Legitimate authorities and rational taxpayers: An investigation of voluntary compliance and method effects in a survey experiment of income tax evasion

Abstract

In order to collect the revenue necessary to fund public goods, a state is often required to both deter tax evasion and encourage voluntary tax compliance on the part of its citizens. While most prior research has focused on explaining tax evasion with standard economic model parameters, there has been growing interest in identifying the determinants of voluntary compliance. We build on this work by proposing a legitimacy-based model of tax compliance that accounts for why some citizens voluntarily comply with their tax obligations and others do not. To test our model, we develop and administer a survey experiment of income tax evasion to a large random sample of undergraduate students. We also investigate the extent to which design-based method effects bias our results, such as order effects, complexity effects, and missing information effects. Substantively, results strongly support the standard economic model of deterrence and weakly support

Different Order, Different Results? The Effects of Dimension Order in Factorial Survey Experiments

“…Moreover, they do not find any signs of primacy or recency effects. Robbins and Kiser (2018) do not find any evidence of order effects in a similar experimental setting.…”

“…Factorial survey experiments are considered cognitively challenging because respondents have to take a number of (varying) dimensions into account before making a judgment. To our knowledge, there have been only two methodological articles that have focused on the order of vignette dimensions (Auspurg and Jäckle 2015; Robbins and Kiser 2018). Auspurg and Jäckle (2015) find some indications of order effects but only when the survey task becomes more complex (in terms of dimensions and number of vignettes to evaluate).…”

“…We focus on one prominent issue in survey research, namely order effects, and analyze whether the order in which the vignette dimensions are presented affects results. So far, there have been only two studies on this using samples consisting of university students (Auspurg and Jäckle 2015; Robbins and Kiser 2018). We use a more heterogeneous respondent sample, which allows the study of possible variations according to respondents’ cognitive abilities (measured by proxies of age and degree of education).…”

Status, identity, and ability in the formation of trust

“…Although order effects have been observed with survey experiments using more than one evaluation task, Auspurg and Jäckle (2017) show that this bias is unique to complex evaluation tasks (i.e. open-ended questions), does not bias dimensions placed at the beginning of a vignette (like age, gender, and race in the current design), and is relatively minor and marginally significant (for methodological research that did not observe order effects under similar conditions, see Robbins and Kiser, 2017). In other words, the present research falls within design parameters known to minimize bias (and robustness checks found in the Supplemental Materials online show how some of these biases did not unduly impact the results).…”

Moral Collapse and State Failure: A View From the Past

“…Levitsky and Ziblatt (2018, p. 212) argue that today politicians often ignore the once important values of mutual tolerance and institutional forbearance. In the US, bipartisan support for public goods has been weakening since the 1960s (Whitman, 2017;Kleinenberg, 2018;Reich, 2018). Surveys on values point to decline in belief in democratic ideals and institutions (Howe, 2017).…”