In early transparency research, no consensus was reached on the conceptualization of the construct, which is why inconsistent operationalizations were used. The understanding of transparency was usually limited to the amount of shared information and was, thus, predominantly measured by unidimensional scales referring mainly to the quantity of shared information. Nevertheless, the scales differed markedly in terms of the number of items, the dimensionality (if any), and the focus or specificity of the scales (Schnackenberg & Tomlinson, 2016). The use of these non-standardized scales can be seen as a reason for the previously unclear empirical results regarding the test of assumed relationships between transparency and theoretically well-established outcomes such as trust and job satisfaction. To address this issue, Schnackenberg and Tomlinson (2016) elaborated a definition and different dimensions of the construct based on an extensive literature review. Following this widely adopted definition, transparency “is the perceived quality of intentionally shared information from a sender” (Schnackenberg & Tomlinson, 2016, p.1788). The constructs and the corresponding scale encompass the dimensions of disclosure, clarity and accuracy (Schnackenberg et al., 2020) as well as the facets of timeliness and relevance of shared information (Hossiep et al., 2021). The transparency scale presented here reflects, besides disclosure, the dimensions of clarity, accuracy, timeliness and relevance of shared information (Hossiep et al., 2021; Schnackenberg et al., 2020). Disclosure refers to the perception that sufficient information is shared by a sender (Schnackenberg & Tomlinson, 2016). Clarity indicates whether the shared information is perceived as comprehensible and can be understood as a measure for congruence of the intended and understood meaning of shared information (Schnackenberg & Tomlinson, 2016). Accuracy refers to the degree to which the shared information is perceived to be correct and not biased at the moment of information sharing (Schnackenberg & Tomlinson, 2016). Timeliness is about the right moment of information sharing; too early and too late information sharing harms the perceived value of received information. For example, very accurate information which is shared too late might be useless for the information receiver. Finally, relevance is the degree of personal importance (professionally and/or privately) of disclosed information (Hossiep et al., 2021). The five dimensions can be perceived differently by different information receivers. For example, employees may be more or less familiar with a technical language and therefore perceptions of clarity of shared information might differ between employees. Thus, transparency is about the degree of perceived transparency rather than its pure existence (Albu & Flyverbom, 2019). The multidimensional view of transparency enables researchers to gain a deeper understanding of the effects of transparency on the one hand, and on the other hand it offers practitioners the possibility to derive well-founded recommendations for action in order to strengthen the perception of transparency, for instance within an organization. For example, disclosure, timeliness, and relevance were found to have a positive impact on job satisfaction, while the other dimensions showed no significant effect.

Item generation and selection

The items of the dimensions of disclosure, clarity and accuracy were developed by Schnackenberg et al. (2020) and were translated by Hossiep et al. (2021) into German. We chose a team translation approach (Harkness, 2003). The translation was conducted by three persons independently of each other and was discussed afterwards. The mother tongue of those persons was German while they were fluent in English. Additionally, two of them were experienced within the organizational research field. Most of the item translations did not differ substantially, especially content wise there were no significant differences within the item translations. Therefore, the joint discussion focused on the consistent use of tenses and terms. The final selection of the item translation was based on the understandability of the items and the linguistic fit of the items to each other. The selection was done in consensus by all involved persons. Finally, the selected items were presented to three uninvolved persons to check the comprehensibility of the questionnaire. No changes were made here. In addition, the German translation was compared with the English original by a bilingual person. Again, no changes were necessary. In addition, items were developed for the two new dimensions timeliness and relevance. The items were derived from the corresponding definition of the dimensions. Special attention was paid to the wording of the items so that they cover all relevant aspects of the dimensions and at the same time match the existing items in terms of phrasing, length, and polarity. The items were also checked to ensure that they could be understood by uninvolved persons.

Samples

The expanded and translated DCA-Transparency scale was tested using six samples. The data collection of five samples took place in 2019 and one additional sample was collected at the beginning of 2021. All participants spoke German at a native level and had a non-self-employed occupation. This is important, because participants needed to fully understand the items and were required to have relevant professional experience to assess their supervisor transparency. The first (N = 325) and second sample (N = 322) served to test the factorial structure of the translated and extended items. Participants were recruited through personal contacts and participation was not compensated. The third (N = 133) and fourth sample (N = 120) were used to quantitatively test distinctiveness to orbiting scales and distinctiveness between transparency dimensions for content validity. Both samples were recruited via the online crowd-sourcing platform Prolific and participants were selected randomly. There was a financial reward for participating in the survey. A longitudinal research design was used for the fifth sample (N = 376). For this, participants were interviewed in three waves. 540 people took part in the first wave, 447 in the second and 376 in the final wave. The sample was representative of German employees in terms of age, gender and gender distribution by scope of employment. The data collection of this sample was supported by the Leibniz Institute of Psychology in Trier, Germany. Participants were recruited by a suitable panel provider and participants were financially rewarded. To ensure representativeness, we used quotas, and participants were rejected if the quotas were already met. Finally, sample six (N = 114) was used to compare transparency dimensions to existing, related scales. Again, participants were recruited on the online platform Prolific and were financially compensated. Participants were selected in an ad-hoc fashion. In samples 1 and 2 missing values were possible and such cases were excluded list-wise. In all other samples respondents were not able to skip items. To ensure data quality, we used attention checks within all surveys and used further indicators to ensure data quality such as processing time and answer pattern. The characteristics of the samples are shown in Table 2.

Table 2

Characteristics of Samples

Note. * Sample 5 represents a longitudinal research design consisting of three measurement points with t₀ = 540, t₁ = 447, and t₂ = 376 participants.

Item analyses

To test whether the supplemented items for timeliness and relevance also form separate factors and whether the translated items have the same structure as in the English original version, an exploratory factor analysis (i.e., maximum likelihood) was performed using Sample 1. We decided to use a maximum-likelihood approach since in contrast to other methods it is best suited to account for the common variance of factors (Worthington & Whittaker, 2006). In addition, an oblique factor rotation method (oblimin) was conducted.

The statistical software IBM SPSS 25 was used. Since one of the accuracy items did not load on the accuracy factor (i.e., Accuracy 3), the survey was repeated with an alternative translation that had already been discussed in the translation process (Sample 2). For factor extraction, parallel analysis (Horn, 1965) was used and indicated five factors. The five factors explain 77% of variance. Except for the corresponding accuracy item, no substantial differences between the samples emerged. The new translation of the accuracy item was then used in all other samples (Samples 2-6). Table 3 shows the pattern matrix. In addition to the exploratory factor analysis, a confirmatory factor analysis was performed using the R-based software JASP version 13 and the “auto” estimator in JASP which provides similar results as the maximum likelihood estimator. The confirmatory factor analysis was based on a different sample (Sample 5). The model fit indicators confirm the suitability of five transparency dimensions: χ²/df = 2.79, CFI = .971, TLI = .967, SRMR = .041, RMSEA = .058. The factor loadings of the confirmatory factor analysis are presented in Table 4.

Table 3

Pattern Matrix of Explorative Factor Analysis

Note. Factor order taken from first sample; factor loadings in brackets are from second sample, factor loadings smaller than .2 are not shown.

Table 4

Unstandardized and Standardized Factor Loadings of Confirmatory Factor Analysis

Note. First wave of sample 5 was used to perform the CFA.

Finally, the correlations of the manifest transparency dimensions based on Sample 5 are shown in Table 5. Factor correlations are generally large and range from r = .521 to r = 758. The relevance factor correlates to a lesser extent than the other factors to each other with factors correlations of r = .521 to r = .597.

Table 5

Descriptives and Correlations of Transparency Dimensions

Note. *** = p < .001; Cronbach’s Alpha on the diagonal. The first wave of Sample 5 (N = 540) was used.

Item parameters

Table 6 indicates items’ mean value, standard deviation, skewness and kurtosis. The items were measured on 5-point Likert scales.

Table 6

Means, Standard Deviations, Skewness and Kurtosis of Items

Note. Scale ranging from 1 (strongly disagree) to 5 (strongly agree), N = 376 (Sample 5).

Objectivity

Implementation objectivity is provided using standardized instructions and common 5-point Likert scales. Since, depending on the context of the survey, the evaluation of the items should be classified as confidential, the participants must be assured of anonymity. This means that no information may be collected that would allow direct conclusions to be drawn about the individual person. This is particularly important if, for example, the information behaviour of one's own superior is to be examined. If anonymity is not given, socially desirable responses may occur in order to prevent potential negative consequences for oneself (Krumpal, 2013). This could cause a distortion of the results. Thus, it is necessary to mention within the instructions that there are no correct and wrong answers and participants should rely on their own perceptions. Objectivity of interpretation is provided by the quantitative response options. For the evaluation, mean values of the individual dimensions and the overall transparency are formed. The perceived transparency of an information sender should be interpreted at the level of the individual dimensions, as this allows to identify deviations from expectations. Regression analyses, structural equation models or mean comparisons are conceivable for more complex analyses. Here, the use of suitable statistical software is recommended. Depending on the field of application and the intended conclusion, enough participants must take part in the survey.

Reliability

In Table 7, McDonald’s Omega of the five transparency dimensions is presented and compared between three different samples. McDonald’s Omega indicates very high to high reliability (conventional cut-off criteria >.7) of each subscale and is comparable between the different samples used (Hayes & Coutts, 2020). This shows that the items of each dimension form consistent dimensions. 

Table 7

McDonald’s Omega of Organizational Transparency Dimensions

Validity

The focus in validating the translated and expanded scale was to replicate and extend the findings from the original study. Especially the quantitative assessment of content validity is often neglected in scale validation (Colquitt et al., 2019). Thus, it was intended to add to the insights of Schnackenberg et al. (2020). To do so, an approach developed by Hinkin and Tracey (1999) and evolved by Colquitt and colleagues (2019) was chosen. Here, individuals who were not involved in the process of scale development (i.e., naive judges) evaluate the fit of the items of the focal scale (transparency) and of orbiting sales to the definition of all constructs on 7-point Likert Scales. Based on these ratings, mean values of the focal items are calculated and statistical mean value comparisons are used to examine whether the items have the significantly highest mean value on their respective definition. If this is the case, this can be interpreted as an indication of content validity (Hinkin & Tracey, 1999; Colquitt et al., 2019). Colquitt and colleagues (2019) further developed indices to qualitatively interpret the degree of definitional distinctiveness. First the definitional distinctiveness of the transparency to orbiting scales was investigated (Sample 3). We chose “informational justice” (Colquitt, 2001; Maier et al., 2017) and the “communication quality between employees and managers” (Mohr et al., 2014) since both constructs are on a similar logical stage as the focal construct and measure comparable perceptions. Following Colquitt et al. (2019) the distinctiveness of organizational transparency to those constructs can be interpret as moderate. In addition, the distinctiveness of transparency dimensions to each other was reviewed (Sample 4). This is especially important, since we added two dimensions, which might have been covered in the items of the existing ones. Again, we followed the recommendations of Colquitt and colleagues (2019) and found a very strong definitional distinctiveness.

To further strengthen content validity of the presented transparency scale, it is investigated whether transparency is relatively highly correlated with theoretically related constructs and not correlated or low correlated with assumed unrelated constructs. As related constructs we chose a scale of informational justice (Maier et al., 2007), a measure of communication quality (Mohr et al., 2014) and the questionnaire to assess communication in organizations (so called KomminO; Sperka, 1997). Those scales are dealing with perceptions about information sharing within organizations and the way of communication between supervisors and employees. Thus, they are measuring a related perception as the extended DCA-transparency scale and should strongly correlate with the focal construct. In contrast, we chose a scale for well-being (Brähler et al., 2007) and an instrument to assess need-for-cognition (Beißert et al., 2014) as unrelated constructs as those constructs describe a psychological state rather than an information-related perception. The correlations presented in Table 8 are further evidence of validity, as the correlation pattern of transparency with related and unrelated constructs is consistent with the predictions.

Table 8

Correlations to Related and Unrelated Constructs

Note. * p < .05, ** p < .01; Communication Quality scale of Mohr et al. (2014); Informational Justice scale of Maier et al. (2007); KomminO scale of Sperka (1997); Well-Being scale of Brähler et al. (2007); Need-for-Cognition scale of Beißert et al. (2014).

Criterion validity was further tested by investigating the relationship of transparency with expected outcomes in a longitudinal research design using the software R (Sample 5). To test the effects, multiple (stepwise) regression was used. The integrative model of organizational trust by Mayer et al. (1995) was used to assess the relationship of transparency dimensions to supervisor trustworthiness and employees’ trust in their supervisor. According to this model, individuals form perceptions regarding a person's ability, benevolence, and integrity. These perceptions form the trustworthiness of a trustee and are thus an important predictor for the formation of trust in the corresponding person (Mayer et al, 1995). Following Schnackenberg and Tomlinson (2016), the transparency dimensions in turn affect individuals’ perceptions of the information senders’ ability, benevolence, and integrity and thus influence individuals’ perception of trustworthiness and ultimately the formation of trust. Thus, trustworthiness can be understood as a mediator in the transparency-trust relationship. The results indicated that all transparency dimensions either directly or indirectly affected employees’ trust in supervisors. Thus, the findings provided further evidence for the distinctiveness of the transparency dimensions and support criterion validity. Besides trustworthiness and trust, employees’ job satisfaction was used as another outcome. Again, certain transparency dimensions (disclosure, timeliness, and relevance) were shown to be antecedents of employees’ job satisfaction. The results are shown in Table 9. For further details please see Hossiep et al. (2021).

Table 9

Examination of Criterion Validity (Main Effects)

Note. * p < .05, ** p < .01, *** p < .001; sample size N = 376 (Sample 5); ^a = the delta R² refers to the difference in explained variance between DCA-transparency and full transparency including timeliness and relevance as additional factors.

Descriptive statistics (scaling)

We included descriptive statistics (i.e., means, standard deviations, skewness, and kurtosis) for each subscale in Table 5.

Further literature

The underlying English scale consisting of the dimensions Disclosure, Clarity, and Accuracy was developed by Schnackenberg et al. (2020). The article on the translation and extension of this scale is by Hossiep et al. (2021).

Hossiep, C. R., Märtins, J., & Schewe, G. (2021). DCA-Transparency: Validation and Extension of a German Scale. German Journal of Work and Organizational Psychology, 65(3), 165-178. https://doi.org/10.1026/0932-4089/a000360

Schnackenberg, A. K., Tomlinson, E., & Coen, C. (2020). The dimensional structure of transparency: a construct validation of transparency as disclosure, clarity, and accuracy in organizations. Human Relations, 74(10), 1628-1660. https://doi.org/10.1177/0018726720933317

• C. Richard Hossiep, Center for Management, Chair of Business Administration – Organization, Human Resource Management and Innovation, University of Muenster, Muenster, Germany
  Universitätsstr. 14-16, 48143 Münster, E-Mail: Richard.Hossiep@wiwi.uni-muenster.de
• Julian Märtins, Center for Management, Chair of Business Administration – Organization, Human Resource Management and Innovation, University of Muenster, Muenster, Germany
  Universitätsstr. 14-16, 48143 Münster, E-Mail: Julian.Maertins@wiwi.uni-muenster.de
• Gerhard Schewe, Center for Management, Chair of Business Administration – Organization, Human Resource Management and Innovation, University of Muenster, Muenster, Germany
  Universitätsstr. 14-16, 48143 Münster, E-Mail: Gerhard.Schewe@wiwi.uni-muenster.de

The data sets of Sample 5 are available at PsychArchives of Leibniz Institute of Psychology (https://doi.org/10.1026/0932-4089/a000360)