Insight for knowledge brokers: factors predicting relationships with federal staffers

Abstract

Background:

In theory and practice, it is understood that personal relationships play a role in the effectiveness of translational models that bridge research and policy. These models can be made more efficient by understanding factors impacting relationships between policy-making players and third-party knowledge brokers.

Aims and objectives:

This study investigates a range of personal and office-level characteristics in predicting initial meetings and sustained relationships between federal staffers and knowledge brokers.

Methods:

A public affairs database, Quorum, was used to pull data on staffers who were contacted between September 2021 and August 2022 during an optimisation phase of the Research-to-Policy Collaboration (RPC). Logistic regression models were used to understand the impact of the characteristics on outcomes such as attending initial meetings and attending meetings facilitated by the RPC.

Findings:

Mid-level staffers and democratic staffers were more likely to meet with RPC staff. Office tenure was predictive of lower odds of meeting with RPC staff. For significant associations, the sample was stratified by political party to determine if the results differed by party.

Discussion and conclusions:

Together, these results suggest there are both personal and office-level characteristics affecting the federal staffers’ engagement with knowledge brokers. This work further informs efforts to bridge the gap between science and policy by informing knowledge brokers which offices and staffers they may want to approach.

In an age where there are increasing levels of distrust in scientific research that can inform policy makers and their work (Agley, 2020; Tseng, 2020), it is important to understand what factors affect the science–policy relationship and how they do so. Both in theory (Bogenschneider and Corbett, 2021) and in some of the leading models of research translation (Bogenschneider et al, 2000; Crowley et al, 2018; Scott et al, 2019) one factor that is shown to be important is personal relationships between the policy making and researcher communities. With funding for research translation scarce (Zurynski et al, 2021), there is a need to make research translation models more efficient and cost-effective. One way in which these research–policy translation models can be made more efficient is by understanding who is more likely to engage with them: the focus of our study.

A key component of the integration of the use of research evidence (URE) into policy making is the development and sustainment of interpersonal relationships between policy makers, researchers and knowledge brokers alike (Bogenschneider et al, 2000; Oliver et al, 2014; Boaz et al, 2019; Scott et al, in progress). Federal policy makers, both in the Senate and the House of Representatives, work on a wide variety of issues which limit their ability to devote time and resources to any one particular issue. Policy makers often rely on trusted relationships for advice and information on the issues they work on to help offset the amount of decisions they need to make (Caldeira and Patterson, 1987; Peoples, 2008). These decision-aiding relationships can occur both with fellow policy makers, staff and stakeholders (Tseng, 2012; DuMont, 2015; Yanovitzky, 2017; Bogenschneider et al, 2019b).

Research–policy relationships in theory

Over the last few decades, there has been an increasing desire to better integrate research into the policy-making process from both researchers and policy makers (Sorian and Baugh, 2002; Huston, 2008). Despite this desire, as Caplan’s Two-Communities Theory (1979) posits, policy makers and researchers live in two different worlds and, as a result, friction can occur, leading to the festering of miscommunication and distrust between policy makers and researchers (Bogenschneider et al, 2013). However, in recent years, there has been an increase in research–policy bridging models and knowledge brokers who can serve as an intermediary community between researchers and policy makers (Franklin et al, 2019; Bogenschneider, 2020; Bogenschneider and Corbett, 2021).

As a result, the Community Dissonance Theory (Bogenschneider and Corbett, 2011) has emerged, building on the Two Communities Theory (Caplan, 1979). Community Dissonance Theory is novel in two ways: (1) it views the research–policy realm as consisting of three archipelagos (researchers, policy makers, research–policy bridging organisations [that is, knowledge brokers]), while also (2) acknowledging that entities within each archipelago have some unique values and norms, too (Bogenschneider and Corbett, 2011; 2021). Differences in backgrounds and institutional settings can cause each entity (for example, an individual researcher) to have their values and norms shaped in substantially different ways than others, both within and outside their archipelago (Bogenschneider et al, 2019a). Effective strategies for bringing researchers and policy makers together must look to deconstruct the forces behind the disconnect between the two communities. As Community Dissonance Theory describes, this is where relationship building, driven by knowledge brokers who can work across both researchers and policy makers, becomes crucial (Bogenschneider et al, 2019a)

Stepping outside the boundaries of the political sphere, research has shown the importance of knowledge brokers and their relationships, specifically in the education and community engagement fields, in much the same way that Community Dissonance Theory has (Bogenschneider and Corbett, 2011; 2021; Bogenschneider et al, 2019a). In order to reach across the boundaries of different communities (or ‘archipelagos’), there are four potential mechanisms that can serve as bridging points for knowledge brokers: identification, coordination, reflection and transformation (Akkerman and Bakker, 2011). Creating opportunities for relationship-building engagement and collaboration across these points remains a challenge for knowledge brokers (Daniels et al, 2010; Ludvigsen et al, 2010). However, similar to the research–policy sphere, there may be institutional or structural factors that can inhibit knowledge brokers in their work (Weerts and Sandmann, 2010).

Research–policy relationships in practice

In both the state and federal policy-making realms, knowledge brokers and research–policy bridging models aim to overcome distrust and miscommunication by increasing the use of research evidence in various aspects of the policy-making process. Beginning in the 1990s, the Family Impact Seminars model began being implemented with state policy makers who worked on child and family related legislation (Wilcox et al, 2005). Consisting of brief reports, seminar series and follow-up meetings and activities with policy makers, the model has earned a nonpartisan reputation as a model for honest knowledge brokerage between child/family researchers and state legislatures (Bogenschneider et al, 2000; Maton, 2016; Owen and Larson, 2017; Bogenschneider, 2020).

Additionally, at the state level, the SciComm Optimizer for Policy Engagement (SCOPE) has demonstrated an effectiveness in increasing URE by state legislators, specifically with regards to COVID-related research (Long et al, 2021a; Scott et al, 2023). SCOPE is an enhanced dissemination model that consists of the delivery of timely and relevant research in the form of nonpartisan fact sheets or policy briefs to policy makers using continuous quality improvement methodology (Scott et al, 2019).

Primarily implemented at the federal level, the Research-to-Policy Collaboration (RPC) supports policymakers’ URE by understanding their offices’ legislative priorities and connecting offices to researchers who study related issues (Crowley et al, 2018; 2021). In this model, a network of researchers with varying areas of expertise is maintained in order to respond to legislators’ requests for relevant research evidence (Scott et al, 2019). Broadly speaking, researchers in the network have expertise in topics most relevant to child and family wellbeing (for example, foster care, early education, substance use). Some of the researchers are focused on issues that tangentially relate to child and family wellbeing (for example, healthcare) while researchers focusing on unrelated topics (for example, nuclear fusion) are not included in the network.

A key overarching focus of these research–policy bridging models is developing relationships with policy makers and responding to their legislative needs with relevant research (Bogenschneider et al, 2000; Crowley et al, 2018; Scott et al, 2019). Each of these models can be applied to a range of policy areas (for example, child welfare, parental substance use disorder) which allows for policy makers to develop relationships with researchers who are focused on a wide breadth of topics. Recently, these models have started to undergo rigorous evaluation, typically through randomised controlled trials (RCTs), to assess their ability to integrate research into policy making (Crowley et al, 2021; Scott et al, in progress). While relationships between policy makers and researchers remain central to these models, staffer and office-level characteristics are often overlooked when implementing and testing these models due to the nature and structure of the RCT design.

Factors affecting engagement

There are a variety of characteristics, both on the person and office levels, that could impact whether an individual engages with the RPC model. An example of a person-level factor is staffer seniority (that is, job title). As staffers spend more time on Capitol Hill (that is, accrue tenure) and their career dynamics evolve (that is, being promoted), they may be entrusted with more important responsibilities and thus may have less availability to meet with knowledge brokers (Romzek and Utter, 1996), or alternatively may rely more heavily on knowledge brokers to offset their workload. Additionally, given there is detectable evidence of gender inequality among congressional staffers, specifically with regards to expectations regarding leadership roles (that is, previous research has suggested men are typically expected to hold more senior positions like chief of staff; Ritchie and You, 2021; Dittmar, 2023), gender may also affect engagement. Given that women are less often placed in senior positions (Erikson and Josefsson, 2021; Ritchie and You, 2021; Dittmar, 2023), this may result in them having fewer opportunities to engage knowledge brokers. Among other factors, previous research using the SCOPE model has found both gender (female) and office tenure were associated with opening more SCOPE-related emails (Pugel et al, 2021). In addition to gender, office tenure may play a role in an office’s engagement as offices tend to develop preferred information sources the longer they are in office and may be resistant to new sources (that is, knowledge brokers). While the Pugel et al (2021) study was focused on email engagement rather than relationships and used a sample of state legislators, it is important to evaluate the effect of these characteristics at the federal level as well.

In addition to office tenure, there are also characteristics of the office that may influence meeting and developing relationships with researchers and external knowledge brokers. One of these characteristics is office capacity (Bednar, 2022), such that offices with increased capacity logically may have the resources and time to spend on sustaining relationships with external knowledge brokers. Previous research has defined key areas of office capacity as chamber of Congress, political party and seniority (Crosson et al, 2018; 2021; O’Neill et al, under review). Chamber of Congress could have a theoretical influence on offices’ engagement with knowledge brokers due to the differences in office size and workload across chambers. For example, Senate offices typically have larger staffs and take on more work per office than their House of Representative counterparts. Additionally, research suggests there has been a decline in trust of scientific evidence among Republicans (Lee, 2021) which could lead to them being less likely to engage with knowledge brokers. Lastly, seniority could influence engagement levels considering that more junior staff are often tasked with the public-facing duties of an office (for example, meeting with a third-party knowledge broker). Under the umbrella of office capacity, it is important to evaluate the impact of these factors on staffers’ ability to engage with knowledge brokers.

Present study

Considering the lack of allocated funding for research translation (Zurynski et al, 2021), there remains a need to make research translation models more cost-effective. These models can be made more efficient by understanding which staffers and policy makers are most likely to meet with and engage with researchers and knowledge brokers via the models. The present study focuses on personal and office-level characteristics of federal staffers that are associated with these outcomes. If these factors are understood, knowledge brokers can more efficiently target specific offices and staffers for research integration as potential users of research evidence. While the present study focuses on the federal level, federal policy makers themselves are often unavailable to members of the public, knowledge brokers included, which often leaves staffers as the people who interact with the public on behalf of the policy maker (Long et al, 2021b). Therefore, in this sense, both office and staffer-level characteristics are important to understand but staffer characteristics are arguably most important to understand.

Methods

Participants were all federal staffers who had been contacted between September 2021 and August 2022 as part of the optimisation phase of the RPC model (Crowley et al, 2018) evaluation (N = 749). Prior to staffer outreach, their respective offices were identified as those who had previously worked on any legislation that focused on topics related to child and family wellbeing (that is, the focus of the RPC model; for example, paid family leave, healthcare, education). Within the identified offices, these staffers were chosen for outreach because they were the staffers who explicitly covered child and family issues within their office.

Measures

Quorum, a public affairs database that compiles the activities of policy makers and their staffers, was used to retrieve staffer-level and office-level characteristics of the sample as described here (Quorum, 2018). These data were determined to be exempt per the Pennsylvania State University Institutional Review Board (IRB).

Dependent variables

Previous studies that have tracked engagement metrics within an outreach-based intervention have largely focused on voter turnout and metrics such as door-to-door canvassing, phone calls and email messaging (Gerber and Green, 2000; Panagopoulos, 2011; Green et al, 2013; Green and Gerber, 2019). In a similar manner, the present study uses engagement-tracking metrics (that is, initial meetings, number of actionable requests, and number of meetings with researchers and RPC staff) after outreach to congressional offices during the implementation of the RPC model.

Initial meeting.

The RPC contacts federal staffers, asking for an initial meeting that initiates the model implementation. Staffer participation in such a meeting was tracked (yes/no).

Number of actionable requests.

Staffers may or may not have needs upon which the RPC can act (that is, research input while crafting legislation) defined as ‘actionable research requests’. Staffers may produce multiple requests either from the initial meeting or through extended collaborations. The number of requests and subsequent details of each were tracked internally. Preliminary descriptive analysis revealed a heavy positive skew in the numbers of actionable requests (range: 0–3, M = 0.11, SD = 0.37). The present study was more interested in whether or not staffers produced any future actionable requests, not how many times they did so. Combined with the limited distribution of the variable, the present study’s focus suggested it would be best to dichotomise the variable with any staffers who initiated one (N = 58) or more (N = 10) of these requests receiving a value of one. Of the 68 staffers who prompted actionable requests, 35 Democrats prompted one request each, one prompted two requests, and three staffers prompted three requests each. For the 29 Republican staffers, 23 prompted one request each while six staffers each prompted two requests.

Number of meetings with researchers and RPC staff.

The staffer may participate in further RPC-facilitated meetings with topic-relevant researchers. The number of these meetings were also tracked internally. The number of meetings also had a heavy positively skewed distribution (range: 0–2, M = 0.02, SD = 0.16) which resulted in its dichotomisation with any staffers who attended one (N = 15) or more (N = 1) meetings receiving a value of one. Similar to the number of actionable requests variable, the limited distribution of the variable combined with the present study warranting a focus on whether or not they participated in further meetings (as opposed to how many) provided justification for the dichotomisation. Of the 16 staffers who met with RPC staff and researchers, 11 were Democratic and five were Republicans.

Independent variables

Both staffer-level and office-level information was collected for each member of staff. Office-level characteristics included office tenure (that is, how many years since the policy maker took office), chamber of Congress and party. Staffer-level characteristics included seniority (that is, legislative aide versus chief of staff) and assumed gender. The staffer seniority variable was grouped into three values including senior, intermediate and low-level positions. Senior positions included chief of staff and legislative director while intermediate positions included positions similar to legislative aides and low-level positions were those such as office assistants. Gender was originally coded as male, female and unknown, but the preliminary analyses revealed only 1.1 per cent of the sample (n = 8) were coded as unknown. Therefore, only for the models which involved the gender variable, these eight staffers were dropped.

Analysis

The number of actionable research requests and the number of meetings between staffers, researchers and RPC staff were both only available to staffers who had participated in an initial meeting. Therefore, analyses of these variables were restricted to staffers who initially met with the RPC (n = 142). Logistic regressions were used to understand the role of staffer and office characteristics on building and sustaining relationships (that is, taking an initial meeting, initiating an actionable research request and attending an RPC-facilitated meeting with researchers) with congressional staffers. As a robustness check, all models with outcomes of number of actionable requests and number of meetings with researchers and RPC staff were re-run with any staffer having a value greater than 1 (10 and 1 staffers, respectively, for the two outcomes) being dropped from the models and the results held with these staffers removed, further justifying the dichotomisation of the variables (see Supplementary table 1). Additionally, secondary analyses were conducted with samples stratified by political party to explore if any significant associations from the primary models differed by party affiliation. Due to low power, each predictor, both in the overall sample and in the secondary analyses, was run in its own model and Holm’s correction for multiple testing (Holm, 1979) was used for the models with the same outcome variable.

Results

We analysed the associations of personal and office-level characteristics with meetings and productive relationships with federal staffers using logistic regression models. Descriptives of the study sample are available in Table 1. Primary results are presented in Table 2. Secondary analyses, stratified by party, are presented in Table 3. Additionally, Figure 1 shows a visual representation of the study sample broken down by the three outcome variables.

Table 1:

Descriptives of study sample

Characteristic	N (%)
Party
Democrat	358 (47.8%)
Republican	375 (50.1%)
Independent	3 (0.4%)
Progressive New Party	2 (0.3%)
N/A	11 (1.5%)
Gender
Male	376 (50.2%)
Female	364 (48.6%)
Undetermined	8 (1.1%)
Chamber
House of Representatives	576 (76.9%)
Senate	162 (21.6%)
N/A	11 (1.5%)
Staffer seniority
Low-level	360 (48.1%)
Intermediate	375 (50.1%)
Senior	14 (1.8%)
	M (SD) - Years
Office tenure	9.86 (8.57)

Table 2:

Predictors of engagement and sustained relationships

	Dependent variables, OR (95% CI)
	Initial meeting	Number of actionable requests	Meeting with RPC + researchers
Office tenure	0.99 (0.97, 1.01)	0.99 (0.95, 1.03)	0.89 (0.78, 0.98)*
Seniority	2: 3.50 (2.34, 5.34)*** 3: 0.69 (0.04, 3.63)	2: 0.91 (0.43, 1.93) 3: 0.33 (0.002, 6.67)	2: 0.70 (0.24, 2.24) 3: 1.91 (0.01, 41.07)
Chamber	1.10 (0.70, 1.68)	1.42 (0.65, 3.13)	0.74 (0.16, 2.48)
Party	1.51 (1.04, 2.18)*	1.02 (0.53, 2.00)	1.76 (0.60, 5.86)
Gender	1.14 (0.79, 1.64)	1.30 (0.67, 2.54)	2.24 (0.77, 7.44)

Notes:

^*p<0.05,

^**p<0.01,

^***p<0.001; OR = Odds Ratio; Reference groups: staffers in offices with 0 years of tenure, low-level staffers, House of Representative staffers, Republican staffers and male staffers.

Table 3:

Predictors of engagement and sustained relationships stratified by party

Office tenure	Dependent variables, OR (95% CI)
	Initial meeting	Number of actionable requests	Meeting with RPC + researchers
Republican	1.02 (0.99, 1.05)	1.00 (0.95, 1.06)	0.87 (0.66, 1.02)
Democrat	0.97 (0.94, 1.00)*	0.98 (0.93, 1.03)	0.90 (0.77, 1.00)
Seniority
Republican	2: 4.94 (2.56, 10.33)*** 3: 3.70 (0.18, 27.88)	2: 0.49 (0.12, 1.75) 3: 0.20 (0.001, 4.61)	2: 0.29 (0.05, 1.92) 3: 1.27 (0.01, 33.46)
Democrat	2: 2.85 (1.70, 4.87)*** 3: 0.55 (0.004, 5.07)	1.27 (0.49, 3.33)	1.22 (0.32, 5.98)
Chamber
Republican	0.94 (0.47, 1.80)	0.93 (0.26, 3.21)	0.29 (0.01, 2.87)
Democrat	1.29 (0.71, 2.29)	1.89 (0.68, 5.45)	1.17 (0.24, 4.57)
Gender
Republican	0.91 (0.52, 1.60)	1.65 (0.59, 4.77)	7.00 (0.95, 142.08)
Democrat	1.19 (0.72, 1.99)	1.09 (0.45, 2.67)	1.17 (0.32, 4.80)

Notes:

^*p<0.05,

^**p<0.01,

^***p<0.001; OR = Odds Ratio; Reference groups: staffers in offices with 0 years of tenure, low-level staffers, House of Representative staffers, Republican staffers and male staffers.

Figure 1:

Study sample broken down by outcome variables

Note: The bottom row of outcome variables are not mutually exclusive.

Overall staffers results

For initial meetings, the staffer’s rank and party affiliation were predictive, but office tenure (p = 0.55), chamber (p = 0.68) and gender (p = 0.61) were not. Specifically, staffers who held mid-level positions such as a legislative aide were 3.5 times as likely to participate in an initial meeting than lower-level staff such as a Congressional fellow (OR = 3.50, p<0.001). And, compared to Republican staffers, Democratic staffers were 1.5 times as likely to participate in an initial meeting (OR = 1.51, p = 0.03).

Among those who participated in an initial meeting with the RPC, we analysed associations between key characteristics and having prompted any actionable research requests and having met with researchers via the RPC. All independent variables were not significant predictors of having prompted any actionable research requests. Lastly, for every year the staffer’s boss (that is, the policy maker) has been in office, the odds of a staffer partaking in a meeting with researchers via the RPC decreased, changing by a factor of 0.89 (OR = 0.89, p = 0.041).

Republican staffers results

For Republican staffers, the only significant predictor of initially meeting with the RPC was seniority such that mid-level staffers were nearly five times as likely as low-level staffers to meet (OR = 4.94, p<0.001). No independent variables significantly predicted initiating an actionable request nor having attended a meeting with researchers via the RPC for Republican staffers.

Democratic staffers results

Staffer seniority was a significant predictor of initially meeting with the RPC, similarly to Republican staffers, such that mid-level democratic staffers were nearly three times as likely as low-level staffers to meet (OR = 2.85, p<0.001). Additionally, for every year the staffer’s boss has been in office, the staffer was slightly less likely to meet with the RPC initially (OR = 0.97, p=0.03). There were no other significant predictors for the initial meeting nor for having participated in meetings with researchers via the RPC.

Discussion

With the understanding that relationships are crucial to the integration of research into policy making (Bogenschneider et al, 2000; Crowley et al, 2018; Scott et al, 2019), this work sought to understand what factors predict meeting attendance of and sustained relationships with staffers. Mid-level staffers and democratic staffers were more likely to partake in an initial meeting. Additionally, office tenure was associated with a decrease in the likelihood of a staffer meeting with researchers via the RPC. When looking only at Republican staffers, mid-level staffers were more likely to meet with RPC staff and researchers. When looking at Democratic staffers only, mid-level staffers were more likely to attend an initial meeting while those working in longer-tenured offices were less likely to meet for each additional year their boss was in office.

Intermediate-level staffers (for example, legislative aide) were more likely to participate in an initial meeting than more experienced staffers (for example, chief of staff), suggesting it may be more productive to reach out to staffers in more junior positions. Despite the high level of congressional staffer turnover (Salisbury and Shepsle, 1981), this finding suggests it may be important for knowledge brokers to get their foot in the door with staffers who have higher seniority on the hill. Doing so appears even more prudent for reaching out to Democratic staffers in offices who are newer to Congress with Democratic staffers less likely to take the initial meeting for each further year their office is in power.

Additionally, Democratic staffers were more likely to take an initial meeting. It is possible that this finding relates to legislative power rather than political party differences. At the time of data collection, Democratic offices were in control of the White House, Senate and House of Representatives and thus had more control over the legislative agenda. Offices in the minority party (that is, Republicans at the time) introduce less legislation because they recognise their bills are less likely to advance (Cox and Terry, 2008). Recognising their bills may be less likely to pass, they may instead use research to oppose legislation, representing a reactive form of URE. This could help to explain why Democratic staffers, in the majority at the time, were more proactive in engaging with the RPC overall. However, this area (that is, proactive versus reactive engagement) needs ongoing evaluation to determine if these findings were driven by party or majority power.

While political party appears to play a role in developing relationships with knowledge brokers, our findings suggest the chamber of Congress of an office does not. Our findings are convergent with related research suggesting that policy makers’ use of research in social media does not differ by chamber (O’Neill et al, under review). Though engagement did not differ between the chambers, they may have different motivations for working with the RPC, as the chambers of Congress operate different procedurally and offices in these chambers have different needs (Heitshusen, 2020). Perhaps House offices engage because they typically have fewer staffers per office (Vital Statistics on Congress, 2021) and could use external help in compiling relevant research while Senate offices are interested in meeting with knowledge brokers because they have the capacity to do so and want to dig deeper with research. Additionally, staffers in newer offices were more likely to participate in meetings with both knowledge brokers and researchers present, which points to the importance of forming relationships earlier in an office’s tenure as they seek to establish preferred information sources.

Limitations and future directions

Though this study provides important evidence to help bridge the gap between science and policy, there are minor shortcomings of the work to note. Some of the data for this study was curated during the peak of the 2022 midterm election season while a majority of offices, particularly those in the House of Representatives, were dealing with campaign and re-election responsibilities in addition to their day-to-day duties. These offices may have been operating differently than they operate during a majority of their time in office (that is, when not in election season). As a result, staffers may have been too busy to meet with knowledge brokers and/or maintain a sustained relationship.

During data curation, while many of the office and staffer-level variables were downloaded from Quorum (Quorum, 2018), the assumed gender variable had to be created manually. Many of the staffer profiles on Quorum did not have pictures associated with them and, as a result, coders were left to make an assumption of the staffers’ gender. The present study used data and variables from the RPC model (Crowley et al, 2018), but these findings can be used when implementing other models such as SCOPE (Scott et al, in progress) and the Family Impact Seminars model (Bogenschneider et al, 2000). While these findings could lead to meetings and sustained relationships between policy makers and knowledge brokers within other research–policy bridging models, this is not guaranteed considering the differences in implementation and outcome measures between models.

In addition to a measurable increase in the amount of research evidence in legislation, SCOPE measures the open and click rates of emails sent to staffers and policy makers (Scott et al, in progress), while, among other measures, the Family Impact Seminars model measures which offices attend their seminars (Bogenschneider et al, 2000): outcome measures not focused on within the RPC model. Regardless of the different outcome measures for different research–policy bridging models, the present study can help to inform knowledge brokers who work with existing models on what offices and staffers their resources are best focused on. Lastly, while the present study’s sample focused on staffers who worked on child and family related issues, future research should look to see if these results generalise to other legislative topic areas (for example, environmental policy) in addition to other knowledge-brokering models.

Conclusion

This work contributes valuable evidence regarding factors that predict meetings and sustained relationships between federal staffers and knowledge brokers. The findings suggest many office-level (that is, office tenure, political party) and staffer-level (that is, seniority and assumed gender) factors impact the likelihood of initial meetings and sustained relationships occurring between staffers and knowledge brokers. Existing research–policy integration models such as the RPC (Crowley et al, 2018; Scott et al, 2019) and the Family Impact Seminars model (Bogenschneider et al, 2000) can use this research to inform their models as to which offices and staffers they approach. With the end goal of increasing the amount of research into the policy-making process, understanding who the best staffers to approach is a crucial part of the process for knowledge brokers and researchers alike.

Key messages.

• Personal relationships between researchers and policy makers play a key role in effective research–policy bridging efforts.

• Staffer seniority and political party predicted an increased likelihood of federal staffers meeting with knowledge brokers.

• These study findings can inform and help research translation efforts target the most effective audiences.