**************************************** The SIPHER-7 personal DCE survey dataset **************************************** The dataset was put together as part of the Systems Science in Public Health and Health Economics Research [https://sipher.ac.uk/] consortium. The SIPHER consortium is supported by the UK Prevention Research Partnership [https://ukprp.org/] (MR/S037578/2), an initiative funded by UK Research and Innovation Councils, the Department of Health and Social Care (England) and the UK devolved administrations, and leading health research charities. The SIPHER-7 Personal DCE survey dataset is based on a Discrete Choice Experiment (DCE) conducted on-line in two waves, each wave with over 3300 respondents from the UK general public aged 18+. The DCE asked respondents a series of choice tasks, between pairs of outcomes described using a suite of seven wellbeing items, from a personal perspective (which outcome they would prefer for themselves). SIPHER-7 is a suite of seven health and wellbeing indicators linked to the UK Household Longitudinal Study (Understanding Society), and consists of: the Effect of Physical Health, the Effect of Mental Health, Loneliness, Household Disposable Income, Employment, Housing Quality, and Neighbourhood Safety. For details of SIPHER-7, please see Tsuchiya & Wu (2021). The wave 1 survey was conducted in autumn 2020; all respondents were re-invited to the wave 2 survey one year later, alongside a fresh sample recruited to make up for attrition. The dataset has four sections: [1] The DCE modelling data For those who are interested in replicating the baseline DCE model (and perhaps exploring variations to it), this section should suffice. [2] The DCE development This section includes all the files associated with the experimental choice design, qualitative and quantitative pilots. [3] The main DCE data files This section includes all the data files from the two main DCE surveys, including time stamp data and sampling weights. [4] The analysis files This section includes Stata syntax files to run the main and additional analyses reported in Ta et al (2024). The below provides further details for each section. ------------------------------------------------------ [1] The DCE modelling data files This section contains one data file in Stata format (ver.17), and a codebook in MS Word. The survey has 6930 respondents across two waves (some individuals have responded twice). Each respondent answered 10 choice tasks (S = 10), where each choice task has two scenarios (J = 2), and each scenario has seven attributes. Six attributes have three or five ordered levels, while one attribute is continuous. "20022023-full-reshaped-income-equi-Pool-exc_w2-desstat-testweights_P1C1-C2.dta": this data file has one row for each scenario for each task for each respondent (J x S x N), resulting in 138,600 observations. The variables include: the respondent identifier, the scenario identifier within a choice task, the attributes of the scenario, the respondent’s choice, respondent’s background such as age, gender, education levels, etc., and the wave identifier. All variables are fully labelled. The following Stata command will estimate the basic DCE model using the wave 1 data only, for example. For a conditional logit model: clogit pchoice i.Phy i.Men i.Lon lnY i.Em i.Hou i.Saf if wave==1, group(csid) For a mixed logit model: mixlogit pchoice lnY if wave==1, rand($randvars) group(gid) id(RespID) nrep(1000) “SIPHER7-DCE-Personal-6.1.1-data codebook.docx”: this codebook that lists all the variables in the data file. ------------------------------------------------------ [2] The DCE development [2.1] Experimental choice design development This section contains the following three files: "SIPHER7-DCE-Personal-6-1-1-design.ngs": this file contains Ngene codes for the initial design (before partial design), which is an efficient design with 120 pairwise choice tasks comprising 240 scenarios. "SIPHER7-DCE-Personal-6-1-1-partial design.ngs": this file contains Ngene codes for the final design drawn from the full profile that was generated from the initial design (i.e., a full set of candidate choice tasks that comprised all parings of scenarios within the initial design that had two out of the six non-income attributes tied. There are 3600 such pairings). "SIPHER7-DCE-Personal-6-1-1-master-design.xlsx": this file contains relevant information and data to the final DCE design used in the survey. [2.2] Quantitative pilot development The quantitative pilot was conducted in August 2020 with 100 respondents using the same internet panel and sampling frame as the main survey. The quantitative pilot provided the priors of the parameters used in the final DCE design. This section contains the quantitative pilot survey in three different file types: "SIPHER7-6-1-1-Quantitative pilot survey.qsf": This is the Qualtrics file that includes the survey questions, the logic conditions and flows embedded in the quantitative pilot survey on Qualtrics. Those who are interested in the Qualtrics settings can import this file into Qualtrics to explore the details of how the quantitative pilot was set-up online. "SIPHER7-6-1-1-Quantitative pilot survey.pdf": This is the exported presentation from Qualtrics that shows how the quantitative pilot survey was presented to respondents. "SIPHER7-6-1-1-Quantitative pilot survey-CODEBOOK.docx": This document shows the logic conditions used in the quantitative pilot survey, the wording of each question and numerical coded values for different answers in each question. [2.3] The main survey development The main survey collected data from the general public through two waves: the first one was in October 2020 and the second one was in November 2021. This section contains five files related to the development of the main survey. "SIPHER7-6-1-1-Main_Survey_Oct2020.qsf": This file is directly exported from Qualtrics and includes the survey questions, the logic conditions and flows embedded in the main survey on Qualtrics. Those interested in replicating the DCE survey can do so by importing this file into Qualtrics. "SIPHER7-6-1-1-Main survey_Oct2020.pdf": This is the exported presentation from Qualtrics that shows how the main survey was presented to respondents. "SIPHER7-6-1-1-Main survey_Oct2020-CODEBOOK.docx": This document shows the logic conditions used in the main survey, the wording of each question and numerical coded values for different answers in each question. "SIPHER7-6-1-1-Information_sheet_online16April2020(mainsurvey).docx": This is the information sheet that appeared at the beginning of the main survey and respondents were given the option to download. ------------------------------------------------------- [3] The main DCE data files The data exported from Qualtrics have information from all of the background questions, the DCE choice tasks and COVID-19 related questions. In addition, as the surveys included time tracking and number of clicks to exclude speeders, the data downloaded also contained information corresponding to number of clicks per question and time spent on each question. Please note that the data in this archive are NOT raw data but pre-analysis clean data as the data from both waves were cleaned by removing observations by speeders (i.e. those finishing the surveys in less than 5 minutes) and variables such as time spent and number of clicks. This section contains 13 files: "SIPHER-7-Personal-6-1-1.xlsx": This workbook contains two wave data from Qualtrics (tab "WAVE1" and tab "WAVE2)) and matching information (tab "matching" and tab "recontacted" to identify those respondents who participated in either wave or in both waves). Data from this workbook are in a format that each respondent's answers were recorded in one row and are not suitable to run regressions on without reshaping (Details are in Section [4]: The analyses). "20022023-full-reshaped-income-equi-Pool-exc_w2-desstat-testweights_P1C1-C2.dta": This is the STATA clean data (i.e. pre-analysis data after being reshaped, treated missing variables, recoded where needed) that are ready for regressions "SIPHER_distribution_weights.xlsx": This file contains sample weights calculated for the dataset for users who are interested in using sampling weights to correct for the different sample composition across the two waves. Other 10 data files that are used to match income values from the DCE choice tasks to the collected data are from the series “201025-Soft-launch-design-income1.dta” to “201025-Soft-launch-design-income10.dta”. ------------------------------------------------------ [4] The analysis files This section contains one do-file and 10 data files. "SIPHER-DO-Main-survey-2waves-mixlogit-revision.do": This file is the full Stata syntax that consists of all commands used in the analyses. The main stages described in this do-file are: (1) importing data from the Excel workbook, (2) reshaping the data to make a compatible format for regressions later (i.e. each person will have 20 observations that capture 2 alternatives that they faced within the 10 DCE choice tasks and their decisions), (3) matching survey data with the income decile thresholds using data from series “201025-Soft-launch-design-income1.dta” to “201025-Soft-launch-design-income10.dta”, (4) cleaning and manipulating data (e.g. recoding some values for missing and not answered, re-labelling variables, etc), (5) analysing: running regressions on different samples and sub-samples from the dataset and plotting of estimated coefficients and willingness to pay, calculating equivalent income in different hypothetical scenarios. Finally, for those who are interested in sample weights, the last part of the analyses explored sample weight across sub-samples in both waves of the main survey. -------------------------------------------------------- References Ta, A., Van Landeghem, B. and Tsuchiya, A., (2024). Eliciting public preferences across health and wellbeing dimensions: An equivalent income value set for SIPHER‐7. Health Economics, 33(12), pp.2723-2741. Tsuchiya A, & Wu C (2021), SIPHER 7: a seven indicator outcome measure to capture wellbeing for economic evaluation, SIPHER Research Paper Series 1, https://www.gla.ac.uk/media/Media_970691_smxx.pdf --------------------------------------------------------- For queries on this collection, please contact Aki Tsuchiya