Panels
Weese, J.D., Lo, W.J., Crawford, B.L., Jozkowski, K.N., & Turner, R.C. (2020, June 11-12). How does inclusion of gestational length and order effects impact factor structure of medical and social abortion item sets? American Association of Public Opinion Research (AAPOR) Annual Meeting.
Abortion attitude questions that include information about the gestational age of a pregnancy often have different response distributions when compared with items that do not include information about the gestational age of a pregnancy (Bumpass, 1997). The order that gestational ages are presented (i.e., order effects) can also influence support for legalized abortion. To expand on Muthen’s (1981) study identifying a two-factor structure for six General Social Survey (GSS) abortion items and Bumpass’ (1997) study comparing order effects of GSS abortion items, our study examines how the gestational age of a pregnancy and order effects may alter the underlying factor structure of the six GSS abortion items. To further investigate, we added the following gestational ages to the original GSS question stem: 6 weeks, 12 weeks, and 20 weeks of pregnancy.
Participants were randomly assigned to receive the items with either ascending gestational ages (i.e., 6, 12, and 20 weeks) or descending gestational ages (i.e., 20, 12, and 6 weeks). Using exploratory structural equation modeling (N=2,419), our results indicate that the latent constructs (medical-related abortion circumstances and social circumstances) found by Muthen (1981) only held for the set of questions with gestational ages of up to 6 weeks for both ascending and descending order groups. However, two items in the social set had multicollinearity issues at 12- and 20-week conditions, indicating that they are redundant and do not provide unique information in the model. In contrast, the medical-related abortion items continued to provide unique information at later gestational periods. Discussion of the use of the item subsets for varying purposes will be presented. Additionally, we will discuss how these findings can inform our interpretations of the GSS items and inform the development of future abortion attitude measures.
Jozkowski, KN., Crawford, B., Turner, R., & Lo, WJ. (2020, June 11-12). Integrating abortion circumstances with gestational lengths and developmental markers: How do gestational time periods inform abortion attitudes?American Association of Public Opinion Research (AAPOR) Annual Meeting.
The General Social Survey (GSS) has measured US attitudes about abortion legality in six circumstances since 1972, with an “any reason” scenario being added later (Smith, Davern, Freese, & Morgan, 2016). Bumpass (1997) examined how gestational timing might alter responses to these questions but only used the “any reason” question. We expand on these findings by administering the seven original GSS abortion items, followed by similar items incorporating gestational ages of 6, 12, and 20 weeks, and any time during pregnancy (maintaining GSS response options of yes, no, don’t know). To investigate potential order effects, participants were randomly assigned to receive items using one of two gestational orders: 6, 12, then 20 weeks versus 20, 12, then 6 weeks. Participants were also administered the seven GSS items with three different fetal development markers included (fetal heartbeat, fetal pain, and fetal viability). Finally, participants received the GSS items with modified instructions asking participants to identify the latest gestational time period when abortion should be legal (never, 6 weeks, 12 weeks, 18 weeks, 24 weeks, 30 weeks, and any time during a pregnancy).
Quota sampling (gender, age, race, party affiliation) was used to obtain a diverse sample through Qualtrics online panels (N = 2,442). Response distributions for the original GSS items were compared with responses for 6, 12, and 20 weeks, any time during a pregnancy, fetal heartbeat, fetal pain, and viability. Initial results indicate that responses to the three GSS questions the highest historical support are most closely align with responses to the gestational ages of 6 and 12 weeks. However, responses to the three GSS items with the lowest historical support correspond closest to a 6-week gestational length. The presentation will conclude with a discussion of how these findings can inform interpretations of the GSS items.
Lo, W.J., Turner, R.C., Crawford, B.L., Jozkowski, K.N., & Weese, J. (2020, June 11-12). Change ~ we can, or can’t we? The sequence effect in survey questions.American Association of Public Opinion Research (AAPOR) Annual Meeting.
Social cognition can help us “understand and predict the behavior of ourselves and others” (Stangor, 2014). Over time, people develop social knowledge about themselves, other people, and social groups through interactions with their daily environment. Schema and attitude are the major two types of knowledge. Once formed, both schema and attitude guide people to make judgments quickly. In contrast, when there is a lack of solid attitude towards social objects, people tend to actively interpret events based on the information they receive and the sequence in which the information is presented. The purpose of this study is to investigate the effect of sequence survey questions on participant’s responses. The survey evaluated participant attitudes toward the legality of abortion while sequencing durations of pregnancy by either ascending or descending weeks. Using the original six GSS abortion items and repeating questions with durations of pregnancy in 6, 12, and 20 weeks, participants were randomly assigned into either the descending order (i.e., 20, 12, 6 weeks) set or ascending order set. After removing those who did not meet our criteria, a total of 1,136 participants were used for further analysis. Based on the latent construct provided by Muthen (1981), six items, served as outcome variables (No = 0, Yes = 1), presented whether abortion should be allowed under six circumstances in the medical and the sociological set. The independent variable is groups with different sequence, and we used age, gender, education level, and political affiliation as covariates for this model. Following the preliminary descriptive analyses, the generalized estimating equations (GEEs) were used to investigate the sequence effects after controlling covariates. The results indicated only one item (i.e., rape) did not show the sequence effect while the rest of items did present significant sequence effect between ascending and descending order groups.
Turner, R.C., Crawford, B.L., Jozkowski, K.N., & Lo, W.J (2020, June 11-12). Relationship between endorsement of GSS abortion circumstance items and gestational length recommendations. American Association of Public Opinion Research (AAPOR) Annual Meeting.
Abortion attitudes in the US have been studied in the General Social Survey (GSS) national poll since 1972 (Smith, Davern, Freese, & Morgan, 2016). Though national trends in abortion attitudes have not varied substantially in the last five decades, proposed legislation has increased since 2011 at the state level (Nash et al., 2018). Seven abortion items on the GSS present seven circumstances for a pregnant woman and ask participants whether abortion should be legal in that situation, with no reference to gestational length within a pregnancy. Given that many abortion laws and the most widely known Supreme Court case (Roe v Wade) include gestational length as a component of their guidelines, we investigated how responses to the seven GSS abortion legality items relate to participants’ attitudes about the same abortion circumstances in regard to gestational length.
Quota sampling was used to collect a national sample using Qualtrics online panels (N = 2,442). The original GSS items were administered, and then repeated using modified instructions asking participants what would be the latest time “it should be possible to obtain a legal abortion” (never, 6 weeks, 12 weeks, 18 weeks, 24 weeks, 30 weeks, any time in pregnancy).
Results indicate that 33-43% of participants who support abortion being legal in medical circumstances (e.g., life at risk, rape) think it should be available “any time” in a pregnancy, and 38-42% believe abortion should be legal up to 12 weeks. Conversely, for those who answer “no” to GSS items with medical-related circumstances, 22-31% indicate abortion should be legal at some stage of pregnancy, with “up to 6 weeks” being the most common response selected after “never.” Items addressing sociological circumstances and demographic subgroup comparisons will be presented, with a caution that results are focused on item interpretation comparisons and not generalizations to national trends.
Crawford, B.L., Jozkowski, K.N., Turner, R.C., & Lo, W.J. (2020, June 11-12). Investigating demographic differences in selecting “don’t know” across different abortion scenarios and time points in the pregnancy. American Association of Public Opinion Research (AAPOR) Annual Meeting.
Research indicates participants without clearly formed opinions or ambivalent attitudes about a construct may accurately select middle or neutral response options when answering items. However, others may select neutral response options due to factors such as satisficing, question ambiguity, or self-protection, which can be related to demographic characteristics such as educational attainment (Krosnick, Berent, Hanemann, et al., 2002; Krosnick & Presser, 2009; O’Muircheartaigh, Krosnick, & Helic, 2000). Assessing abortion attitudes can produce conflicting response patterns for participants who do not have a clearly formed abortion attitude or who have clearly formed belief systems that are in conflict. As such, failure to include responses of “don’t know” in a survey or the exclusion of “don’t know” responses from analyses, may mask or fail to account for legitimate instances of complexity, internal conflict, or ambivalence regarding abortion.
To that end, our study examines patterns of “don’t know” responses across a series of questions modeled after seven General Social Survey (GSS) abortion items. The survey first provides the original GSS questions followed by the same question stem with additions of gestational age of the pregnancy (6, 12, and 20 weeks) and fetal development markers (heartbeat, pain, and viability) added to the question stem. Data were collected through Qualtrics online panels (N=2,442). Quota-based sampling was used to ensure diversity across gender, race/ethnicity, age, and political affiliation. Results indicate that responses of “don’t know” varied across different weeks and fetal development markers. We found a general increase in “don’t know” responses at later gestational ages of the pregnancy and across fetal development markers. We will also discuss how responses of “don’t know” varied across gender, political affiliation, and abortion knowledge. The presentation will conclude with a discussion of the implications of our findings for creating new measures to assess abortion attitudes.
Individual
Simmons, M.K., Crawford, B.L., Turner, R.C., Lo, W.J., & Jozkowski, K.N. (2020, June 11-12). Complexity in abortion attitudes: The case for more nuanced measures of abortion-related political identity. American Association for Public Opinion Research (AAPOR) Annual Meeting.
Measures of abortion attitudes typically examine peoples’ support or opposition to legal abortion in specific circumstances, or by asking whether they identify as pro-choice or pro-life, and more recently also include response options for equally both or neither. Scholars argue current measures of abortion attitudes do not adequately capture complex belief systems that delineate between the personal and political. This study examines differences in self-reported abortion-related political identities across multiple measures. We administered a web-based pilot survey to 576 English and Spanish-speaking adults in the U.S. assessing abortion-related attitudes and political identities (e.g., pro-life, pro-choice) via Qualtrics’ national panels. Quota sampling ensured demographic diversity across race, gender, age, political affiliation and language. Quality checks were used to screen for invalid responses. We compared participants’ self-reported abortion identities from a single-item, categorical measure with those from a two-item, Likert-style measure assessing the degree that people independently identify as pro-life and pro-choice. We also examined open-ended responses regarding abortion beliefs as they relate to the abortion-related political identities. Approximately 14% of the sample indicated they were “equally both” on the single item measure, whereas almost 63% indicated dual identities using the two Likert-style items. According our open-ended findings, participants who identified as both pro-life and pro-choice to some extent endorsed both identities to indicate delineation between personal preferences and others’ behavior, as well as moral versus political attitudes. The two-item approach appeared to better capture the co-occurring belief systems espoused in the open-ended responses. Results indicate single-item measures of abortion-related political identity that offer a “equally both” option may not capture participants who identify with both labels to different degrees. Barring adoption of two-item measures of abortion-related political identities, we recommend modifying categorical, single-item measures to include response options that capture more nuanced co-occurring identities and beliefs (“Both, leaning more pro-life/pro-choice”).
Simmons, MK, Crawford, B, Klein, T, Turner, RC, Lo, W, & Jozkowski, KN. (2020, June 11-12). Abortion-related political participation and the Supreme Court nomination of Justice Kavanaugh. American Association for Public Opinion Research (AAPOR) Annual Meeting.
Abortion-related activism appears to fluctuate according to political climate and in response to national policy changes and political events. For example, the confirmation of Justice Kavanaugh to the Supreme Court of the U.S. appeared to mobilize stakeholder organizations on both sides of the abortion debate. However, research assessing the extent that such political events actually mobilize the general public (non-stakeholders) is lacking. To address this gap, we examined U.S. adults’ lifetime abortion-related political engagement, as well as whether and how Justice Kavanaugh’s nomination influenced abortion-related political behaviors. Qualtrics’ national panels were used to recruit English and Spanish-speaking U.S. adults to participate in a web-administered survey assessing abortion-related political participation. We used quota sampling to ensure diversity across demographics (age, race, gender, language, political affiliation). Use of data quality checks (CAPTCHA, reading checks) allowed us to screen for potential bots and straight-liners. We used logistic regression models to examine predictors of prior engagement and mobilization around the time of Justice Kavanaugh’s nomination across a variety of political behaviors, including calling politicians, campaigning, posting to social media, donating money to organizations, and attending rallies. Approximately 40% of participants reported having engaged in pre-nomination abortion-related political behaviors. Unlike previous studies, younger people, Hispanic participants, and people who identified as pro-choice were most likely to have ever engaged. Mobilization during the survey period was minimal. People who were already engaged in abortion-related political behaviors were most likely to have increased their frequency of participation during this period. Our findings suggest that lifetime abortion-related political behavior beyond voting is somewhat common. However, calls to action made by stakeholder groups have limited influence, and are likely to motivate those with established interest in the issue. Expanding measurement of political engagement to include other behaviors may provide a more holistic view of political engagement surrounding issue-specific movements.
Montenegro, M.S., Valdez, D., Turner, R.C., Jozkowski, K.N., Crawford, B.L., & Lo, W.J. (2020, June 11-12). Latinx attitudes toward abortion: Descriptive findings of abortion attitudes among the Latinx community. American Association for Public Opinion Research (AAPOR) Annual Meeting.
As the demographic landscape of the US shifts, there is a growing need to examine minority groups’ (e.g., Latinx) public opinion on social issues. This is particularly relevant for social issues disproportionately affecting certain demographic subgroups such as abortion and the Latinx population. Indeed, Latina women experience unintended pregnancy and seek abortion at greater proportions than their white counterparts. Thus, understanding the collective public opinion of this population for this issue is particularly relevant and timely as an increasing number of laws have limited women’s access to abortion. This study examines Latinx people’s attitudes toward abortion legality and morality in the US. We administered a web-based survey to English and Spanish-speaking Latinx adults in the US. Quota-based sampling was used to achieve demographic diversity. The survey instrument included questions about current social issues, with an emphasis on abortion attitudes. Specifically, we separately assessed participants’ attitudes toward abortion’s legality and morality under specific circumstances. Preliminary results indicate Latinx are less likely to support abortion legality and consider abortion moral. Yet, there were no significant differences based on language. Men expressed more favorable attitudes toward abortion legality compared with women. Unsurprisingly, Republicans expressed less support for abortion legality and morality for social situations compared with Democrats and Independents. Those between the ages of 39 and 49 were significantly more opposed to abortion legalization than 18-34-year-old and those 50 years and up. Although the Latinx community tends to oppose abortion legality and morality, our results indicate that their attitudes toward abortion do not differentiate much from the general population and do not differ by language preference. These results suggest that other variables, besides ethnicity, may be more influential when assessing abortion attitudes. Future research is needed to examine how levels of acculturation interact with other variables and influence abortion attitudes.
Montenegro, M.S., Valdez, D., Turner, R.C., Jozkowski, K.N., Crawford, B.L., & Lo, W.J. (2020, June 11-12). Religious identity and abortion attitudes among Latinx in the US. American Association for Public Opinion Research (AAPOR) Annual Meeting.
The Catholic Church opposes abortion. Given that many Latinos/as identify as Catholic, their religious affiliation may influence their abortion attitudes. There is also a growing number of Catholic Latinos/as converting to an Evangelical faith—which is also associated with less support for abortion. Latinos/as’ faith may be fundamental to understanding the role of religious identity in shaping abortion attitudes. We examine religious identity and abortion attitudes among Latinos/as by examining differences between Latino/a Catholics, Christians, and those who identified as born-again. We also compare foreign and US-born Latino/as’ attitudes, to assess the role of acculturation within the US. We administered a web-based pilot survey to English and Spanish-speaking US Latino/a adults (n=224); additional data collection is underway for a sample of n=1,000 Latino/a participants. Quota-based sampling was used to achieve demographic diversity. We compared participants’ responses to items measuring attitudes toward abortion legality and morality across their self-reported religious affiliations and countries of birth. We tested for religious differences among US and foreign-born Latinos/as using MANOVA contrasts. Preliminary results indicate that Latinos/as are less likely to support abortion legality for social circumstances and more likely for medical reasons. They are more likely to agree abortion is moral under medical circumstances and less likely for social reasons. Catholics, Christians, and people who identified as born-again responded similarly. However, those who see the Bible as the literal Word of God were significantly less likely to support abortion legality and morality. No significant differences were found based on language and place of birth. Initial results indicate Bible literalism may be a stronger predictor of abortion attitudes than religious identity and country of birth for Latino/as. However, by using a larger sample in general and a cross-validation sample that includes an evangelical Christian classification, we will further investigate these relationships and potential interaction effects.
Kaplan, A.M. Crawford, B.L., Turner, R.C., Montenegro, M., Valdez, D., Lo, W.J., Jozkowski, K.N. (2020, June 11-12). The influence of data collection methods on differing response patterns: Acculturation as a mediator to knowledge and attitude towards Roe v. Wade. American Association for Public Opinion Research (AAPOR) Annual Meeting.
Background: The contribution and influence of the Latinx community on social and political climate in the US continues to grow as they become a larger proportion of the population. When studying this subgroup, it is important to consider acculturation--the process of a minority group adapting and incorporating majority group attitudes or culture--to provide a contextualizing lens for findings. In this study, we use an acculturation framework to compare how different quota sampling designs influence how acculturation mediates the relationship between attitudes toward abortion among samples of Latinx US adults. Specifically, we looked at differing response patterns related to knowledge of and attitudes toward Roe v. Wade across two samples of US Latinx adults, considering language use and generation status as proxies for acculturation.
Methods: We administered two web-based surveys using Qualtrics’ panels to English and Spanish speaking Latinx adults in the US (N1Spanish = 625; N1English= 181; N2Spanish = 738; N2English = 500). The two samples were collected using different quota sampling designs. Descriptive statistics were used to assess how samples differed on demographic variables (e.g., gender, age, political affiliation, race/ethnicity) and acculturation variables--generational status and language use.
Results: We conducted three regression models to assess the influence of survey design on abortion attitudes using acculturation as a mediator, while controlling for demographic variables. Preliminary findings suggest that generational status significantly predicts attitudes in one data set and language significantly predicts attitudes in the second data set.
Conclusion: Findings suggest the degree that acculturation factors, language and generational status, influence knowledge of and attitudes toward Roe v. Wade may be a product of differing data collection protocols. It is important to be purposeful in data collection methods related to acculturation variables among Latinx, as it may influence how acculturation affects certain knowledge and attitude variables.
Valdez, D., Turner, R.C., Jozkowski, K.N., Crawford, B.L., & Lo, W.J. (2020, June 11-12). Reliability generalization as a formative tool for evaluating survey functionality by demographic sub-groups: A long-term analysis of the General Social Survey abortion attitude scale (1978-2016). American Association for Public Opinion Research (AAPOR) Annual Meeting.
Background: Although polls and surveys are written with generalizability in mind, they rarely function universally— i.e. reliably— among all groups in a sample. Indeed, many demographic indicators influence how groups understand (and engage with) surveys, including age, race, and even sexual orientation. It is uncommon, however, to test for survey functionality by specific subgroups within a survey. Consequently, we sometimes overlook cases where highly reliable scales are less reliable for certain groups/demographics.
Purpose: The purpose of this study is to examine if reliability for even the most validated of surveys can fluctuate depending on sub-groups within a sample.
Method: We conducted a reliability generalization (RG) of the General Social Survey abortion scale from 1978-2016. RG is a meta-analytic method that evaluates reliability changes of survey scales over time and by group. We directed our analysis to broadly examine reliability differences among the following groups: (1) the original GSS race variable (1980-2016), (2) race with additional categories (2000-2016), and (3) language--English and Spanish (2006-2016).
Results: The GSS abortion scale was highly reliability (<.75) across all years and tested groups. However, reliability (McDonald’s Omega) was highest among white and English-Speaking subgroups. All races, other than white, had significantly lower levels of reliability. Additionally, the Spanish-speaking, Latino sample consistently had the lowest reliability of all tested groups.
Conclusion: Although the GSS abortion items are used widely and frequently to assess abortion attitudes, our findings suggest the scale is still prone to sample sub-group reliability fluctuations. It is important to conduct cursory analyses (including a reliability analysis) to determine if groups are engaging with a survey in a manner consistent with how they are intended to be interpreted. Stark differences in reliability may indicate, early on, if items should be altered to address potential interpretation issues.
Maier, J., Montenegro, M., Kaplan, A., Willis, M., Valdez, D., Jozkowski, K.N., Crawford, B., Lo, W.J., & Turner, R. (2020, June 11-12). Would v. should: Examining word selection in assessing abortion attitudes. American Association for Public Opinion Research (AAPOR) Annual Meeting. Toronto, Canada.
People’s abortion attitudes are complex; there have been calls for new ways of measuring attitudes to capture this nuance. In doing so, however, survey questions need to be carefully designed in a way that yields uniform understanding. Even seemingly insignificant wording differences can affect participants’ responses. In this research, we examined how the words would/decidíra and should/debería function in a multi-language (English and Spanish) open-ended survey on abortion attitudes. We administered a salient belief elicitation survey examining abortion attitudes online (n = 172) and in-person (n = 73). Participants were randomly assigned to survey versions, which included identical questions except for the word would/decidíra versus should/debería. For example, “What are the top three reasons you should/would decide to have an abortion?” Additionally, participants who took the survey in-person were interviewed to understand how they interpreted survey questions. Data analysis included: 1) an inductive content analysis of the open-ended survey responses to see if answers differed by survey version; and 2) a thematic analysis of the cognitive interview data focused on how participants were understanding would/decidiría and should/debería. Preliminary results indicate that would/decidiría yielded a greater number of survey response categories compared with should/deberia. This finding was supported by the interviews as participants conceptualized would/decidira as allowing for more choice or options. Alternatively, should/debería was largely thought to include the most important reasons or circumstances for abortion, which may account for the lower number of response categories. However, some participants also indicated little difference in their interpretation of would/decidiría and should/debería. Future research can examine how other auxiliary verbs, such as might or may, function in abortion-related surveys.