Conservatism is an uncompromising understanding of globalisation and modernisation (Kemmelmeier 2015; Saroglou 2015). It rejects diversity in ideas, interpretations, beliefs and perceptions of everything. Differences are considered taboo, even prohibited, because establishment must be interpreted literally and continuously occasionally. In addition, conservatism reflects past cultural justifications that must be maintained (Marty 2015). This suggests that conservatism tends to be an ideology over the status quo rather than a scientific ideology. This kind of attitude leads to intolerance towards pluralism and considers pluralism as a real threat to past cultures that must be held (Merrotsy 2016). Conservatism upholds conventional values, and there is no room for differences. Similarities can sometimes be distinguished as being exclusive. Similarities are considered a good thing, but differences are considered deviations from traditions that already exist and are inherited continuously from the past (Clobert 2021). Positive things can be well maintained if they receive support from the ideology underlying their thinking. In addition, negative things are seen as products of deviations from traditions and habits passed down from generation to generation (Hall et al. 2022). In this context, conservatism does not allow room for change. This attitude will stagnate people’s lives, religion, economy, culture and politics.

In the field of religion, conservatism will reject anything different from mainstream religious traditions (Zuhdi 2018). Nothing needs to be changed in both faith and religion. Everything must be understood continuously and passed down continuously from time to time. Special conditions and situational contexts are unacceptable in terms of religious conservatism. Religious conservatism makes a person more aggressive and can trigger internal and external disharmony. Even religious experts can be misguided when giving a new opinion contradicting the mainstream view. This certainly causes someone with different thoughts to be more likely to choose silence when they have a different understanding than mainstream thinking. Plus, conservatism is religious and tends to interpret religious teachings based on the literal meaning of scriptures and religious traditions (Kaňuková 2019). There is no room for new interpretations, even those based on scientific logic or philosophy. They speak out militantly and stand up for what they believe is right. This certainly causes religion to be the best locus for the growth of identity politics, which is not very beneficial for multicultural realities and not very good for people with high diversity.

Ideology refers to the norms and values that bind social identity (De Witte 2004; Dubois 2001; Thompson 2001). Ideology also refers to shared or common values considered reasonable and generally accepted in society. Currently, ideology is represented in various domains, for example, religion, understanding, flow, norms, principles and even doctrine (Birdir et al. 2022; DePaula 2023; Hossain, Rjiba & Saadi 2022; Sun, Zhu & Guo 2022; Wu et al. 2022). Despite thereby, ideology until now has not been defined exclusively and objectively (Berkes 2017; Devine 2015; Olwig & Mels 2020). However, some definitions can help us understand the essence of ideology. Some scholars, for example, define ideology as critical thinking (Stern 2022). In another sense, some people think ideology is identity thinking about social life (Kobayashi 2020; Sinha 2022). Reese, Rosenmann and Cameron (2019) identify ideology as the essence of reality in social life. This diversity of definitions can be seen philosophically as a representation that cannot be separated from one’s awareness of seeking and formulating ‘common values’ related to social, political, and even religious life. The pertinent question therefore is: What is the relationship between ideology and conservatism?

The relationship between ideology and conservatism can be traced from two groups. Firstly, ideology is the foundation of conservatism, one of the political ideologies within a wider spectrum of ideologies. Ideologies such as liberalism, socialism and other conservatism also provide a foundation for conservatism’s beliefs and values (Bourke 2018). Secondly, the principles of conservatism, In this context, conservatism has fundamental principles that form the basis of belief in various things, such as politics and religion. This principle includes effort and belief to establish a tradition and anti-change (Bourke 2018; Collins 2013). These two classifications emphasise that the relationship between ideology and conservatism can be traced from the foundations, principles and values believed to be true and then used as collective ideals. In the context of religion, doctrines and values that arise as interpretations of doctrines become a standard that does not accept change and then become sacred as a collective standard, which must be established and maintained.

Rasch is a statistical analysis approach based on the paradigm of internal rationality and objectivity concerning reflective measurement norms (Andrich 2005, 2010). This is characterised by the principle of measuring latent variables that are required to have similarities with physical measurements using Log-Odds Unit (Logits). Its main characteristic is the value of unidimensionality related to reflective norms, which measures the items offered on a single scale. Furthermore, this approach requires the statistical correspondence of items to people and people to items (Bond & Fox 2013). In addition, this study used Conditional Maximum Likelihood Estimation (CMLE) and Joint Maximum Likelihood Estimation (JMLE). The CMLE estimates probability parameters based on logit items and people, and JMLE estimates patterns based on items and people. Rasch approximation should be used as the parameter barrier zone for each scale category. This refers to Andrich Thresholds logit and logit category areas. It measures both regular and conventional logits. Some differences in types, mean intervals and median gaps can be understood as probabilities in measuring ideological conservatism (Bond & Fox 2013, 2015). As a result, this study highlights raw scores as ordinal data and converts them into interval data.

In this research methodology, an online survey approach was used to collect data from 572 Islamic students at three universities in Sumatra, Indonesia. The survey was designed to investigate various aspects related to religious conservatism and their worldview. Firstly, survey preparation involved creating clear and relevant questions, focussing on developing the scale of religious conservatism. An online survey platform was chosen in consideration of its ability to handle a significant number of respondents. A respondent database that included contact information and respondent characteristics was carefully prepared. Thereafter, the online survey was distributed through invitations to respondents with an explanation of the purpose of the research and the importance of their participation (Harlow 2010). The survey results were comprehensively analysed to support the findings of this study in understanding the dynamics of the relationship between these factors in the context of Islamic students in Sumatra. In data context, this study applied the data filtering and cleaning stage by excluding 23 unfit data for further analysis for the following reasons: too many blank answers, the answers were the same between one item and another, and the answers were identical to other respondents. In total, 549 students were included in the final analysis. Samples were selected according to the category and sex of the students. It consisted of 174 men and 358 women, 98 freshmen, 56 sophomores, 73 juniors and 27 seniors (Table 1).

This study used the Rasch analysis approach to check validity and reliability. According to Rasch rules, measurement attributes will be changed in Logits values (Andrich 2005; Engelhard & Wind 2019; Van Der Linden 2010). The resulting scale was unidimensional and statistically fit. In general, there are four steps in this research. Firstly, generated items of the conservatism scale refer to three previous studies to produce draft items (Groff 2022; Huber & Huber 2012; Luckmann 2022). The results obtained at this stage included 29 unidimensional items; secondly, they were validated internally, both linguistically by experts and by some target populations. This stage allows for identifying feasible items to be proposed after eliminating and revising several items considered invalid. Thirdly, randomly selected 549 students at three universities were analysed psychometrically using the Rasch Model approach. Then the data obtained were analysed in terms of performance based on the item-respondent attributes and vice versa using the Rasch model approach (Zufriani, Pitriani & Damni 2022). In all, 15 items qualified following the Rasch model approach after excluding unnecessary items that overlapped with the logit values of other items. Fourthly, the participants were measured broadly, comparing by gender (male and female) and student category (first year, sophomore year, junior and senior). The data visualised the probability of conservatism on the Wright map. The classification of genders and students will enrich the research results. Rasch Analysis uses Winstep software version 5.3.4 (by Winstep Software Technologies) and JASP 0.16.4.0 (a statistical and graphical analysis software that is open source and free).

Items were arranged based on three previous studies (Adira et al. 2021; Huber & Huber 2012; Wildman et al. 2021). This unidimensional scale framework produced 29 Indonesian items after the literature review stage, as shown in Table 2.

Table 2 is the product of applying the scale formulation model offered by Sholihin, Shalihin & Addiarrahman (2023) related to Islamic beliefs. In this regard, the procedure for constructing the scale involved a systematic approach, beginning with the careful selection of relevant statements reflecting various aspects of Islamic beliefs. These statements were subjected to preliminary testing among a small sample to identify and address any potential ambiguities. Subsequently, the chosen statements were organised into a logical sequence, and a scoring system utilising a Likert scale was established. The scale underwent pilot testing on a larger sample to assess its reliability and validity, with statistical measures such as Cronbach’s alpha employed for internal consistency evaluation. Item analysis helped refine the scale by identifying and addressing any redundant or weakly correlated items. Expert input was sought to ensure clarity and appropriateness. The finalised scale consists of statements addressing diverse dimensions of religious beliefs within Islam, providing a comprehensive tool for measuring attitudes and views on various aspects of the faith. Regular reviews and updates will, however, be essential to maintain the scale’s relevance and reliability over time.

Then, the scale is validated holistically, both in content and linguistics and empirically tested on the population. Psychometrically, a scale is declared valid when it is qualitatively declared valid by the expert and satisfied with the items offered. The scale can then be expanded in the next step. Content experts suggested eliminating six wrong items: items 2, 4, 11, 15, 21 and 26. Linguists suggest improvement and that these items should be replaced with new items because of interpretation problems that may occur on account of the grammar used for the five items, namely, items 1, 3, 8, 14 and 19. Meanwhile, from the target population, seven items must be adjusted so that they are more accessible to identify students, namely, items 1, 3, 12, 13, 16, 17 and 25. The remaining scale consists of 23 items distributed to students in a wider scope of educational institutions in Indonesia.

In this scale, the items were constructed based on inputs from 549 respondents. In this context, there were 23 input data and 23 estimation data. Generally, items and people on a conservative level perfectly fit the criteria the Rasch model requires. The value of person indicates (Pn) and Item (Im) of Outfit (Out) – Infit (In); Mean square (MNSQ) is close to 1, Pn-In MNSQ = 1.03; Pn-Out MNSQ = 0.99; Im-In MNSQ = 0.99 dan Im-Out MNSQ = 0.99. The people and items outfit-fit Z-standard (ZSTD) were close to 0, Pn-In ZSTD = –0.1, Pn-Out ZSTD = –0.1, Im-In ZSTD = –0.3 and Im-Out ZSTD = –0.1. The scale has a high-reliability value, namely 0.77 for person reliability and 0.99 for item reliability. The mean and standard deviation (SD) based on standardised residuals are 0.01 and 1.00, respectively; this indicates that the data are normally distributed (Andrich 2005, 2010; Bond & Fox 2013). The details of the statistical description of the scales used are shown in Table 3.

In particular, its item characteristic is smooth in measuring conservatism. All items demonstrated linearity in the measurements, as shown in Table 4. The point measure correlation (PMC) values are all positive and in the range of 0.4–0.85 (Junker 2015). Except for items 16 = 0, 36, 17 = 0, 25 and 20 = 0.32, the suitability of the items is also extraordinary, referring to the value of the outfit and infit mean square. The most appropriate thing to agree on is conservative is item 8 (−0.96 logits), and the easiest is item 23 (0.86 logits).

The initial number of remaining items was 23. Point Measure Correlation eliminated three items outside the 0.4–0.85 range (Items 20, 22 and 25) and one item by Outfit MNSQ outside the 0.5–1.5 range of (item 29). Four items overlapped with the logit values of other items (items 6, 9, 16 and 18). The final 15 items were eligible, based on the Rasch model approach. Each final logit item on the conservatism scale are shown in Table 5. The most challenging thing to agree on is conservative is item 8 (–1.02 logits), and the easiest is item 23 (0.99 logits).

The choice of items (strongly agree-agree between agree and disagree-disagree-strongly disagree) significantly affected the overall scale. This is indicated by the Andrich Threshold value, which has shown a concordance between non-negative-positives (Andersen 2005). In addition, Outfit and Infit MNSW show fit statistics at 0.5–1.5. The details regarding the choice of each item are presented in Table 6.

The expected values were well distributed following the Rasch approach. The categories of observations with operational distances for each item corresponded to the distribution of respondents, and the hierarchy of items on the measurement scale demonstrated this. This corresponds to Figure 1.

Items and respondents explained the variance in each measurement item according to the Rasch approach. The natural variance explained by size was 100%, as expected in the 100% model. The empirical variance explains 40.5%, which is not much different from what is expected: 41.4%, people 14.3% and expected 14.6%, and items 26.2% and expected 26.8%. The item’s natural variance values showing this and the person were 3.6 and 6.6 eigenvalues. The unexplained variance value also reinforces this at the first level, which is not greater than 2 (Larsen & Warne 2010). Further details of the variance are shown in Figure 2.

The conservative level can be classified as moderately low, as shown in Figure 2. The mean person value is outside the mean item. In addition, the mean person score was lower than two SDs of the log items. This shows that the measurement respondents reject actions and perceptions referring to religious conservatism. This low conservative value can be divided into two levels (1.91), referring to the separation value. A detailed map of this religious conservatism variable is presented in Figure 3.

The students’ average logit was 0.70. The civil servant category had the highest logit (0.96), followed by female (0.75 logit) and male (0.55 logit). The male’s 2.57 logit is less than the female 2.70 logit, as shown in Table 7. This evidence is strengthened by the median value (g = 0.71, m = 0.71, f = 0.71, and prefer not to say [PNS] = 0.93), reliability of the pattern (g = 0.78, m = 0.81, f = 0.77 and PNS = 0.48), standard error (g = 0.03, m = 0.05, f = 0.03 and PNS = 0.10) and SD (g = 0.60, m = 0.61, f = 0.59 and civil servant = 0.41). Commonly referred to as the separation value (g = 1.91); the level of conservatism can be divided into two levels. The gender category can be divided into six levels: men have three levels, women have two and civil servants have one level (m = 2.03, f = 1.83 and civil servant = 0.93). Specifically, this can be observed in Table 7.

Significant differences existed between gender groups (m, f and civil servants). The probability value with analysis of variance (ANOVA) and Welch is < 0.05 (< 0.001). Each category compared to one other category also shows the significance of the difference: PNS to m (P_tukey = 0.018) and m when f (P_tukey = < 0.001) except civil servant the f (P_tukey = 0.329). The details are listed in Table 8.

First-year students had a –0.81 logit level of religious conservatism with a mean standard error of 0.18. This is 0.04 logit, which is higher than the average value. Sophomores had 2.37 logs, which is 0.29 logs lower than the median, with a standard error of 0.21. Juniors have 2.64 logs, 0.02 logs lower than average with a standard error of 0.13, and seniors have 3.18 logits, 0.52 logits higher than average with a standard error of 0.26. Subsequently, several groups were formed from the split values. 2.13 shows three groups for first years, 1.99 shows two groups for sophomores, 1.25 shows one group for Juniors and 1.13 shows two groups for Seniors. Details can be seen in Table 9.

There was a significant difference between the student categories (1^st, 2^nd, 3^rd and 4^th). The probability value with ANOVA was < 0.05 (< 0.001). Several categories compared to one other category also showed significant difference: 1^st to 3^rd (P_Tukey = < 0.001), 1^st until 4^th (P_Tukey = 0.017), 2^nd until 3^rd (P_Tukey = < 0.001), 2^nd up to 4^th (P_Tukey = 0.007), except 1^st until 2^nd (P_Tukey = 0.938) and 3^rd until 4^th (P_Tukey = 0.993). Specifically, this can be observed in Table 10.

The theoretical implications derived from this study’s analysis of respondents’ responses to items related to religious conservatism are profound and multi-faceted. Firstly, the high agreement among respondents on items related to the Quran signifies the centrality of the sacred text in Islamic beliefs, aligning with the foundational dimensions of the Islamic faith. This finding resonates with previous research that demonstrates the influential role of scriptures in Islam, Judaism and Christianity. Secondly, the difficulty in reaching consensus on items related to Muslim groups reflects a potential link to Muslim prejudice against non-Muslims and individuals of different faiths. This aligns with existing literature highlighting societal challenges related to religious diversity. Thirdly, the significant gender-based differences in responses suggest that women tend to align less with religious conservatism compared to men. This aligns with previous research indicating women’s increasing willingness to express themselves with a more liberal ideology. Sociologically, these response patterns can be explained by the Quran’s multifaceted role as a moral guide, a unifying identity for Muslims and a determinant of mindset and daily routines. The study further identifies variations in responses among Muslim students, particularly men showing a more supportive attitude towards religious conservatism. This aligns with the broader context of Indonesia facing the rise of religious conservatism, exemplified by movements like hijrah, emphasising a rigid interpretation of the Quran and displaying exclusivity against religious plurality. Overall, these findings contribute to a nuanced understanding of the complex interplay between religious beliefs, gender dynamics and societal attitudes in the context of Islamic conservatism.