Developing Measures Using Mixed Methods Posted on January 20, 2009 Dr. Nalini Negi talks about how mixed methods helped her capture data accurately.

So I became involved in working with the organization that does workers' rights issues. So through that, I started visiting the day laborer corners on a nearly weekly basis for a year and a half. And I would talk to the workers. I would do informal interviews.

I did a lot of ethnographic sort of stuff like participant observation and really taking in what the corner really meant to the workers, but also hearing their stories and what they thought about their lives and how they negotiated work within the context of this informal market.

And through that, I really started thinking, "What is the best way to approach research with this population?" There really isn't too much that you can find in the literature about this population. So and because also I come from a participatory sort of world view, I really wanted the research questions to emerge from the ground.

So I took all my ethnographic work that I was engaged in for that year and a half, and then I did a focus group in which I explicitly asked the workers about what they themselves identified to be risks and protective factors for their wellbeing. And then through the identification of those, I included this in a survey.

So I took the themes from the identified risk and protective factors in the focus group as well as my ethnographic work and selected measures that were appropriate to fit. For example, one of the things that came up was social isolation as being a significant risk factor or an identified risk factor. So I included that in my survey, and then I quantitatively assessed whether these things really made sense.

So I had really a three phase sort of study. The first phase being the ethnographic stuff, and then the second phase was the survey. And I administered the survey to 150 participants of which I received 147 that were full. The three that weren't full were because employment came, and they really had to go. And I was really careful to tell the workers if a contractor comes or some sort of private person, employer then — and you have to leave the interview, that's fine, but 147 really did fill it out.

So then I did the quantitative analysis and then went back into the field and did another focus group and asked the participants, "Does this really make sense in your experience?" like a member-checking focus group and, "If it doesn't, why not?"

And then the quantitative results, they could've been on their own, but I really wanted to go back into the field and integrate the voice of the participants and how, if it really made sense or it didn't make sense within the context of their lives. And that really taught me a lot.

For example, discrimination was not significant in the quantitative analysis, but when I asked the workers if they felt that discrimination wasn't or was it significant factor for their wellbeing or substance use and abuse. They were adamant that it was, but what I noticed that it was really a measurement error.

It was a measurement selection in terms of the measurement wasn't really tapping into the concept of discrimination the way that the workers themselves identified discrimination to be like.

The measurement that I selected for discrimination was asking more about their micro experiences of discrimination, and then the way that they were talking about it was much more sophisticated, meaning they were thinking through like policy, specifically local ordinances that were affecting them but then also integrating like meso-level factors not so much very micro- which was exactly what the measures looked at.