Picking religion questions when survey space is limited Kelsey Jo Starr Jonathan Evans Research Analyst Research Analyst
Our guiding principles • An independent, nonpartisan “fact tank” • Strict nonadvocacy stance • No commercial or market research • Conduct opinion polling, demographic research, content analysis and other empirical social science research • Promote a deeper understanding of issues at the intersection of religion and public affairs, among other topics 2
Research Question • Those ose condu nducti cting ng sur urveys ys on non-religion religion topics cs often n want nt to include lude que uesti tion ons s on religion gion for repor ortin ting g or for us use as statis istica tical l contr ntrols ols in explainin laining social, ial, political tical or econom onomic c attitu tudes des. • But ut which ch shou ould ld you us use if you u only y have e space e for one e or two o relig igion ion que uesti tion ons? s? 3
About the surveys Af Africa ca Latin in Am Ameri erica ca • 25,000+ interviews conducted • 30,000+ interviews conducted across 19 countries across 18 countries & Puerto Rico • Conducted Dec. 2008-April 2009 • Conducted Oct. 2013-Mar. 2014 • Face-to-face • Face-to-face Predo dominant minantly y Mus uslim m Countri ntries es Central ntral and Ea East stern ern Eu Europe pe • 32,000+ interviews conducted • 29,000+ interviews conducted across 26 countries across 18 countries • Conducted Oct. 2011-May 2012 • Conducted June 2015-July 2016 • Face-to-face • Face-to-face United Un ed States es West stern ern Eu Europe pe • 35,000+ interviews • 24,000+ interviews conducted across 15 countries • Conducted June-Sept., 2014 • Conducted April-Aug., 2017 • Cell and landline • Cell and landline 4
Religion’s importance varies widely around the world % who say religion is very/somewhat important in their life Africa (regional median) 97 97 Predominantly Muslim Countries (regional 93 93 median) Latin America (regional median) 93 93 United States 78 78 Central and Eastern Europe (regional median) 72 72 Western Europe (regional median) 46 46 5
ARE THERE E ONE OR TWO O QUESTIONS IONS THAT T CA CAN BE USED CONSISTENTL TENTLY Y AROUND ND THE WORLD LD? 6
Approach • Regiona onal logist stic c regress ssion ion models s predicti cting: ng: Support for legal gay marriage • Support for legal abortion • • Thinking homosexual behavior is immoral • Thinking abortion is immoral • Tested ed re religio ion n questio stions ns that t have e been en include uded around d the worl rld: d: • Importance of religion in individual’s life (religious salience) • Religious service attendance • Prayer • Belief in God • Religious affiliation • Fasting • Donating funds to religious organization (e.g. tithing, zakat) • Belief in the evil eye • Belief in reincarnation • Controlling for age, gender, education, satisfaction with country’s direction, persona onal economic omic situat ation, ion, countr try fixed ed effects ects 7
Africa: morality of abortion No religion Full model variables 50% 76% Base predicted probability Relative influence of each factor on the likelihood of a respondent to say that abortion is morally wrong +14 Belief ief in God +11 Religious gious salie ience nce +9 Daily prayer +3 Monthly attendance -11 11 Christi tian an (vs. unaffilia iliated ed) -12 12 Muslim im (vs. unaffilia iliated) ed) +5 Pays tithing/zakat +4 Fasts +4 Believes in evil eye -1 Believes in reincarnation +2 +2 Male -1 -1 Young (18 to 34) -4 -2 Educated Note: Affiliation with tribal religion or other religion/don’t know/refused also included in models 1 & 3. Source: Survey conducted Dec. 2008-April 2009 in 19 African countries. PEW RESEARCH CENTER 8
Africa: morality of abortion No religion Full model variables Only affiliation 50% 76% 72% Base predicted probability Relative influence of each factor on the likelihood of a respondent to say that abortion is morally wrong +14 Belief ief in God +11 Religious gious salie ience nce +9 Daily prayer +3 Monthly attendance -11 11 +4 +4 Christi tian an (vs. unaffilia iliated ed) -12 12 +6 +6 Muslim im (vs. unaffilia iliated) ed) +5 Pays tithing/zakat +4 Fasts +4 Believes in evil eye -1 Believes in reincarnation +2 +2 +2 Male -1 -1 -1 Young (18 to 34) -4 -2 -2 Educated Note: Affiliation with tribal religion or other religion/don’t know/refused also included in models 1 & 3. Source: Survey conducted Dec. 2008-April 2009 in 19 African countries. PEW RESEARCH CENTER 9
Africa: morality of abortion No religion Full model variables Only affiliation Only salience Only God 50% 76% 72% 60% 58% Base predicted probability Relative influence of each factor on the likelihood of a respondent to say that abortion is morally wrong +14 +19 Belief ief in God +11 +17 Religious gious salie ience nce +9 Daily prayer +3 Monthly attendance -11 11 +4 +4 Christi tian an (vs. unaffilia iliated ed) -12 12 +6 +6 Muslim im (vs. unaffilia iliated) ed) +5 Pays tithing/zakat +4 Fasts +4 Believes in evil eye -1 Believes in reincarnation +2 +2 +2 +1 +1 Male -1 -1 -1 -2 -2 Young (18 to 34) -4 -2 -2 -4 -4 Educated Note: Affiliation with tribal religion or other religion/don’t know/refused also included in models 1 & 3. Source: Survey conducted Dec. 2008-April 2009 in 19 African countries. PEW RESEARCH CENTER 10
Latin America: legality of abortion No religion Full model variables Only affiliation Only salience Only God 20% 7% 12% 12% 18% Base predicted probability Relative influence of each factor on the likelihood of a respondent to say that abortions should be legal -2 0 Catholic olic (vs. . unaffiliat liated) ed) -2 Daily prayer -2 Monthly attendance -4 -5 Religious gious salie ience nce -6 -7 Prot otestant tant (vs. unaffilia iliated) ed) -9 -11 11 Belief ief in God +5 Believes in reincarnation +2 Believes in evil eye 0 Pays tithing -3 Fasts +11 +5 +8 +8 +10 Educated 0 +1 +1 +1 +1 Male -1 0 0 0 0 Young (18 to 34) Note: Affiliation with other religion/don’t know/refused also included in models 1 & 3. Source: Survey conducted Oct. 2013-Mar. 2014 in 18 Latin American countries and Puerto Rico. PEW RESEARCH CENTER 11
Latin America: legality of abortion Affiliation, No religion Affiliation and Affiliation and God and Full model variables salience God salience 20% 7% 15% 21% 22% Base predicted probability Relative influence of each factor on the likelihood of a respondent to say that abortions should be legal -6 -3 -2 -4 Catholic olic (vs. . unaffiliat liated) ed) -2 Daily prayer -2 Monthly attendance -5 -4 -5 Religious gious salie ience nce -11 -9 -6 -8 Prot otestant tant (vs. unaffilia iliated) ed) -10 -9 -9 Belief ief in God +5 Believes in reincarnation +2 Believes in evil eye 0 Pays tithing -3 Fasts +9 +10 +11 +11 +5 Educated 0 +6 0 0 +1 Male 0 0 0 -1 0 Young (18 to 34) Note: Affiliation with other religion/don’t know/refused also included in models 1 & 3. Source: Survey conducted Oct. 2013-Mar. 2014 in 18 Latin American countries and Puerto Rico. PEW RESEARCH CENTER 12
Conclusions • In several eral regions, ns, belief ef in God is the consi sist stentl ently best st proxy y for all ot other er re religio ion questio stions. ns. However er, , the interpre reta tati tion on or re r relevance nce of belie ief f in God can vary betw etwee een n regions. ns. For exampl ple, e, • In a Muslim context, asking about belief in God can be offensive or is often asked as part of Islam’s first pillar – belief in one god, Allah, and his prophet Muhammed – which is double-barreled. • The regions included are predominantly Christian and Muslim, and in other religious contexts the impact may be different. • Salience ence of religi gion on is frequ quently ently the secon cond d best st proxy y in each h region on (if not ot the best), st), and in no region on yet et tested ed does s it not ot wo work. On its own, , re religious ous affilia iliati tion on typical cally y does s not ot work as the proxy for r overa rall • religiosit iosity. But when combined with salience or belief in God, it gets much closer to replicating • the full model. 13
Future research • Use data a from ot other er sources ces to see if withi hin-regi egion on patterns erns hold • Use data from ot other er re regions ns to see if general eral patt tterns erns hold in Asia (amon mong g non- Ab Abrah ahami amic c fa faiths) ths) • Run model els s that t includ ude e all regions ons at once 14
Contact Information Kelsey Jo Starr Jonathan Evans Research Analyst Research Analyst kstarr@pewresearch.org jevans@pewresearch.org @IAmStarrStuff 15
Recommend
More recommend
