Streamable Services: Affinities for Streaming in Pre-Pandemic Congregational Worship

Abstract

Following the COVID-19 pandemic in March 2020, many congregational leaders had to scramble to set up streaming or recording systems in order to continue their worship services without putting congregants at risk, but some congregations had already set up such systems in the years leading up to the pandemic. Previous research has found that these capabilities were not evenly distributed throughout the population of congregations, but this work has primarily focused on how technological divides are the result of a lack of economic resources. However, economic resources were not the only factor associated with whether congregations had streaming options or not. Using Wave 4 of the National Congregations Study (NCS) conducted in 2018–2019, I find that, prior to the pandemic, Catholic congregations and Protestant congregations with more enthusiastic worship services were more likely to have streaming or recording systems even after controlling for economic resources, technological knowledge, and other organizational features. The elective affinities between certain worship practices and online streaming meant that some congregations were in a better position to meet the unexpected challenges posed by the coronavirus pandemic. These findings highlight the important role congregations’ cultural beliefs and practices can play in shaping their activities.

1. Introduction

The COVID-19 pandemic brought disorder to every aspect of social life, and religious life was no exception. Facing a deadly and highly communicable disease, congregations were forced to cease in-person activities to protect their parishioners. However, in doing so congregations had to confront the challenge of maintaining their religious communities without meeting in person. Communal worship is the central activity of religious congregations, and is essential for maintaining group solidarity (Chaves 2004; Draper 2014). As such, going without a worship service for an extended period of time would greatly imperil a congregation’s future. A solution to this problem is to stream or record the worship service to allow congregants to engage in communal religious activity without physically congregating. There was already a trend towards online worship prior to the pandemic, and the era of social distancing certainly accelerated this trend (Campbell 2012; Drescher 2011; Gorrell 2019; Roso et al. 2020; Thumma 2012). With the pandemic making such technologies especially salient, scholars began investigating how congregations adapted their worship services to meet the challenges of the pandemic, and which congregations had pre-existing streaming and recording capabilities (Francis et al. 2020; Johnston et al. 2021; Holleman et al. 2022). This line of research primarily focused on how differences in worship streaming or recording capabilities were the product of inequality in financial resources.

However, the focus on financial resources has largely ignored the role that congregations’ existing practices and beliefs about worship may have played in developing streaming or recording capabilities. Previous research within organizational sociology suggests that established organizational practices can limit the sorts of innovations organizations adopt (Meyer and Rowan 1977; DiMaggio and Powell 1983; Thornton et al. 2012). Recent post-pandemic research suggests that this holds true for religious congregations as well as they work to maintain their religious communities in the era of social distancing (Ben-Lulu 2021; Johnston et al. 2021; Francis and Village 2021). Adding a new routine such as streaming to a worship service can be difficult, time-consuming, and risky—particularly if it does not mesh well with the organization’s existing cultural practices (Bertels et al. 2016). However, when an affinity exists between an organization’s existing practices and a new innovation, incorporating a new routine is much easier (Rindova et al. 2011).

In this article, I examine two specific cultural affinities that exist with streaming or recording worship among religious congregations. I use data from the fourth wave of the National Congregations Study (NCS)—a nationally representative study of religious congregations conducted in 2018–2019 (Chaves et al. 2020b)—to investigate how religious tradition and enthusiastic worship practices were related to streaming or recording capabilities prior to the pandemic. I find a robust association between both worship practices and religious traditions with streaming or recording of worship services even after controlling for financial resources, other technological capabilities, and organizational characteristics. Specifically, I find that Catholic congregations were substantially less likely to stream or record their services than Protestant congregations, and that Protestant congregations with enthusiastic worship practices such as shouting amen, raising hands, and using drums during worship were more likely to stream or record their services than Protestant congregations with less enthusiastic services. These findings show that financial and social capital are not the only factors predicting the usage of worship streaming technology, as certain cultural affinities between worship and streaming also play an important role.

2. Background

2.1. Cultural Affinities within Organizations

In The Protestant Ethic and the Spirit of Capitalism, Weber ([1905] 2002) famously claimed that the theological beliefs of particular European Protestant groups formed the basis for capitalism. As Weber argued, the Calvinist understanding of a calling had an “elective affinity” with capitalistic enterprise.1 Faithful Calvinists saw their earthly labors and prudent investment of money as part of their own personal calling, and the profits of such endeavors served as evidence of their salvation.

While empirical evidence for Weber’s particular claims about the origins of capitalism remains murky (c.f. Barro and McCleary 2003; Hayward and Kemmelmeier 2011; Norris and Inglehart 2004), the core idea of elective affinities between cultural ideas and material relations has been useful in many fields, including the study of organizations. Scholars advocating a neo-institutional approach study organizations and institutions as embedded cultural entities using cultural myths and practices to maintain legitimacy (DiMaggio and Powell 1983; Friedland and Alford 1991; Meyer and Rowan 1977). Organizations are not perfectly rational and materialistic, and as organizational myths and ceremonies are institutionalized, they become durable and can limit or enable future innovation.

These concepts have been studied further in more recent scholarship (Harrison and Corley 2011; Weber and Dacin 2011). Scholars of organizations call these cultural myths and practices “logics”, or “socially constructed, historical patterns of cultural symbols and material practices […] by which individuals and organizations provide meaning to their daily activity, organize time and space, and reproduce their lives and experiences” (Thornton et al. 2012, p. 2), and argue that organizations can develop durable logics to help them maintain their legitimacy in organizational fields and respond to challenges they face. Crucially, changing these established logics is challenging and risky (Barnett and Pontikes 2008; Haveman 1992), particularly when attempting to incorporate new logics that do not mesh well with the organization’s existing ethos (Bertels et al. 2016).

In this way, an organization’s established logics can limit its available actions. For example, Rindova et al. (2011) studied how executives at Alessi, an Italian manufacturer of consumer products, tried to shift strategies over time. In some instances, the company executives marketed the company’s products as works of art, and at other times they marketed them as more functional tools. These different strategies demanded different technologies and manufacturing processes, and sometimes trying to implement these new routines conflicted with both the organization’s existing capacities and how it was perceived in the market. In an attempt to innovate, Alessi executives tried to import cultural ideas about their product from different industries (for example, the art industry) to help devise new organizational strategies in manufacturing and marketing the product—with mixed success. For Alessi and other organizations, the cultural affinity between existing logics and the proposed change in practices is an important concern. A poor match between the two makes failure more likely.

2.2. Online Worship during the Pandemic

While scholars of religion seldom use the term “organizational logics”, the concepts of neo-institutional theory may still be useful in the study of religious organizations (Demerath et al. 1998). Just as businesses have established logics around manufacturing and marketing their products, religious organizations’ cultural ideas are manifested in their own practices—the most notable of which is weekly worship. Collective worship is the core activity of congregations, and scholars since at least Durkheim ([1912] 1995) have remarked on the power that communal religious rituals have to promote solidarity within communities (Baker 2010; Draper 2014). Over the last few years, live-streaming and other worship-related technologies have become increasingly common in worship services (Roso et al. 2020; Thumma 2012), and scholars have investigated how both religious leaders and adherents can use these technologies to maintain their communities online (Drescher 2011; Campbell 2012; Frost and Youngblood 2014; Gorrell 2019). Through this technology, worship services can be streamed or recorded so that people who are not physically present can still experience worship. Such technological innovation has most often been associated with megachurches, but these practices have increasingly diffused to smaller congregations (Ellingson 2007).

The need for online worship became especially acute with the onset of the COVID-19 pandemic. A deadly and transmissible respiratory disease, the novel coronavirus can spread rapidly through indoor communal activities such as worship (Hamner et al. 2020). Moreover, those who regularly attend worship services are, on average, older than the general population, and elderly people are at an elevated risk of death or serious illness from COVID-19 (Voas and Chaves 2016; Zhou et al. 2020). The sudden need for online worship forced many church leaders who were not already streaming or recording their services to find solutions if they wanted to maintain their religious communities during the pandemic (Johnston et al. 2021). Congregations with pre-existing streaming options already had the equipment, personnel, and expertise to stream their worship during COVID lockdowns, and so were better prepared for the pandemic (Seabright and Raiber 2020). Furthermore, congregations with expertise in worship streaming and recording were likely to produce a better quality online worship experience, making it less likely that congregants would disconnect from worship and the congregation (Village and Francis 2022a). This has led some scholars to examine which congregations had streaming or recording capability and where this stratification dovetailed with other areas of inequality (Holleman et al. 2022; Francis et al. 2020). In particular, this line of research found that small and rural congregations were particularly vulnerable to the effects of the pandemic and comparatively unlikely to have streaming or recording services.

2.3. Affinities for Streaming in Worship

While research on how congregational streaming and recording capabilities vary based on their financial resources highlights important areas of inequality in the religious landscape, it overlooks the cultural factors required to make these changes and overstates the financial burden of setting up streaming or recording systems. With streaming on free platforms such as Facebook Live becoming increasingly available, setting up basic streaming and recording services may be well within the means of even relatively small congregations. However, even if a congregation has the financial resources and technical know-how to set up streaming or recording, its ability to translate its worship to an online format may be limited by the cultural practices and ideas embedded in the worship itself.

Much like Weber’s proposed elective affinity between Calvinist theology and capitalistic enterprise, an elective affinity may exist between particular liturgical traditions and streaming or recording worship. Certain religious traditions have theological beliefs about worship that either preclude or enable worship streaming. For example, research on online worship among Reform Jewish congregations found that these congregations had the ability to shift their worship online because, unlike some other branches of Judaism, Reform Judaism does not have proscriptions against worship services occurring online (Ben-Lulu 2021; Cohen 2014). Similarly, clergy in England had different responses to shifting their services online in 2020 depending on their theological orientation towards the Eucharist (Francis and Village 2021). In these cases, the limiting factor for whether or not a congregation adopted online worship was not money or experience, but cultural understandings of what worship is.

Among the major religious traditions with substantial presence in America, Catholicism stands out for having worship that may not translate to an online format. Much like worship in some of the traditions that Francis and Village studied in England, the Catholic mass centers on the administration of the Eucharist. Catholics believe in transubstantiation, or the transformation of the Eucharistic elements into the literal body and blood of Christ upon being blessed by the priest. This physical ritual cannot be transmitted through the internet, and watching a Catholic mass from afar does not fulfill the same religious function as attending in person does. Prior research has found that American Catholics stand out for their low rates of streaming (Holleman et al. 2022), but little research has been done investigating whether this is the result of a cultural misalignment between streaming and Catholic worship, or if it is simply the product of other differences between Catholic and Protestant congregations.

Within Protestantism, one liturgical tradition that may have an affinity with streaming and recording is contemporary worship. A liturgical style that is becoming increasingly common in US congregations, contemporary worship consists of several important qualities, including a commitment to adapting worship to the concerns of contemporary people, greater expressiveness in worship, and a predilection for informality (Roso et al. 2020; Lim and Ruth 2017). Classical music from organs and choirs and homilies from robed religious leaders are replaced by praise music and expository sermons delivered by pastors in jeans. Importantly, since even before the internet, contemporary worship has relied on electronic technology both to enhance the worship experience for people physically present and to share recordings and video of the worship service to people who are not present (Ruth and Lim 2021). Two reasons for this connection between electronic technology and contemporary worship are the desire to connect with contemporary people where they are, and a displacement of traditional Christian sacraments.

At the heart of some early streams of contemporary worship was a pragmatic concern to reach contemporary people where they were. Prominent early practitioners of contemporary worship found electronic technology to be helpful in connecting with a generation that grew up watching TV (Ruth and Lim 2021), and subsequent celebrity megachurch pastors followed suit (e.g., Johnson 2018). Following the success of these early adopters, smaller churches began incorporating similar practices in an attempt to reach people within their own communities (Ellingson 2007). This connection has grown stronger in recent years, as churches increasingly try to adapt their worship to reach an online audience (Drescher 2011; Gorrell 2019).

Beyond just a pragmatic desire to connect with everyday people, the affinity between contemporary worship and streaming and recording is also rooted in the ways this style has de-emphasized traditional Christian sacraments. Congregations with a contemporary worship style shifted the sacramental focus of worship away from rituals such as communion to music (Lim and Ruth 2017). This shift in focus is evident in church buildings that eschew traditional ritual objects in favor of an auditorium-style arrangement designed to enhance the musical performance (Adler et al. 2022; Kilde 2002; Loveland and Wheeler 2003). A service that leans heavily on concert-style musical performances likely translates more easily to an online format. In contrast, worship styles that center around physical rituals such as partaking in the Eucharist may be more difficult to shift online (Francis and Village 2021; Village and Francis 2022b).

3. Data

3.1. The National Congregations Study

To investigate the cultural affinities toward streaming among congregations, I used the National Congregations Study (Chaves et al. 2020b). The NCS is a nationally representative repeated cross-section of American religious congregations which has been conducted four times. For this article, I am only using the most recent wave, which collected data in 2018 and 2019. To obtain a nationally representative sample of congregations, the NCS used a hyper-network sampling method. The 2018–2019 wave of the NCS was collected in conjunction with the 2018 General Social Survey (GSS). GSS respondents who reported attending religious services at least once in the past year were asked to name the congregation they attended. The nominated congregations were then located, and a key informant from the congregation was administered a survey. In most cases, this key informant was the primary leader of the congregation, and if not, they were either another religious leader, staff member, or lay person who was knowledgeable about the inner workings of the congregation.

In total, 740 congregations were nominated by GSS respondents, located by NCS staff, and completed the survey. In order to increase the NCS sample size, a panel component was added to the 2018–2019 NCS where congregations interviewed in Wave III of the NCS in 2012 were re-interviewed. This added an additional 522 congregations to the sample, resulting in a total sample size of 1262 congregations from all religious traditions. All of the NCS interviews were completed before the onset of the COVID-19 pandemic, and as such the data provide a snapshot of the state of congregational worship and streaming practices right before the pandemic struck. While the NCS sampled both Christian and non-Christian congregations, the analysis here is restricted to only the 1195 Christian congregations in the sample, as the worship practices which serve as one of the major independent variables may have completely different meanings and connotations when used in different religious traditions. The response rate for the 2018–2019 NCS was 69% when not accounting for the GSS’s response rate, or 49% when accounting for the 2018 GSS’s own response rate (Chaves et al. 2020a).

The hyper-network sampling design of the NCS means that congregations are selected with probability proportional to size, as congregations with more members have more adults who could be sampled by the GSS. This feature can be undone through the use of survey weights which weight congregations inversely proportional to the number of adults in the congregation. As this article is focused on congregations as the unit of analysis, all descriptive statistics are reported using these survey weights (WT_ALL4_CONG_DUP in the public dataset).

3.2. Variables

3.2.1. Dependent Variables

The key dependent variable is streaming or recording worship services. In the 2018–2019 NCS, key informants were asked a number of questions about the most recent worship service at their congregation. Two of these yes or no questions were: “Was this service broadcast or streamed live so that people outside your building or campus could see it?” and “Was any part of this service recorded so people can listen or watch at a later time?” If the key informant responded “yes” to either of these questions, the congregations was coded as “1” for having streaming or recording. Otherwise, the congregation was coded as “0”.

3.2.2. Independent Variables

The two main independent variables in this study are religious tradition and contemporary worship practices. Religious tradition is measured by a categorical variable dividing congregations into four categories roughly corresponding to the RELTRAD categorization scheme (Steensland et al. 2000): Catholic, Conservative or Evangelical Protestant, Black Protestant, and White Liberal or Mainline Protestant.

There are no direct measures of contemporary worship in the NCS, but there are many measures of worship enthusiasm which can serve as a proxy for this kind of worship. I follow Roso (2022) by constructing a scale of six enthusiastic worship practices asked about in the NCS: attendees shouting “amen” or similar interjections, attendees jumping or dancing during the service, attendees applauding, the use of drums, speaking in tongues, and attendees raising their hands. For each element, a congregation was coded as “1” if the key informant reported that the practice occurred in the most recent service, and “0” if they did not. The results were summed together to create an additive scale of enthusiastic worship practices that ranged from 0 to 6. The scale has a Cronbach’s α of 0.79, indicating acceptable internal reliability.2

3.2.3. Controls

A relationship between enthusiastic worship practices or religious tradition and streaming or recording practices might simply reflect the fact that congregations of some religious traditions or with certain worship practices tend to differ in their levels of financial resources, technological knowledge, or other organizational features. To ensure that the relationships identified are not reduceable to factors other than organizational beliefs and practices about worship, a number of controls are included.

To control for liberal/conservative theological differences, a measure of theology is included. Theology is measured by asking the key informant if, “theologically speaking”, the congregation is “more on the conservative side”, “right in the middle”, or “more on the liberal side”. Those who responded that the congregation is either conservative or liberal were then asked if the congregation is “extremely”, “moderately”, or “slightly” liberal or conservative. These answers were used to create a 7–point scale where 1 is “extremely conservative” and 7 is “extremely liberal”.

To account for a congregation’s financial and personnel resources, I used several controls to measure the congregation’s size, income, building ownership, and status as a multisite congregation. To measure size, I used the natural log of the response to the question “how many adults—people 18 years or older—would you say regularly participate in the religious life of your congregation?” To measure income, I used the natural log of the response to the question “what is the total amount of money your congregation received in income from all sources during your most recent fiscal year?”3 To measure building ownership, I used the responses to the question “Does [the building your congregation uses for the primary worship service] belong to your congregation, or does it belong to another group that loans or rents space to you?” Congregations for whom the key informant responded “belongs to [my] congregation or denomination” were coded as “1”, and others were coded as “0”. To measure being a multisite, I used the variable MULTISITE_2 in the NCS dataset which uses answers from a sequence of two questions to categorize congregations as either multisite or not multisite. Key informants were asked if their congregation has “worship services that take place every week at more than one location, but all locations are considered part of the same congregation”. Those who answered “yes” were then asked to clarify if “all of the services take place in the same building or campus”. Congregations whose key informant also answered “no” to this second question were coded as “1” for being a multisite, while all others were coded as “0”.

To account for technological resources available to a congregation prior to the pandemic, I used indicators of whether or not the congregation had a Facebook page, a website, or used a projection system in worship. Key informants were asked whether their congregation has a Facebook page, whether their congregation has a website, and if any visual projection equipment was used in the congregation’s most recent main worship service. For each of these questions, an answer of “yes” was coded as “1”, and an answer of “no” was coded as “1”.

Finally, controls for features of the congregation’s history, location, and membership were also included. Year Founded was measured as a continuous variable based on the key informant’s response to the question: “In what year was your congregation officially founded?” Region was measured through a categorical variable that places congregations into one of four groups of aggregated US Census regions: New England or Mid-Atlantic; East North Central or West North Central; South Atlantic, East South Central, or West South Central; and Mountain or Pacific. Rural location was measured as a binary variable indicating whether or not the congregation is located in a rural census tract as defined by the US Census Bureau. Attendee age was measured by a question asking the key informant “of the regular adult participants, about what percentage are over 60 years old?” Attendee poverty was measured by a similar question asking the key informant “of the regular adult participants, about what percentage live in households with income under $35,000 a year?”

Table 1. Descriptive statistics.

	Mean/ Proportion	SD	Range	Non-Missing Values
Enthusiasm scale	3.04	2.03	1–6	1136
Amen	0.714	--	0–1	1188
Applause	0.622	--	0–1	1190
Drums	0.423	--	0–1	1171
Jump	0.304	--	0–1	1188
Raising hands	0.680	--	0–1	1178
Tongues	0.294	--	0–1	1173
Religious Tradition
Catholic	0.067	--	0–1	1195
White Evangelical	0.469	--	0–1	1195
Black Protestant	0.234	--	0–1	1195
Mainline Protestant	0.230	--	0–1	1195
Facebook page	0.718	--	0–1	1183
Website	0.702	--	0–1	1184
Projection in worship	0.491	--	0–1	1190
Owns building	0.834	--	0–1	1192
Adults in congregation	122	326	7–30,000	1195
Percent of adults >60 years old	41.12	26.48	0–100
Percent of adults in households with <$35,000 annual income	25.66	26.82	0–100	979
Congregation’s income	$387,722	$1,604,495	0–$79,000,000	903
Multisite	0.065	--	0–1	1195
Year founded	1947	57.7	1588–2018	1157
Liberal theology	2.87	1.56	1–7	1136
Region
Northeast/Mid-   Atlantic	0.121	--	0–1	1195
East North Central/   West North Central	0.250	--	0–1	1195
South Atlantic/   East South Central/   West South Central	0.472	--	0–1	1195
Mountain/Pacific	0.156	--	0–1	1195
Rural census tract	0.269	--	0–1	1195

Note: these numbers are calculated by weighting the data such that the congregation is the unit of analysis. See Chaves et al. (2020a) for details on weights for the NCS.

reports descriptive statistics for all the variables used in the analysis, weighted using WT_ALL4_CONG_DUP such that the congregation is the unit of analysis.

4. Methods

To assess the extent to which worship style has a relationship with streaming or recording practices net of technological and financial resources, I constructed a series of multiple regression models. Because the dependent variable of interest, whether or not the congregation streamed or recorded services, is binary, I used logistic regression and reported the coefficients as log odds ratios. Cases with missing data on key variables other than the congregation’s income or attendee poverty were listwise deleted from analysis. Because of the relatively large amount of missing data on financial variables (292 missing values for income and 216 missing values for attendee poverty), I used v3.15.0 of the mice package in R to impute missing values on these variables based on the values of the other independent variables.4 One consideration in performing regression on survey data is whether or not to calculate weighted regression coefficients. If features of the sample accounted for by survey weights interact with the key relationships of interest, then not using weights could result in biased coefficient estimates, but using weighted regression when it is not necessary results in inflated standard errors (Gelman 2007). Auxiliary analyses (see Supplementary Materials) reveal that there were no statistically significant interaction effects between the weighting variable and the key relationships of interest, and that there are no substantively meaningful differences between the coefficients of interest in the weighted and unweighted regressions. As such, to avoid inflating standard errors, I used unweighted logistic regression.

5. Results

5.1. Congregational Streaming and Recording by Religious Tradition

Figure 1 reports the percentage of congregations that streamed or recorded their services prior to the pandemic, broken down by religious tradition. Consistent with previous research on the subject, I found that streaming and recording were common—but not ubiquitous—among congregations on the eve of the COVID-19 pandemic (Holleman et al. 2022). Only about 1 in 5 (21%) congregations live-streamed their services, but just over half (51%) recorded their services. Overall, 54% of congregations either streamed or recorded their worship services in 2018 and 2019. Certainly, these figures increased through 2020 as the pandemic prevented congregations from meeting in person.

However, there are stark differences by religious tradition. Most notably, Catholic congregations were substantially less likely to stream or record their services than Protestant congregations. Depending on the tradition, between 48% and 64% of Protestant congregations reported streaming or recording their services, but only 12% of Catholic congregations did so. This suggests that it is much more difficult to translate a Catholic worship service online than a Protestant worship service. The focus on collective rituals and the Eucharist as the literal body and blood of Christ means that viewing the service through a screen is not a functional replacement for attending mass.

5.2. Enthusiastic Worship and Streaming Practices

Congregations with more enthusiastic or contemporary worship practices also had an affinity for streaming, as shown in Figure 2. There is some variation through the middle of the distribution of contemporary worship, but the overall pattern is clear: congregations with more contemporary worship streamed or recorded their services more often. While only one quarter (24%) of the congregations with the fewest contemporary worship practices streamed or recorded their services, 7 in 10 (70%) of congregations with the most enthusiastic services did so.

This suggests that there is a strong affinity between a contemporary worship style and streaming or recording. Contemporary worship services’ focus on music performances and longer sermons rather than collective rituals such as the Eucharist translates easier in an online medium. Consequently, congregations with more contemporary worship elements are more likely to stream or record their services.

5.3. Multiple Regression Results

While the bivariate results are strongly suggestive, there are a number of potential confounders. The differences in streaming by tradition may be a product of religious traditions varying in their level of enthusiastic practices. Catholic and mainline Protestant congregations have less enthusiastic services than evangelical Protestant or, especially, Black Protestant congregations (Roso et al. 2020). Additionally, these differences may reflect differences in congregations’ financial, human, or technological resources. Megachurches and smaller churches emulating a contemporary megachurch style may have more enthusiastic or contemporary worship services in addition to having the resources and technical know-how to set up streaming and recording systems. Congregations from different religious traditions also vary substantially in average size, income, rural/urban location, and other important features that may be related to streaming capability. If so, the bivariate relationships between contemporary worship or religious tradition and proclivity for streaming may simply reflect the fact that richer and more technologically savvy churches tend to stream more. To account for this, I constructed multiple logistic regression models including controls for financial resources, technological capacity, and other features of congregations, the results of which are reported in

Table 2. Logistic regression predicting streaming or recording of the congregation’s main worship service.

	Model 1	Model 2	Model 3	Model 4
(constant)	−1.242 *** (0.155)	−10.584 ** (3.707)	−1.987 *** (0.193)	−9.949 ** (3.725)
Religious Tradition
Catholic (ref.)	(ref.)	(ref.)	(ref.)	(ref.)
Evangelical Protestant	2.653 *** (0.199)	3.734 *** (0.362)	2.270 *** (0.207)	3.622 *** (0.366)
Black Protestant	2.140 *** (0.236)	3.988 *** (0.405)	0.972 *** (0.282)	3.366 *** (0.441)
Mainline Protestant	2.056 *** (0.208)	3.231 *** (0.348)	2.153 *** (0.216)	3.218 *** (0.348)
Enthusiastic worship			0.380 *** (0.0512)	0.208 *** (0.630)
Facebook page		1.281 *** (0.237)		1.191 *** (0.240)
Website		0.741 ** (0.269)		0.749 ** (0.272)
Projection in worship		0.877 *** (0.190)		0.691 *** (0.198)
Owns building		0.158 (0.330)		0.191 (0.568)
Adults in congregation (natural logged)		0.575 *** (0.107)		0.574 *** (0.108)
Percent of adults >60 years old		−0.003 (0.004)		−0.000 (0.004)
Percent of adults in households with <$35,000 annual income		−0.002 (0.004)		−0.005 (0.004)
Congregation’s income (natural logged)		0.114 ^ (0.058)		0.105 ^ (0.059)
Is multisite		−0.598 ^ (0.341)		−0.687 * (0.048)
Year founded		0.001 (0.002)		0.000 (0.002)
Liberal theology		0.004 (0.949)		−0.006 (0.065)
Region
Northeast/Mid-Atlantic (ref.)		(ref.)		(ref.)
South Atlantic/East South   Central/West South Central		0.297 (0.288)		0.357 (0.291)
East North Central/West   North Central		−0.041 (0.293)		0.074 (0.298)
Mountain/Pacific		−0.355 (0.321)		−0.276 (0.325)
Rural census tract		0.003 (0.255)		0.001 (0.256)
N	1037	1037	1037	1037

Note: ^ p < 0.1; * p < 0.05; ** p < 0.01; *** p < 0.001.

. Model 1 estimates the bivariate relationship between religious tradition and streaming or recording practices, and Model 2 adds controls for financial resources, technological capabilities, and other organizational features. Model 3 estimates the relationships between both religious tradition and enthusiastic worship practices with streaming/recording, and Model 4 incorporates the controls from Model 2.

As shown by Model 1, the differences between Catholic and each Protestant tradition in their streaming or recording practices are highly statistically significant in the bivariate (p < 0.001). These differences are not reduceable to differences in size, financial resources, technological resources, or other technological features, as shown in Model 2. The relationships between the control variables and streaming or recording are generally in the expected directions. Congregations with technological capacities such as having a Facebook page, website, or projection system were substantially more likely to stream or record their services (p < 0.01). Similarly, large congregations and congregations with more income were more likely to have these capabilities (p < 0.05). Somewhat surprisingly, multisite congregations are marginally significantly less likely to stream than non-multisite congregations after controlling for other factors (p < 0.1), perhaps suggesting that having worship services at multiple locations acts as an alternative means of reaching more people than streaming—at least when controlling for factors such as congregational size and income. There are no statistically significant differences in streaming or recording related to attendee age, attendee income, founding year, theological orientation, region, or rural/urban location net of the other variables included in the model.

Model 3 includes measures for enthusiastic worship alongside measures of religious tradition. This confirms what is evident in Figure 2: congregations with more enthusiastic worship are substantially more likely to stream than those with less enthusiastic worship. Furthermore, this relationship is not reduceable to religious tradition, as the correlation remains strong and statistically significant (p < 0.001) even when controlling for religious tradition. Likewise, the large difference between Catholic congregations and every Protestant tradition is not merely a reflection of the fact that Catholic congregations have much less enthusiastic services. There remains a large and highly statistically significant difference between Catholic congregations and Protestant congregations in streaming or recording (p < 0.001). Model 4 includes additional controls for technological resources, financial resources, and organizational features. Once again, the relationship between enthusiastic worship and streaming or recording services remains strong and statistically significant (p < 0.01). Even net of these other factors, there remains a strong cultural affinity for streaming or recording worship among certain forms of worship and religious traditions.

5.4. Interaction between Religious Tradition and Enthusiastic Worship

The relationship between enthusiastic worship and streaming or recording may be fundamentally different across religious traditions. Catholic worship in particular has a strong focus on communal rituals, and even the most enthusiastic Catholic mass will include partaking of the Eucharist. As such, enthusiastic worship’s strong affinity with streaming or recording may hold only among Protestant congregations.

Table 3. Logistic regression predicting streaming or recording with Catholic interaction.

	Model 5	Model 6	Model 7
(constant)	0.309 * (0.141)	0.190 (0.148)	−7.423 (3.603)
Enthusiastic worship	0.269 *** (0.041)	0.313 *** (0.313)	0.267 *** (0.060)
Catholic	−2.065 *** (0.180)	−1.440 *** (0.292)	−2.756 *** (0.403)
Catholic x enthusiastic worship		−0.309* (0.121)	−0.335 * (0.136)
Facebook page			1.190 *** (0.238)
Website			0.756 ** (0.238)
Projection in worship			0.741 *** (0.191)
Owns building			0.183 (0.337)
Adults in congregation (natural logged)			0.591 *** (0.110)
Percent of adults >60 years old			0.001 (0.004)
Percent of adults in households with <$35,000 annual income			−0.005 (0.004)
Congregation’s income (natural logged)			0.099 (0.060)
Is multisite			−0.678 ^ (0.348)
Year founded			0.001 (0.002)
Liberal theology			−0.030 (0.059)
Region
Northeast/Mid-Atlantic (ref.)			(ref.)
South Atlantic/East South   Central/West South Central			0.393 (0.289)
East North Central/West   North Central			0.114 (0.297)
Mountain/Pacific			−0.119 (0.328)
Rural census tract			0.047 (0.255)
N	1037	1037	1037

Note: ^ p < 0.1; * p < 0.05; ** p < 0.01; *** p < 0.001.

reports results from models testing this hypothesis by estimating models similar to those in Table 2, but including a Catholic vs. non-Catholic indicator and a Catholic x enthusiastic worship interaction term. For ease of interpretation, individual dummy variables for each Protestant tradition are not included.

Model 5 shows what was evident from Model 3—Protestant congregations and congregations with more enthusiastic worship were more likely to stream their worship services prior to the pandemic. Model 6 adds an interaction term, and shows that there is a strong interaction between religious tradition and enthusiastic worship. The negative interaction term means that, among Catholics, the positive relationship between enthusiastic worship and streaming or recording is much weaker. It is so much weaker, in fact, that the relationship is reduced to zero. Model 7 adds in all the control variables used in Model 4 and confirms that this interaction remains strong and significant even after accounting for financial resources, technological capacity, and other organizational features.

To aid in interpreting the interaction, Figure 3 graphs the predictions of Model 6. The large baseline differences between Protestants and Catholics are clear, as even the least enthusiastic Protestant congregations are twice as likely to stream or record their services as any Catholic congregation. Furthermore, there is a clear and strong positive relationship between enthusiastic worship and streaming or recording among Protestant congregations, but there is no relationship present among Catholic congregations. While highly enthusiastic Protestant congregations are as much as 1.5 times more likely to stream or record than Protestant congregations without any enthusiastic practices, there is no such difference among Catholics.

6. Discussion and Conclusions

In this article I have demonstrated robust relationships between both religious tradition and enthusiastic worship practices with streaming or recording worship services which persist even after accounting for financial resources, technological know-how, and other organizational features. These strong statistical relationships point to cultural affinities between certain liturgical practices and streaming or recording technologies that are not reduceable to material resources. Congregational preparedness for the pandemic was not just about money or personnel. It also was about the extent to which the congregations’ core activity of worship was amenable to an online format. Culture matters.

This study is not without its limitations. These data were collected entirely before the pandemic, and so cannot speak to the ways in which congregations adapted their practices during the pandemic. Furthermore, the analyses are limited to just the variables that are in the dataset. Of particular note is the exclusion of more direct measures of sacramental activity or contemporary worship. The fourth wave of the NCS has no indicator for whether or not the congregations’ most recent worship service included communion, and though prior NCS waves did include such a measure, they did not have measures of worship streaming or recording. Further research is needed on how the presence of particular sacraments within a worship service can enable or dissuade streaming or recording of that service. Another limitation of this article is that it can only speak to the cultural practices of Christian congregations. Of course, congregations from all religious traditions had to adapt their worship to an online format during the pandemic (e.g., Ben-Lulu 2021), and more research is needed to explore the affinities between worship and streaming or recording among non-Christian congregations.

Moreover, this article is largely agnostic to the causal direction of the key relationships examined. I chose to present the models as if liturgical practice is an independent variable shaping streaming and recording practices. While this is plausible, it is also likely that streaming or recording encourage a certain kind of worship, a possibility future research can examine more thoroughly. Given the sudden shift to online worship, did worship services change to accommodate the new mode, and has that change persisted even after most social distancing regulations have been lifted? While congregations are sometimes slow to change their worship practices, they do change them (Roso 2022), and an affinity between particular forms of worship and streaming and recording suggests that such a change will occur. In what Swidler (1986) describes as “unsettled times”, there may be opportunities for religious leaders to forge new logics and practices around worship (Johnston et al. 2021). Time will tell if these new practices persist after the pandemic.

Despite these limitations, this article reveals important dynamics of congregations’ activities. Enthusiastic worship practices and religious tradition are just two areas where elective affinities exist between established organizational practices and the incorporation of new technology, but there may be important culture affinities in other areas as well. When studying the activities of religious congregations, scholars cannot look to just financial resources and the expertise of personnel—they must also account for cultural factors. Even large, financially well-endowed congregations may struggle to implement an innovation that clashes with existing practices. As religion continues to change and decline in America (Voas and Chaves 2016), congregations will need to continue to adapt their practices to meet the needs of their faithful, but the cultural factors within the congregations will shape the kinds of adaptations that they are able to implement.