One task of a longitudinal project that we think we have executed fairly well is maintaining a sufficient sample across the phases. Of the original 168 couples, we were able to confirm the marital status of 164 of them in 1994. Of the 105 couples who were still married at that time, 102 of the wives and 100 of the husbands agreed to participate again in at least some part of the project. Based on our experiences, we think there are two equally important aspects to maintaining a sample over time: the first is gaining and maintaining commitment to the project, and the second involves tracking people down.

Gaining and Maintaining Commitment to the Project
In the first three years    In more recent years

Searching for the Participants
  Keeping up with addresses    Accounting for information
Starting a notebook     The actual search 

Gaining and Maintaining Commitment to the Project

One of the reasons the PAIR Project has been successful is the surprisingly high level of commitment that many of the participants seem to have to the project. This level of commitment was striking to many of the research assistants who have worked on the project in more recent years. Obviously, this is due largely to the relationships that the earlier project team established with the PAIR couples.

One example of this commitment is evident in what often occurred when we tried to contact the couples for the most recent phase of data collection. Many couples had moved since the initial phases of the project.
We frequently had their parents’ phone numbers, however. When we called the parents, the response was often surprising. Considering all the "surveys" and sales that occur over the phone, we thought that many of the
parents would be reluctant to reveal their children’s phone numbers to strangers. Occasionally, some parents were reluctant, but it was much more common for them to remember the PAIR Project and tell us something like "Oh, they’ll definitely want to talk to you." We think this says something about the role the Project played in the lives of the participants. Apparently, the participants had talked enough about the Project that their parents remembered that their children had participated in it, even though the previous interviews were up to a decade earlier. Furthermore, what the participants said about the PAIR Project was apparently positive enough that the parents felt comfortable telling us where their children were.

Since we found the relatively high commitment to the project to be so helpful in the later stages of the project, it makes sense to consider what was done earlier in the project to instill such commitment.

Establishing rapport

While efforts at establishing rapport were initially aimed at procuring valid data, some of the procedures were designed to increase commitment on the part of the participants.

Effects of going to the couple’s homes. The initial interviews took place in a setting chosen by the couples, most often their own homes. Although not intended as a rapport building tactic,meeting in the couples’ homes did seem to increase trust and openness from participants. Not only were the home meetings more personal, but in many cases this was treated by the participants as "entertaining guests." For several couples, the PAIR Project interviewers were the first guests they had in their homes after the wedding. Some couples even baked cakes or cookies for the event.

Details of the interview. There were also several details of the interview that possibly helped establish initial rapport:

	Interviewers were sure to thank all the participants before and after the interviews.
	The participants were assured that their own perspective was important. For example, each couple was told, "Our questions are guided by an intense interest in trying to find out as much as possible about each of you as individuals and about your relationship from the time you first met until you eventually married."
	The participants were assured that we wanted them to be honest and thoughtful. For example, before the interview, all participants were told, "Please feel free at any time to ask us about any aspect of the study as we go along. It is very important to us that you understand all of the questions and that you do the best you can to answer them honestly. Some of the questions are personal in nature. We would like you to answer every question, if possible, but we will respect your right to decline to answer any question as you wish." The interviewers were told not to pressure participants to respond to questions they did not want to answer. The notion was that we would rather have no answer than a deceptive answer.
	Interviewers were trained to not be judgmental. They were instructed not to reveal their own beliefs, and if a participant asked them questions like, "Is that normal?" they were instructed to turn those questions back to the participants with answers like, "Well, that is what we’re trying to find out."
	The primary sections of the interviews were conducted separately for the husbands and wives in separate rooms. This allowed each participant to make a connection with an interviewer. If the interview was conducted with the spouses together, there would be a chance that one spouse would dominate the interview, and the other person may feel that his or her perspective was not heard.
	There were several open-ended sections to the interviews. This seemed to serve the function of getting participants to "open up" to interviewers. This also prevented participants from feeling that we did not ask the most important questions. For example, one section of the interview prompted the participants to tell us about other things we should know about that we had not asked about.

Since the interviews were spaced a year apart in the first three phases, efforts were taken to "keep in touch" with the participants between interviews. Birthday cards and anniversary cards were sent to the participants. We also sent out newsletters to keep the couples updated on our progress and interested in the project. 

Besides just maintaining contact with the couples, the newsletters served as a way for the couples to identify with the other couples and with the PAIR Project itself. The newsletters served to promote identification in at least two ways. First, the newsletters helped them see that they were indeed an important part of the larger group. One newsletter, for example, included every baby name in the sample so that couples could see their child’s name. Second, the newsletters also allowed couples to stay informed about the other couples. One woman told us that she likes to "hear how things are going with the other couples."

One challenge to doing a newsletter is that the participants are interested in "results" of the project, but there is a possibility that some findings may affect the way they respond to future interviews. Thus, stories in the newsletters need to be written to be informative without including any details that may compromise key questions for the project. For example, one story, entitled "Living Together Before Marriage" in the 1991 newsletter described the effects of premarital cohabitation. This story stated:

Other types of information that have appeared in the newsletters include:

	"Where we now stand" information about what phases of the interviews had been completed, when the couples could expect to hear from us again, the information about the project moving from Penn State to Texas, etc.
	Descriptions of the sample — e.g., an early newsletter had information about the religious affiliations of the participants, which counties the participants resided in, and the percentages of the couples who had premarital sex. A more recent newsletter showed all the places that participants had moved to, number of children born to the PAIR project participants (both in their marriage to their original partner and in any subsequent relationships), and the number of couples that had divorced each year.
	Benefits of the project — each newsletter has included a small bit of "propaganda" about how important the information provided by respondents is. This section typically mentions articles in the popular press that have used the PAIR Project data, and one newsletter even printed a section from such an article out of the New York Times.  In the most recent newsletter, we included some quotes from participants who had a favorable impression of the project. For example, we made sure that everybody knew that one person thought the project was so interesting that he wished we could call more often. The key to this "propaganda" being successful is that is must be true. If somebody thought we were making up information to manipulate them to participate, it would certainly make them resist participating.
	Where are you now? — Each newsletter requested updated information on participants that had moved. This helped with tracking people down because we were often notified of moves.

Continuing successful activities

Many of the practices from the early part of the project seem to have been so successful that we have continued them. For example, a newsletter was produced just before the most recent data collection, and another is in the works.

New choices for compensation

In the past, all couples received money for compensation. In the last phase, for couples who chose to accept slightly less money, we offered two new compensation choices — t-shirts and mugs with the PAIR Project logo on them. Although it’s difficult to measure how much this would add to commitment, the basic notion behind this (besides trying to be nice) is that the more choices people have about something, the more they tend to be committed toward it. We also thought that, among people who selected the t-shirts and mugs, having those artifacts would further increase their commitment. A person who wears a PAIR Project t-shirt in public is likely to be a person who will continue to participate.

Our search process was conducted by a number of people throughout the last several years. Ideally, one person could be assigned the role of finding the couples because finding particular participants gets complicated enough that it is difficult for somebody to "pick up" leads from a person who has been previously working on the case. Perhaps, if one person cannot do all the searching, a good compromise would be to assign particular participants to individual research assistants. We have found that searching by committee often results in repeating unfruitful lines of inquiry, and it is difficult for a new person taking over a case from somebody else to determine the next logical step. Nevertheless, regardless of whether one person is responsible for cases or whether a committee is responsible, the procedures described below for searching and documenting progress have
been fairly successful at finding participants.

Keep up with addresses as much as possible

For example, our newsletters encouraged people to notify the PAIR Project in case of a move. Obviously, not everyone is going to send notification, but for every participant who does, it helps with later search efforts.

When it is time to begin a new phase of data collection, the first step is to take account of what information you do and do not have about the couples. This step was begun in 1991 for the PAIR Project. We used a form for couples to send back to us. This was mailed to the last known address, and obviously, for those who still lived there, we were able to confirm that our information was current. The form was forwarded to a few people, and we were able to get new information that way.

Also, the post office can arrange to notify you if a piece of mail has been forwarded and can provide you with the new address. Talk with your local post office for details on how to do this.

This is critical because people need to be able to look at the information that is available about the participants, see what has already been done, and figure out what might be a good next step. We used a "couple tracking data sheet" and a "search summary sheet."

This was mostly intended as a summary of the information we had about each couple for whom we were looking. It included the county where the couple was wed, their marriage license number, the spouses' parents' names, the spouses' drivers' license numbers and social security numbers, the last known address of the couple, and the phone number of other persons contacted and their relationship to the couple.

Search Summary Sheet

The key here is to record who has been called and what was discovered. This should act as a paper trail so that anybody can follow the logic of the search. Often, it’s important to document things that do not seem important. For example, in our search for one couple we thought we had the phone number of one participant’s parents. Although we called this number many times, we did not get an answer. Fortunately, we had been recording the calls to that person; eventually we noticed that we were calling around the same time during the day. We were able to contact that person by calling in the evening. It’s easy to fall into a pattern of calling during particular times because schedules tend to be somewhat fixed from week to week.

It’s also important to keep track of both phone calls and mailings in the same place. (Sometimes one gets a possible address but not a phone number.)

A big predictor of whether participants can be found later is the type of information you have about them. The types of information listed below are things you should certainly ask for in the first interview. However, do not be afraid to collect redundant information. For instance, we had all the information from the marriage licenses from each couple, but we also had the wedding announcements (when there was one) from the local newspapers. This seemingly redundant information from the wedding announcement was the clue that led us to find one woman who had gotten divorced. We initially had no luck finding this woman, but then one research assistant noticed that the first name on the wedding announcement in her file was different from the name on the marriage license. Apparently, this woman had changed both her first and last names when she got married. When she divorced, she changed her name back to her premarital name. Once we attempted to look for her under her former name  (which was listed in the wedding announcement), we found her easily.  Other important information to collect is:

	Name and Social Security Number (SSN). Credit bureaus can find most people with this information. Without SSN, most names are too common to be used by credit agencies.
	Other contacts. Other people who will know where the participants are in the future. It’s important to try to do this with each spouse separately so that you can get some information from both sides of the family (or from friends of each of them). Not only does this double your chances of finding a couple through another source, but it is critical if the couple divorces.

The search process

Before the search process, we attempted to create a flow chart that would depict the set of decisions one ought to make depending on what information we knew about the couples and what sources of information had already been consulted.  The flow chart didn’t work because too many cases are unique. For example, two couples were found through the Army and one through the Florida Health and Rehabilitation Services. Which one of these should we put first on the grand flow chart?

Instead, it seems best to begin with a series of resources. By checking couples through these resources, it often led to other clues that we could follow. Thus, each couple is treated as a unique case — more like an investigation than a rote set of steps.

Although the rigidity of the process shouldn’t be exaggerated, there was a logic to how we proceeded. This logic is dictated by what we thought would remove the most couples from the search pile (i.e., we began with sources that we thought would be the most general, revealing information about the greatest number of people).

Typically, the first source was the "other contacts" whom each participant had named as people who would know where the participant would be in the future.Although this helped us find many couples, it was not foolproof. Often, the people the participants told us to contact had moved themselves, for example.

After this initial search, one can turn to large national sources or to local sources. If we had had the SSN’s of the participants, it probably would have made sense to contact TRW and/or other national sources first. In all likelihood, since most people have at least some credit history, this would have allowed us to find most of the couples easily. However, since we didn’t have that information, we had to begin with more local sources.

The first thing we did was order the current telephone books from the counties involved in the original data collection phase. We found several couples that way.  Even without SSN’s, one of the local credit bureaus was able to find a few couples for us based on their names and their last known addresses. They were also able to find SSN’s on some couples. You can find local credit bureaus under "credit reporting agencies" in the Yellow Pages. County court houses were also helpful  — we were able to find out that several couples had divorced and also got some current addresses from county records. (People’s addresses at the time of the divorce are listed as part of the divorce decree papers.)

With the local credit bureaus and county courthouse workers, it was important to establish rapport (just like you would with a participant). One of our research assistants, Sabrina, got to know a particular person at a local agency. Sabrina kept dealing with the same person and always wrote thank you notes for the help she received. When this person was contact a year later by another RA, that person said, "Oh, you must know Sabrina," and she gave some information to us for free!  If you have such a rapport with a local agency, you may wish to use them rather than a national credit agency because the local bureaus typically get the same information from the national agencies.

[ National Resources ]
This page created & maintained by Shanna Smith, ella@utxsvs.cc.utexas.edu