The unidimensionality of the questionnaire was evaluated through the use of Rasch Principal Component Analysis (PCA) to scrutinize the overall scale as well as the distinct subscales. Unidimensionality refers to the extent to which items in the instrument measure a single construct ( DiStefano & Jiang, 2020; Khine, 2020; Ningsih et al., 2021). Moreover, an examination of the raw variances revealed how each subscale aligned with unidimensionality criteria: the Global scale demonstrated 48.0%, the Key Opinion Leader (KOL) subscale 54.2%, Word of Mouth (WoM) 44.5%, Funding Philanthropic Brand Trust (FPBT) 60.8%, Philanthropic Intention (PI) 52.8%, Philanthropic Brand Preference (PBP) 57.2%, and Pride of Affiliation (PoA) 56.4%. The PCA analysis determined that the global scale and FPBT calculated eigenvalues of 4.6 and 2.7 logits respectively, surpassing the 2.0 logits threshold. These results suggest that the global scale could comprise multiple dimensions that extend beyond a single construct. Conversely, eigenvalues of 2.0 logits or less for the subscales suggested a reflection of unidimensionality within the subscale items.

The reliability analysis indicated that the global scale’s item separation reliability was recorded at 0.77 logit, which is below the accepted standard of 0.80. Similarly, KOL and WoM scales yielded reliabilities of 0.77 logit and 0.79 logit, respectively. These findings indicate that the reliablity of both scales shown at minimum levels. On the other hand, the FPBT scale displayed a reliability of 0.80 logit, which is considered acceptable, while the PI scale showed a reliability of 0.85 logit, classifying it as good. In contrast, the PBP and PoA scales scored lower with reliabilities of 0.57 logit and 0.35 logit, respectively, indicating that both were not reliable. Furthermore, the analysis revealed that the global scale had an item separation index of 1.85 logit, while the KOL and WoM scales were close, with indices of 1.81 logit and 1.92 logit, respectively. The FPBT scale had a separation index of 1.99 logit, and the PI scale outperformed with 2.35 logit. Although these indices suggest a differentiation into three levels of item difficulty for these scales, the PBP scale’s index at 1.15 logit implies the possibility of only two difficulty levels, and the PoA scale’s index of 0.73 logit indicates a single level of difficulty. Overall, the separation indices of the global scale and the various subscales did not surpass the 3 logit threshold, suggesting shortcomings in their ability to differentiate item difficulty effectively. These results highlight the need for further review and potential revision of these scales to enhance their reliability and discriminative capacity.

Wright Maps are utilized to assess the distribution of individuals’ proficiency levels and the difficulty levels of items on a common scale. Figure 1 illustrates the distribution of agreement levels for items and persons. The left side of the map illustrates the distribution of persons’ measured proficiency levels, while the right side shows the distribution of item difficulty levels, organized from the most agreed item to the most disagreed item. As suggested by Rusland et al., (2020), logit data falling within the range from the Mean (M) to one standard deviation below the mean (S) are considered to represent a moderate level of agreement. Data points above M indicate a higher level of agreement, whereas data below S are categorized as reflecting a lower level of agreement.

Figure 1 illustrates that participants struggled with certain questions, such as Q17, Q31, Q42, and Q44, suggesting a reluctance to discuss social issues and online donations. This difficulty may point to a preference for adhering to cultural norms and valuing recognition and support from their community in donation decisions. However, the study found broad agreement on questions related to KoL, especially Q2 and Q4. Participants viewed KoL as compelling figures in the philanthropic sector and trusted their influence on donation behaviors. The survey also revealed strong affirmative responses to questions about the effectiveness of WoM, including the impact of personal discussions and stories about donations (Q12 and Q13) and the utility of donation information for social issues (Q9). Furthermore, agreement was evident on items tied to FPBT, demonstrating a belief in technology’s ability to facilitate donations (Q22) and trust in charities working for the public interest (Q18). Responses to PI items showed a strong endorsement of donating as an ethical obligation (Q28), a focus on the ethical credibility of organizations when giving (Q24), and alignment of donations with personal moral values (Q25). Such agreement highlights the importance of ethics in driving donation decisions. Finally, participants agreed on items associated with PBP and PoA, displaying a preference for well-regarded charities (Q33) and pride in affiliating with their chosen organizations (Q38). These attitudes reflect the importance of charity credibility and the personal satisfaction that motivates philanthropic actions for the good of society.

The analysis of Differential Item Functioning (DIF) using Rasch-Welch tests was conducted to assess the potential presence of item biases attributed to demographic variables ( Huang et al., 2020; Lee et al., 2020; Ningsih et al., 2021; Zulaiha & Mulyono, 2020). DIF was determined when the DIF contrast value, representing the difference in item difficulty, exceeded 0.5 logits and was statistically significant (Rasch-Welch probability value < 0.05) ( Chan & Subramaniam, 2020; Kreijns et al., 2020; Linacre, 2018; Mulyono et al., 2020; Suryoputro et al., 2023) Evaluation of DIF among participants’ demographic characteristics revealed that several items exhibited statistically significant DIF concerning participants’ age, gender, and educational background (see Table 1).

The DIF analysis revealed distinct age-based disparities; younger participants (aged 11 to 26) more readily acknowledged a “moral obligation to donate money” (Q26) and cultural influences on donations (Q31) compared to older respondents (27 to 42). Conversely, the latter group aligned more with statements involving tangible solutions in donation recommendations (Q11) and the need for communicative services by charities (Q17). Gender disparities emerged, with females showing higher agreement with cultural influences on donations (Q31) than males. Regarding the education background, high school graduates found cultural influences easier to agree with than master’s degree holders (Q31), while bachelor’s holders aligned more with the effectiveness of KOL items (Q1 and Q2). Master’s degree holders found it easier to support communicative charity services (Q17) and discuss support for charity campaigns (Q34) compared to bachelor’s degree holders.

A research evaluation panel from Doctor of Research in Management (DRM) Program BINUS Business School, Research ethics and publication center approved the current preliminary study on August 25th, 2023. The ethical approval was obtained from DRM Research Ethics committee with letter no 004/HoP.DRM/I/2024. The participants of the current study were selected on voluntary basis and written informed consent was collected prior to the study. The participants were anonymized and their identity had been removed from the dataset.