Post-Election Survey, January - February 2021

Probability ( DDI Alliance Controlled Vocabulary for Sampling Procedure)

Random digit dialing was employed for selecting respondents.

Primary Sampling Units

Telephone numbers (both – landline and mobile) were randomly generated following this logic:
1) all numbers started with “8” (general code for dialing within Lithuania);
2) then city or mobile codes from the database (containing all the existing codes in Lithuania, and consisting of two or three digits) were randomly added;
3) and then remaining random five or six numbers were added, so that finally generated telephone number would consist of nine digits.
In total 120.000 randomly generated numbers were uploaded into the calling systems and used for the survey. Two different calling systems were used as part of interviewers were working from the office and another part was working remotely.

Sample Selection

Interviewers were calling randomly generated telephone numbers, and (in case there was eligible answer) asked respondents if they have a right to vote in Lithuania (i.e. he or she is 18+ y. o. and has Lithuanian citizenship). If yes, then respondents were invited to participate via the WEB survey platform. If respondents agreed to participate via the WEB survey, they were asked to provide their mobile telephone number or email address to send a WEB survey link. If respondents did not agree to participate via the WEB survey platform, they were invited to answer the survey questions continuing the same telephone call (CATI survey). In both cases (WEB/CATI) the main survey started with screening eligibility questions (citizenship/18 y. o. or more). The calling system automatically recorded the statuses of the phone numbers, if they are answered / not answered / wrong telephone number. If respondent answered, then interviewer noted the status – refusal / screened / complete / appointment.

The post stratification weights were constructed in order to match demographic characteristics of the population and correct for the official election results. There are two weights available WEIGHT1 and WEIGHT2. The first refers to weighting with regard to demographic characteristics and the second refers to weighting with regard to demographic characteristics and official election results. Both were constructed using the function rake() available in the R package survey. Five known population characteristics were used to construct weights: Gender, Age, Education, Settlement size, and Voting for a candidate list (the latter only for constructing WEIGHT2). Due to large realized sample imbalances some of the weights produced were extreme: as low as 0.15, and as high as 18.51. Therefore, both weights were trimmed using the function trimWeights() available in the R package survey (with the limits: lower – 1/5, upper - 5). Thus, WEIGHT3 is a trimmed version of the WEIGHT1, and WEIGHT4 is a trimmed version of the WEIGHT2.

Demographic characteristics of the population were taken from the Lithuanian official statistics portal: https://osp.stat.gov.lt.
Voting results were taken from the Central Electoral Commission of the Republic of Lithuania: www.vrk.lt/en/2020-seimo/rezultatai.

Data users are strongly advised to include post stratification weights into their analyses. The most recommended option is WEIGHT4, however, other options could also be considered depending on the goals of the analysis.

Pagrindinė tyrimo kalba – lietuvių kalba (lit).

Standartizuotas klausimynas respondentams buvo pateikiamas lietuvių kalba (lit).

Dataverse kolekcijoje "Rinkimų studijos" talpinamos apklausos, tiriančios Lietuvos rinkėjų rinkiminę elgseną ir socio-politines nuostatas nacionalinių ir vietos rinkimų metu.

Porinkiminė apklausa, 2012 m. lapkritis - gruodis

Porinkiminė apklausa, 2016 m. lapkritis - gruodis

Tikimybinė (DDI aljanso Sampling Procedure kontroliuojamas žodynas )

Atrenkant respondentus naudotas atsitiktinis skaitmeninis rinkimas

Postratifikaciniai svoriai buvo sudaryti siekiant atitikti demografines gyventojų charakteristikas ir pakoreguoti oficialius rinkimų rezultatus. Galimi du svoriai WEIGHT1 ir WEIGHT2. Pirmasis reiškia svorius, atsižvelgiant tik į demografines charakteristikas, o antrasis - svorius, atsižvelgiant į demografines charakteristikas ir oficialius rinkimų rezultatus. Abu svoriai sudaryti naudojant funkciją rake(), esančią R pakete survey. Svoriams sudaryti buvo naudojamos penkios žinomos gyventojų charakteristikos: lytis, amžius, išsilavinimas, gyvenamosios vietos dydis ir balsavimas už kandidatų sąrašą (pastarasis naudojamas tik WEIGHT2). Dėl didelio realizuoto imties disbalanso kai kurie iš gautų svorių buvo kraštutiniai: mažiausiai nuo 0,15, daugiausiai iki 18,51. Todėl abu svoriai buvo apkarpyti naudojant funkciją trimWeights(), kuri yra R pakete survey (su ribomis: apatinė - 1/5, viršutinė - 5). Taigi, WEIGHT3 yra apkarpyta WEIGHT1 versija, o WEIGHT4 yra apkarpyta WEIGHT2 versija.

Demografinės gyventojų charakteristikos buvo imamos iš Lietuvos oficialaus statistikos puslapio: https://osp.stat.gov.lt.
. Balsavimo rezultatai buvo imami iš Lietuvos Respublikos vyriausiosios rinkimų komisijos: www.vrk.lt/en/2020-seimo/rezultatai.

Duomenų naudotojams primygtinai rekomenduojama į savo analizes įtraukti postratifikacinius svorius. Labiausiai rekomenduojama parinktis yra WEIGHT4, tačiau, atsižvelgiant į analizės tikslus, galima apsvarstyti ir kitas parinktis.