The data contains the experimental response from subjects recruited through Prolific in the completion of a tax form for a series of business expenses associated with a small business owner. Upon completion of the initial tax filing tasks, subjects were potentially shown a report with further information about their tax return - specifically a feedback report with nudges to act on the information provided. The primary research questions were: Can correct nudges embedded within tax filing software increase compliance? What effect may incorrect nudges embedded within tax filing software have on compliance? Are particular forms of nudges more effective in raising compliance than others?Good tax design and administration are central to the functioning of the economy. Taxes are important determinants of economic behaviour, and good implementation can significantly increase economic and social welfare. The role of the Tax Administration Research Centre is to deliver research that enhances tax policy and provides lasting benefit to the economy. There are many research tools that can contribute to this goal but the greatest success will be achieved by combining a range of research methodologies and disciplines. The Centre will unite researchers from two institutions with distinguished reputations for research into tax administration and tax design. Complementary abilities and methodologies will be brought together to address a wide range of intersecting research projects. The research methodologies will include economic modelling, econometric analysis, experimentation, numerical simulation, and qualitative analysis. In undertaking its work the Centre will make extensive use of the HMRC/HMT Datalab to permit innovative empirical work to be undertaken. Some examples of the research projects to be conducted at the Centre are: Risk-based auditing and taxpayers' responses will investigate the reaction of taxpayers to audits, and how this will change the pattern of compliance behaviour. The project will study whether taxpayers learn about the audit strategy over time, and how a risk-based strategy should be updated to take this into account. The project will link with laboratory experiments on compliance and with research on the role of tax advisers. The results will increase the return on resources allocated to auditing. Decomposing the elasticity of taxable income will determine how the response of individuals to taxes summarised by the "elasticity of taxable income" can be separated into the channels through which individuals respond (e.g. reduced work effort, increased pension contributions or charitable contributions, moving income abroad, taking income in other forms). The results will enable greater focus of tax interventions and improved design of tax structure. Consequences of pre-population will study the implications of pre-populated tax returns for compliance. If the pre-populated form shows an income level below actual income this might convey an impression to the taxpayer that they can successfully evade. The research will implement an experiment that randomises the allocation of pre-populated returns. The outcome will advise on how best to proceed with the implementation of pre-population. Large business (Intermediaries) relationship with HMRC will use qualitative depth interviews address the extent to which new initiatives have altered working practices in intermediary firms and their clients' businesses, the HMRC understanding of tax avoidance practices, and the appropriateness of current policy strategies currently. This will enhance understanding of the effects of HMRC operational policy and improve administration. Understanding the determinants of customer experience will link HMRC survey data on customer experience with objective measures of service delivery standards, third-party information, and tax return data in Datalab. The linked data will be used to analyse the systematic determinants of subjective customer experience and will show which aspects of HMRC delivery generate a positive customer experience, and which do not. The Centre will enhance tax administration and tax design. It has ambitious plans to become the leading international centre with a central role in research and in building research capability in tax analysis through the training of PhD students and training of research staff.

Data was collected through a bespoke website. Subjects were recruited through the academic system Prolific, which contains a large number of people pre-registred to undertake online studies. We pre-filtered through Prolific for UK residents with self-employment or entrepreneurial experience. Subjects were forwarded to the experimental web site and given a set of instructions about the tax filing task. The main element being that they were incentivised to enter correct details as each mistake would lead to a reduction in their payment for the task. Having completed a comprehension check, subjects were informed about the tax filing "software" and the associated tax guidance for completing the task. The main task comprised of 16 expenses for which subjects had to categorise if the expense was allowable, and if so, under what tax category. There were two main technical treatments, in one subjects had to additional calculate by category totals and enter them. In the second, subjects simply had to assign an allowable value to each of the items. There were 5 treatments with regard to the feedback (nudges) information displayed after the initial submission: NONE - no nudge RES - a set of nudges based on errors in the subjects tax declaration OTHER - a specific nudge in relation to the Other Business Expenses category was always displayed AIA - a specific nudge in relation to the AIA category was always displayed COST - a specific nudge in relation to the Cost of goods category was always displayed Key measures include if a subject returned to reconsider their tax return, which items they changed, and whether any changes lead to more or less correct items being reported in their final tax submission.