Understanding the contribution of effective communication strategies to program performance in humanitarian organizations Open Access

Faced with increased humanitarian and development needs, humanitarian organizations must improve performance while ensuring appropriate program accountability. However, the large number and diversity of actors, system complexities and high levels of uncertainty make it difficult for humanitarian organizations to deliver efficient programs (Dolinskaya et al., 2011; Ergun et al., 2010; George, 2003). Recent reforms in humanitarian and development work have focused on increasing coordination and communication as mechanisms to improve performance (Ebrahim, 2005; Unerman and O’Dwyer, 2006). While performance is directly influenced by objective factors (e.g. organizational capacity and the skills, background and experience of the individuals involved), it is also influenced by subjective factors (e.g. the quality of the communication between stakeholders (Jakki and Nelvis, 1990; Goldhaber, 1990). For example, an efficient exchange of information between stakeholders may reduce uncertainty in the decision-making process (Schweitzer et al., 2002). The ability of humanitarian organizations to communicate, coordinate and manage stakeholders’ concerns properly are critical to improving the effectiveness and efficiency of humanitarian programs (Apte, 2009; Barnard, 1938; Beamon and Balcik, 2008; Ramadan and Borgonovi, 2015; Maiers et al., 2005). Hence, improving communication strategies not only within organizations but also with other stakeholders may lead to programs that are more efficient.

As Heaslip (2012) argues, “a major barrier to delivery of aid is poor communication,” not only in terms of language but also in the way how people interact with each other and with the changing environment. Therefore, developing standard rules of communication seems to be important in the context of trust-building and information-sharing in the context of building disaster resilience (Tatham and Kovács, 2010; Papadopoulos et al., 2016).

Typically, organizations communicate with stakeholders using different communication strategies: one-way communication (OC) and two-way (interactional and transactional) communication (Beebe and Masterson, 1996; Barnlund, 2008; Vlăduţescu, 2013). While previous studies have investigated how these specific strategies affect operations, none has systematically hypothesized how these three strategies relate to each other and interact to improve program performance (PP) within a humanitarian setting. This study helps to close this gap in the literature, by examining how different communication strategies and the interactions between them influence the perceived PP of humanitarian operations.

We propose a theoretical model to integrate the relationships between different communications strategies and their impacts on PP. We use structural equation modeling (SEM) to test our model within one specific humanitarian organization working in Somalia: the Danish Refugee Council (DRC), an independent non-profit organization devoted to humanitarian principles. DRC implements the Humanitarian Accountability Framework (HAF), which includes OC and two-way communication standards, in an attempt to improve its communication and PP. Our results provide practical evidence in support of our theoretical model, highlight the significant impact of OC and two-way communication on PP, and demonstrate the mediating action of OC on the relationship between two-way communication and PP. In addition, we show how effective communication plays an important role in improving coordination, which can in turn have positive effects on the performance of humanitarian operations.

This paper continues as follows: the next section reviews the relevant literature and describes the basic concepts of organizational communication, highlighting its importance in the humanitarian sector. The succeeding section presents our research questions, describes our field site and reviews the data-collection methods used. Next, we present our model, our analysis and results. Finally, we discuss our principal findings, their implications and future research opportunities.

Organizational performance represents the ability of an organization to be effective and efficient (Barnard, 1938). Effectiveness refers to the ability of an organization to fulfill the requirements of its stakeholders, while efficiency emphasizes how the resources are used to meet the organizational goals properly – that is, with the least waste of time and effort (Neely et al., 1995). However, it is a challenge to characterize performance in the humanitarian sector. First, because different kinds of disasters and emergency or development programs require different types of measurement (Beamon, 1999; Micheli and Kennerly, 2005; Wakolbinger and Toyasaki, 2011). Second, because humanitarian organizations are required to be accountable to multiple actors (beneficiaries, donors, other organizations and their own staff) which have different performance metrics (Ramadan and Borgonovi, 2015). Finally, because of the difficulty of obtaining accurate data in chaotic environments, where conflicts between long- and short-term organizational goals are frequent (Abidi et al., 2013). Beamon and Balcik (2008) draw on the concepts of effectiveness and efficiency to propose a framework for understanding performance in a humanitarian context. They consider three performance dimensions: resource management (related to the level of efficiency), output (related to the level of effectiveness) and flexibility (related to the ability to respond to a changing environment). For relief operations, Blecken et al. (2009) propose a performance measure as a function of available resources, delivery delay and donation-to-delivery time. Ramadan and Borgonovi (2015) suggest a performance framework that stresses the importance of resource allocation, efficiency, effectiveness, impact, quality and partnerships. For their part, Davidson (2006) and Haavisto and Goentzel (2015) build on goal-setting theory to evaluate performance as the way in which organizational activities progress toward pre-established organizational goals.

Thus, the literature considers different performance measures depending on the variables under evaluation and the context of the study, highlighting the difficulty of establishing common criteria for measuring and operationalizing performance in humanitarian organizations. For instance, Blecken (2010) reports that only a small fraction of humanitarian organizations have consistent performance measures and proper monitoring and evaluation processes aimed at satisfying donor requirements.

As mentioned above, organizational performance is affected by both objective and subjective factors. Objective factors are quantifiable aspects such as organizational structure (e.g. size or age), the number and skills of staff, utilization levels, information technology infrastructure, processes and innovation (Gavrea et al., 2011; Kates and Galbraith, 2007; Wu and Wang, 2006), while subjective factors refer to qualitative aspects such as interpersonal relationships and communication (Beebe and Masterson, 1996; Jakki and Nelvis, 1990). Interestingly, previous studies have shown that many problems affecting organizational performance in the humanitarian sector are rooted in subjective factors. For instance, Dubey et al. (2016) find that the high employee turnover faced by many humanitarian organizations is due to factors such as lack of role clarity which can occur as a result of poor internal communication. Similarly, Urrea et al. (2016), and Altay and Pal (2014) show how adequate connections between humanitarian organizations and their willingness to exchange information facilitates information flow and the transfer of knowledge and thereby increases PP. Finally, Tatham and Kovács (2010) and Papadopoulos et al. (2016) develop the importance of the concepts of inter-organizational, intra-organizational and interpersonal trust in developing a trusting environment among humanitarian partners conducive to enhanced information exchange, improved coordination and increased operational performance. Given the important role played by communication, this paper explores how interactions between different stakeholders and the mechanisms through which they communicate with organizations may affect performance in humanitarian operations.

March and Simon (1958) define communication as a flow of information among members of an organization, in which a series of messages is sent and received (Drenth et al., 1998) using different channels (Shannon, 1948), which themselves (Wrench et al., 2008) may be verbal (spoken word), non-verbal (gestures, looks) or mediated (achieved using a specific sort of technology). Even if communication is defined in these simple terms, it constitutes a key element that maintains critical relationships inside and outside organizations, creating competitive advantages that lead to better organizational performance (Baker, 2002; McCroskey et al., 2006). For instance, improved communication tools and technologies may help increase the efficiency and effectiveness of organizational processes (Tucker et al., 1996; Lucas, 1996).

As Myers and Myers (1982) argue, if it is to facilitate the sharing of information with other stakeholders and persuading them to act, good communication must fulfill three roles: inform, coordinate and regulate; align individual goals with collective goals; and foster innovation. These three roles make communication a critical aspect of organizational performance.

An appropriate communication model, driven by an efficient exchange of information, may be an effective strategy for improving the performance of humanitarian organizations (Apte, 2009; Goldhaber, 1990; Leiras et al., 2014; Maiers et al., 2005; Schweitzer et al., 2002) and for addressing the challenges posed by uncertainties, time pressures, limited resources and donor pressures (Brown, 2008; Dolinskaya et al., 2011; Ergun et al., 2010; George, 2003; Halachmi, 2002; Wakolbinger and Toyasaki, 2011).

The availability of information and effective communication not only help coordinate logistical efforts and increase the acceptance and implementation of new practices (Sheffi, 2005; Sarkis et al., 2012), but also help manage relationships between the different actors involved. Child (1972) suggests that effective communication and the integration of actions help in the allocation of tasks and responsibilities between individuals, positively influencing organizational performance as a result. Two forms of communication, and the interactions between them, impact on task-allocation and information flow: OC and two-way communication, which in the latter case may be either interactional or transactional (Barnlund, 1970; Beebe and Masterson, 1996; Vlăduţescu, 2013).

The goal of OC is to persuade the audience by imparting a specific message; stakeholders act, in other words, as simple recipients of information (Crane and Livesey, 2003). This kind of communication is usually formal, i.e. it reflects a set of internal rules and policies established by the organization (Beebe and Masterson, 1996; Shannon and Weaver, 1949), which flow downward (from superiors to subordinates) (Baker, 2002). OC appears in manuals, procedure guidelines, newsletters and instructions that are made available to the different members of organizations. This communication reduces the inefficiencies caused by information deficits. It can bridge the spatial distance between members of organizations and mitigate the negative effects of team dispersion on PP (Bardhan et al., 2013; Inderfurth et al., 2013).

According to Barnard (1938), when the message is believed to come from an official organizational source, or communications center, the communication itself reflects authority. For this authority to be echoed in the message, the channels of communication should be formal and clearly understood, the lines of communication short and direct, and the people serving at the communication centers should be adequate to the task and their role recognized. When all of these aspects are in place, OC can be an effective tool for the coordination of actions (Barnard, 1938), and may have a positive impact on performance (Bardhan et al., 2013). Finally, Goldhaber (1990) supports the argument in favor of a positive relation between effective communication and PP, arguing that “the better informed employees are, the better employees they will be.” Therefore, communication practices within an organization institutionalize the ways in which people interact with each other, how communication flows, and the ways in which power relationships are defined, creating as a result the basis for successful PP. Accordingly, we formulate our first hypothesis as:

In contrast to OC, two-way communication involves stakeholder dialogue and has two principal goals. The first is to gather information from different audiences and the second is to establish better understanding among different parties (Crane and Livesey, 2003). There are two modes of two-way communication: interactional and transactional communication (TC) (Beebe and Masterson, 1996; Vlăduţescu, 2013).

Interactional communication (IC) refers to a kind of communication where participants are both senders and receivers of information (Schramm, 1954). In an organizational setting, this communication is usually formal, horizontal (among people who are not in a hierarchical relationship) and with an external focus (Baker, 2002), i.e. with stakeholders outside the organization. This type of communication is usually considered asymmetrical, because its main objective is to extract information from stakeholders and use it to design the message(s) that will subsequently be conveyed in the hope of persuading them. IC has the fundamentally instrumental purpose of gaining compliance (Crane and Livesey, 2003).

At a practical level, organizational initiatives aligned with this type of communication include the handling of complaints from external parties. For example, HAP International (2013) suggests that in order to achieve effective communication, humanitarian organizations should ensure that: stakeholders (e.g. donors, beneficiaries, staff) have access to timely, relevant and clear information about the organization and its activities; and organizations should classify, handle and document stakeholders’ complaints. These practical mechanisms oblige humanitarian organizations to continuously create processes and systems that enable better communication and increased commitment by stakeholders (Cavill and Sohail, 2007).

Processes that gather information from external stakeholders using IC can be beneficial to organizations. As an example, Best and Andreasen (1977) assert that when stakeholders do not air their voices, an organization loses the opportunity to recognize and address problems, hampering program success and leaving the beneficiaries, donors and staff dissatisfied. Moreover, Bies and Moag (1986) and Lapré (2011) suggest that an organization’s complaints procedure leads to interaction with the customer, contributing subsequently to customer satisfaction and effective PP. Consequently, Davidow and Dacin (1997) support the HAF, arguing that while seemingly paradoxical, it is in the best interests of organizations to encourage consumers to complain, and then to react promptly and appropriately to address the issues raised.

Therefore, two-way IC in organizations is a critical ingredient that exerts a positive influence on PP. Based on these arguments we hypothesize that:

The second kind of two-way communication corresponds to TC, in which the sending and receiving of messages between the communicators occurs simultaneously (Barnlund, 2008; Vlăduţescu, 2013). At the organizational level, TC usually takes place upward (from subordinates to superiors), with an internal focus, and may either be formal – occurring through established feedback processes – or informal, that is, mediated through interpersonal relationships (Baker, 2002). This communication is considered to be symmetrical, because it involves “genuine” dialogue in which both participants persuade, and allow the other to persuade (Crane and Livesey, 2003). According to Hirokawa (1979), this type of communication is useful for managers seeking to assess previous downward communication and to test possible employee responses. In addition, it is helpful for employees, as it allows them to have a voice regarding different organizational policies and procedures. Moreover, accurate intra-organizational communication capabilities help to overcome organizational barriers (Bendoly et al., 2008) and to increase levels of trust between members of organizations, which then leads to program success (Brinkhoff et al., 2015).

Feedback, which may be defined as the receiver’s observable response(s) to a source’s message (McCroskey et al., 2006), is an important form of TC. Feedback helps develop interpersonal relationships and gain compliance and understanding, which can potentially influence individual and organizational performance (Ilgen et al., 1981; Taylor et al., 1984). Hence, the successful implementation of a program also requires the provision of clear performance-based feedback from staff and other stakeholders (Carroll and Schneier, 1982; Ilgen et al., 1979). Nevertheless, whether these benefits are realized depends on how recipients react to the feedback. One possible – and desirable – reaction to feedback is satisfaction. Satisfaction with feedback leads to favorable reactions that lead in turn to further program success (Dorfman et al., 1986; Keeping and Levy, 2000). Giles and Mossholder (1990) argue that satisfaction as a measure of employees’ reactions is a more encompassing indicator of reactions to feedback than more specific, cognitively oriented criteria, such as its perceived utility or accuracy (Keeping and Levy, 2000). In addition, satisfaction with feedback leads to increased trust and higher recognition of future prospects within organizations (Brinkhoff et al., 2015). Similar streams of research recognize that performance feedback has cognitive and motivational elements that can enhance employee motivation and performance (Ambrose and Kulik, 1999). For instance, expectancy theories predict that job motivation will improve as employees understand the relationship between performance and rewards (Lawler, 1994; Vroom, 1964).

TC, represented by feedback, is also central to goal-setting theory. As Locke (1996) states, goal-setting is more effective when there is feedback that shows progress toward the goal. Earley et al. (1989) argue that humanitarian organizations should include TC as a method of discussing outcomes (ratings) and processes (strategies to enhance future performance), both of which enhance motivation to foster better performance. Here, we are interested in understanding how clearly defined TC processes, such as feedback, affect PP in humanitarian organizations. Thus, we hypothesize that:

As described above, OC is principally a top-down process with an internal focus and involving no feedback. This lack of feedback leads to two problems in terms of communication: accuracy and adequacy (Hirokawa, 1979). Accuracy refers to how well the information is received, while adequacy points to the sufficiency with which the message meets the need for information. Problems with the accuracy and adequacy of information can reduce the necessary connotation of authority that should be reflected in the messages sent. Hence, in OC, authority can be lost if the means by which the messages are conveyed (e.g. procedures, manuals, templates, processes and communication channels) are not up-to-date and do not represent the current situation of the receiver of the information (Barnard, 1938).

Nonetheless, a proper OC model should adjust information, progressively correct deviations and improve both the accuracy and adequacy of the message (Beebe and Masterson, 1996; Vlăduţescu, 2013) by making use of interactional and transactional two-way communication models. Organizations that use IC should be able to capture accurate information about the concerns of external stakeholders and use it to design or improve the formal manuals or newsletters that they should distribute in order to interact with them. Such practices ensure that the accuracy and quality of the information provided by OC will be improved. Similarly, the use of TC should allow organizations and top managers to take advantage of feedback sessions to help them understand people’s concerns and any misunderstandings that may exist about current rules and guidelines. These conversations should foster improvement of the current rules and the creation of new ones that facilitate future accuracy and ensure the different policies and procedures used by the organization are adequate.

Thus, by integrating the positive effects of interactional and TC into the OC model, we also hypothesize that:

Thus, OC can be interpreted as a mediator variable in the relationship between two-way communication and organizational performance.

Given that our set of hypotheses is based on subjective factors, we conceptualize the basic relationships between the different communication modes and PP in a single model (see Figure 1). In addition, in order to avoid confounding effects in our ensuing estimations it is necessary to control for some subject-specific characteristics (including levels of education attainment/experience, status within the organization, age and gender) that may have a direct impact on perceptions of PP.

Our study gathers data from the DRC in Somalia, an independent non-profit organization devoted to humanitarian principles, which works to protect refugees and internally displaced people by promoting long-term solutions to the problems of forced migration. DRC Somalia tries to be accountable not just to donors and staff but also – and in particular – to the beneficiaries the organization seeks to support. In 2007, DRC was certified by the Humanitarian Accountability Partnership International (HAP International), the leading consortium of development and humanitarian agencies. Since then, it has become one of the leading humanitarian organizations promoting accountability in Somalia. It is also one of the few humanitarian organizations that is continuously working toward improving communication with its stakeholders using the HAF model, which focuses on the effective implementation of humanitarian and development programs. It also aims to ensure accountability to all stakeholders.

We chose to focus on DRC Somalia for our case study for the following reasons: DRC Somalia implemented its HAF using both OC and two-way communication mechanisms, including stakeholder feedback, improved communication channels and complaint response systems. These mechanisms have been widely disseminated within DRC Somalia and stakeholders have been well aware of them. In addition, DRC expanded the Complaint Response Mechanism (CRM) for its Somalia program. Inclusion of the CRM led to the creation of different complaints handling guidelines that were sent to multiple stakeholders as a tool for improving organizational performance on the basis of stakeholder concerns. With the adoption of the CRM, a designated focal person responsible for ensuring timely and comprehensive responses to stakeholders’ concerns was appointed in each DRC office in Somalia. On the basis of this OC and two-way communication, staff are able to propose long-term opportunities for improvement. Finally, DRC has promoted a range of campaigns encouraging staff to listen objectively to the suggestions of those giving feedback, maintain a co-operative attitude and avoid preconceived ideas during feedback sessions. Thus, the decision to select this particular organization as a case study enabled us to identify and evaluate the critical variables and links contained in our theoretical model (Stuart et al., 2002; McCarthy and Golicic, 2005).

We conducted our study in Somalia (i.e. in Somaliland, Puntland, and Southern and Central Somalia). We used a cross-sectional design and collected most of the data using structured questionnaires that were self-administered by respondents. We also carried out some direct structured interviews to obtain responses from one director and a few specialized staff members. Traditionally, organizations adopt both hard and soft measures to evaluate operational performance. Hard measures refer to objective concepts such as net income, order cycle time and costs. However, due to the challenges of evaluating and tracking operations in the humanitarian sector, we used soft measures to estimate the main variables of our model. These soft measures involved managerial perceptions of variables such as stakeholder satisfaction, logistical efficiency and productivity (Brewer and Speh, 2000; Chow et al., 1995; Fugate et al., 2010). The questionnaire contained questions including indicators (using a five-point Likert scale) to measure our four main factors: OC (eight indicators), IC (nine indicators), TC (nine indicators) and PP (ten indicators). We used reflective indicators to estimate each specific factor. The indicators were defined and aligned with the main theoretical concepts associated with each factor. For example, for the evaluation of PP we included items that accounted for subjective evaluations of project delays, service levels, outputs and program quality. The indicators for OC measured information related to awareness of communication channels and access to procedure guidelines covering work-related information. For IC, the indicators involved concepts related to the use, awareness and usefulness of complaints mechanisms. Finally, for the evaluation of TC, we included items intended to evaluate respondents’ awareness of the feedback processes and their impact on employee motivation and organizational decision-making processes. More details of the constructs and the specific items used are presented in  Appendix 1. The study comprised a sample of 107 respondents (71 percent of the population) with 84 males and 23 females drawn from all departments of DRC Somalia. The diversity of the staff group captured in the sample allowed proper representation of the variability in stakeholder perceptions. However, to control for the influence of subject-specific effects on PP, we collected demographic information on variables that might create more differences in subjects’ perceptions, such as age, gender and highest education attainment, and on contextual factors, including job classification and geographical region. Table I shows the population and the sample size for each job category in the organization, while  Appendix 1 presents additional information and descriptive statistics on the control variables, along with additional details concerning the questionnaire.

To assess whether DRC Somalia was an appropriate organization for evaluating our model, we tested the awareness of interviewees regarding the implementation of the HAF model. Two subsections in the questionnaire included questions about orientation received at the time of hire, and their ability to obtain or access policy documents. Results showed that on average 74.3 percent of the participants perceived a high level of orientation at the time of hire, while 63.5 percent perceived a high level of access to relevant communication policies (see  Appendix 1).

Before testing our hypotheses, we examined the data for normality, statistical power and reliability. For the normality test, examination of univariate indices of skewness and kurtosis revealed no troublesome values (absolute values lower than 3) for all four variables (see Table II). We conducted a power analysis for a path coefficient of a predictor that accounted for at least 5 percentage points of unique variance in the outcome, where we ran a two-tailed test assuming a squared multiple correlation of 0.20 and α=0.05. Under these conditions, for a small model with four variables and a sample size of 107 respondents, we obtained a statistical power of 0.98, which is acceptable for the proposed analysis.

We ran an exploratory factor analysis of all items to check whether the interrelationships among items that are part of each unified variable did indeed represent the general construct we were seeking to account for. We extracted the four principal components using eigenvalue extraction methods, excluding items with a cross-loading values difference lower than 0.2. Results accounting for the formation of four components were satisfactory: KMO=0.79, the sum of squared extractions accounted for 59.5 percent of the variance, average value of loadings was 0.72 and correlation between components were all positive and lower than 0.5. Results are typically considered satisfactory when KMO>0.7, percentage of variance explains >50 percent, average value of loadings >0.6 and correlations <0.7.

Next, to check that the instruments yielded consistent results, we ran a reliability analysis (Mugenda and Mugenda, 1999). This approach provides information about the internal consistency between individual items in each component. Cronbach’s α was computed as an initial indicator of reliability and since all the items in the constructs generated Cronbach’s α above 0.7, the measurements were considered reliable. Table III presents all the loading values, IDs and Cronbach’s αs for the configuration obtained with the factor analysis, as well as the dependent variable Pp, and the independent variables OC, IC and TC.

We removed some of the initial items incorporated to estimate each specific factor of our model from our analysis because of the low contribution they made to explain the specific factor. We observe that factors 1, 2, 3 and 4 corresponded to our four variables of interest: PP, OC, IC and TC, respectively.

However, these results only included general computations between pairs of variables. In order to gain a better understanding of the general process and effects, we needed to consider all the variables at the same time, using a multivariate regression approach, that is, SEM. Before analyzing the complete model using SEM, we ran a confirmatory factor analysis (CFA) to provide an additional measure of reliability by ensuring that the measures of the variables were consistent and that the data fit the model. Table IV presents the CFA model fit summary and  Appendix 2 the measurement model, validity values, non-response bias and the results for the final CFA model (Armstrong and Overton, 1977). These CFA results show that the fit of the model satisfied the acceptance criteria for all measures.

In addition, given the empirical nature of our study, we needed to check for potential evidence of endogeneity. Although endogeneity can never be completely eliminated in this type of cross-sectional study, it was important to reduce the factors might potentially cause it (Podsakoff et al., 2003; Ketokivi and Schroeder, 2004). The effects of endogeneity in an empirical analysis may be reduced in two principal ways: either theoretically or empirically (Guide and Ketokivi, 2015). From the theoretical perspective, the potential effect of endogeneity was reduced by defining and explaining the direction of the relationships of the constructs used in our model (causality). In this study we define causality with reference to some theoretical concepts drawn from the communication literature, which we used to build our hypotheses.

From the empirical perspective, endogeneity problems may be caused by common-method bias, where part of the observed variance is due to the measurement method rather than to the constructs that are being measured (Bagozzi et al. 1991). Common-method bias may have several causes (Podsakoff et al., 2003), one of the most important being the propensity of subjects not to answer each question independently, in an attempt to be consistent in their answers and achieve social acceptance. Another reason for common-method bias is the fact that both the dependent and independent variables in a model are obtained from the same population, leading subjects to look for some consistency in their answers for the dependent and independent variables. See Podsakoff et al. (2003) for further details of the phenomenon. We used two strategies to control for the presence of common-method bias: study design, and statistical controls. From the perspective of study design, we protected our respondents’ anonymity and told them that there were no right or wrong answers. This meant they were more likely to answer the questions freely and it was less probable they would edit their responses to make them appear more socially desirable. In addition, we selected respondents from different locations, different positions within the organization, and who belonged to different gender and age groups, in an attempt to reduce some context-specific biases.

From a statistical point of view, we used two methods to test the conditions under which the data were obtained in order to assess the extent to which common-method bias might be a problem. Initially, we used the Harman single factor test (Harman, 1960) as a diagnostic technique. Although this method does not provide inferential statistical validity, it does offer some insight into the extent to which common-method bias can be a problem. In this case, the total percentage of variance explained by a single factor is 32.9 percent, which is lower than the reference point of 50 percent. The second method we used was the common latent factor test. Although this method does not provide information about the specific causes of common-method bias (a matter that is not a focus of our study), it adequately accounts for presumed measurement error in the method (Podsakoff et al., 2003). Results from the common latent factor test give a common latent variance lower than 20 percent. There is, therefore, no clear evidence for the presence of common-method bias in our data, and there was no need to retain the common latent factor when we analyzed our model using SEM.

Finally, another cause of endogeneity may be sample selection bias (Heckman, 1979). However, given the large sample size in our study (relative to the size of the total population), we did not predict any important effect of this kind.

Given the positive results of the CFA and the endogeneity tests, we were able to make use of our set of constructs to test the model conceptualized in Figure 1. As linearity and non-multicollinearity are requirements if SEM is to be used, we ran a curve estimation analysis for different possible relations between the dependent and independent variables (linear, logarithmic, quadratic, power, exponential and logistic) presented in our model. Results showed that the linear model had the highest R², and therefore, the model was sufficiently linear to be tested using a covariance SEM (see details in  Appendix 3). In addition, we tested the model for possible multicollinearity issues among the independent variables. A context-specific analysis of Variance Inflation Factor (VIF) gave values lower than 3 in all possible variable combinations; suggesting that we do not have significant multicollinearity issues (see details in  Appendix 3).

We evaluated the hypotheses shown in Figure 1 using SEM implemented in AMOS© 19. The α_s and β_s presented in Figure 1 represent the main effects of the model. Figure 2 presents the general scheme of the SEM, accounting for the items that satisfied all the CFA tests, and including control variables. We used a set of control variables that were intended to control for the context of subjects (e.g. hierarchical status in the organization) and their individual attributes (e.g. gender, age and highest level of educational achievement), that might affect subjects’ perceptions. We also accounted for some meaningful modification indices, which are not, however, presented in the figure for reasons of clarity. For further details about the modification indices included in the model (see  Appendix 2).

Finally, as in our model OC could be interpreted as a mediating construct, we needed to determine whether it really did operate as a mediator of IC and TC on PP. Therefore, we followed the mediating analysis process proposed by Baron and Kenny (1986). We determined the direct effect of IC and TC on PP and the direct effect of IC, TC and OC (see Table V).

To determine whether OC acted as a mediator in our model, different conditions had to be satisfied:

1. parameter estimations for the direct effect on PP had to be significant;

2. parameter estimations for the direct effect on the mediator (IC) had to be significant;

3. parameter estimations for OC had to be significant; and

4. the estimations for IC and TC in the last column had to be smaller (in absolute value) than the direct effect estimations (first column).

Since all these conditions were satisfied, we can conclude that there was full mediation of OC in our model. Therefore, we were able to keep OC as a mediating variable for our analyses and we preserve the proposed structure of our model (see Figure 1).

Table VI shows different typical measures of model fit used to test our SEM. Results indicated an acceptable level of model fit, as the model satisfied all goodness-of-fit tests. The model summary (Table VII) reveals that 66 percent of variation in perceived PP is explained by the variability in the three factors: OC, and two-way (interactional and transactional) communication. This result reflects the good level of explanatory power that our structure of independent variables and models had for PP in the organization under study.

Additionally, Table VII shows the main effects, the confidence intervals and standardized effects of our hypothesized model; all the independent variables and controls were accounted for simultaneously. The results also show that the three different types of communication had a significant and positive effect on PP. These results also supported our five hypotheses.

Our findings mean, on the one hand, that having a comprehensive process of OC facilitates the alignment of individual goals with collective organizational goals, enhancing prospects for effective and efficient delivery. Given the uncertain nature of their operating environments, humanitarian organizations can benefit from using OC strategies, because they serve to bring teams closer and help align work throughout the organization (Bardhan et al., 2013; Inderfurth et al., 2013). This alignment may reduce organizational inefficiencies and lead to improved performance. Well-structured, appropriately timed, OC aimed at the correct target(s) is recognized to have reduced deficiencies in programs, which might have resulted from gaps in information. It is thus imperative that staff have up-to-date information and know where and how to access it in order to enhance effective and timely implementation of tasks. They should be in the know before executing the how.

On the other hand, two-way communication mechanisms (IC and TC) have a positive and significant effect both on PP and on OC. This means that there is a positive association between the gathering of information from an organization’s different audiences, such as beneficiaries, contractors and donors, and the establishment of improved understanding among different (internal and external) parties by ensuring continued feedback and enhanced PP in humanitarian organizations and the construction of improved internal communication frameworks.

At a practical level, the existence of appropriate two-way communication mechanisms helps humanitarian organizations deal with their high number of existing stakeholders (Dolinskaya et al., 2011; Ergun et al., 2010). If the relationships with both external (e.g. donors, beneficiaries) and internal (e.g. staff) stakeholders are well managed, then those mechanisms may also enhance the coordination and alignment of goals, which might in turn lead to improved performance.

Effective IC provides an avenue for understanding an organization’s audience and is useful in informing and improving organizational policies so that they fit the operational context. The existence of an effective complaint-handling mechanism may improve beneficiary confidence in organizational processes and enhance their participation in program design and implementation. These results are aligned with the findings of Bies and Moag (1986) and Lapré (2011), who argue that proper complaints mechanisms will make organizational staff more aware of beneficiaries’ needs and add value to project design.

Effective TC is indicative of the creation of an environment where informal and formal feedback systems are effective. These friendly environments increase staff confidence that they will be heard, leading to favorable reactions that enhance employee satisfaction. Increased staff satisfaction is likely to improve PP (Brinkhoff et al., 2015).

Finally, concerning the control variables in our model, we may conclude that gender, age and status do not play an important role in the perceptions people have about PP. However, staff education levels exert a positive and significant effect. This finding suggests that people with a higher level of educational achievement tend to assess PP more positively. This might be because people with higher educational levels are better informed about the dynamics of the system and have superior knowledge about how things should work as a result of changes that have been implemented. According to this view, if people do not recognize the expected performance outcomes of a system/policy, their mental models might exclude important interconnections that in fact exist, causing them to be reactive and pessimistic (Sterman, 2000; Bendoly et al., 2010; Gonçalves and Villa, 2016).

While extensive research has been conducted on the importance of communication to PP, empirical research focusing on humanitarian organizations is still underdeveloped. To help close this gap, this study proposes a general model for explaining organizational performance in the humanitarian sector, based on the interrelations of different communication strategies. An important aspect of this structured communication framework is the enhancement of OC and two-way (interactional and transactional) communication, in order to allow humanitarian organizations to use their communication channels properly and to improve operations on the basis of lessons learned and the concerns of stakeholders. Such communication frameworks also allow humanitarian organizations to design more robust programs and respond better to humanitarian crises.

This paper evaluates a theoretical model using data from the DRC in Somalia. Results of the systematic review of the organization’s communication strategies reported in this study provide important insights for non-profit humanitarian and for-profit organizations alike. In addition, our econometric estimations show that the interrelations between the different types of communication explain a high percentage of the variation in the performance of programs.

The inability of humanitarian organizations to manage the flow of quality data properly and regularly update internal and external information makes it difficult to improve coordination, leading to reduced performance, ineffective resource utilization and the duplication of efforts (Balcik et al., 2010; Leiras et al., 2014; Altay and Pal, 2014; Diedrichs et al., 2016). Therefore, effective communication should play a vital role in achieving better coordination that in turn might improve the performance of the humanitarian effort and of organizations engaged in providing a response.

Our results suggest that organizations should ensure that all stakeholders (beneficiaries, staff, donors, etc.) have access to timely, relevant and clear information about the organization and its activities. The current dissemination and communication strategies used by some humanitarian organizations may not only be ineffective but may even exclude specific stakeholders. Staff responsible for an organization’s communication strategy should design and implement an effective interactive communication structure capable of disseminating information both internally and externally. Hence, the creation of information centers, manuals, procedures, instructions and newsletters can reduce inefficiencies caused by information deficits, helping to share appropriate information with stakeholders (Bardhan et al., 2013; Inderfurth et al., 2013). In addition, the inclusion of technology to share accurate information throughout an organization in a timely manner may also provide strategic advantages that lead to improved PP (Wrench et al., 2008; Serrato-Garcia et al., 2016).

Similarly, humanitarian organizations should work on properly handling the concerns of external parties. Organizations should be aware that unresolved complaints can lead to a decline in their image, raise stress levels among staff and harm stakeholders’ commitment, all of which ultimately affect the success of current and future programs (Crane and Livesey, 2003). Similarly, ensuring the existence of platforms that allow dialog and interactions with other organizations and stakeholders may allow humanitarian organizations to build up swift trust and improve coordination.

In addition, our results highlight the importance of interpersonal relationships and a “genuine” dialogue between members of the organization and their impact on the performance of humanitarian and development programs (Baker, 2002). Hence, clear performance-based feedback leads to the establishment of lessons learned that might improve the quality of response in future programs. This improvement can be achieved in two ways: first a good feedback process provides new insights for the stakeholder who is receiving it, and second it helps set a tone of personal responsibility.

Finally, our results show that the lack of feedback presented in OC is compensated by the lessons learned from staff member experiences and from stakeholders’ complaints. Therefore, OC strategies should be progressively corrected in order to improve both the accuracy and the adequacy of the messages being conveyed (Beebe and Masterson, 1996; Vlăduţescu, 2013).

The main limitation of our study is the restricted possibilities of generalizing its findings, given that the research was conducted within a single organization in one geographical region. Future research should first focus on applying this model to different humanitarian organizations, enabling researchers to draw more general conclusions about the principal effects reported here. Second, similar studies could be carried out of commercial organizations, which face different challenges from those faced by their humanitarian counterparts and for which coordination with different stakeholders might be less essential. This would allow researchers to understand whether the importance and effects of OC and two-way (interactional and transactional) communication on organizational performance are the same for both kinds of organization. Third, future research should investigate the evolution over time of the effects of the different communication strategies on PP. This would allow a better understanding of the dynamics involved in PP in the humanitarian sector to be built up. Finally, an additional challenge faced in this paper is associated with the subjective evaluations provided by our respondents, where their degree of familiarity with each of the constructs/variables under study might have influenced their decisions. Thus, if our metric is, for example, perceived PP then the subjective assessment might also capture changes in actual performance, though there would be known way for us to know this. Consequently, subjects might perceive change when actual PP has not varied. Therefore, instead of using primary data to account for the experiences and perceptions of staff, the model could be run using secondary data and indicators collected from the organizations over time.

As humanitarian emergencies become more frequent, humanitarian organizations face increased pressure to improve performance and ensure their programs are more effective and efficient. This study investigates the ways in which different communication strategies may affect the PP of humanitarian organizations. Communication plays a very important role in the success of any humanitarian effort. Coordination among organizations of varied nature, backgrounds and objectives is also difficult but essential. This paper proposes and uses a communication framework composed of three related strategies: OC and two-way (interactional and transactional) communication, using a structural model to test the effect of these strategies on PP. The paper shows how proper communication strategies that address stakeholders’ concerns can lead to improved PP. In addition, the results of a SEM showed a significant relationship between the three communication strategies and PP, suggesting that structured and targeted internal and external communications with different stakeholders positively influence the success of humanitarian programs. We show how effective communication plays a vital role in achieving better coordination that may in turn improve the performance of the humanitarian effort as well as of organizations themselves. Finally, our theoretical model shows how OC can be used as a full mediator between two-way communication and PP.

Table AI 

This part of the study sought to capture the views and knowledge of the respondents on the orientation they received when beginning employment at DRC Somalia (see Table AII). Interviewees were well-informed about the different policies and procedures involved within the organization, especially those concerning human resources policies and procedures including the code of conduct, staff development and job description (91.6 percent). However, opportunities remain for improvement in different aspects such as security orientation (61.7 percent) and the integration of the HAP into DRC’s practice (67.3 percent).

Table AIII illustrates perceived accessibility of a range of important policy and process documentation within the organization. These results reflect a good level of document accessibility for all DRC staff, though there is a need to improve procedures for the dissemination of documentation.

Tables AIV and AV show that participants are young and have limited experience in DRC, but that most have high levels of educational attainment.

Table AVI shows that correlation among the independent variables of our model is not high. This result facilitates analysis and interpretation of our results.

Figure A1 and Table AVII represent the general structure and estimations of the CFA. Estimations of the standardized regression loading weights are relatively high (>0.5), which facilitates the SEM analyses. Modification indices for the CFA are: e13-e12, e16-e15, e11-e10, e7-e8, e21-e12, e8-e22, e1-e18, e3-e24, e1-e24, e2-e25, e23-e15, e21-e20, e22-e9. These are not presented in the diagram for the sake of clarity.

In order to test if the model could be represented by a tau equivalent model (all the factor loadings of each latent should have the same numerical value) or by a parallel model (all the factor loadings and all the disturbance variances should be equal), we ran a model comparison accounting for the possible invariances. Results show that the congeneric model (no constraints) is the best way to represent our data (p-values=0.00) (see Table AVIII).

We also analyzed our model to account for possible differences across groups of respondents. The questionnaire was run with people with different rank positions within the organization. In particular, we observed (see Table I) that there was a considerable difference between the percentages of people in high status positions (directors, senior management and middle management) who answered the questionnaire and those with low status (specialized staff, support staff and lower level staff). Therefore, to check for potential non-response biases in our data, we ran a comparison test between these two groups (Podsakoff et al., 2003). Table AIX shows that there is at least one non-significant z-score in the comparison of the items in each construct. Therefore, it could be argued that there are no significant differences in the responses provided by subjects with different status within the organization.

In addition, using the convergent validity measurement, three items (Pp2, Pp8 and Pp9) were excluded from the final model because of their low contribution to the Average Variance Extracted (AVE) in program performance (see Table AX). We accepted AVE values slightly lower than 0.5 because in our final CFA model all correlations were below 0.7 and the average factor loadings values among all variables was 0.7, which gave validity to our model.

Table AXI shows that the linear model had the highest R², meaning that among all the possible type of equations considered, the linear model was the best way to represent the relationship between the dependent variable (program performance) and each independent variable (OC, IC and TC).

Table AXII shows a formal detection-tolerance test for multicollinearity using the VIF test between each pair of independent variables. Values of VIF below 3 represent a low level of multicollinearity between each pair of variables. Based on these results, we do not expect large changes (<21 percent) in the standard error of our estimates compared to the case where all independent variables were completely uncorrelated between themselves. These small changes in the size of the standard errors will affect neither the significance nor the direction of our estimates. In addition, we would not expect significant changes in our estimations as a result of sample selection, because in our case our sample size was sufficiently large (75 percent of the whole population).