Prayer Intensity, Technological Mediation, and Civic Engagement: Comparing Catholic, Lutheran, and Orthodox Contexts

Abstract

Technological advancements keep influencing religious landscapes in unpredictable ways. This implies changes at the level of personal spirituality and also at the level of community building and civic engagement across different denominations. In this paper, we present survey data (N = 443) from Malta (Southern Europe), Serbia (Balkans), and Denmark (Northern Europe), which assessed participants’ prayer intensity, Christian identity, Christian belief, and civic engagement behaviors, among other variables. The participants in our sample were all Christians: the participants from Malta were mostly Catholic, those from Serbia were mostly Orthodox, and those from Denmark were mostly Lutheran, reflecting the dominant Christian contexts and denominations in all three countries. We conducted multiple regression analysis showing how prayer intensity predicts civic engagement, even when adjusting for other covariates, notably those tapping Christian identity and Christian belief. The relationship was significant across all three countries. Moreover, we conducted further multiple regression analyses with two prayer intensity sub-indices: one tapping technologically mediated prayer (e.g., using apps or podcasts) and the other tapping non-technologically mediated prayer (e.g., praying directly to God or going to mass). In this model, only non-technologically mediated prayer predicted civic engagement in Malta and Denmark, and no sub-index predicted civic engagement in Serbia. Our discussion focuses on the implications of these patterns for engagement and community building, with a particular focus on religious collectives across denominations and the impact of technology.

1. Introduction

The impact of technological change continues to be noted in a variety of domains, ranging from religious to secular ones. Technological dynamics are distinguished by two main factors, namely (a) the obstinate dynamics of technique, which infiltrate various aspects of daily life and make themselves indispensable (Ellul [1954] 1964), and (b) the fact that, over time, technological innovations function not as tools but as geographies (see Straccamore et al. 2023).

In religious milieus, technology operates in different ways, making the “politics of space” (Kong 2001, p. 405) an increasingly important field of study. Issues abound, namely because of “uneven access to techno-religious spaces” (Kong 2001, p. 404) and related matters at the interface between religious expression and technological innovation. The comfort brought about, or technical expertise required, by technology is often held to be at odds with religious expression, with both spheres of human endeavor held as advancing different projects (see Buhagiar and Sammut 2020) altogether. As the debate on this interface continues unabated, it is worth noting that classic debates on the relationship between different religious expressions and civic engagement remain complex, both offline and online. Studies often link religiosity with pro-sociality (e.g., Lewis et al. 2013; Putnam and Campbell 2012) and civic engagement (e.g., Pancer 2015; Kim and Jang 2017) in nuanced and qualified ways. However, the measures used in such studies matter a great deal (e.g., individual vs. collective measures of religious expression, different types of civic engagement, different contexts of study, etc.).

In adopting an extended view of religion (Kalmykova 2025)—that is, one that takes religion as an embodied and practice-based contextual phenomenon—it is important to define civic engagement with a view to finding common ground with engagement behaviors that are broadly appreciated in religious contexts. Civic engagement has been defined as any set of actions concerned with the welfare of others or of the community more broadly (Adler and Goggin 2005; Putnam and Campbell 2012). Some authors also view civic engagement as being synonymous with the common good (Adler and Goggin 2005; Pancer 2015), that is, with “the concerns, interests and common good of a community” (Barrett and Pachi 2019, p. 3). Whereas definitions of civic engagement differ in the extent to which they factor in political considerations, notions such as voluntary work tend to feature throughout (Ekman and Amnå 2012).

In the civic engagement literature, one less studied but highly important variable is that of prayer (in terms of importance or intensity), as per the following studies. For instance, in recent work, “prayer importance was shown to mediate the relationship between beliefs accepting the existence of God and social activities and to increase the strength of service-oriented activities” (Tatala et al. 2024) (p. 1). Similarly, Jang et al. (2023) found that prayer can be related to transcendent accountability to a higher power (when it comes to ethical decision-making) and that, in turn, transcendent accountability is positively related to pro-community attitudes.

Interestingly, Storm (2015) writes that “religion increases volunteering primarily through bonding rather than bridging social networks. […] solitary and collective religious rituals may both have an effect on civic participation, but whereas the effect of service attendance is mediated through bonding social networks, the effect of prayer is mediated more through bridging networks” (p. 14). It is worth noting that Storm’s (2015) focus was largely on non-Christian religions, which raises questions about different dynamics—both between religions and within denominations in the same religion. Another chief source of difficulty concerns the multitude of variables in the literature that are used for similar ends and the nuances between different kinds of measures. This issue plagues the literature on civic engagement as much as it does that on religious expression.

Four humble ways to attenuate the negative consequences of this diversity are the following: (a) conducting context-sensitive studies, favoring an extended, cultural, and embodied understanding of religion (see Kalmykova 2025); (b) utilizing outcome variables (in surveys) that tap behavioral intentions/actions, thus focusing on the tangible civic outcomes that individuals or communities work toward (see Buhagiar and Sammut 2020); (c) using predictor variables that conceptually cohere and reflect people’s actual lives, hence enhancing the lived validity of the study (see Hartley 1998; Akkerman et al. 2021); and (d) looking at intra-religious (e.g., intra-Christian) differences, which allows researchers to distinguish between patterns that are inter-denominational and others that are specific to some denominations. The last point advances a research focus that increases the chances of there being projects (Buhagiar and Sammut 2020) and underlying worldviews (Koltko-Rivera 2004) that are similarly shared across the denominations in question, limiting issues relating to the translation of ideas from one religion to another.

2. The Present Study

Accordingly, in line with the above four points, we present a quantitative survey study that (a) analyzed data from Malta (majority Catholic), Serbia (majority Orthodox), and Denmark (majority Lutheran) and, therefore, explicitly relied on contexts with a majority religious culture (even if some participants were Christian but belonged to other Christian traditions); (b) civic engagement behaviors (Doolittle and Faul 2013) as an outcome variable (behavior-related output); (c) used predictor variables that coherently focused on a belief–identity–expression triad, that is, Christian belief (Nikodem and Jurlina 2018), Christian identity—using Pulis’s (2019) scale-adaptation of the Centre for Applied Research in the Center for Applied Research in the Apostolate’s (2008) questions, and Christian prayer intensity (Pulis 2023); and (d) studied both inter-denominational similarities and differences.

In the study we present below, we conducted multiple regression analysis across the three sub-samples (Serbia, Malta, and Denmark) to test whether prayer intensity (and Christian identity and Christian belief) predicts civic engagement, even when adjusting for other covariates (H1). Moreover, we conducted further multiple regression analysis with two prayer intensity sub-indices: one tapping technologically mediated prayer (e.g., using apps or podcasts) and the other tapping non-technologically mediated prayer (e.g., praying directly to God or going to mass). The purpose of the second set of models across the three sub-samples (Serbia, Malta, and Denmark) was to see whether only non-technologically mediated prayer predicts civic engagement when disaggregating the index in this way (H2). This study builds on preliminary work on prayer and civic engagement in Malta and Serbia, which had only involved descriptive statistics and simple bivariate observations (Pulis et al. 2024). Accordingly, this study makes a novel contribution by including a new sample from Denmark and by applying multiple regression models across all three countries (Malta, Serbia, and Denmark). This makes for a much more in-depth examination of the relationships at hand. Another unique contribution of this paper, therefore, lies in that we test H1 and H2 using multivariate modeling across contexts with distinct Christian heritages.

Before proceeding to present the methodology, the findings of the study, and our discussion of their implications, we note important data points on the contexts of the study concerning civic engagement and religiosity. Firstly, the civic society participation index recently gave Serbia, Malta, and Denmark the scores of 0.53, 0.85, and 0.97, respectively, with higher scores indicating higher levels of civic participation in various aspects of public life (V-Dem Institute 2025). Similarly, the democracy index (World Population Review 2025) scored Serbia, Malta, and Denmark, respectively, at 6.33, 7.93, and 9.28, with higher scores indicating more robust democratic processes. Interestingly, it has been reported that the population of Serbia tends to express a preference for individual volunteering, very low levels of charity, and low public participation, all unfolding in an unstable political climate and alongside the presence of a few notable NGOs (see ILOSTAT 2024). In contrast, the population of Malta has been reported to express preferences for organized volunteering and high levels of charity, amidst highly extensive and overbearing influence by political parties and an increasingly expanding civil society (European Parliament 2020). Moreover, Denmark has been reported as exhibiting high levels of both individual and organized volunteering and high levels of charity, amidst a robust civic tradition and the presence of diverse community-based and political entities (European Parliament 2020).

Ultimately, what makes these three contexts opportune for our study are the religious demographics in each of the countries. Serbia’s population is ∼85% Orthodox Christian, with around a third identifying as highly religious and around a fifth who pray daily and with compulsory primary education in traditional religion or civic education (Sikt 2018; Gavrilovic 2021; Ćumura and Barbanti 2023). Similarly, Malta’s population is ∼85% Christian Catholic, with a third who attend Sunday mass and with Catholicism enshrined in the constitution. In Malta, although religious participation is decreasing over time, Catholic cultural influence remains high (e.g., Discern 2018). Finally, Denmark’s population is ∼75% Evangelical Lutheran Church (ELC, or Folkekirken), with most people exhibiting low religiosity in a highly secular state. In Denmark, only ∼10% agree that religion is important for them or pray daily, and the Church is seen more as a community center (Jensen 2015; Pew Research Center 2019). While the ELC is stipulated as the state church in Denmark’s constitution, religion tends to be related to a more individual (rather than an institutional) focus in Denmark, reflecting ongoing transitions in the church (see Jensen 2015).

3. Method

Having reviewed the literature on the main variables of the study and having provided information on the relevant contexts when it comes to both civic engagement and religious expression/prayer, we now turn to the study’s methodology. We provide details on procedures, measures, and analyses below.

3.1. Procedure

A survey in the form of an online questionnaire was built in English and translated to Maltese, Serbian, and Danish. Piloting showed that the survey had an average completion time of 7 min. Distribution took place online with participants from Malta and Serbia between December 2023 and January 2024 and with participants from Denmark between June and July 2024. Importantly, all three countries have a majority religious culture (Catholicism, Orthodoxy, and Lutheranism), making inter-denominational comparisons using survey methodology a meaningful endeavor. The survey was completed via LimeSurvey (open-source software).

The questionnaires were distributed using convenience and snowball sampling via varied sets of original “seed” participants (Kirchherr and Charles 2018). This encouraged maximum variation among the participants while enabling us to access the Christian populations in the contexts of the study (responses from Denmark were also collected using the application Pollfish (www.pollfish.com, accessed on 1 April 2024) due to a low response rate). Multiple snowball streams avoided having homogeneous participant pools, thus ensuring samples that were diverse enough for “interpretive consistency” (Collins et al. 2016, p. 87) when conducting inferential statistics. Our study was therefore “correlational–comparative” (Gelo et al. 2008, p. 271) because it studied multivariate correlations within samples (Malta, Serbia, Denmark) and compared multiple regression models across the same samples (Brewer and Kuhn 2010, p. 126).

3.2. Measures

The questionnaire consisted of the following measures. All measures are detailed below, including reliability statistics where relevant.

3.2.1. Demographics

First, participants’ (1) demographics were collected. These namely consisted of age, gender, religion, education, and employment.

3.2.2. Christian Prayer Intensity

Secondly, the questionnaire proceeded with an index tapping (2) Christian prayer intensity (Pulis 2023) (henceforth, prayer intensity). This measure was based on a lightly modified version of Pulis (2023). The light modifications were intended to make the items relevant to all three contexts in question. Here, we asked participants to indicate engagement per prayer type using a 7-point Likert index (where 1 = “Never”, 2 = “Once a year”, 3 = “Once a month”, 4 = “Once a week”, 5 = “Multiple days per week”, 6 = “Once a day”, and 7 = “More than once a day”), following a heading stating “How do you live your Christian faith?” (15 items). The prayer types in question concerned various modes, such as participation in prayer groups and mass, individual prayer rituals, and overall prayer life, and ranged from traditional prayers (e.g., Our Father) to contemporary practices, such as using prayer apps or tuning into Christian podcasts (one item concerning the rosary/Jesus prayer was omitted from inferential statistics as it would have been irrelevant to Lutheran participants). Only items particular to Christianity that tapped frequency of prayer in the context of “How do you live your Christian faith?” were present in this index.

For further analysis, this index was split into two sub-indices: (a) technologically mediated prayer (TMP), which included five items—prayer using mobile apps, podcasts, videos, social media, and music; and (b) non-technologically mediated prayer (NTMP), which included nine items—traditional prayer, prayer before/after meals, mass, bible prayer, conversation with god, quiet chapel, confession, prayer group, and spiritual book (the item “Other type of prayer”, present in the overall index, was not included in any of the two sub-indices due to its open nature). There was no crossover between the subscale items; that is, each item only featured in one subscale or the other, apart from the overall scale. Both the overall index (α > 0.9 for all three samples) and also the two sub-indices, that is, TMP (α > 0.8 for all three samples) and NTMP (α > 0.9 for all three samples), showed excellent reliability in all three samples.

3.2.3. Christian Identity

The third measure tapped (3) Christian identity (Pulis 2019; adapted from Center for Applied Research in the Apostolate 2008). This index had ten items concerning Christian views on war/peace, marriage, the nature of work, and so on and asked participants about the extent of their agreement with these items. This index, thus, measured how attuned participants were to Christian identity and values and about matters related to both doctrine and sociopolitical issues. All items were scored on a Likert scale ranging from 1 (Strongly disagree) to 6 (Strongly agree). This index showed differing levels of internal reliability (Serbia: α = 0.78; Malta: α = 0.80; Denmark: α = 0.59).

3.2.4. Christian Belief

The fourth measure tapped (4) Christian belief (Nikodem and Jurlina 2018). This index tapped participants’ views on the soul, the Bible, the afterlife, the idea of God-as-personal, and so on (seven items in total). All items were scored on a Likert scale ranging from 1 (Strongly disagree) to 6 (Strongly agree). All three samples exhibited excellent internal reliability on this index (α > 0.8).

3.2.5. Cosmic Religion

We also included a fifth measure that tapped (5) Cosmic Religion (Nikodem and Jurlina 2018), that is, ideas related to the supernatural but which exclude the idea of a personal God (seven items in total). Items were about ideas such as there being something supernatural, there being a cosmic intelligence, God being a mix of good and evil, and similar non-Christian beliefs. This item was included as a “counter-weight” to the Christian items in order to assess the extent to which participants also entertained non-Christian beliefs in their worldviews and the extent to which these predict civic engagement behaviors alongside the Christianity-related predictors. The three samples differed in internal reliability, with all values being acceptable (Serbia: α = 0.81, Malta: α = 0.69, Denmark: α = 0.78).

3.2.6. Civic Engagement Behaviors

The final measure tapped (6) civic engagement behaviors (Doolittle and Faul 2013). This was a subscale in Doolittle and Faul’s (2013) broader civic engagement scale and focuses on the actions people take to improve their community. The six items of this subscale tapped specific behaviors such as involvement in structured positions, participation in communal discussions, volunteering, and contributions to charity. All items were scored on a Likert scale ranging from 1 (Never) to 6 (Always). All three samples exhibited excellent internal reliability on this measure (α > 0.8).

3.3. Analysis

The data were analyzed using descriptive statistics and inferential statistics. After obtaining descriptive statistics and conducting exploratory bivariate statistics (see Section 4), a multiple regression was conducted for the Malta, Serbia, and Denmark samples, involving civic engagement behaviors as the outcome variable and the following predictor variables: (a) gender, (b) age, (c) prayer intensity, (d) Christian belief, (e) Christian identity, and (f) Cosmic Religion. This first exploratory model was intended to see how the prayer–belief–identity triad predicts civic engagement behaviors.

Following the above, a second set of multiple regression models was conducted for the Malta, Serbia, and Denmark samples, this time involving civic engagement behaviors (CEB) as the outcome variable and the following predictor variables: (a) gender, (b) age, (c) non-technologically mediated prayer intensity, (d) technologically mediated prayer intensity, (e) Christian belief, (f) Christian identity, and (g) Cosmic Religion. Based on the reviewed literature on the technology–religion interface, we had reason to believe that technological and non-technological prayer may predict civic engagement behaviors in different ways, and, hence, we disaggregated the prayer intensity index into two sub-indices in the second set of multiple regression models. All multiple regression models were conducted using robust estimators in view of our sample sizes. More specifically, we used the heteroscedasticity-consistent HC3 estimator (Davidson and MacKinnon 1993).

Our study was guided by two main hypotheses. H1 predicted that Christian prayer intensity, Christian belief, and Christian identity all predict CEB, even when adjusting for covariates. Moreover, H2 predicted that, when disaggregating Christian prayer intensity into TMP and NTMP, only NTMP predicts CEB (but not TMP). Below, we present our findings, which partly support H1 across all three countries and which fully support H2 in Malta and Denmark but not in Serbia.

4. Findings

Having provided an overview of the employed methodology, this section now presents the findings of this study. We start by presenting descriptive statistics and proceed with presenting exploratory bivariate statistics. We then present our multiple regression models before proceeding to discuss our findings. Our analysis focuses solely on the Christian participants that completed our questionnaire. The original sample sizes (Christians and non-Christians) consisted of 199 participants (Serbia), 174 participants (Malta), and 220 participants (Denmark). However, the below analyses focus exclusively on the Christian proportions of these samples. Accordingly, the following Christian-only samples were taken as final for analytic purposes: N = 164 (Serbia), N = 143 (Malta), and N = 136 (Denmark).

4.1. Descriptive Statistics

Our total sample size of Christian participants was N = 443. The participants from Serbia (n = 164 Christians) had the following characteristics: Overall, 95% (156 participants) were Orthodox Christians, and the participants were largely female (100 participants; 61%) and had a mean age of 39.0 (SD = 13.6). Most of the participants had a tertiary level of education (65 participants; 40%) and were employed full-time (102 participants; 62%). Overall, the participants’ mean civic engagement behaviors (CEB) score was 3.4 (on a 1–6 scale, thus slightly below the midpoint of 3.5) (SD = 1.4).

Moreover, the participants from Malta (n = 143 Christians) had the following characteristics: Overall, 97% (139 participants) were Catholic Christians, and the participants were largely female (82 participants; 57%) and had a mean age of 43.3 (SD = 13.4). Most of the participants had a tertiary level of education (110 participants; 77%) and were employed full-time (106 participants; 74%). Overall, the participants’ mean CEB score was 3.9 (thus slightly above the midpoint of 3.5) (SD = 1.1).

Finally, the participants from Denmark (n = 136 Christians) had the following characteristics. Overall, 86% (117 participants) were Lutheran Christians (Folkekirken), and the participants were largely female (69 participants; 51%) and had a mean age of 49.5 (SD = 16.0). Most of the participants had a tertiary level of education (111 participants; 82%) and were employed full-time (66 participants; 49%). Overall, the participants’ mean CEB score was 3.8 (thus slightly above the midpoint of 3.5) (SD = 1.3).

4.2. Exploratory Bivariate Statistics

In terms of demographic predictors, in Serbia, employment (full-time: mean = 3.6, SD = 1.3; non-full-time: mean = 3.1, SD = 1.4) predicted civic engagement behaviors (CEB), t(161) = 2.03, p < 0.05. In Malta, no demographic variable predicted CEB. In Denmark, gender predicted CEB (females: mean = 4.0, SD = 1.2; males: mean = 3.5, SD = 1.3), t(133) = 2.1, p < 0.05. Also in Denmark, education (tertiary: mean = 3.9, SD = 1.2; non-tertiary: mean = 3.3, SD = 1.4) predicted CEB, t(134) = 2.3, p < 0.05. Moreover, across the overall sample (Malta, Serbia, and Denmark), all individual items in the prayer intensity index correlated positively with CEB, the only exception being the item about prayer using mobile apps in the Denmark sub-sample (no correlation in this case).

Moreover, in the Serbia sub-sample, positive correlations were found between prayer intensity (full index) and CEB (r = 0.35, p < 0.001), Christian belief and CEB (r = 0.26, p < 0.001), Christian identity and CEB (r = 0.25, p < 0.01), technologically mediated prayer (TMP) and CEB (r = 0.33, p = < 0.001), and non-technologically mediated prayer (NTMP) and CEB (r = 0.35, p < 0.001).

In the Malta sub-sample, positive correlations were found between prayer intensity (full index) and CEB (r = 0.60, p < 0.001), Christian belief and CEB (r = 0.46, p < 0.001), Christian identity and CEB (r = 0.48, p < 0.001), technologically mediated prayer (TMP) and CEB (r = 0.44, p = < 0.001), and non-technologically mediated prayer (NTMP) and CEB (r = 0.60, p < 0.001); and a negative correlation was found between Cosmic Religion and CEB (r = −0.20, p < 0.05).

Similarly, in the Denmark sub-sample, positive correlations were found between prayer intensity (full index) and CEB (r = 0.41, p < 0.001), Christian belief and CEB (r = 0.27, p < 0.001), Christian identity and CEB (r = 0.40, p < 0.001), technologically mediated prayer (TMP) and CEB (r = 0.28, p = < 0.001), and non-technologically mediated prayer (NTMP) and CEB (r = 0.43, p < 0.001).

4.3. Multiple Regression Models: Set 1

Based on the above exploratory bivariate statistics, we conducted a first set of multiple regression models (for the Malta, Serbia, and Denmark sub-samples), involving civic engagement behaviors (CEB) as the outcome variable and the following predictor variables: (a) gender, (b) age, (c) prayer intensity, (d) Christian belief, (e) Christian identity, and (f) Cosmic Religion. This first exploratory model was intended to see how the prayer–belief–identity triad predicts civic engagement behaviors.

Table 1. Multiple Regression Model 1a: Serbia.

Model	Model Summary					ANOVA
	R	R2	Adj. R2	Std.Err.of Est.		SS	df	MS	F	p
Serbia	0.379	0.143	0.110	0.518	Regression	42.256	6	7.043	4.269	<0.001
					Residual	252.395	153	1.650
					Total	294.651	159
Variable				HC3 Estimator
		β		B	SE	CI (LL)		CI (UL)	t	p
Intercept				5.09	0.804	0.410		3.589	2.485	0.014
Gender		0.035		0.000	0.076	−0.151		0.151	0.000	0.999
Age		0.054		0.005	0.009	−0.012		0.023	0.606	0.545
Prayer Intensity		0.379		0.400	0.129	0.146		0.655	3.104	0.002
Christian Belief		0.029		0.029	0.131	−0.230		0.287	0.218	0.827
Christian Identity		0.007		0.009	0.177	−0.341		0.359	0.051	0.959
Cosmic Religion		0.152		0.206	0.135	−0.061		0.474	1.527	0.129

Note. This table presents Multiple Regression Model 1a: Serbia. Significant predictors (p < 0.05) of CEB are indicated using bold and underline.

,

Table 2. Multiple Regression Model 1b: Malta.

Model	Model Summary					ANOVA
	R	R2	Adj. R2	Std.Err.of Est.		SS	df	MS	F	p
Malta	0.631	0.398	0.371	0.874	Regression	67.779	6	11.296	14.791	<0.001
					Residual	102.341	134	0.764
					Total	170.120	140
Variable				HC3 Estimator
		β		B	SE	CI (LL)		CI (UL)	t	p
Intercept				1.272	0.710	−0.133		2.676	1.791	0.076
Gender		−0.080		−0.178	0.154	−0.483		0.126	−1.158	0.249
Age		0.065		0.005	0.006	−0.006		0.017	0.911	0.364
Prayer Intensity		0.601		0.544	0.108	0.329		0.758	5.015	<0.001
Christian Belief		−0.052		−0.049	0.103	−0.253		0.155	−0.476	0.635
Christian Identity		0.177		0.225	0.154	−0.080		0.530	1.460	0.147
Cosmic Religion		0.163		0.261	0.130	0.005		0.518	2.014	0.046

Note. This table presents Multiple Regression Model 1b: Malta. Significant predictors (p < 0.05) of CEB are indicated using bold and underline.

and

Table 3. Multiple Regression Model 1c: Denmark.

Model	Model Summary					ANOVA
	R	R2	Adj. R2	Std.Err.of Est.		SS	df	MS	F	p
Denmark	0.496	0.246	0.211	1.133	Regression	53.968	6	8.995	7.003	<0.001
					Residual	165.685	129	1.284
					Total	219.653	135
Variable				HC3 Estimator
		β		B	SE	CI (LL)		CI (UL)	t	p
Intercept				1.237	0.757	−0.261		2.735	1.634	0.105
Gender		0.072		0.000	0.277	−0.549		0.549	0.000	1.00
Age		0.155		0.012	0.007	−0.001		0.026	1.783	0.077
Prayer Intensity		0.341		0.364	0.111	0.144		0.584	3.275	0.001
Christian Belief		−0.083		−0.080	0.110	−0.297		0.137	−0.732	0.466
Christian Identity		0.256		0.439	0.191	0.061		0.818	2.295	0.023
Cosmic Religion		0.052		0.068	0.127	−0.184		0.319	0.534	0.594

Note. This table presents Multiple Regression Model 1c: Denmark. Significant predictors (p < 0.05) of CEB are indicated using bold and underline.

present the findings. For the following models, in plain language for readers without a statistical background, the terms mean the following: (i) β (standardized beta) shows how strong the predictor is relative to other predictors in the model (e.g., if β = 0.4, then that predictor has a moderate positive effect on the outcome variable compared to other variables in the model); (ii) B (unstandardized beta) tells us how much the outcome changes when a predictor increases by one unit (e.g., if B = 0.4, then for every 1 point increase in that predictor, the outcome variable increases by 0.4 units, unstandardized); (iii) p tells us whether the relationship is statistically significant (i.e., a p that is lower than 0.05 is statistically significant); and (iv) R² indicates how much of the variation of the outcome variable is explained by the predictors in the model (e.g., an R² of 0.50 means 50% of the variation in the outcome variable is explained by the variables included).

Set 1 (i.e., models 1a, 1b, and 1c) tested H1, which posited that prayer intensity, Christian belief, and Christian identity all predict CEB, even when adjusting for covariates. In Model 1a (Serbia), the predictors accounted for 14.3% of the variance (R² = 0.143; Adjusted R² = 0.110) in CEB, and only prayer intensity (β = 0.379) positively predicted CEB, t(153) = 3.104, p < 0.01, thus partly supporting H1. In Model 1b (Malta), the predictors accounted for 39.8% of the variance (R² = 0.398; Adjusted R² = 0.371) in CEB; and out of the Christian triad, only prayer intensity (β = 0.601) positively predicted CEB, t(134) = 5.015, p < 0.001, thus partly supporting H1. In this model, Cosmic Religion also positively predicted CEB (β = 0.163), t(134) = 2.014, p < 0.05. Finally, in Model 1c (Denmark), the predictors accounted for 24.6% of the variance (R² = 0.246; Adjusted R² = 0.211) in CEB; and out of the Christian triad, prayer intensity (β = 0.341) positively predicted CEB, t(129) = 3.275, p < 0.01, and Christian identity also positively predicted CEB (β = 0.256), t(129) = 2.295, p < 0.05, thus partly supporting H1.

4.4. Multiple Regression Models: Set 2

We followed up the Set 1 models with a second set of multiple regression models. Set 2 models involved civic engagement behaviors as the outcome variable and the following predictor variables: (a) gender, (b) age, (c) TMP, (d) NTMP, (e) Christian belief, (f) Christian identity, and (g) Cosmic Religion. This second set of models was intended to see what changes disaggregating prayer intensity into technologically mediated (TMP) and non-technologically mediated (NTMP) prayer intensity would bring about in predicting CEB.

Table 4. Multiple Regression Model 2a: Serbia.

Model	Model Summary					ANOVA
	R	R2	Adj. R2	Std.Err.of Est.		SS	df	MS	F	p
Serbia	0.383	0.147	0.108	1.28594	Regression	43.296	7	6.185	3.740	<0.001
					Residual	251.355	152	1.654
					Total	294.651	159
Variable				HC3 Estimator
		β		B	SE	CI (LL)		CI (UL)	t	p
Intercept				1.998	0.817	0.385		3.612	2.447	0.016
Gender		0.039		0.000	0.068	−0.134		0.134	0.000	1.00
Age		0.058		0.006	0.009	−0.012		0.024	0.648	0.518
TMP		0.169		0.174	0.153	−0.129		0.476	1.136	0.258
NTMP		0.235		0.236	0.177	−0.114		0.586	1.332	0.185
Christian Belief		0.025		0.025	0.134	−0.239		0.290	0.188	0.851
Christian Identity		0.005		0.006	0.178	−0.345		0.358	0.034	0.973
Cosmic Religion		0.155		0.211	0.134	−0.055		0.476	1.567	0.119

Note. This table presents Multiple Regression Model 2a: Serbia. No significant predictors of CEB were found.

,

Table 5. Multiple Regression Model 2b: Malta.

Model	Model Summary					ANOVA
	R	R2	Adj. R2	Std.Err.of Est.		SS	df	MS	F	p
Malta	0.627	0.394	0.362	0.88076	Regression	66.946	7	9.564	12.328	<0.001
					Residual	103.174	133	0.776
					Total	170.120	140
Variable				HC3 Estimator
		β		B	SE	CI (LL)		CI (UL)	t	p
Intercept				1.204	0.723	−0.226		2.634	1.665	0.098
Gender		−0.048		−0.106	0.154	−0.411		0.198	−0.691	0.490
Age		0.041		0.003	0.006	−0.009		0.016	0.533	0.595
TMP		0.033		0.027	0.093	−0.156		0.210	0.290	0.773
NTMP		0.574		0.494	0.113	0.271		0.718	4.375	<0.001
Christian Belief		−0.044		−0.042	0.111	−0.261		0.177	−0.377	0.707
Christian Identity		0.174		0.221	0.159	−0.094		0.536	1.386	0.168
Cosmic Religion		0.172		0.276	0.129	0.020		0.532	2.135	0.035

Note. This table presents Multiple Regression Model 2b: Malta. Significant predictors (p < 0.05) of CEB are indicated using bold and underline.

and

Table 6. Multiple Regression Model 2c: Denmark.

Model	Model Summary					ANOVA
	R	R2	Adj. R2	Std.Err.of Est.		SS	df	MS	F	p
Denmark	0.511	0.261	0.221	1.12594	Regression	57.382	7	8.197	6.466	<0.001
					Residual	162.270	128	1.268
					Total	219.653	135
Variable				HC3 Estimator
		β		B	SE	CI (LL)		CI (UL)	t	p
Intercept				1.219	0.763	−0.290		2.728	1.599	0.112
Gender		0.084		0.00	0.294	−0.582		0.582	0.000	1.000
Age		0.127		0.010	0.007	−0.004		0.024	1.450	0.150
TMP		−0.107		−0.123	0.123	−0.366		0.119	−1.006	0.316
NTMP		0.448		0.431	0.125	0.184		0.678	3.457	<0.001
Christian Belief		−0.088		−0.085	0.109	−0.300		0.131	−0.780	0.437
Christian Identity		0.257		0.442	0.191	0.065		0.819	2.317	0.022
Cosmic Religion		0.084		0.110	0.131	−0.149		0.370	0.842	0.401

Note. This table presents Multiple Regression Model 2c: Denmark. Significant predictors (p < 0.05) of CEB are indicated using bold and underline.

present the findings of Set 2.

Set 2 (i.e., Models 2a, 2b, and 2c) tested H2 which predicted that, when disaggregating prayer intensity into TMP and NTMP, only NTMP predicts CEB (but not TMP). In Model 2a (Serbia), the predictors accounted for 14.7% of the variance (R² = 0.147; Adjusted R² = 0.108) in CEB, but no variable significantly predicted CEB, thus providing no support for H2. In Model 2b (Malta), the predictors accounted for 39.4% of the variance (R² = 0.394; Adjusted R² = 0.362) in CEB; and among the prayer intensity sub-indices, only NTMP (β = 0.574) positively predicted CEB, t(133) = 4.375, p < 0.001, thus fully supporting H2 (as TMP did not predict CEB). In this model, Cosmic Religion once again positively predicted CEB (β = 0.172), t(133) = 2.135, p < 0.05. Finally, in Model 1c (Denmark), the predictors accounted for 26.1% of the variance (R² = 0.261; Adjusted R² = 0.221) in CEB; and among the prayer intensity sub-indices, only NTMP (β = 0.448) positively predicted CEB, t(128) = 3.457, p < 0.001, thus fully supporting H2 (as TMP did not predict CEB). In this model, Christian identity once again positively predicted CEB (β = 0.257), t(128) = 2.317, p < 0.05.

5. Discussion

In this paper, we presented data comparing Serbia (mainly Orthodox), Malta (mainly Catholic), and Denmark (mainly Lutheran) on the predictors of civic engagement behaviors (CEB). We tested two hypotheses. H1 posited that Christian prayer intensity, Christian belief, and Christian identity (the Christian triad) all predict CEB, even when adjusting for covariates. H2 posited that, when disaggregating Christian prayer intensity into technologically mediated prayer intensity (TMP) and non-technologically mediated prayer intensity (NTMP), only NTMP predicts CEB (but not TMP). In our findings, H1 was only partly supported across all three sub-samples, because out of the Christian triad, only prayer intensity predicted CEB. Moreover, H2 was fully supported in Malta and Denmark, where NTMP predicted CEB but TMP did not; however, H2 was not supported in Serbia, as neither NTMP nor TMP predicted CEB.

These findings show that prayer is a trans-denominational predictor of civic engagement across different contexts (both religious and geographic). This means that the reflexive and ritualistic aspects of prayer do more to reinforce the importance of communal engagement (CEB) than any avowal of belief or identity does. When considering that the bivariate statistics had shown that all types of prayer (except using mobile apps in Denmark) actually predict CEB, this finding becomes more powerful and pressing to consider. Perhaps this is evidence for the power of transcendent accountability (Jang et al. 2023) mentioned above. Given the items tapping prayer intensity, these findings also show that both private and communal prayer (Storm 2015) are possibly relevant when it comes to their influence on civic behaviors.

Some contextual/denominational differences were noted as well, principally the significant effect of Cosmic Religion in Malta and that of Christian identity in Denmark. One interpretation of this finding is the following: in view of the remaining influence of Catholicism in Malta, the effect of both Christian (prayer intensity) and non-Christian (Cosmic Religion) variables on CEB might indicate a polarized public when it comes to religious matters and their expression in the community. In contrast, in view of Denmark’s highly secular publics, it may be more likely the case that minorities who hold Christian identity more strongly end up going that extra step to take part in CEB, precisely because of their minority status. We acknowledge that this is a speculative interpretation, and future research may indeed shed light on this matter and aspects related to the contents of H1.

Perhaps more interesting are the findings related to H2. These findings are novel in that they demonstrate that physical communal space matters. Whereas technique strives to make itself indispensable by transferring human engagement to non-physical platforms (Ellul [1954] 1964), this is not without its negative consequences. Based on our findings, although technological innovations do tend to function as geographies (see Straccamore et al. 2023), they are not enough to bring people together (e.g., in prayer) in ways that promote CEB. The “politics of space” (Kong 2001, p. 405) remain in need of, and conducive to, actual physical encounters, where possibly the “uneven access to techno-religious spaces” (Kong 2001, p. 404) may be felt less.

Our findings support the idea that religious expression/prayer and technological innovation may promote different societal projects (Buhagiar and Sammut 2020) when it comes to civic engagement. Studies linking religiosity with pro-sociality (e.g., Lewis et al. 2013; Putnam and Campbell 2012) and with civic engagement (e.g., Kim and Jang 2017; Pancer 2015) have tended to focus on non-technologically mediated forms of religious expression, and, in that sense, our findings on NTMP align with this literature. This is because prayer using different media (or mediation) per se seemed to not have any pro-social effects over and above prayer pursued via non-mediated/embodied means. The mere introduction of a technological interface may, therefore, act to dilute those aspects of prayer that promote civic engagement in other non-virtual contexts. Whether this is fully attributable to the technology itself (or rather to the absence of embodied community that accompanies it) is a matter for future research.

We also note that our extended view of religion (Kalmykova 2025) allowed us to find a balance between studying contexts with a majority denomination whilst being flexible when it came to sample characteristics (e.g., one may be Orthodox in Malta, and this did not exclude such participants from our sample). We understand this as a strength of our work because of its de-individualized focus: it was the different Christianity-inspired contexts that mattered the most for our correlational-comparative work. Our findings show that it is NTMP that promotes “service-oriented activities” (Tatala et al. 2024) across some denominational contexts, and that promotes and enables people to form meaningful social networks (see Storm 2015) for civic engagement purposes.

6. Conclusions

In conclusion, we note the limitations of our study. Firstly, the sample sizes were modest and were not obtained via random sampling. This may have biased our sample, despite the efforts at diversifying our initial snowball seed participants, and therefore our work is largely exploratory. Secondly, the correlational nature of our study does not allow for causal inferences to be made. Thirdly, we note that the prayer intensity subscales may benefit from more work. For example, “mass” was featured in NTMP but not in TMP, whereas “videos” or “social media” in TMP may have been interpreted by some participants as potentially also referring to online mass. This can create ambiguity, which no quantitative measure is fully immune to, but which nonetheless can be addressed in future research. This is chiefly because physical-versus-online mass and similar dichotomies are a major area of consideration for scholars and church bodies that try to offer (or that seek to oppose or defend) online worship. Fourthly, whilst our choice to go for two sets of multiple regression was justified on the basis of H1 and H2, having six models (especially with disaggregated indices in Models 2a, 2b and 2c) may have increased the risk of Type I error and may also have complicated the interpretation of the variables involved. Following discussion among the team, we felt that this limitation was offset by the benefits of testing H1 and H2 directly. However, future research may wish to conduct similar studies using larger samples and other analytical approaches, such as structural equation models.

In terms of future research, scholars may also wish to look into political engagement more directly to see whether the patterns observed for civic engagement also remain when the outcome variables rely on items that are more explicitly political. This helps answer the question concerning which specific political projects (Buhagiar and Sammut 2020) may be promoted by different kinds of prayer, if any. Similarly, idiographic work focusing specifically on qualitative reasons behind the prayer–engagement nexus would greatly enhance this field of inquiry.

In conclusion, we remind our readers of our “four humble ways”, explained above, which favor: (a) conducting context-sensitive studies, favoring an extended, cultural, and embodied understanding of religion (see Kalmykova 2025); (b) utilizing outcome variables (in surveys) that tap behavioral intentions/actions, thus focusing on the tangible civic outcomes that individuals or communities work toward (see Buhagiar and Sammut 2020); (c) using predictor variables that conceptually cohere and reflect people’s actual lives, hence enhancing the lived validity of the study (see Hartley 1998; Akkerman et al. 2021); and (d) looking at intra-religious (e.g., intra-Christian) differences, which allows researchers to distinguish between patterns that are inter-denominational and others that are specific to some denominations, by retaining similar worldviews (Koltko-Rivera 2004) in the sample, hence limiting issues relating to the translation of ideas from one religion to another. Based on the above findings, it seems that the four humble ways yielded a highly insightful study—of this, the reader can be the judge.