Author: Áine M. Humble

Setting up the Project (continued)

Sampling procedure
My entry into the field for this study was quite different from the entry that occurred in my dissertation research. For my dissertation, I actually used a non-purposive sampling approach. Using publicly-available marriage license records from three county offices, I used a random sampling procedure to contact couples, with the intent to find 10 couples from each county for a total of 30. In the end, I interviewed 21 couples. This sampling procedure was not used to locate a sample that I could generalize to the larger population. The marriage license records were a very convenient way to located recently married couples. I didn't have to pay for expensive newspaper ads, something for which, as a graduate student, I was very grateful. More importantly, I wanted to have a sample that varied in terms of gender construction - that is, how men and women negotiated and carried out the work for their weddings. I knew I would be less likely to find such variation if I used a procedure such as snowball sampling. Purposive sampling was not on my mind so much as attaining maximum variation in my sample, and my dissertation advisors did not advise me otherwise. Moreover, I did not use a form of purposive sampling called maximum variation sampling because I could not very well advertise for "couples who thought they planned their wedding together but really did not" - this had to be discovered naturally through the actual interviews. At any rate, since that time I have realized that most qualitative researchers advocate that purposive sampling must be used with qualitative research; in fact, one prominent qualitative researcher told me that my sampling procedure was incorrect. But I digress - this is a topic for another paper that I do hope to write one day!

When it came time to carry out my study on couples who were remarrying, I did not have publicly available marriage records available where I was carrying out this second study, so I turned to using a purposive sampling procedure, as had been recommended in the qualitative research literature. I used word-of-mouth and newspaper ads, looking to interview couples in which at least one person had remarried within the past three years. My application for ethics approval went smoothly, and I was ready to interview couples. Given that I had already done a study in which I had interviewed people on a similar topic, I felt confident about how the interviews would go.

Difficulties with finding participants
The only problem was - where were these couples? A few hiccups occurred along the way. For example, a friend who fit the criteria offered to be interviewed. Turning a feminist lens on a close friend's relationship was not what I was comfortable with, so I turned down her offer as graciously as I could. I interviewed a few couples initially but the responses and inquiries petered out despite a popular columnist's article in the local newspaper. Some women or men expressed initial interest in the project, but eventually declined as those who were divorced sometimes realized they were not as comfortable talking about their first marriage as they thought they might be. Although I assured them that I would only ask one question asking them to describe their first wedding (their wedding, not their marriage) and how they planned for it and that they could refuse to answer any question they did not feel comfortable answering, they were firm in their decisions. Fair enough, I moved on.

I came to realize that it would be more difficult to find respondents than I had previously thought, and I was humbled by this realization. I was collecting data in a different country (Canada) from my dissertation study (the US), and I was very much aware that Canada had a significantly smaller population compared to the US. Moreover, I realized I was in a region of Canada (the Atlantic provinces) that had a smaller population compared to other parts of the country. And even though I lived in an urban area (Halifax, Nova Scotia), it was not a large city and my province had a considerable rural population. Thus, I knew I had a smaller regional population from which to find my sample. This was important to be aware of, as I was interviewing people in person and I had a small travel budget.

I knew, too, that the population of divorced individuals remarrying would be smaller than my previous population of couples marrying for the first time, but I had not thought carefully enough about this. Canada has a current divorce rate of approximately 34%, with about 70% of divorced men and 58% of divorced women remarrying (Ambert, 2005a [PDF- 209 KB]). I underestimated the number of divorced individuals who had reentered an intimate relationship in the form of cohabitation rather than remarriage. Thus, I heard of many divorced individuals who were back in committed relationships and living with their partners but not married. In Canada, although cohabitation is more common with younger adults (typically ones who have not yet married) than with older cohabiters, older cohabiters are more likely to be divorced than never-married (Ambert, 2005b [PDF- 2 MB]). If individuals had not had the formal ritual of a wedding, which was what I was studying, I could not interview them.

I took a break from data collection, regrouped, and tried again at a later date, this time placing newspaper ads in several newspapers (I spent much more on promotion of the study than original planned; I had greatly underestimated promotional costs - an important lesson to take in the future). The second time around, I received a much better response to the ads (note: I did not use snowball sampling because it was still important to have a sample that varied in terms of gender construction). In the end I interviewed 14 couples - a little lower than I would have preferred, but sufficient for my purpose.

Back to Project Home Page