There are two questions about UK general elections: vote7 (“Did you vote in this (past) year’s general election?”) and vote8 (“Which political party did you vote for?”). Due to the length of the fieldwork in UKHLS (up to 30 months to complete one wave, see https://www.understandingsociety.ac.uk/documentation/mainstage/survey-timeline) it might happen that some people participating in the wave currently in field are interviewed before the General Election. As a result, those respondents were asked about that General Election in later waves. In other words, the same event was captured in multiple waves of the UKHLS data. Additionally, the irregularity of the election calendar in recent years has further complicated the process.
The General Elections held in the UK between 2010 and 2019 were then captured in the following UKHLS waves:
1) the General Election on the 6th of May 2010 in wave 2,
2) the General Election on the 7th of May 2015 in wave 7,
3) the General Election on the 8th of June 2017 in waves 8, 9, 10,
4) the General Election on the 12th of December 2019 in waves 11, 12.
Here is a detailed summary of how the vote7 question was asked across waves:
Wave 2: if sample month was May 2010 to December 2010 // if inrange(b_month,5,12)
Wave 7: if interview was after the General Election (the 7th of May 2015) and in 2015 // if g_doi>td(07/05/2015) & g_doi<td(01/01/2016)
Wave 8: ((if mode was face-to-face AND has agreed to self-completion) OR mode was telephone or web) AND If interview was after the General Election (the 8th June 2017) AND issued in sample months from June 2017 to December 2017 / if ((h_indmode==1 & h_scflag_dv==1) | h_indmode==2 | h_indmode==3) & inrange(h_month,18,24) & h_doi>td(08/06/2017)
Wave 9: (Mode is face-to-face and has agreed to self-completion OR mode is telephone OR web) And interview is after the General Election (the 8th June 2017) AND issued in sample months from June 2017 to May 2018 // if ((i_indmode==1 & i_scflag_dv==1) | i_indmode==2 | i_indmode==3) & i_doi>td(08/06/2017) & inrange(i_month,6,17)
Wave 10: (Mode is face-to-face and has agreed to self-completion OR mode is telephone OR web) AND issued in sample months January 2018 to May 2018 // if ((j_indmode==1 & j_scflag_dv==1) | j_indmode==2 | j_indmode==3) & inrange(j_month,1,5)
Wave 11: (Mode is face-to-face and has agreed to self-completion OR mode is telephone OR web) AND the interview was after the General Election (the 12th of December 2019) // if ((k_indmode==1 & k_scflag_dv==1) | k_indmode==2 | k_indmode==3) & k_doi>td(12/12/2019)
Wave 12: (Mode is face-to-face and has agreed to self-completion OR mode is telephone OR web) // if ((l_indmode==1 & l_scflag_dv==1) | l_indmode==2 | l_indmode==3)
Note that the “month” variables (e.g., h_month, i_month) do not necessarily match standard calendar months (1 = January, 2 = February, etc.). They reflect a wave-specific 24-month fieldwork period, so, for instance, “18” correspond to June of the second year of fieldwork for a given wave.
To make analysis easier, we have created a set of derived variables picking up the information about voting in each election from all waves in which the question was asked into a single variable. The code/syntax creating these variables is available here: https://www.understandingsociety.ac.uk/documentation/mainstage/syntax/user-deposited-syntax/creating-uk-general-election-variables/. The variables in the file also correct for minor routing errors.
