AMES Australia provides a comprehensive range of settlement services for recently arrived refugees and migrants to help them settle in to Victoria. These include English language, vocational training and employment services. The AMES Australia client base is made up people from non-English speaking backgrounds with most having low to intermediate levels of English. When undertaking any research or evaluation involving current or former clients, there is a need to determine whether surveys or interviews will be conducted in English or first language. This paper will detail the methodology used for an outcome evaluation of education and training programs conducted during 2015/16. It is the intention that the findings from this research will generate an evidence base around the efficacy of these programs and to identify areas for continuous improvement. This evaluation followed up a select cohort of clients who had undertaken educational and training programs six months after the course was completed. Bilingual research assistants (RAs) were employed to conduct telephone interviews in first language and English. RAs were allocated clients with the same first language and from different cultural backgrounds. When a language was not available among the group of RAs, the survey was conducted in English. The survey was not directly translated into community languages; there was an expectation that the RAs translate each question while administering the survey. In some instances RAs explained key concepts relating to questions in first language. The participant chose to answer in English or first language. All answers were recorded in English. The lead researchers were cognisant of the methodological, epistemological and ontological challenges arising from translation across languages. There was a certain level of trust, that the bilingual RAs could capture the cultural context and meaning. When the translator (RA) and lead researchers are different people the process of knowledge construction involves another layer and has a significant bearing on the data collection outcomes. The challenge for the researchers was to not only use the evidence generated by the quantitative data collected, but to use the RAs as key informants to improve survey questions, explore how to increase the response rate and to identify some of the complexities of administering a survey to clients in first language. This paper will address the benefits and challenges of using bilingual research assistants and the consequence this has on the validity of the research is terms of how it is produced and received.