2016 Time Series: Updates & Errata
Current Dataset Release: September 4, 2019 version
September 4, 2019
The September release includes the following additions or changes:
• Additional sample variables and sample variable placeholders have been added (V163003-V163008)
• Additional pre/post summary variables have been added (V162063x & V162064x)
• Previously unreleased cases have been added to V162129 (web cases) and V162367 (face-to-face cases)
• Missing data previously coded as -1 (Inap) in summary variables have been recoded to -2 (Missing)
• Minor edits to variable and value labels
• Erratum from February 5, 2019 affecting variables V162032x and V161121-V161125 have been resolved and updated in this release.
On the Auxiliary File: 2016 Methodology File, for case (V160001_orig) 302252 the variable nrfu_mainstatus should be changed from “1. Responded to Time Series” to “2. Non-respondent (to Time Series).”
February 5, 2019
Note 1: V162032x variable label is not fully descriptive. V162032x describes reported vote participation only for respondents who were registered to vote. The variable label for V162032x should read “POST: SUMMARY – Post vote status for registered respondents.”
Note 2: Data for V162065x, providing turnout-registration data for all respondents having known status, will be available in the next release. Until then, it can be created as follows from V161024x and V162032x:
V162065x = -1 ;
if V161024x eq 1 and V162032x eq 1 then V162065x = 0 ;
if V161024x eq 1 and V162032x eq 2 then V162065x = 1 ;
if V161024x eq 1 and V162032x eq 3 then V162065x = 2 ;
if V161024x eq 2 and V162032x eq 1 then V162065x = 0 ;
if V161024x eq 2 and V162032x eq 2 then V162065x = 1 ;
if V161024x eq 2 and V162032x eq 3 then V162065x = 2 ;
if V161024x eq 3 and V162032x eq 2 then V162065x = 1 ;
if V161024x eq 3 and V162032x eq 3 then V162065x = 2 ;
if V161024x eq 4 and V162032x eq -7 then V162065x = 2 ;
if V161024x eq 4 and V162032x eq -6 then V162065x = 2 ;
if V161024x eq 4 and V162032x eq -1 then V162065x = 2 ;
(V162065x codes: -1. Inap, 0. Did not register and did not vote, 1. Registered but did not vote, 2. Registered and voted).
In the 2016 Time Series codebook, variables V161121-V161125 include 1 reference to the Democratic Presidential candidate (Hillary Clinton) but should read “Donald Trump” (Republican Presidential candidate).
December 18, 2018
This release includes previously unreleased religion and Hispanic origin summary variables. Also added: coded versions of the Most Important Problem questions, interview language, randomization, screener (FTF & Web; subset of complete interviews from the 2016 Methodology File), interviewer description, candidates & races (House, Senate & Governor) variables. Corrections (please see the user guide for further details) have been made to several religion, occupation and industry, post thermometer, and interview time and date variables. Variable V161317 has been recoded per the erratum notice from October 2, 2018.
One case (original case ID 302252) from previous releases has been deleted from the file after the case was determined to have been ineligible.
Defense spending 7-point scales (variables V161181, V161182, V161183) were correctly administered to respondents although the labels for the endpoint categories in the codebook documentation represent an error in the face-to-face instrument. Endpoint labels in the codebook represent what appeared onscreen to interviewers during administration of the survey. However, although incorrect, they do not represent the endpoint labels which were correct in the Respondent Booklet (used by face-to-face respondents to make a number selection from the scale) and correct in the Web instrument . The 2016 data for these variables should be considered representing questions administered comparably to other years.
Web cases in 2016 Time Series Study variable V161317 require a recoding of values as follows: code 0 to 4, code 1 to 0, code 2 to 1, code 3 to 2, and code 4 to 3.
The ANES 2016 Time Series data have been re-released with several additional variables, including occupation and industry codes, the timing of the respondent’s voting decision, interviewer observations about the dwelling units in the face-to-face sample, a summary score for “need to evaluate,” and reasons for non-registration to vote.
The May 2, 2017 version of the initial release includes new variables containing interviewer observations, including the interviewer’s observation of the respondent’s skin tone. The data contained in these variables were collected for face-to-face interviews only. Interviewer observations were recorded following both the pre and post election surveys.
Updates to the ANES 2016 Time Series study are available. The April 3, 2017 version includes important changes to the data for the following variables: Party ID (V161155 & V16158x), sexual orientation of R (V161511), pre-election survey media variables, party to deal with most important problem (V162117, V162119, V162121), country needs free thinkers (V162168), discrimination in the US (V162357-66), self discrimination (V162367). Corrections were also made to the number of post election cases with no-post interview. Many variable labels and code labels have also been updated.
We thank several data users, including Alan Abramowitz, Ismail White, and Gary Jacobson, for noting a problem with the Party Identification data. The problem reflects a data processing error in which Internet questionnaire randomization was not correctly accounted for. The face-to-face data were not affected. We are preparing a corrected dataset for release as soon as possible.