 134)
finds that 71 percent of the sample thinks taxes are too high, “taxpayers who perceive their
tax burdens as too high are no more likely to oppose additional spending than are those
who do not.” Most recently, Ballard-Rosa, Martin and Scheve (2017) find that Americans’
tax preferences would raise significantly less tax revenue than current policy. Instead of un-
realistic fiscal preferences, it could be that a lack of fiscal context is contributing to extreme
attitudes.
Second, many studies narrowly focus on specific items within the federal budget such as
spending on education or corporate tax on the revenue side. While such item-specific ap-
proaches can be useful to explain preferences on a particular budgetary line, it is difficult to
generalize these preferences to the budget at-large, and it is unclear what the composition
of the budget would be if these preferences were translated into actual budget items.
Third, existing studies on preferences over the budget and budget items often lack context,
giving respondents little information about the actual content of the budget. In a typical
opinion survey respondents do not have to choose between real alternatives. Participants
are then left to make choices without having to consider how their choices would affect
other areas of the budget, which in turn could affect them personally. It is easy to say that
you want to “cut spending” on something without the distribution of the resources or the
actual amounts spent. Similarly, it can be easy to say that “the rich” or corporations should
pay higher taxes when you do not really know how much companies or wealthy individuals
currently pay, or how much money would actually be generated if they were asked to pay
more.
Real budgets are dynamic and multidimensional. The decision to decrease/increase spend-
ing in one area affects how much governments are able to decrease/increase spending and
taxation. For this reason, we argue that a holistic approach that considers both sides of the
ledger in context of an actual budget is preferable to a “singular” approach. Moreover, tra-
ditional public opinion surveys are not suited for capturing these dynamic, multidimensional
decisions. Hainmueller, Hopkins and Yamamoto (2013) argue this point and offer conjoint
analysis as one way to study multidimensional policy preferences. Yet, conjoint analysis has
its own restrictions. Assigning too few attributes can cause respondents to use an included
attribute to draw conclusions about an omitted characteristic (Hainmueller, Hopkins and
Yamamoto 2013:25). Malhotra (1982:419) states “it is important to determine the number
of choice alternatives and attributes consumers can simultaneously process without suffering
8
A Holistic Approach to Studying Budget Preferences
the negative effects of overload.” Malhotra’s results indicate that there was not information
overload at between five and ten attributes, but respondents suffered information overload
with over ten attributes. Hainmueller, Hopkins and Yamamoto (2013:25) acknowledges this
limitation, stating “when faced with a conjoint table that includes too many attributes,
respondents might disregard all but the first attribute.”
Despite many scholars having argued for a holistic approach to studying budget preferences,
most still study either taxes or spending.3One of the main reasons for not studying revenue
and spending together is because of methodological limitations. Over the last several years,
scholars have made significant progress studying multidimensional preferences utilizing tools,
such as conjoint analysis. However, since conjoint analysis is very limited in the number of
attributes it can include, it also tends to constrain scholars to one side of the budget. In the
next section, we introduce interactive budget simulation tools as a solution.
Interactive Budget Tools
While public opinion surveys are useful in measuring attitudes toward a specific, stand-
alone issue, it is difficult to link these attitudes to particular policies because surveys tend
not to provide budgetary context. In other words, when thinking about budget preferences,
a complete preference model needs to account for both sides of the budget, coupled with
budgetary context without which survey respondents lack the necessary prior information
to make informed decisions.
As Bonica (2015: 2) elegantly observes,
“It is worth considering that the incoherency present in responses to conven-
tional budgeting survey questionnaires is more a symptom of the test than the
takers. In any other context than a survey, it would likely seem unreasonable –
perhaps even unfair – to ask individuals about their public budgeting preferences
without first providing even a minimal amount of information on relative spend-
ing amounts needed to place them in context, and then declare the uninformed
and often incoherent preferences that result as evidence of their irrationality and
incompetence.”
3There is a growing movement to account for a more complete preference model by studying both sides of
the budget ledger together. Bj¨orn Bremer and Reto B¨urgisser (see B¨urgisser and Bremer 2017), for example,
consider individual tradeoffs between taxes and spending using a conjoint experiment. But again, researchers
sacrifice important budgetary information for simplicity.
9
A Holistic Approach to Studying Budget Preferences
At its core, an interactive budget tool is an online simulation designed to give a holistic
perspective of a budget and allow respondents to adjust items within the budget while
providing real-time feedback for how their choices affect the overall budget. In the most
basic of interactive budget tools, individuals go to a web page and are presented with some
basic information, such as the size of the current budget (or simulated budget). For example,
in our tool there is a slider at the top of the page that reads: “2017 Budget: You are in
deficit.” Under the text, the red slider shows the extent to which the budget is in deficit
(surplus). The slider is interactive in that it adjusts as the participant makes decisions on
taxes and spending. There is a list of line items from the budget with revenue on one side
and spending on the other side. One of the many benefits of a budget tool is that it can
be as simple or complex as one would like. There can be a whole battery of line items that
mimics a specific budget, or it can be simple with very few line items to address specific
questions.
These interactive budget tools were initially designed for learning, teaching, and citizen
engagement. Using a budget tool, policy makers can build a two-way relationship with
citizens based on facts and get informed input on service levels and priorities. Through
these tools policy makers are also able to foster public understanding and put users in the
decision makers’ shoes, including seeing the difficult tradeoffs inherent in budgets.
Local governments have long recognized the value of simulations to both provide useful
information to assist staff and elected officials to make budgets, but also as a meaningful
form of public engagement. Until recently, simulations were paper-based and both staff-
intensive and could usually only involve about 30 people. However, as internet technology
has improved, local governments and states (i.e. Canada, UK and Australia) are increasingly
using simulations to both convey complex information and to receive informed input. One
of the many uses of budget tools includes providing input on how to close a gap between
projected revenue and expenses with cost-saving options, such as closing libraries for one
day each week. Budget tools have also been used to help identify projects that should be
included in a long-term infrastructure plan alongside the taxes need to pay for the increase;
they have been employed to demonstrate transparency and accountability by accompanying
a proposed budget from city managers or mayors while embedding layers of information in
an effective way to share with residents. Furthermore, in states that have local referendums,
these simulations have been used to prepare for an election by creating understanding of
why an increase in bond revenue may be needed.
Finally, budget simulations are great pedagogical tools. They can easily keep students in-
10
A Holistic Approach to Studying Budget Preferences
volved and attentive. Moreover, the interactive element can spark student interest in the
budgetary process, while demonstrating the basic features of budgets.
Scope of Application
Just as with any methodological tool, there are both benefits and limitations to the budget
tool. First, the budget tool can provide a large amount of information that can be laid
out in an intuitive way for respondents. It is also great for researchers studying the trade-
offs between taxes and spending. This can be as broad or specific as the research question
demands. For example, researchers can look at the total ratio of spending cuts to tax
increases or they can analyze taxes increased on the wealthy compared to spending cuts on
welfare programs. It can also be used as a tool to measure issue salience by measuring how
much they are reducing and raising specific items.
These tools are flexible. Researchers can use real approximations of budgets; they can input
values that speak more directly to their research questions; it also provides scholars who
study state, local, and federal governments a variety of options. Something like gauging
public support for a local sales tax initiative can be easily executed employing an interactive
budget tool. They can also be simple or provide labyrinth of detail, depending on the
particular research question.
There are also important ways in which an interactive budget can be structured so that
researchers can examine specific issues. Suppose a researcher is interested in preferences
for welfare programs. Very specific welfare items can be nested within a much broader
welfare category. Further within something like this, researchers can vary question wording
to examine the effects of priming on preferences. Calling it welfare might elicit different
behaviors than calling it support for low-income individuals, for example. Finally, scholars
can include more items than traditional methods.4
Yet, a clear drawback with using a fully flexible budget tool is that it gives respondents
too many choices. When given too many options, respondents have been known to satisfice
which could lead to biased and inefficient measures. It also asks survey respondents to make
unrealistic decisions. As Haskell (2001: ) states, “the main point of having people in leader-
ship positions in legislatures is to have them do that which citizens are in no position to do –
construct some rational whole out of a given year’s budget.” It is somewhat problematic that
4An alternative method is conjoint analysis but it is less flexible in the number of items (and estimates
more sensitive too many items) and cannot measure the intensity of preferences - only relative importance)
11
A Holistic Approach to Studying Budget Preferences
by exploiting an interactive budget, we are asking subjects to make particularly complicated
decisions that they would never be forced to make in reality. Further, it is impossible to
provide an exact replica of a government budget, while at the same time keeping the tool
parsimonious and interpretable for analysis. Therefore, although interactive budgets can
provide a breath of information and detail that surveys and experiments cannot provide, it
is important that the budgets are also simple enough to provide accurate measures.
Moreover, it can be less useful for measuring less salient budget items. For example, re-
searchers could find that there is little variation on spending cuts to widely popular items,
such as the fire department, which renders analysis on these issues problematic. They can
also be costly to construct. Although interactive tools can be affordable, fully customizable
budgets can be expensive. However, prices will come down as more researchers adopt these
methods.
Application of Budget Tools for Social Science Research
We fielded an online survey with an embedded link to the budget tool between August
7 and 8, 2018. We conducted the survey with 919 voting-age Americans using a panel
provided by Mechanical Turk (mTurk). Survey participants were randomly allocated to one
of three treatments. In our baseline survey, we provided participants with an embedded
link to our full online budget simulation containing both taxes and spending together. This
meant that they could choose to adjust both taxes and spending. For another treatment, we
gave survey takers the tax-side of the budget first. Once they submitted the tax-side, their
choices were locked-in, and we then gave them the spending-side of the budget. Finally, for
our third treatment, we asked respondents to manipulate the spending-side of the budget
first. Those decisions were locked and followed by the tax-side of the budget. To be clear,
once the decisions were made and the budget was submitted in treatments two and three,
respondents could not go back and adjust their decisions.
In the first part of the study we presented respondents with an online survey that pro-
vided consent and basic demographic information such as age, employment status, gender,
and income. In addition to the pre-survey questions, the subjects were primed with the
statement
“The U.S. federal government is in debt. In 2018 the government plans to spend
approximately $487 billion more than it collects in taxes and other revenues. This
is commonly referred to as a “budget deficit”. This 1/2 trillion dollars will be
12
A Holistic Approach to Studying Budget Preferences
added to the total debt which is currently about 21 trillion dollars. To put this
in perspective, the current total debt of the U.S. government is equal to approxi-
mately $65,000 for every American citizen. Generally speaking, governments can
reduce deficits by decreasing spending and/or increasing taxes and revenue.”
After they completed the survey respondents were given a prompt which stated:
“With the size of the deficit in mind, we would like to know how you personally
would change federal taxes and spending. We have created a “budget model”
based on the actual federal budget. In the model you will be asked to increase
or decrease taxes in any category where you believe that the federal government
taxes too much or too little. Likewise for the spending side, you will be asked to
increase or decrease spending in any category where you believe that the federal
government spends too much or too little. The basic idea is to allow you to adjust
the budget in the ways you prefer.”
They were then given an embedded link, which took them directly to the budget tool.
The full budget tool contains both revenue and spending items. Figure 1 is a screen capture
of the online budget tool used for this study. The revenue-side contains six income brackets
and an additional six categories including social security payroll and corporate income tax.
Effective tax rates and revenue amounts were given for each revenue item.5The limited
budgets contain the exact same tax and spending items. On the spending-side of the budget
we included nine broad spending categories and a total of 27 subcategories. Each budget
item included an information bubble with a short description to help clarify the contents of
the item. Respondents could change any item in increments of 1% by clicking on “+” and
“-” signs.
5See Appendix Table 3 and Table 4 for the complete list of revenue and spending items.
13
A Holistic Approach to Studying Budget Preferences
Figure 1: Screen Capture of Online Budget Tool
Similar to the baseline treatment, we provided respondents in treatment two with a pre-
treatment survey with the same prime followed by an embedded link to the budget tool.
However, for this treatment we only gave participants the tax-side of the budget first followed
by the spending-side. Figure 2 is a screen capture of the tax-side only. Specifically we
stated,
“In the model, you can adjust the budget by changing the tax side of the budget.
The tax side allows you to increase or decrease the amount of money coming
from different tax sources. You will see the total amount of money from that
source as well as the current tax rate, if applicable. Click the arrow to the right
to expand a category to see its sub-categories. This is where you will make your
14
A Holistic Approach to Studying Budget Preferences
changes to taxes.
The bar located at the top of the model indicates whether your budget is in a
deficit or surplus. As a friendly reminder, we are asking you to submit a budget
that you support. This means that you can submit a budget with a deficit or
surplus.
We provide a timer to help protect your valuable time as a respondent. Please
note that you are allowed to go over the allotted time.”
Figure 2: Screen Capture of Online Budget Tool-Taxes Only
Once they submit the tax-side of the budget, we present them with an information bubble
15
A Holistic Approach to Studying Budget Preferences
explaining that budgets are made up of both revenue and spending. Figure 3 is a screen
capture of the information bubble.
Figure 3: Screen Capture of Instruction Bubble Following Submission
Finally, in treatment three was the opposite of treatment two. Thus, we first provide par-
ticipants with brief information, specifically stating,
“In the model, you can adjust the budget by changing the spending side of the
budget.
The spending side allows you to increase or decrease the amount of money that
is spent on certain items. Click the arrow to the right to expand a category to
see its sub-categories. This is where you will make your changes to spending.
The bar located at the top of the model indicates whether your budget is in a
deficit or surplus. As a friendly reminder, we are asking you to submit a budget
that you support. This means that you can submit a budget with a deficit or
surplus.
We provide a timer to help protect your valuable time as a respondent. Please
16
A Holistic Approach to Studying Budget Preferences
note that you are allowed to go over the allotted time.”
Figure 4: Screen Capture of Online Budget Tool-Taxes Only
Again, we locked these decisions once they submitted their budget, and we then provided
them with an information bubble about the next phase of the study. Just as the previous
treatment, we told participants that the spending-side is only displayed as a reference and is
not adjustable. In all three treatments, once the participants submitted their final budgets,
we redirected them to a post-treatment survey that asked questions about the participants
experience, such as how difficult was it to submit a budget that they supported.
By giving participants the tax-first treatment and spending-first treatment and locking those
decisions, the design more closely resembles a traditional survey in which researchers give
survey-takers only one side of the budget in isolation. We can then compare these treatment
to our baseline treatment. In other words, we can test the singular approach against the
holistic approach.
17
A Holistic Approach to Studying Budget Preferences
Our budget tool has several desirable attributes as a research instrument to measure prefer-
ences. First, it allows respondents to see both sides of the budget. Second, respondents see
how their individual choices on a single item impact the overall budget. Third, the budget
tool contains all the major categories and subcategories in the budget and allows respondents
to make budget decisions on each of them.
The number of items in our budget totaled 39 line items. As mentioned above, one potential
limitation to budget tools is that it overburdens respondents with information. We chose not
to reduce the budget further in order to match the budget as closely as possible to the actual
federal budget. Instead, we also include a variable in our analysis measuring the difficulty of
the simulation because scholars have shown that respondents face cognitive constraints and
optimization problems when asked to perform complicated tasks such as budgets (Benartzi
and Thaler 2007).
The budget tool logged how respondents interact with the individual lines in the federal
budget, producing a wealth of fine-grained information for researchers. In particular, the tool
stores information about which items of the budget are changed and the degree of change
for each item. Stylistically, these characteristics of the data are analogous to traditional
survey questions employing Likert scales. For example, data provides information on whether
a respondent reduces a spending item, like foreign military aid, and the extent to which
it is reduced. However, in contrast to survey items in a questionnaire, the budget tool
environment gives respondents a “big picture” of how each item fits into the total budget
and how their choices affect changes in the budget deficit. Importantly, this open-choice
environment allows researchers to track which items respondents prioritize in the context of
other budgetary choices.
The average respondent spent less than ten minutes to complete the simulation. Once
respondents submitted their preferred budgets they were redirected back to the a post-
survey questionnaire for which they answered a short battery of questions on demographics
and political orientation.
What Do American Budget Preferences Look Like?
Perhaps the greatest benefit of using interactive budget tools for social science research is
the rich, fine-grained data collected. In this section, we demonstrate the wealth of data
by first providing descriptive data on Americans’ budget preferences. We do this first by
looking at the total increases or decreases on each item of the federal budget. We follow
18
A Holistic Approach to Studying Budget Preferences
this by examining the percent change in each item for our entire sample. We also examine
the distribution of increases vs. decreases. In other words, how many items on average did
participants increase and/or decrease?
In general, data on whether an item (or items) has been adjusted is useful for studying
budget preferences at the macro-level, such as explaining mass support for tax increases or
spending cuts, while the dollar amount adjustment on a particular item is more appropriately
used for budgetary questions at micro-level, like explaining support for health care spending.
The reason for this macro-micro distinction relates to how the budget simulation is designed.
Respondents are able to adjust items at 1% increments, but in terms of actual dollar amounts,
a 1% incremental change varies by item depending on the initial allocations in the budget.
For example, in terms of the actual dollar size of the federal budget, a 1% decrease in defense
spending roughly equates to an 8% cut in education spending because spending on defense
is currently about eight times larger than education spending. Therefore, measures using
the total dollars adjusted (e.g. total spending cuts or total revenue increase) to summarize
budget preferences at the macro-level are more sensitive to small adjustments on big budget
items like defense and less sensitive to substantively important adjustments on smaller items
like education.
Next, we examine how budget preferences can be influenced by the survey design. We argue
that holistic budgets can overcome the weaknesses of the singular approach. By levering
a controlled experiment we are able to demonstrate that the holistic approach elicits more
balanced budget priorities.
American Attitudes Toward the Federal Budget? We examine here Americans’ at-
titudes towards the federal government. We provide summary statistics for each spending
and tax category in Table 1. Commencing from the last column, start amount refers to
the beginning amount of federal spending before adjustment. However, though this table
provides quite a lot of data, it is not the most intuitive of tables. We, therefore, provide the
average percent of revenue from each tax item and average change in spending from each
spending item in Figure 5.
19
A Holistic Approach to Studying Budget Preferences
Table 1: Descriptive statistics in Billions of US Dollars
Statistic N Mean St. Dev. Min Max Start Amount
Medicare 888 590.9 7.,7 0 1,239 590
Medicaid 888 476.8 57.6 0 988.8 480
Affordable Care Act Subsidies 888 58.5 35.1 0 465.1 61.2
Exempt Health Insurance 888 249.8 40.6 0 345.8 260
Military 888 510.6 154.2 0 1,236 603.2
Veterans 888 179.0 16.5 85.1 301,5 177.4
Retirement 888 985.9 83.4 0 1,473 995.6
Disability Spending 888 148.2 14.2 28.1 252.2 148.4
Family Nutrition 888 184.7 28.6 0 361.3 190.2
Housing Assistance 888 46.4 19.5 0 326.4 48.6
Unemployment Insurance 888 33.6 10.7 0 162.1 34.5
Mortgage Deduction 888 68.1 22.3 0 210 75
Infrastructure 888 34.5 14.1 0 215.7 32.2
Water and Air 888 21.1 6.3 0 83.2 20.8
K-12 888 46.5 24.2 0 350.8 40.8
Higher Ed 888 32.8 20 0 302.8 29.4
Arts 888 19.2 11.1 0 187.2 19.5
Job Training 888 13.8 7.3 0 123.8 12.9
Humanitarian Aid 888 21.8 10.7 0 97.5 27.1
Foreign Military Aid 888 9.5 4.3 0 38.4 12.4
Diplomacy 888 10.8 3.6 0 33.5 12.9
Fed Pensions 888 134.1 21.1 0 202.7 141.8
Legal 888 28.7 8.4 0 73.1 31.8
Border Spending 888 28 11.5 0 202.8 31.2
Research 888 30.6 14.1 0 211.7 31.6
Farm Spending 888 21.5 5.7 0 68.7 22.9
Community Spending 888 22 8.2 0 136.8 22.8
Tax Over $741K 888 842.6 216 0 2,755 738.7
Tax Below $25,000 888 -35 58.6 -66.2 544.1 -35.6
Tax $88k-$151k 888 270 143 0 2,383 228.7
Tax $50k-$88k 888 105.7 57.2 0 752.6 95
Tax $25k-$50k 888 -14.3 46.9 -312.6 433.7 -16.9
Tax $151k-$741k 888 841.3 298 0 4,439 687.8
Payroll Tax 888 1,182 124.3 0 2,044 1,200
Other Taxes 888 168.8 27.8 0 783 164.6
Gas Tax 888 226.4 21.9 0 347.9 227.4
Estate Tax 888 23.7 7.8 0 113.1 22.9
Corporate Tax 888 391 100.7 0 1,282 341.9
Social Security Tax Cap 888 1.9 2.8 0 6.1 6.1
20
A Holistic Approach to Studying Budget Preferences
On the tax side, a few observations stand out. First, Americans largely support progressive
taxation, confirming (Ballard-Rosa, Martin and Scheve 2017). Interestingly, they tend to
cut taxes more on those who make between $25,000-$50,000 than those making less than
$25,000. This could be due to the fact that individuals making less than $25,000 start at an
effective tax rate of -3.5%. Moreover, they made the largest decreases to payroll taxes, and
opted to not lift the cap on social security spending.
On the spending side, people tended to cut rather than increase spending. Because of the
way that we primed our subjects, we expected this. However, there were also some enor-
mous cuts to programs such as foreign military spending, humanitarian aid, and diplomacy
spending. Although these spending items make up a very small portion of the federal bud-
get, participants cut foreign military aid by 55%, humanitarian aid by 20%, and diplomacy
spending by 16%. They also cut military spending by 15%; however, military spending
makes up a much larger part of total federal spending. We also observe large increases to
education and infrastructure spending.
Figure 5: Average Change for Tax Items (left) and Average Change for Spending Items
(right)
We now move on to consider the number of items participants adjusted. We examine whether
individuals demonstrate a diverse set of preferences or whether individuals are more singu-
larly focused on more salient features of the budget. Examining Figure 10, we observe that
people have quite disperse preferences for taxing and spending. While Sanders (1988) and
Bonica (2015) demonstrate that individuals have a range of preferences on many issues,
Ballard-Rosa, Martin and Scheve (2017) suggest that most people have a strong reaction
to a small set of tax related issues, such as raising taxes on the poor. Similarly, Kolln and
Wlezien (2016) uncover that individual preferences for spending line up on just a few salient
21
A Holistic Approach to Studying Budget Preferences
issues.
Overall, we find most respondents have preferences over multiple items in the federal budget.
Figure 10 contains two histograms showing the number of items adjusted on the revenue and
spending sides. On average respondents decreased taxes on 2.54 items and increased taxes
on 7.38 items. In fact, the average respondent in the sample adjusted more than 49.6
percent of the budget items, 19.36 items out of the total 39. Finally, Figure 7 displays total
revenue raised in the simulation on the y-axis and total expenditures on the x-axis with
colors representing changes to spending only, revenue only, and changes to both taxes and
spending. From the figure, it is clear that most people prefer a balance between changes to
taxes and spending.
(a) Number of Revenue Items Increased and De-
creased
(b) Number of Spending Items Decreased
Figure 6: Number of Spending Items Increased or Decreased
22
A Holistic Approach to Studying Budget Preferences
Figure 7: Decreased Spending Versus Increased Revenue
In summary, we examined the rich data that budget simulations can provide. From our
data, we are able to get a good glimpse into Americans’ preferences for taxing and spending.
In short, Americans support progressive taxation. They also make large spending cuts
to any kind of international spending, such as foreign military aid. At the macro level,
American’s preference are quite dispersed between taxing and spending. Most respondents,
take a balanced approach to the budget, meaning they make changes to both sides of the
budget. Indeed, very few participants only adjusted taxes or spending. This, thus, leads us
to ask the question: What are the effects of giving respondents only one side of the budget
(singular approach) versus giving respondents both sides of the budget. We turn to this in
the following section.
1 Holistic Approach versus Singular Approach?
In this section, we test whether the “holistic” approach – providing survey-takers with both
sides of the budget – elicits more balanced or moderate attitudes toward taxes and spending
than the “singular” approach – providing survey-takers with only one side of the budget.
In the first section we use means tests to demonstrate that on most items in our federal
23
A Holistic Approach to Studying Budget Preferences
budget respondents cut significantly less spending or increase significantly less taxes when
given both sides of the budget than when given only one side of the budget.
Here we briefly review our treatments. In the taxes and spending treatment, participants
were given both treatment simultaneously. They were able to adust both taxes and spending
together. In another treatment, we gave participants spending first. Once they submitted
their spending choices, those choices were locked, and we gave them taxes. For treatment
three, we gave them taxes first. Theoretically, becuase we locked the “spending first” and
“tax first” decisions, we are mimicking the singular approach. Therefore, for our analysis we
will compare choices on the revenue side of the budget when given the holistic approach to
“tax first,” and for spending, we will compare the holistic budget to “spending first.”
(a) Mean differences for Income Taxes (b) Mean differences for Other Taxes
Figure 8: Mean Differences for Other Taxes
We turn now to our group means tests for the tax treatments. For this analysis, we use the
macro approach defined above. The tax and spending variables range from -1 (decrease),
0 (no change), and 1 (increase). A couple interesting things emerge here. First, as we
mentioned above, there is a general trend towards progressivity in both treatments, demon-
strated in Figure 8a. Moreover, there are significant differences between the two treatments
on incomes between $50,000 and $151,000, and the trend tends to move towards zero when
we give them both sides of the budget. This means that when we give respondent taxes first,
they increase significantly more at those income levels. However, on incomes below $50,000
and above $741,000, preferences seem to be quite stable between preferences. Finally, in
Figure 8b, we can see there is still a general trend towards 0.
24
A Holistic Approach to Studying Budget Preferences
(a) Education and Defense Spending
(b) Economic Security and Health Spending
(c) Other Spending Categories and Diplomacy Spending
Figure 9: Average Percent Decrease in Spending by Category
25
A Holistic Approach to Studying Budget Preferences
On the spending side, there is a lot of variation between treatments. In fact, on Arts,
Economic Security, and Health Spending, participants switch from spending cuts to slight
increases. The general trend, however, is to to move right from deep cuts to more moderate
cuts, as we would expect. For example, in the spending only treatment 26 percent of par-
ticipants cut border spending, whereas in the taxes and spending treatment nine percent of
participants cut border spending. On education in Figure 9a, though, individuals in both
treatments increase spending, but they increase spending significantly more in the taxes and
spending treatment.
This tells us prima facie that when researchers gauge preferences for taxes and spending,
those preferences will vary quite significantly when they give subjects a complete picture of
the federal budget. Next, we will investigate this using individual level models.
Individual Level Models
In this section, we move on to our individual level models. We test whether respondents
will have different preferences when given a singular budget or a holistic budget. More
specifically, we suggest that by giving individuals a complete picture of the federal budget
they will make more moderate/balanced choices than when we give them only one side of
the budget. We formally state our hypothesis as:
Hyp1: Respondents will make significantly larger cuts to spending items in the “spend-
ing first” treatment than in the “taxes and spending treatment.”
Hyp2: Respondents will increase taxes significantly more in the ‘taxes first” treatment
than in the “taxes and spending treatment.”
Dependent Variables: We measure respondents’ support for each tax item as a simple
ordinal variable ranging from -1 (decrease), 0 (no change), and 1 (increase). We create
equivalent variables for the spending side. These variables (one for each category) serve as
the primary dependent variables for comparing coefficients across treatments.
Independent Variables: The key independent variables in our analysis are “treatment
group 1” on the spending side and “treatment group 2” on the tax side. In addition to our
main variables of interest, we also include several control variables to account for respondents’
26
A Holistic Approach to Studying Budget Preferences
ideology, party affiliation, trust in federal government, political awareness, importance of
federal debt, task difficulty, and whether they balanced the budget.
Results
Because we are interested in the treatment effect on each individual category and our depen-
dent variables are categorical, we run 39 separate ordered logit models and plot the coefficient
for “treatment group 1” for the spending variables and the coefficients for “treatment group
2” for the tax variables in Figures 10a and 10b. The reference category for both taxes and
spending is the “taxes and spending” treatment. To the extent that there are significant
treatment effects, we do not expect the coefficients to cross zero. Moreover, we expect the
coefficients to be negative in the spending treatment, because respondents should cut spend-
ing more in the “spending first treatment” compared to the “taxes and spending treatment.”
On the other hand, we expect the coefficients on the tax side to be positive, as we predict
that participants will increases taxes more in the “taxes first” treatment compared to the
“taxes and spending treatment.”
27
A Holistic Approach to Studying Budget Preferences
(a) Ordered Logit Regressions for Tax Side
(b) Ordered Logit Regressions for Spending Side
Figure 10: Individual Level Regression Analyses of Treatment Effects
28
A Holistic Approach to Studying Budget Preferences
On the tax side, our results, shown in Figure 10a, demonstrate that respondents in the “taxes
first” treatment increase taxes significantly more on incomes between $88,000 and $741,000
than in the “taxes and spending” treatment. Moreover, the result is highly significant on
other taxes and the social security cap. For instance, for those in the “taxes first” treatment,
the odds of increasing taxes on incomes between $151,000 and $741,000 versus not changing
or decreasing taxes is .51 times greater, all else being equal, displayed in Figure 11. Finally,
taxes on income between $55,000 and $100,000 is significant at .1. This provides some
support to our means test above and our underlying hypothesis. However, the results are
mostly not significant on the tax side, which lead us to the conclusion that Americans’
preferences for taxes are relatively stable to our treatments. This is not necessarily surprising
and also confirms Ballard-Rosa, Martin and Scheve (2017).
Figure 11: Ordered Logit Regressions for Tax Side with Odds Ratios
The results are much different, however, when considering the spending side. Our results in
Figure 10b demonstrate that respondents who receive spending first cut spending significanly
more than in the “taxes and spending” treatment on all spending items, except for military
spending and the employer health care exemption. In other words, for those in the “spending
first” treatment, the odds of decreasing spending on Affordable Care Act subsidies versus
not changing or increasing spending is .41 times greater, all else being equal. The largest
effect is on the home mortgage deduction variable, meaning that for those in the “spending
first” treatment, the odds of decreasing spending versus not changing or increasing spending
is 1.2 times greater, all else being equal.
29
A Holistic Approach to Studying Budget Preferences
Figure 12: Ordered Logit Regressions for Spending Side with Odds Ratios
To sum our results: Americans have a general preference for progressive taxation. Though
their decisions do not vary much between receiving only one side of the budget rather than
both, they do vary on incomes between $88,000 and $741,000. On spending, overall the
trend was to cut, and they cut some items, like any kind of foreign aid, much more than
others. However, they do show a trend towards increased education spending. Finally, their
decisions are much more extreme on spending cuts when given only one side of the budget
compared to a much more balanced approach when given both sides of the budget.
Discussion
The literature on budget preferences has expanded tremendously in recent years. However,
there appears to be a growing disconnect between what actually constitutes a budget and
research on budget preferences. Examine only spending or revenue preferences requires
researchers to either make potentially unrealistic assumptions about individual preferences
over the other side of the budget or ignore them altogether.
Echoing early research in the field, we argue that research can generally benefit from study
budget preferences from a holistic approach. A holistic approach enables researchers to study
budget preferences within each side of the ledger and across the two sides. Put differently,
singular approaches have much to offer for answering specific questions about revenue or
30
A Holistic Approach to Studying Budget Preferences
spending, but they are limited in explaining more macro-level attitudes over the spending
v. revenue trade-off.
This paper makes two contributions to the ongoing development in this research programme.
Methodologically, we introduce budget tools as a new way for social science researchers to
study and measure budget preferences from a holistic approach. Budget tools enable re-
searchers to obtain micro-level data on preferences toward individual line items as well as
macro-level data on attitudes toward cross-ledger preferences, such as the size of government
and budget deficits. From the respondents perspective, budget tools provide important infor-
mation about the constituent line items and presents a bird’s view of how revenue is raised
and spent. We believe this added context leads to more refined measures of preferences
than more traditional instruments like surveys, which essentially ask questions in a com-
partmentalized vacuum. We test this proposition by leveraging a randomized experiment in
which survey-takers receive one of three budgets: A budget with taxes first; a budget with
spending first; and a budget with taxes and spending together. Theoretically, our results
demonstrate that indeed the different treatments render different preferences for taxes and
spending, and that when given only one side of the budget, respondents tend to make more
extreme decisions, whereas when respondents are given both sides of the budgets, they make
more informed and balanced choices.
Finally, budget simulations provide a great deal of flexibility for scholars studying preference
for taxes and spending. For future research, we will examine whether these treatment effects
vary by ideology or party affiliations. We will also remove the prime in the instructions to see
how that affects the treatments. Future research can also begin to examine how partaking
in these types of exercises can increase political efficacy and civic engagement. The possible
utility of these tools are numerous, providing benefits to policymakers, policy advocates, and
scholars endeavoring to understand mass attitudes toward budgetary policies.
31
A Holistic Approach to Studying Budget Preferences
References
Ballard-Rosa, Cameron, Lucy Martin and Kenneth Scheve. 2017. “The Structure of Ameri-
can Income Tax Policy Preferences.” The Journal of Politics 79(1):1–16.
Barnes, Lucy. 2014. “The size and shape of government: preferences over redistributive tax
policy.” Socio-economic review 13(1):55–78.
Bartels, Larry M. 2005. “Homer gets a tax cut: Inequality and public policy in the American
mind.” Perspectives on Politics 3(1):15–31.
Beck, Paul Allen, Hal G Rainey and Carol Traut. 1990. “Disadvantage, disaffection, and race
as divergent bases for citizen fiscal policy preferences.” The Journal of Politics 52(1):71–93.
Beck, Paul Allen and Thomas R Dye. 1982. “Sources of public opinion on taxes: The Florida
case.” The Journal of Politics 44(1):172–182.
Benartzi, Shlomo and Richard Thaler. 2007. “Heuristics and biases in retirement savings
behavior.” Journal of Economic Perspectives 21(3):81–104.
Blinder, Alan S and Douglas Holtz-Eakin. 1983. Public opinion and the balanced budget.
Technical report.
Bonica, Adam. 2015. “Measuring public spending preferences using an interactive budgeting
questionnaire.” Research & Politics 2(2):1–9.
Braithwaite, John and Toni Makkai. 1994. “Trust and compliance.” Policing and Society:
An International Journal 4(1):1–12.
B¨urgisser, Reto and Bj¨orn Bremer. 2017. “Pre-Analysis Plan: Public Preferences Towards
Fiscal Policies: Survey Experiments on Budgetary Priorities and Trade-Offs.”.
URL: http://dx.doi.org/10.7910/DVN/TKTNZG
Citrin, Jack. 1979. “Do people want something for nothing: Public opinion on taxes and
government spending.” National Tax Journal 32(2):113–129.
Eismeier, Theodore J. 1982. “Public preferences about government spending: Partisan,
social, and attitudinal sources of policy differences.” Political Behavior 4(2):133–145.
Feldman, Stanley and John Zaller. 1992. “The political culture of ambivalence: Ideological
responses to the welfare state.” American Journal of Political Science pp. 268–307.
Franko, William, Caroline J Tolbert and Christopher Witko. 2013. “Inequality, self-interest,
and public support for ‘Robin Hood’ tax policies.” Political Research Quarterly 66(4):923–
937.
Freed, Lloyd A. and Hadley Cantril. 1967. “The Political Beliefs of Americans: A Study of
Public Opinion.” Rutgers: Rutgers University Press.
32
A Holistic Approach to Studying Budget Preferences
Gilens, Martin. 2001. “Political Ignorance and Collective Policy Preferences.” The American
Political Science Review 95(2):379–396.
Hainmueller, Jens, Daniel J Hopkins and Teppei Yamamoto. 2013. “Causal inference in
conjoint analysis: Understanding multidimensional choices via stated preference experi-
ments.” Political Analysis 22(1):1–30.
Hansen, John Mark. 1998. “Individuals, institutions, and public preferences over public
finance.” American Political Science Review 92(3):513–531.
Haskell, John. 2001. Direct democracy or representative government?: dispelling the populist
myth. Boulder, CO: Westview Press.
Hetherington, Marc J. 2005. Why trust matters: Declining political trust and the demise of
American liberalism. Princeton University Press.
Kolln, Ann-Kristin and Christopher Wlezien. 2016. “Conjoint Experiments on Political
Support for Governmental Spending Profiles.”.
Ladd Jr, Everett Carll, Marilyn Potter, Linda Basilick, SALLY DANTELS and Dana
Suszkiw. 1979. “The polls: Taxing and spending.” Public Opinion Quarterly 43(1):126–
135.
Levi, Margaret. 1998. “A state of trust.” Trust and governance 1:77–101.
Lipset, Seymour Martin. 1979. The first new nation: The United States in historical and
comparative perspective. Transaction publishers.
Lowery, David and Lee Sigelman. 1981. “Understanding the tax revolt: Eight explanations.”
American Political Science Review 75(4):963–974.
Malhotra, Naresh K. 1982. “Information load and consumer decision making.” Journal of
consumer research 8(4):419–430.
McClosky, Herbert and John Zaller. 1984. The American ethos: Public attitudes toward
capitalism and democracy. Harvard Univ Pr.
Mueller, Eva. 1963. “Public attitudes toward fiscal programs.” The Quarterly Journal of
Economics 77(2):210–235.
Murphy, Kristina. 2004. “The role of trust in nurturing compliance: a study of accused tax
avoiders.” Law and Human Behavior 28(2):187.
Rokeach, Milton. 1973. The nature of human values. Free press.
Rudolph, Thomas J. 2009. “Political trust, ideology, and public support for tax cuts.” Public
Opinion Quarterly 73(1):144–158.
Rudolph, Thomas J and Jillian Evans. 2005. “Political trust, ideology, and public support
for government spending.” American Journal of Political Science 49(3):660–671.
33
A Holistic Approach to Studying Budget Preferences
Sanders, Arthur. 1988. “Rationality, self-interest, and public attitudes on public spending.”
Social Science Quarterly 69(2):311.
Scheve, Kenneth and David Stasavage. 2010. “The conscription of wealth: mass warfare and
the demand for progressive taxation.” International Organization 64(4):529–561.
Sears, David O and Jack Citrin. 1982. Tax Revolt. Boston: Harvard University Press.
Welch, Susan. 1985. “The “more for less” paradox: Public attitudes on taxing and spending.”
Public Opinion Quarterly 49(3):310–316.
Wlezien, Christopher. 1995. “The public as thermostat: Dynamics of preferences for spend-
ing.” American journal of political science pp. 981–1000.
34
A Holistic Approach to Studying Budget Preferences
Appendix
Survey
Prompt given before starting the budget tool exercise: “The U.S. Federal Govern-
ment is in debt. In 2017 the government is going to spend approximately $519 billion more
than it will collect in taxes and other revenues. This 1/2 trillion dollars will be added to the
debt total which is currently approximately 19 trillion dollars. To put this in perspective, the
current debt of the U.S. Government is equal to approximately $45,000 for every American
citizen. With the size of the deficit in mind, we would like to see how you personally would
change Federal taxes and/or spending. We have created a ?budget model? based on the
actual Federal budget.
In the model you will be able to cut spending in any category where you believe that the
Federal government spends too much money and/or increase revenues in any category where
you think taxes should be raised. The basic idea is to allow you to change taxes and/or
spending in the ways you prefer. The model will only allow spending cuts and/or increases
in Federal taxes.You will have 10 minutes to complete the budget model. After completing
the model, you will be returned to the survey and asked some follow-up questions.”
Survey Questions and Coding
Ideology: “When it comes to politics, do you usually think of yourself as”: 1) Very
conservative; 2) Conservative; 3) Somewhat conservative; 4) Moderate or middle of the
road; 5) Somewhat liberal; 6) Liberal; 7) Very liberal. We coded the dummy variable for
Liberal as “1” if the respondent chose 5,6, or 7; otherwise “0”. We coded the dummy variable
for Conservative as “1” if the respondent chose 1,2, or 3; otherwise “0”. Moderate serves as
the reference category.
Party ID: “Generally speaking, do you usually think of yourself as”: 1) Strong Republi-
can; 2) Weak Republican; 3) Lean Republican; 4) Independent; 5) Lean Democrat; 6) Weak
Democrat; 7) Strong Democrat. We coded the dummy variable for Democrat as “1” if the
respondent chose 5,6, or 7; otherwise “0”. We coded the dummy variable for Republican
as 1 if the respondent chose 1,2, or 3; otherwise “0”. Independent serves as the reference
category.
Political Trust: “How often can you trust the federal government in Washington to do
what is right?” 1) Never; 2) Sometimes; 3) About half the time; 4) Most of the time; 5)
Always
35
A Holistic Approach to Studying Budget Preferences
Table 2: Descriptive Statistics
Variable Obs Mean Std. Dev. Min Max
Spending Count 1,991 7.221 6.244 0 27
Revenue Count 1,991 3.841 2.944 0 12
Moderate 2,901 0.347 0.476 0 1
Conservative 2,901 0.349 0.477 0 1
Trust 2,901 2.535 1.064 1 5
Income 2,901 6.170 3.276 1 16
Education Level 2,901 3.887 1.438 1 6
Age 2,899 48.097 16.874 18 82
Female 2,217 0.535 0.499 0 1
White 3,184 0.750 0.433 0 1
Democrat 2,900 0.350 0.477 0 1
Republican 2,900 0.333 0.472 0 1
Age: “In what year were you born?” Continuous measure of age ranging from 18 to
82.
Income: Over the past year, what was your family?s approximate annual income? 1) Less
than $10,000; 2) $10,000 - $19,999; 3) $20,000 - $29,999; 4) $30,000 - $39,999; 5) $40,000 -
$49,999; 6) $50,000 - $59,999; 7) $60,000 - $69,999; 8) $70,000 - $79,999; 9) $80,000 - $99,999;
10) $100,000 - $119,999; 11) $120,000 - $149,999; 12) $150,000 - $199,999; 13) $200,000 -
$249,999; 14) $250,000 - $349,999; 15) $350,000 - $499,999; 15) $500,000 or more; Prefer not
to say (coded as missing).
Education: “What is the highest level of education you have completed?” 1) Did not
graduate from high school; 2) High school graduate; 3) Some college, but no degree; 4) 2-year
college degree; 5) 4-year college degree; 6) Post-college graduate degree.
Female: “What is your gender?” 1) Female; 0) otherwise.
White: “What racial or ethnic group or groups best describes you?” 1) White; 0)
otherwise.
Appendix Tables & Figures
36
A Holistic Approach to Studying Budget Preferences
Table 3: Revenue Items Included in Balancing Act Budget Tool
Main Line Items Initial Revenue Amount*
Income Tax on Low Incomes (below $25,800) -$35.6
Income Tax on Lower Middle Incomes (between $25,800-$50,200) -$16.9
Income Tax on Middle Incomes (between $50,200-$88,100) $95.0
Income Tax on Upper Middle Incomes (between $88,100-$151,400) $228.7
Income Tax on High Incomes (between $151,400-$741,000) $687.8
Income Tax on Top 1% of Incomes (over $741,000) $738.7
Social Security Payroll Taxes $1,220
Corporate Income Taxes $341.9
Estate Tax $22.9b
Gas Tax (18.4 cents per gallon) $227.4
Other Revenues $164.5
Eliminate Income Cap on Social Security $0
*Amounts reported in billions USD.
37
A Holistic Approach to Studying Budget Preferences
Table 4: Spending Items Included in Balancing Act Budget Tool
Main Line Items Initial Spending
Amount by Main Line
Items*
Subcategories under main line items
Economic Security $347.7 Family and Nutrition Assistance, Housing Assistance, Unemploy-
ment Insurance, Home Mortgage Interest Deduction
Social Security $1,100 Average Retirement Benefit, Disability Benefits
Education $102.6 K-12 and Vocational Education, Higher Education, Arts, Museums,
and Other Social Services, Job Training, Research, and Other Labor
Services
All Other Spending $77.3 NASA and General Research, Farm Income Stabilization, Commu-
nity Development
Defense $780.6 Military, Veterans Benefits
General Government $204.8 Federal Employee Retiree/Disability, Legal, Judicial, and Correc-
tional System, Border Security and Domestic Law Enforcement
Health Care $1,400 Medicare, Medicaid, Affordable Care Act Subsidies, Employer Paid
Health Insurance Exemption
Foreign and Diplomatic Aid $52.4 International Development/Humanitarian Assistance, Foreign Mil-
itary Aid, Embassies and Diplomatic Affairs
Infrastructure and Development $53.0 Highways, Railroads, Other Surface Transportation, Water and Air
Transportation
*Amounts reported in billions USD.
38
