January 12, 2012

Mormons in America - Certain in Their Beliefs, Uncertain of Their Place in Society

Survey Methodology

The analysis in this report is based on telephone interviews conducted Oct. 25-Nov. 16, 2011, among a national sample of 1,019 Mormon adults 18 years old and older; 694 interviews were conducted on landlines and 325 were conducted on cell phones. The survey was conducted by Social Science Research Solutions (SSRS).

Sample Design

Mormons constitute a rare population in the U.S.; Pew Research Center surveys conducted in 2011 find that 1.9% of U.S. adults describe themselves as Mormons. In order to sample Mormons efficiently, the study design involved oversampling certain regions of the country where Mormons are most numerous (as a percentage of the overall population) and recontacting self-identified Mormons from previous surveys conducted by the Pew Research Center and by SSRS. When data collection was completed, the sample was weighted to correct for the geographic stratification and to account for the use of recontact sample in the study (more details on weighting are provided below). This ensures that Mormons from the various geographic regions of the country and from a variety of demographic groups are represented in the sample in their proper proportions.

Geographic Stratification

ma-method-1

To identify U.S. counties with the largest Mormon populations, results from Pew Research Center surveys conducted between 2008-2011 were combined and analyzed along with results from surveys conducted by SSRS over the same period. Based on this analysis, each county in the U.S. was categorized into one of five geographic strata based on the estimated percentage of adults who are Mormon. The “very high” stratum consists of counties (located mostly in Utah and Idaho) that previous Pew Research Center surveys suggest are home to 23% of the Mormon population but only 1% of the total U.S. adult population, and in which Mormons comprise upwards of 50% of the total population. The “very low” stratum includes counties that are home to an estimated 29% of the Mormon population and 87% of the total U.S. population, and in which Mormons comprise less than 2% of the total population. 8 Similarly, other counties are grouped into “high,” “medium” or “low” strata based on the share of the county’s total population estimated to be Mormon.9

ma-method-2

After the geographic stratification was complete, interviews were allocated such that the “very high” stratum was oversampled (35% of all interviews were conducted among Mormons residing in the “very high” stratum counties). The “high” and “medium” strata were sampled roughly in proportion to their share of the Mormon population, with 18% of all interviews conducted among Mormons residing in the “high” stratum counties and 13% of interviews conducted among Mormons in the “medium” stratum counties. The “low” and “very low” strata were undersampled, with 13% of interviews conducted among Mormons in “low” stratum counties and 21% conducted among Mormons in the “very low” stratum. After data collection was completed, the sample was weighted such that each of the five strata is ultimately represented in its proper proportion in the final data.

Recontact Sample

Though the geographic stratification described above dramatically increases the efficiency with which Mormons can be reached and interviewed relative to a simple random sample, geographic stratification alone is not sufficient to conduct a nationally representative survey of Mormons at an affordable cost. Thus, to supplement the geographic stratification, the study also recontacted households from previous Pew Research Center and SSRS surveys since January 2008 in which a Mormon had been interviewed. The recontact sample was used strategically for two specific purposes. First, the recontact sample was used to boost coverage of Mormons residing in the “very low,” “low” and “medium” strata. The study design called for the “very low” stratum to be covered entirely by the recontact sample and for the “low” and “medium” strata to be covered primarily by the fresh sample (i.e., interviews with Mormons contacted and screened for the first time as part of the current study) and supplemented with some recontact sample.10 This strategic use of the recontact sample to cover areas of the country with lower Mormon incidence rates helped to decrease the total number of screening interviews required for the project and thus reduced the cost of the study.

The second consideration underlying the use of the recontact sample for this study was the need to achieve a good balance of landline and cell phone interviews. As the cell-phone-only population has continued to grow, the proportion of surveys conducted by organizations like the Pew Research Center and SSRS among cell-phone samples has also grown. This means that the older recontact sample (e.g., from 2008) includes less cell-phone sample and thus less representation of the cell-phone-only population as compared with the more recent recontact sample (e.g., from 2011). To help offset this disparity and to boost coverage of cell-phone-only Mormons, the study attempted to recontact and interview all Mormons interviewed on cell phones by SSRS over the past three years, regardless of the stratum in which they reside.

ma-method-3

In total, 697 interviews were conducted among the fresh sample, and 322 were conducted among the recontact sample. Interviews with Mormons residing in the “high” and “very high” strata were conducted almost exclusively among fresh sample, with the exception of a small number of recontact interviews conducted on cell phones. Interviews in the “medium” stratum were conducted primarily using fresh sample (71%), but with a significant amount of recontact sample incorporated as well (29%). Interviews in the “low” stratum were evenly divided between fresh sample and recontact sample. And interviews with Mormons residing in the “very low” stratum were conducted almost exclusively among recontact sample.

Identification of Mormons

For both freshly sampled households and those in the recontact sample, the survey began with a screening interview. Respondents reached by landline were selected by randomly asking for the youngest adult male or female currently at home. Interviews on cell phones were conducted with the person who answered the phone, if that person was an adult 18 years of age or older. The screening interview consisted of a few short warm-up questions (about the respondent’s level of satisfaction with their community and their life), followed by a question about the respondent’s religious affiliation: “What is your present religion, if any? Are you Protestant, Roman Catholic, Mormon, Orthodox such as Greek or Russian Orthodox, Jewish, Muslim, Buddhist, Hindu, atheist, agnostic, something else, or nothing in particular?” Those who described themselves as Mormons in response to this question were then administered the main survey, while the interview was discontinued for non-Mormons.11

After identifying themselves as Mormons, qualified respondents were asked a separate question, “And is that The Church of Jesus Christ of Latter-day Saints, the Community of Christ, or some other Mormon church?” All self-identified Mormons were eligible for the survey, regardless of whether they identified themselves as part of the LDS Church. In practice, nearly all self-identified Mormons (99% in the current sample) describe themselves as part of the LDS Church.

Questionnaire Design

The study had two main goals. First, it sought to learn about Mormons’ perceptions of American society and of their own place within it at a time when Mormons and Mormonism are receiving increased attention in the news media and popular culture. Second, it sought to assess the degree to which Mormons resemble or are distinctive from the broader public in their social and political attitudes and in their religious beliefs and practices. As such, the survey included a mix of new questions specific to Mormons and Mormonism and “trend” questions that have previously been asked of the general population in Pew Research Center surveys. The development of the survey questionnaire was informed by the advice and feedback received from a panel of advisers with expertise in the study of the U.S. Mormon population.

As with other surveys the Pew Research Center has conducted among relatively rare populations, the initial questions in the current survey were chosen to be of a general nature in order to establish rapport with respondents. After these items, respondents were asked about their religious affiliation (as described above), and the survey then proceeded with a series of questions on social and political topics (e.g., about satisfaction with the direction of the country; views about a variety of prominent political figures; attitudes about homosexuality, immigration and the size of government; life priorities; the morality of various activities such as extramarital sex and drinking alcohol; and an assessment of the degree to which various groups face discrimination in American society today).

At this point in the interview, respondents were told that the survey was designed specifically for Mormons. They were told, “Just to give you a little more background before we continue, the Pew Research Center conducts many surveys on religion and public life in the United States. Earlier, you mentioned that you are a Mormon, and we have some questions about the views and experiences of Mormons living in the United States. I think you will find these questions very interesting.” The interview then continued with questions probing how Mormons perceive their place in American society and with questions about their religious beliefs and practices, before concluding with a set of questions about their demographic characteristics.

Data collection was preceded by two pretests of the survey instrument. The first pretest was conducted on Oct. 11, 2011, among 47 Mormon respondents residing in Utah, Idaho and Nevada. The second pretest was conducted on Oct. 18, 2011, among 17 Mormon respondents residing in Utah, Idaho and Wyoming.

Survey Administration

Interviewing for the study was conducted Oct. 25-Nov. 16, 2011, by SSRS. All interviews were conducted using a computer-assisted telephone interviewing (CATI) system. Interviews averaged 24.7 minutes in length. The questionnaire was translated into Spanish and respondents who were unable to complete the interview in English were offered the option of completing it in Spanish. In total, two interviews were conducted in Spanish.

The administration of the survey utilized a seven-call design, in which all numbers that were not immediately determined to be government offices, businesses, other types of institutions or out of service were attempted at least seven times over different times of day and days of the week before being abandoned. A voice mail message about the content of the study was left the first time an answering machine or voice mail system was encountered. Respondents were offered the opportunity to call in and complete the interview, as well as the opportunity to schedule an appointment to be interviewed if they were reached at an inconvenient time. All cell phone respondents were offered a $5 reimbursement.

Weighting

A two-stage weighting design was applied to ensure an accurate representation of the national population of Mormons. The first stage of weighting is called the design-weight phase. The second stage of weighting is called the post-stratification phase.

Design-Weight Phase

The first step in the design-weight phase is the stratification correction. The percentage of interviews conducted in each stratum was divided by the actual proportion of the Mormon population that lives in that stratum, as estimated by the combined set of Pew Research Center and SSRS interviews conducted over the past three years. This was done separately for the landline and cell-phone samples. This step ensures that respondents in the various strata are represented in their proper proportions in the final data and that the views of Mormons living in areas heavily comprised of Mormons are not given undue weight.

The second step in the design-weight phase is the within-household selection correction. This corrects for the unequal probabilities that are introduced by some households having more adults than others. Landline-sample households with a single adult receive a weight of one, whereas landline-sample households with two or more adults receive a weight of two. Landline-sample respondents with missing household composition data and cell phone respondents were given a weight of one.

The third step in the design-weight phase is the dual-frame correction. This adjusts for the greater probability of selection of households that have both a landline phone and a cell phone, which are twice as likely to be sampled as households that have only one kind of phone or the other. Dual users (i.e., those with both a landline and a cell phone) receive a weight of 0.5, whereas those who have only a landline or only a cell phone receive a weight of one.

The fourth and final step in the design-weight phase is the recontact-propensity correction. This step accounts for the potential bias associated with recontacting (panel bias) and is applied to the prescreened sample only. It uses demographic attributes as measured in the original survey in which a respondent was interviewed to model the probability that a household in the recontact sample yielded a successful interview in the current project. The propensity regression analysis found that households in the original sample where the respondent was white, childless, politically independent (as opposed to Democratic) and registered to vote are more likely than others to have been successfully recontacted; respondents who have these characteristics were weighted “down” accordingly, while respondents who do not have these traits were weighted “up.”

Post-Stratification Phase

Following the design-weight phase, the data were put through a second weighting stage, called the post-stratification phase. This involved using a sample balancing method to match the demographic characteristics of the current sample to known population targets for age, gender, education, geographic region, race/ethnicity, population density and phone status and usage. The population targets were derived from analysis of the demographic characteristics of Mormons interviewed in Pew Research Center surveys between 2008 and 2011 and in SSRS surveys over the same time period. The estimates from the Pew Research Center surveys and SSRS surveys were developed separately and then averaged together to compute the post-stratification weighting targets. An analysis of the demographic characteristics of the (weighted) current sample and how it compares with Mormons interviewed in recent Pew Research Center surveys is presented below.

Design Effect and Margin of Error

ma-method-4

Surveys that use a complex sampling design, rather than a simple random sample, ordinarily will have a margin of sampling error larger than in a simple random sample of the same size. In addition, the post-stratification weighting can also increase the margin of error. The extent to which the margin of error is inflated by the design and the weighting is called the study’s “design effect,” and it must be taken into account when reporting a margin of error and conducting tests of statistical significance. The overall design effect for this study, taking into account both the design-weight and post-stratification phases of the weighting process, is 1.97. The margin of error for the full sample of 1,019 Mormons (at the 95% level of confidence) is +/-4.5 percentage points. The margins of error for subgroups are larger. Sample sizes and corresponding margins of error for many of the subgroups analyzed throughout this report are provided in the accompanying table.

Mormon Demographics

As mentioned above, estimates from interviews with U.S. Mormons conducted as part of other Pew Research Center surveys from 2008-2011 were used in order to create demographic targets for post-stratification weighting of the current sample. This section discusses how the demographic characteristics of Mormons in the current sample compare with the demographic characteristics of Mormons in previous Pew Research Center surveys. With few exceptions, the demographic characteristics of the current sample closely match the demographic characteristics from previous surveys. Because the current sample was weighted on several demographic characteristics, it will be similar though not identical to the previous surveys on those measures.

Gender, Age and Marital Status

In recent Pew Research Center surveys, the gender distribution of U.S. Mormons has been nearly evenly split, 49% male and 51% female. This is also the case in the current sample (50% male, 50% female).

Two-thirds of Mormons in recent Pew Research Center surveys (66%) have been under 50 years old. By comparison, the current survey of Mormons is somewhat older (58% are under 50).

ma-method-5

Among Mormon respondents in recent Pew Research Center surveys, two-thirds (67%) have been married, 9% divorced or separated and 17% had never been married. Similarly, in the current sample, two-thirds of Mormons are married (67%), 9% are divorced or separated and 16% have never been married.

Race and Ethnicity

Among U.S. Mormons in recent Pew Research Center surveys, 84% have been white and non-Hispanic, while 8% were Hispanic, 2% were black non-Hispanic and 5% of another race. In the current sample, 88% are white, 7% are Hispanic, 1% are black and 4% are of another race.

ma-method-6

Comparatively, among the general public, about seven-in-ten (68%) people are white and non-Hispanic, while 11% are black, 14% are Hispanic and 7% are of other racial and ethnic backgrounds.

Income and Education

In recent Pew Research Center surveys, three-in-ten U.S. Mormons (29%) had a household income of less than $30,000, while 30% had a household income of $75,000-$100,000 (14%) or over $100,000 (16%). In the current sample, 28% have a household income less than $30,000 and 26% report household incomes of $75,000 or more.

ma-method-7

The vast majority of U.S. Mormons in previous Pew Research Center surveys had a high school diploma or higher (93%). About one-third (32%) had completed some college, and almost another third (30%) had a college degree and/or some post-graduate education. The levels of educational attainment among Mormons in the current sample closely match these estimates.

Region

A majority of U.S. Mormons in previous Pew Research Center surveys resided in the West (73%), as do 71% of Mormons in the current sample. Just over a third (37%) of Mormons in previous surveys lived in the state of Utah, while about two-thirds (63%) lived in other states. Similarly, in the current sample, about one-third (34%) of Mormons live in Utah and two-thirds (66%) live in other states.

ma-method-8

Response Rates and Sample Disposition Reports

The disposition of all telephone numbers in the sample is available in the full report PDF, broken down by stratum and sample type. The response and cooperation rates for this study were calculated using AAPOR’s RR3 and COOP3 formulas, respectively. In the case of the prescreened sample, the response rates reported in the tables below reflect only the current study. The final response rate for these cases, and thus for the full study, must take account of both the response rates provided below and the response rates of the original surveys from which recontact sample was obtained. The original surveys from which the recontact sample was drawn attained average response rates of 12% for the SSRS omnibus landline samples, 8% for the SSRS omnibus cell-phone samples, and 6% for the SSRS address-based design samples. The average response rate for recent Pew Research Center cell phone surveys is 9%. Taking these into account, the final combined weighted response rate for the full study is 20.4%.


Footnotes:

8 The “very low” stratum includes those counties in which neither the Pew Research Center nor SSRS has interviewed a Mormon respondent since 2008. (return to text)

9 Until recently, Pew Research Center surveys did not cover Alaska and Hawaii. As a result, most counties in these states could not be categorized into strata based on their estimated Mormon incidence rates. Instead, all counties in Alaska and Hawaii were included in the “medium” stratum, with the exception of Anchorage Borough, which was placed in the “high” stratum. (return to text)

10 There is an imperfect correspondence between respondents’ phone numbers and the geographic area in which they actually reside. As a result, a small number of interviews in the “very low” stratum actually come from fresh sample. (return to text)

11 In an effort to find and interview Mormons with maximum efficiency, non-Mormons were asked whether there are any other adults in the household whose religion is different than the respondent’s own religion. In those households where the respondent indicated that there is another household member who is Mormon, interviewers asked to speak with the Mormon household member and attempted to complete the interview with that person. This approach was followed for landline sample respondents for the entirety of the field period; for the cell phone sample, this approach was followed early in the field period before being discontinued. Ultimately, 14 interviews with Mormon respondents were completed via this “handoff” technique. (return to text)

Photo Credit: © Walter Bibikow/JAI/Corbis