Empirical research on how to effectively recruit and retain staff may help organisations to operate more effectively. Such research may also help Animal Advocacy Careers to offer more useful services to the animal advocacy movement. Accordingly, Google Scholar searches were conducted to identify existing reviews and meta-analyses of academic research on recruitment and retention. In total, 52 relevant research items were reviewed and included. Promising actions to improve recruitment outcomes include the use of structured interviews, general mental ability, and conscientiousness to select candidates, improvements to the usability and aesthetics of websites and job ads, and being personable and informative to candidates. Promising actions to improve the retention of staff include the provision of socialisation, education, and support (e.g. mentoring) for new staff and focusing on employee commitment rather than control. Some of the other identified actions seem likely to have positive effects on recruitment and retention but, given the associated costs, may still not be worthwhile; salary increases probably fit into this category.
Animal Advocacy Careers (AAC) is an organisation that seeks to address the career and talent bottlenecks in the animal advocacy movement, especially the farmed animal movement. The services that AAC could plausibly provide to the movement can be thought of as falling into the following categories:
AAC has already begun offering a service aimed primarily at recruitment into the movement; our introductory online course and workshop. A clearer understanding of the existing academic research into the effectiveness of various interventions for recruitment and retention could therefore help AAC to improve its services. It could also help AAC prioritise between recruitment and retention or the other categories of possible services.
Of course, animal advocacy organisations and other organisations seeking to do good in the world are likely already grappling with questions about how to recruit and retain high-quality staff. A clearer understanding of the existing academic research on recruitment and retention may also help those organisations in their efforts.
To address these needs, Animal Advocacy Careers undertook a brief overview of recruitment and retention research.
The research questions were:
1) Which actions can be taken that seem likely to increase the quality of recruited staff to organisations?
a) Which actions can be taken that seem likely to improve the outputs and results of hired candidates?
b) Which actions can be taken that seem likely to increase the number of applicants?
2) Which actions can be taken that seem likely to increase employee retention?
3) What are the predictors of employee turnover and retention (and what does this imply about which actions can be taken)?
Searches of Google Scholar were conducted. The first 5-20 pages of results were skimmed or reviewed for each search term. There were not strict inclusion and exclusion criteria and the likely relevance of research was assessed predominantly by the phrasing of the title.
We focused on existing reviews and meta-analyses rather than on identifying all relevant individual studies, so the interventions identified in the results section below are not exhaustive.
The interventions in the results tables below are ordered in terms of effect size. Brief intuitive comments on possible cost-effectiveness are provided. The results section below uses the acronym DEI (diversity, equity, and inclusion).
The summaries below are intended to be easily readable and intuitive. If you’d like to dive further into the methodology, see the appendix below.
52 research items were reviewed and included in the results spreadsheet. The following caveats about the identified research should be borne in mind:
1a: Which actions can be taken that seem likely to improve the outputs and results of hired candidates?
The reviewed research suggests a number of actions can be taken to improve the outputs and results of hired candidates.
1b: Which actions can be taken that seem likely to increase the number of applicants?
The reviewed research suggests a number of actions can be taken to increase the number of applicants. Of course, this is only useful insofar as it improves the outputs and results of hired candidates. However, some research focuses on applicant numbers rather than job performance. All else being equal, having more applicants to choose from should, on average, lead to higher quality hires.
2: Which actions can be taken that seem likely to increase employee retention?
The fairly extensive research on the predictors of employee turnover (see the section below) does not seem to have encouraged much research on the effectiveness of particular actions to reduce turnover. Hence, the table in this section should be supplemented by the findings in the section below.
3: What are the predictors of employee turnover and retention?
Several relevant meta-analyses were identified for this question, though one (Rubenstein et al. 2018) was notably more recent and comprehensive than the others.
Although Rubenstein et al.’s (2018) focus was on predictors of turnover, rather than interventions to reduce turnover, we can draw inferences about interventions that seem likely to be effective at reducing turnover.
It may be possible to reduce turnover by looking out for the following traits, abilities, and characteristics during hiring processes:
Though these are not likely to be quick and easy to implement, organisational changes that may reduce turnover include:
Limitations and suggestions for further research
Appendix: Results spreadsheet and scoring system
If you have a strong interest in this research, we encourage you to read the summaries on the full results spreadsheet.
To generate the results tables above, a scoring system was used, which is reflected in the “Scores by intervention, question, and item” and “Summary scores by intervention and question” tabs of the results spreadsheet.
Each research item was assigned a “naive score” for each question that it provided evidence for. The scores were given on a possible range from -5 to +5. These rankings are based on our interpretation of the information and evidence provided, rather than the ranking that we believe that the author of a research item would choose.
Each research item was also assigned a “strength of evidence” (SoE) multiplier for each question that it provided evidence for. The possible SoE multipliers range from 0 to 1, where 0 means that there is no relevant evidence or arguments and 1 means that there is very strong evidence. This was used to create a “question weighted score,” which can be interpreted roughly as the “naive score” is for individual research items, e.g. 1 means “means very low positive impacts” and 5 means “very high positive impacts.”
The full list of references is available in the second tab of the results spreadsheet.
 The search terms used were:
Several variations of these search terms were also used, such as adding “AND (“human resources” OR business OR corporate OR profit)” and removing “AND (“meta analysis” OR “systematic review”).”
 The following criteria were used to decide which items to include:
We did not exclude all items that failed to meet some of the inclusion criteria, if they seemed to perform especially well on others. These inclusion criteria were pre-planned.
 See here.
 One approach would be to qualitatively seek to hire clever individuals. More systematically, organisations could use assessments as part of an application process; Le et al. (2007) noted that “GMA is most commonly measured in job candidates by assessments such as the Wonderlic Personnel Test, the Wesman Personnel Classification Test, or the Watson-Glaser Critical Thinking Appraisal Form.” An alternative is to use candidates’ GPA scores, which, as Imose and Barber (2015) discuss, are a proxy for general mental ability and conscientiousness. See also 80,000 Hours’ discussion of GMA.
 This is discussed to some extent in Le et al. (2007) and Imose and Barber (2015).
 This would probably just mean qualitatively seeking to hire hard-working individuals, e.g. placing a lot of weight on evidence that they have done or achieved lots of things. Administering personality tests in a job interview seems cumbersome.
 See here.
 Nikolaou (2014) found that “job seekers still seem to use job boards more extensively” than social media. Since this sort of comparison is not well-suited to meta-analysis, it is likely that there is other evidence available on this question that was not identified by this brief overview.
 The most recent identified meta-analysis was from 1985 and several more specific reviews reporting to focus on interventions to reduce turnover seemed to base their recommendations primarily on the research on the predictors of turnover.
 If you have a strong interest in turnover and retention, we recommend you read the full paper, or at least our summary in the results spreadsheet.
 After examining a single study on volunteer management practices’ impact on retention and a single study on volunteer recruitment, it seemed that recruitment and retention issues relating specifically to volunteers are substantially separate and should not be integrated into this review.
 -5 means that if this was the only relevant evidence on this issue, we would expect this to have strong negative impacts on recruitment or retention, 0 means that we would expect it to have no impact on recruitment and retention (i.e. useless but not harmful), 1 means very low positive impacts, 2 means quite low positive impacts, 3 means moderate positive impacts, 4 means quite high positive impacts, 5 means very high positive impacts.
 The multiplier was used roughly as follows:
0 = no relevant evidence or arguments (in practice, 0 will never be used, since a paper would not be included if it provided no relevant evidence or arguments).
0.2 = a plausible theoretical argument, with no supporting empirical evidence.
0.4 = a highly convincing and important theoretical argument, or an argument with some supporting evidence, such as from a single observational study, from multiple studies with weak relevance, or from multiple studies where the findings are somewhat contradictory.
0.6 = a highly convincing and important argument with some supporting evidence or a plausible argument with moderate evidence from multiple studies.
0.8 = a highly convincing and important argument with moderate evidence from multiple studies or a plausible argument with strong evidence from multiple studies.
1 = a highly convincing and important argument with strong evidence from multiple studies.
 For each item, the naive scores were multiplied by the SoE multiplier, to generate a “weighted score.” This metric is of little interest for individual research items, since, for example, items with a naive score of 1 and a SoE multiplier of 5 would have the same weighted score as items with a naive score of 5 and a SoE multiplier of 1, even though such items have opposite implications for the question. Hence, the relevant columns in the “Findings tables” spreadsheet are hidden. When the sum of the weighted scores is divided by the sum of the SoE multipliers, this creates the “question weighted score.”