| Data collection method: |
MOBILISE— Determinants of ‘Mobilisation’ at Home & Abroad Wave 1 Nationally Representative Survey Technical Report The data collection strategy of the MOBILISE Project consists of a multi-wave nationally representative panel survey of the 18+ citizen populations of our four “home” country cases (Ukraine, Poland, Argentina, and Morocco). The primary aim of our survey is to collect information on citizens’ attitudes towards, dispositions around, and personal experiences of protest and migration specifically. We also collect other political, sociological, and demographic factors pertinent to our study. Questionnaire
To this end, the originally designed questionnaire (attached in Appendix), composed by the MOBILISE Project PIs/CoIs and reviewed by an external Expert Advisory Board, was divided into 5 sections: 1) Media Consumption, 2) Political dispositions and engagement, 3) Migration experience and intention, 4) Other issue positions, policy preferences and item opinions, and 5) Demographics. The questionnaire also had embedded within it a list experiment (testing issues around repression) and a framing experiment (testing the possible trade-off between migration and protest). Temporal Scope Wave 1
As part of the first wave, nationally representative surveys were conducted in Ukraine, Poland, Argentina, and Morocco respectively. The fielding of the first wave was timed, as best as possible given the circumstances, to immediately precede (Ukraine & Poland) or directly follow (Argentina) major national elections: presidential (Ukraine and Argentina) and parliamentary (Poland). The aim was to connect the survey field to electoral cycles. Table 1. Survey Dates by Country
Country Period Wave 1 Survey
Ukraine March-April 2019
Poland October 2019
Argentina February-March 2020 Sample
The sampling design of the first wave of each of our panel surveys has two aims: 1) to capture a random nationally representative sample (with a number of observations that approximates N = 1,600 in all country-cases), and 2) to increase the likelihood of capturing a large enough sample of both would be migrants and protesters (the subjects of our study), for which we oversampled 18+ residents of the five largest major urban localities in each country (with a number of observations that approximates N = 400 in all country cases). Mode
The surveys for the first wave were all planned to be conducted employing the computer-assisted personal interview (CAPI) mode. That is, they were designed as face-to-face surveys where canvassers use a tablet device to record the answers. This mode was followed thoroughly for Ukraine and Poland. Due to the reasons detailed below, the CAPI mode was only partially followed in Argentina and Morocco. The onset of the COVID-19 pandemic in February-March 2020, and the enactment of ensuing government mitigation policies, such as lockdowns, resulted in the Argentinian Wave 1 fieldwork being paused three days. Upon its recommencement the remaining 159 interviews were completed employing Computer Assisted Telephone Interviewing (CATI) mode. In Morocco, fieldwork was paused on March 19th with an N of 1,688 interviews completed. Our partners did not have the capacity at the time to move to CATI mode and the remainder of the survey was suspended. Table 1 summarize the dates, mode, nationally representative sample and oversample sizes, the urban localities compensated for, as well as our ‘local partners’, that is, the companies fielding the surveys in each country.
Table 1: Summary Information of Wave 1 Nationally Representative Surveys in Each Country Cases
Country Survey Company Fieldwork Dates Mode Sample Size
(Nat. Rep. Sample) Sample Size
(Oversample) Urban localities oversampled
Ukraine KIIS* Mar. 4 – Apr. 12, 2019 CAPI 1,600 400 Kyiv, Odesa, Dnipro, Kharkiv and Lviv.
Poland CBOS** Aug. 20 – Sep. 24, 2019 CAPI 1,618 400 Cracow, Łódź, Poznań, Warsaw and Wrocław.
Argentina ELDASA Feb. 24 – Mar. 29, 2020 (survey paused from March 21st to March 24th) CAPI (1,841) following first COVID-19 mitigation related lockdown CATI (159) 1,500 500 Ciudad Autónoma de Buenos Aires, Córdoba, Rosario, La Plata, Mar del Plata.
Morocco GSC*** March 1st – 19th 2020 (surveying was interrupted with onset of COVID-19 lockdown) CAPI face to face (1,688) 1,350 338 Casablanca, Fès, Tanger, Rabat, Salé, Marrakech
* Kyiv International Institute of Sociology
** Fundacja Centrum Badania Opinii Społecznej
*** Global Survey and Consulting Country Specific Additional Information Ukraine Wave 1
Response Rate I=Complete Interviews 2,000
Number of Potential Interviewees
Contacted 6,599
Response Rate 0.325
Sampling Procedure With the database of the Voting Commission, three-staged Stratified random sampling was used. The first stratification was by regions (24 oblasts and the city of Kyiv). The second strata differentiated between urban and rural population. The third and last strata identified voting precincts in proportion to the adult population. In total, the population of Ukraine was divided into 49 strata and 200 precincts in which 8 respondents had to be surveyed. To begin fieldwork our partners in KIIS selected election precincts, with a probability proportional to the number of registered voters within the district (PPS, probability proportional to size). The starting point in each sampling unit was then selected randomly and the subsequent apartments were selected following a random walk system. The respondent within each household is selected using an “each third person” method. For the oversample (N=400) of would be migrants and protesters in urban localities, the same procedure was followed: First, the random selection of electoral precincts, then a random selection of households, and a selection of respondents with the each-third-person method. Between March 27th and April 1st, 2019. 6,599 total contacts were made with 2,000 completed interviews. In case respondents were not at home or could not answer the survey in the moment of the visit, canvassers returned to the household a minimum of three times before the dwelling was replaced.
1,995 (99.75% of) of interviews were conducted before 1st round of the Presidential election of 2019 and only 5 interviews from oversampling were conducted on April 1st.
Representativeness The sample is representative of 18+ years old permanent Ukrainian residents who are neither currently serving in the army, nor imprisoned or staying in medical facilities (hospitals, live-in medical facilities). The sample does not include the territories which are temporarily not controlled by the Ukrainian government: Crimea, some districts of the Donetsk and Luhansk Oblasts.
Weights Women and older people are overrepresented in the sample. As a corrective, two sets of ‘probability’ weights were estimated by our partners in KIIS. One corrects the nationally representative sample only, and the second one corrects both the nat. rep. sample, and the oversample. Poland Wave 1
Response Rate I=Complete Interviews 2,018
Number of Potential Interviewees
Contacted 7,004
Response Rate 0.288
Sampling Procedure With the database of the Universal Electronic Population Registration System (PESEL database) from the Ministry of Digitisation, three-staged Stratified random sampling was used. The first step obtained roughly 60 strata distinguishing rural areas, as well as towns and cities pending on the number of inhabitants. Then in each stratum specific precincts were drawn with probabilities proportional to their respective population sizes. In the third and last step a pre-set number of respondent clusters were selected. Each cluster consisted of 10 individuals from the selected unit, which were in turn drawn in a simple, non-returnable way. For the oversample (N=400) of would be migrants and protesters in urban localities, the same procedure was followed: First, each city was divided into strata, then a random selection of districts within each strata with probabilities proportional to the respective sizes. Finally, respondents in the randomly drawn districts were further divided on the basis of their demographic characteristics so as to reflect the structure of the surveyed population in each city. Between August 20th and September 24th, 2019. 7,004 total contacts were made with 2,018 completed interviews. 1,834 (90.88% of) interviews were between August 20th and September 16th with 184 interviews conducted between the 17th and the 24th of September 2019.
Representativeness The sample is representative of 18+ years old permanent Polish residents.
Weights To keep the representativeness of the sample our partners at CEBOS determined weights on the basis of an iterative rim weighting procedure, taking into account the distributions of the following variables: voivodship (province) of residence, type of domicile (rural/urban), sex, age group and education.
CEBOS Report The full report by our partner CEBOS can be found here. Argentina Wave 1
Response Rate I=Complete Interviews 2,000
Number of Potential Interviewees
Contacted 5,787
Response Rate 0.346
Sampling Procedure With the last Census data from the Instituto Nacional de Estadística y Censos (INDEC), our partners at ELDASA implemented a four-stage probability sampling framework. Taking localities that make up 70% of the Argentine population as a representative base, the first step distinguished 3 localities in 6 provinces, obtaining roughly 18 strata. Then each stratum was randomly assigned a number or ‘sample points’ based on the map ordinate and the abscissa of the corresponding municipality or commune. In total 103 sample points were planned and distributed. In the third step clockwise homes were selected, where a respondent was interviewed in each of them. Finally, to select interviewees within each household two criteria were adopted: Only permanent residents were interviewed and the questionnaire was asked to the adult (18+) who most recently celebrated their birthday. The sampling of the booster was based on an even allocation of 500 respondents among the five largest Argentine cities: Ciudad Autónoma de Buenos Aires, Córdoba, Rosario, La Plata and Mar del Plata. The sampling procedure was similar to the one of the nationally representative sample. Sampling units were randomly selected within each city. Households and respondents were also selected following a random systematic selection of households and respondents at the final stage. Due to the Covid-19 pandemic, our partners changed from CAPI to CATI mode to implement the survey. Face to face interviews were conducted until March 3rd, 2021, and resumed by phone on March 25th, 2021. To call participants phone numbers were selected from telephone directories that corresponded to the already selected sampling units.
Representativeness The sample is representative of the non-rural 18+ years old permanent Argentine residents.
Weights To keep the national representativeness of the sample our partners at ELDASA determined ‘probability’ weights. To do so, they considered: sex, age groups, and level of education MOBILISE — Determinants of ‘Mobilisation’ at Home & Abroad Wave 2 Nationally Representative Survey Technical Report The data collection strategy of the MOBILISE Project consists of a multi-wave nationally representative panel survey of the 18+ citizen populations of our four “home” country cases (Ukraine, Poland, Argentina, and Morocco). The primary aim of our survey is to collect information on citizens’ attitudes towards, dispositions around, and personal experiences of protest and migration specifically. We also collect other political, sociological, and demographic factors pertinent to our study. To capture any shifts in attitudes overtime, this second wave was not only conducted across the same set of countries, but it also captured panel observations. That is, with the help of our local partners, the MOBILISE Project surveyed a considerable number of individuals which participated in the first wave also. Questionnaire
To this end, the originally designed questionnaire (attached in Appendix), composed by the MOBILISE Project PIs/CoIs and reviewed by an external Expert Advisory Board, was divided into 5 sections: 1) Media Consumption, 2) Political dispositions and engagement, 3) Migration experience and intention, 4) Other issue positions, policy preferences and item opinions, and 5) Demographics. The questionnaire also had embedded within it questions concerning the impact and perceptions around the COVID-19 pandemic, as well as a framing experiment (testing the possible trade-off between migration and protest). Temporal Scope Wave 2
Table 1. Survey Dates by Country
Country Period Wave 2 Survey
Ukraine October 2020 – April 2021
Poland November 2020 – January 2021
Argentina December 2021- February 2022 Sample
Sample descriptions Table 1: Summary Information of Wave 2 Nationally Representative Surveys in Each Country Cases
Country Survey Company Fieldwork Dates Mode Sample Size
(Nat. Rep. Sample) Sample Size
(Oversample) Urban localities oversampled
Ukraine KIIS* January 8th 2021 — February 25th 2021 CATI 1,387 253 Kyiv, Odesa, Dnipro, Kharkiv and Lviv.
Poland CBOS** November 26th 2020 — January 1st 2021 CAPI (31%)
CATI (69%) 1,105 400 Cracow, Łódź, Poznań, Warsaw and Wrocław.
Argentina ELDASA December 26th, 2021 — February 2nd, 2022 CAPI 1,210 302 Ciudad Autónoma de Buenos Aires, Córdoba, Rosario, La Plata, Mar del Plata. * Kyiv International Institute of Sociology
** Fundacja Centrum Badania Opinii Społecznej Country Specific Additional Information Ukraine Wave 2
Response Rate I=Complete Interviews 1,640
Number of Potential Interviewees
Contacted 9,142
Response Rate 0.116
Sampling Procedure With the database of the Voting Commission, three-staged Stratified random sampling was used. The first stratification was by regions (24 oblasts and the city of Kyiv). The second strata differentiated between urban and rural population. The third and last strata identified voting precincts in proportion to the adult population. In total, the population of Ukraine was divided into 49 strata and 200 precincts in which 8 respondents had to be surveyed. To begin fieldwork our partners in KIIS selected election precincts, with a probability proportional to the number of registered voters within the district (PPS, probability proportional to size). The starting point in each sampling unit was then selected randomly and the subsequent apartments were selected following a random walk system. The respondent within each household is selected using an “each third person” method. For the oversample (N=400) of would be migrants and protesters in urban localities, the same procedure was followed: First, the random selection of electoral precincts, then a random selection of households, and a selection of respondents with the each-third-person method. Between March 27th and April 1st, 2019. 6,599 total contacts were made with 2,000 completed interviews. In case respondents were not at home or could not answer the survey in the moment of the visit, canvassers returned to the household a minimum of three times before the dwelling was replaced.
1,995 (99.75% of) of interviews were conducted before 1st round of the Presidential election of 2019 and only 5 interviews from oversampling were conducted on April 1st.
Representativeness The sample is representative of 18+ years old permanent Ukrainian residents who are neither currently serving in the army, nor imprisoned or staying in medical facilities (hospitals, live-in medical facilities). The sample does not include the territories which are temporarily not controlled by the Ukrainian government: Crimea, some districts of the Donetsk and Luhansk Oblasts.
Weights Women and older people are overrepresented in the sample. As a corrective, two sets of ‘probability’ weights were estimated by our partners in KIIS. One corrects the nationally representative sample only, and the second one corrects both the nat. rep. sample, and the oversample. Poland Wave 2
Response Rate I=Complete Interviews 1,505
Number of Potential Interviewees
Contacted 2,018
Response Rate 0.746
Sampling Procedure With the database of the Universal Electronic Population Registration System (PESEL database) from the Ministry of Digitisation, three-staged Stratified random sampling was used. The first step obtained roughly 60 strata distinguishing rural areas, as well as towns and cities pending on the number of inhabitants. Then in each stratum specific precincts were drawn with probabilities proportional to their respective population sizes. In the third and last step a pre-set number of respondent clusters were selected. Each cluster consisted of 10 individuals from the selected unit, which were in turn drawn in a simple, non-returnable way. For the oversample (N=400) of would be migrants and protesters in urban localities, the same procedure was followed: First, each city was divided into strata, then a random selection of districts within each strata with probabilities proportional to the respective sizes. Finally, respondents in the randomly drawn districts were further divided on the basis of their demographic characteristics so as to reflect the structure of the surveyed population in each city. Due to the Covid-19 pandemic, our partners managed to re-interview 1,448 participants, which represents a 72% re-interview rate. This means that for the second nat. rep. survey in Poland there are 57 new participants.
Representativeness The sample is representative of 18+ years old permanent Polish residents.
Weights To keep the representativeness of the sample our partners at CEBOS determined weights taking into account the distributions of the following variables: voivodship (province) of residence, type of domicile (rural/urban), sex, age group and education. Argentina Wave 2
Response Rate I=Complete Interviews 1,512
Number of Potential Interviewees
Contacted
Response Rate 0.518 (panel respondents) Sampling Procedure With the last Census data from the Instituto Nacional de Estadística y Censos (INDEC), our partners at ELDASA implemented a four-stage probability sampling framework with the main objective of re-interviewing as many of the original respondents as possible. Taking localities that make up 70% of the Argentine population as a representative base, the first step distinguished 3 localities within 6 provinces, obtaining roughly 18 strata. Then each stratum was randomly assigned a number or ‘sample points’ based on the map ordinate and the abscissa of the corresponding municipality or commune. In total 103 sample points were planned and distributed. In the third step clockwise homes were selected, where a respondent was interviewed in each of them. Finally, to select interviewees within each household two criteria were adopted: Only permanent residents were interviewed and the questionnaire was asked to the adult (18+) who most recently celebrated their birthday. The sampling of the booster was based on an even allocation of 302 respondents among the five largest Argentine cities: Ciudad Autónoma de Buenos Aires, Córdoba, Rosario, La Plata and Mar del Plata. The sampling procedure was similar to the one of the nationally representative sample. Sampling units were randomly selected within each city. Households and respondents were also selected following a random systematic selection of households and respondents at the final stage. Due to the Covid-19 pandemic, our partners managed to re-interview 784 participants, which represents a 52% re-interview rate.
Representativeness The sample is representative of the non-rural 18+ years old permanent Argentine residents.
Weights To keep the national representativeness of the sample our partners at ELDASA determined ‘probability’ weights. To do so, they considered: sex, age groups, and level of education. MOBILISE PROJECT
Determinants of ‘Mobilisation’ at Home & Abroad TECHNICAL REPORT
Online Protesters Survey Belarus
24 March 2024 Cite as
Onuch O., Doyle D., Ersanilli E., Sasse,G., Toma S., Van Stekelenburg J., Espinoza F., Perez Sandoval J., Gonzalez Santos F., Michiels S., (2024). “MOBILISE Project Determinants of ‘Mobilisation’ at Home & Abroad. Technical Report Online Protesters Survey Belarus August 2020- January 2021.” Funding Note
We acknowledge funding associated with this research. Data described here were collected as part of the Open Research Area funded MOBILISE project (www.mobiliseproject.com) Grant Ref ESRC ES/S015213/1. Executive Summary
• This technical report outlines the MOBILISE Belarusian Protester Survey, conducted between August 18, 2020, and January 29, 2021. The survey aimed to gather data from both protesters and non-protesters during the protest wave in Belarus, utilizing online platforms due to pandemic restrictions.
• Targeted advertising on social media platforms like Facebook and Instagram was employed to reach users aged 18 and above residing in Belarus. Different advertisements targeted various demographics, emphasizing middle-aged individuals and women.
• Participants confirmed their age, citizenship, and residency before answering questions about protest participation. Protesters were further asked about their involvement in anti- or pro-government events. The survey also covered socio-economic status, political behaviour, and demographics.
• With a completion rate of over 40%, the survey garnered responses from 40,344 participants, with 17,174 completing it in full according to our specification (see below). Analysis focused on these completions for better comparability.
• Although not nationally representative, the survey's findings align with trends observed in nationally representative surveys conducted before, during, and after the protest wave in Belarus.
• The MOBILISE Belarusian Protester Survey offers valuable insights into the characteristics and attitudes of protesters and non-protesters during a significant period of social and political turmoil in Belarus.
• References are provided for further examination of the survey methodology and outcomes.
Aim of Survey
We note that this survey was not intended to be nationally representative – its aim was to collect as many protester responses as possible during the protest wave – but also collect enough non- protester responses to be able to compare between these two groups.
This approach was taken as:
(1) in the middle of an on-going pandemic it is neither practical nor is it ethical to collect face- to- face on-site protest survey data (the PIs’ past method of choice), and because
(2) a nationally representative survey would result in not only a very small n of protest participants – but this sub- sample would itself likely be highly unrepresentative of the actual protest population (for extended discussions of this please see: Ogan, Giglou, and d’Haenens 2017; Onuch 2011, 2014, 2014a, 2014b; Rosenfeld 2017; Walgrave, Wouters, and Ketelaars 2016).
And thus, we make no claims about the Belarusian population as a whole. Nonetheless, we note that the general patterns found among our survey respondents, hold and are replicated in nationally representative surveys collected before (Toal, O’Loughlin, and Bakke 2020), during the height (Greene 2020), and denouement of the protests (ZOiS 2020). Mode and Fielding
The Survey was fielded 18 August 2020 - 29 January 2021.
The MOBILISE Belarusian Protester Survey was run online employing the SurveyMonkey platform and its sampling design was social media generated using existing best practices (Andrews, Nonnecke, and Preece 2007; Samuels and Zucco 2013). Using mainly Facebook and Instagram adverts (whilst also allowing for link sharing across personal networks and Telegram channels), each collector (each targeted advert) had its own link – which we also cross-referenced with a question posed to the respondent about where they found the link.
On Facebook/Instagram we employed Facebook’s own advert/recruitment algorithm but we specifically targeted 18+ users, who reside in Belarus, speak Belarusian, and/or had Belarus as main listed interest. Because there is an established literature about the nature and efficacy of such recruitment (Kaye and Johnson 1999; Samuels and Zucco 2013; Van Selm and Jankowski 2006; Wright 2005), namely that middle aged people and women are more likely to engage – we also ran two separate adverts targeting men and youth to account for the oversampling of the two former groups.
Two targeted advert texts were used: (a) “Are you Belarusian?…” and (b) “Have you participated in Protests?...” followed up with “…please take this short Survey.” We ran these adverts in Belarusian, Russian, and English.
Figure 1 Example Collector list See Dataset for Full List
Collectors
Belarussians Abroad Interest Belarus Language All Countries Blr Facebook Blr Share Link End Of Survey Blr Survey Change Lang To Blr Male Oversampling Migrants PL LT Migrants UA RU Migrants UK ES DE Migrants V2 (Language Targeting In Ads) Oversampling Young Russ Facebook Russ Facebook Groups Russ Media Russ Share Link End Of Survey Russ Snowball Russ Survey Change Lang To Russ Russ Test Russ Twitter Expectedly, the most successful recruitment language was Russian. Respondents could also select/switch the language they completed the survey in on the welcome page. The survey was available in all three languages.
Users can assess the surveys by Collectors which is a variable in the dataset.
The survey began by asking a series of inclusion/exclusion criteria confirming questions: age (only those 18+ could continue/ are included), citizenship (only Belarusian citizens are included), and residency (only those who provided details of their location).
Following these inclusion/exclusion questions the respondents were asked if they participated in any of the recent protests that began on August 9 (online, in-person, or both online and in-person). Those who selected any one of these three options were then asked if they participated in anti-Lukashenka or pro-Lukashenka protest events. All of these self-confirmed protest participants were then directed to the protest survey - a series of items about the protests specifically. And following their completion were taken to the second portion of the survey – which asked a variety of typical political behaviour, evaluation, and attitudinal questions, as well as, a standard series of questions about socio-economic status and experience, and demographics.
All those who did not report being protest participants in that first question were taken directly to the second half of the of the survey – which was thus, answered by all respondents allowing for our comparison between protesters and non-protesters.
We also recruited Belarusians living abroad. They too answered the initial questions and were either taken to the protest specifics items or not. But once they completed the protest questions all respondents were again asked if they are abroad and those that answered positively were taken to questions specific for migrants. If they responded that they were not migrants they went straight to the main section of the survey which all respondents were asked to answer.
The survey took 25-35 minutes to complete in full (making it a rather long online survey). As is typical with online surveys more than 10 questions long the drop off rate is higher than in face-to-face CAPI or Telephone CATI surveys. We note though that majority of respondents who do drop off do so on the first page of the survey. Completion Rate and Inclusion
The completion rate for the survey is over 40% - a very respectable outcome when compared to other online surveys of this kind. In total 40,344 (37,785 Russian 2,486 Belarusian, and 73 English) respondents began the survey.
We remind that we had several inclusion criteria: having given consent, having answered the item on sex, having provided their year of birth, providing the respondents country location at the time of survey, and being of Belarusian citizenship (remaining n= 37,076).
Of those, 17,174 completed the survey in full (which we consider as having provided at least one response among the final battery of questions in the survey - those asking basic demographic questions).
Outside of age, sex, citizenship, and location of residence (all asked at the very beginning of the survey) the majority of demographic characteristics were collected at the last stage of the survey as is known best practice when surveying populations and especially when the topic is considered sensitive.
For this reason, in our analyses we only analyse those respondents who completed the survey in full as to allow for better comparison. Whilst users of these data may take a different approach this is the approach we recommend. MOBILISE PROJECT
Determinants of ‘Mobilisation’ at Home & Abroad TECHNICAL REPORT
Online Protester Survey Poland 24 March 2024 Cite as
Onuch O., Doyle D., Ersanilli E., Sasse,G., Toma S., Van Stekelenburg J., Espinoza F., Perez Sandoval J., Gonzalez Santos F., Michiels S., (2024). “MOBILISE Project Determinants of ‘Mobilisation’ at Home & Abroad. Technical Report Online Protesters Survey Belarus December 2020-March 2021.” Funding Note
We acknowledge funding associated with this research. Data described here were collected as part of the Open Research Area funded MOBILISE project (www.mobiliseproject.com) Grant Ref ESRC ES/S015213/1. Executive Summary: • This technical report details the MOBILISE Polish Protester Survey, conducted between December 26, 2020 and March 5, 2021. The survey aimed to gather data from both protesters and non-protesters during the protest wave in Poland, utilizing online platforms due to pandemic restrictions.
• Targeted advertising on social media platforms like Facebook and Instagram was employed to reach users aged 18 and above residing in Poland. Different advertisements targeted various demographics, emphasizing middle-aged individuals and women.
• Participants confirmed their age, citizenship, and residency before answering questions about protest participation. Protesters were further asked about their involvement in anti- or pro-government events. The survey also covered socio-economic status, political behaviour, and demographics.
• With a completion rate of over 44%, the survey garnered responses from 14,360 participants, with 11,610 meeting our inclusion specifications, and 6,819 completing the survey in full according to our team’s specifications. Analysis focused on these completions for better comparability.
• Although not nationally representative, the survey's findings align with trends observed in nationally representative surveys conducted before, during, and after the protest wave in Poland. Aim of Survey:
We note that this survey was not intended to be nationally representative – its aim was to collect as many protester responses as possible during the protest wave – but also collect enough non- protester responses to be able to compare between these two groups. This approach was taken as:
(1) in the middle of an on-going pandemic it is neither practical nor is it ethical to collect face- to- face on-site protest survey data (the PIs’ past method of choice), and because
(2) a nationally representative survey would result in not only a very small n of protest participants – but this sub- sample would itself likely be highly unrepresentative of the actual protest population (for extended discussions of this please see: Ogan, Giglou, and d’Haenens 2017; Onuch 2011, 2014, 2014a, 2014b; Rosenfeld 2017; Walgrave, Wouters, and Ketelaars 2016). And thus, we make no claims about the Polish population as a whole. Nonetheless, we note that the general patterns found among our survey respondents, hold and are replicated in nationally representative surveys collected by our project. Mode and Fielding:
The Survey was fielded 26 December 2020 and 5 March 2021. The MOBILISE Polish Protester Survey was run online employing the SurveyMonkey platform and its sampling design was social media generated using existing best practices (Andrews, Nonnecke, and Preece 2007; Samuels and Zucco 2013). Using mainly Facebook and Instagram adverts (whilst also allowing for link sharing across personal networks and Telegram channels), each collector (each targeted advert) had its own link – which we also cross-referenced with a question posed to the respondent about where they found the link. On Facebook/Instagram we employed Facebook’s own advert/recruitment algorithm but we specifically targeted 18+ users, who reside in Poland, speak Polish, and/or had Poland as main listed interest. Because there is an established literature about the nature and efficacy of such recruitment (Kaye and Johnson 1999; Samuels and Zucco 2013; Van Selm and Jankowski 2006; Wright 2005), namely that middle aged people and women are more likely to engage – we also ran two separate adverts targeting men and youth to account for the oversampling of the two former groups.
Two targeted advert texts were used: (a) “Are you Polish?…” and (b) “Have you participated in X Protests?...” followed up with “…please take this short Survey.” We ran these adverts in Polish and English. Expectedly, the most successful recruitment language was Polish. Users can assess the surveys by Collectors which is a variable in the datasets.
Figure 1 Example Collector list See Dataset for Full List
Collector
All-ages protest question
Main fb ads link PL
PL in PL 18-45yo
PL in PL 46+yo
Poles in Germany
Poles in Spain
Poles in UK
Protesters in PL 18-45
Protesters in PL 46+ Respondents could also select/switch the language they completed the survey in on the welcome page. The survey was available in two languages. The survey began by asking a series of inclusion/exclusion criteria confirming questions: age (only those 18+ could continue/ are included), citizenship (only Polish citizens are included), and residency (only those who provided details of their location). Following these inclusion/exclusion questions the respondents were asked if they participated in any of the recent protests that began on August 9 (online, in-person, or both online and in-person). Those who selected any one of these three options were then asked if they participated in a specific protest event. All of these self-confirmed protest participants were then directed to the protest survey - a series of items about the protests specifically. And following their completion were taken to the second portion of the survey – which asked a variety of typical political behaviour, evaluation, and attitudinal questions, as well as, a standard series of questions about socio-economic status and experience, and demographics. All those who did not report being protest participants in that first question were taken directly to the second half of the of the survey – which was thus, answered by all respondents allowing for our comparison between protesters and non-protesters. We also recruited Poles living abroad. They too answered the initial questions and were either taken to the protest specifics items or not. But once they completed the protest questions all respondents were again asked if they are abroad and those that answered positively were taken to questions specific for migrants. If they responded that they were not migrants they went straight to the main section of the survey which all respondents were asked to answer. The survey took 25-35 minutes to complete in full (making it a rather long online survey). As is typical with online surveys more than 10 questions long the drop off rate is higher than in face-to-face CAPI or Telephone CATI surveys. We note though that majority of respondents who do drop off do so on the first page of the survey. Completion Rate and Inclusion:
The average completion rate for the survey is over 44% - a very respectable outcome when compared to other online surveys of this kind. In total 14,360 (14,257 Polish, and 103 English) respondents began the survey. We remind that we had several inclusion criteria: having given consent, having answered the item on sex, having provided their year of birth, providing the respondents country location at the time of survey, and being of Polish citizenship (remaining n= 11,610). Of those, 6,819 completed the survey in full (which we consider as having provided at least one response among the final battery of questions in the survey - those asking basic demographic questions). Outside of age, sex, citizenship, and location of residence (all asked at the very beginning of the survey) the majority of demographic characteristics were collected at the last stage of the survey as is known best practice when surveying populations and especially when the topic is considered sensitive. For this reason, in our analyses we only analyse those respondents who completed the survey in full as to allow for better comparison. Whilst users of these data may take a different approach this is the approach we recommend. References
Andrews, Dorine, Blair Nonnecke, and Jennifer Preece. 2007. “Conducting Research on the Internet: Online Survey Design, Development and Implementation Guidelines.”
Kaye, Barbara K., and Thomas J. Johnson. 1999. “Research Methodology: Taming the Cyber Frontier: Techniques for Improving Online Surveys.” Social Science Computer Review 17(3): 323–37.
Ogan, Christine, Roya Imani Giglou, and Leen d’Haenens. 2017. “Challenges of Conducting Survey Research Related to a Social Protest Movement: Lessons Learned from a Study of Gezi Protests Involving the Turkish Diaspora in Three European Countries.” The Information Society 33(1): 1–12.
Onuch, Olga. 2011. “Revolutionary Moments and Movements: Understanding Mass-Mobilisation in Ukraine (2004) and Argentina (2001).” University of Oxford.
Onuch, Olga. 2014a. Mapping Mass Mobilizations: Understanding Revolutionary Moments in Ukraine and Argentina. London: Palgrave MacMillan.
Onuch, Olga. 2014b. “The Puzzle of Mass Mobilization: Conducting Protest Research in Ukraine, 2004–2014.” Reviews & Critical Commentary: Council of Europe.
Onuch, Olga. 2014c. “Who Were the Protesters?” Journal of Democracy 25(3): 44–51. doi:10.1353/jod.2014.0045.
Rosenfeld, Bryn. 2017. “Reevaluating the Middle-Class Protest Paradigm: A Case-Control Study of Democratic Protest Coalitions in Russia.” American Political Science Review: 1–16. doi:10.1017/S000305541700034X.
Samuels, David J., and Cesar Zucco. 2013. “Using Facebook as a Subject Recruitment Tool for Survey-Experimental Research.” https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2101458.
Van Selm, Martine, and Nicholas W. Jankowski. 2006. “Conducting Online Surveys.” Quality and quantity 40(3): 435–56.
Walgrave, Stefaan, Ruud Wouters, and Pauline Ketelaars. 2016. “Response Problems in the Protest Survey Design: Evidence from Fifty-One Protest Events in Seven Countries.” Mobilization.-San Diego, Calif., 1996, currens 21(1): 83–104.
Wright, Kevin B. 2005. “Researching Internet-Based Populations: Advantages and Disadvantages of Online Survey Research, Online Questionnaire Authoring Software Packages, and Web Survey Services.” Journal of computer-mediated communication 10(3): JCMC1034. |
