Qualitative Examination of Satisfaction with Three Expert System Interventions to Reduce Cancer

Expert Systems are computerized population programs that can provide tailored interventions for behavior changes. These systems have been used in various population samples throughout the United States, however, no one has qualitatively examined the experiences of participants. In this research, participants had three cancer risks (i.e. poor diet, sedentary lifestyle and smoking) and were provided interventions in one of three types of Expert Systems (i.e. Telecommunications , Modular or Integrated). The experiences and satisfaction of 56 participants across the United States using these Expert Systems were examined, with special attention given to demographic differences. Qualitative methodologies were employed to design and administer structured telephone interviews. Data were transcribed and analyzed using the qualitative management program, NVivo 7 and complimentary quantitative data were analyzed using SPSS. Eight themes were drawn from the data representing participants' experiences including: Reasons to Participate , Expectations, Likes, Style, Reaction to Feedback, Trust, Satisfaction and Suggestions. While participant data revealed pros and cons of participating in each Expert System, the Integrated group displayed greater levels of behavior change and higher rates of satisfaction. This information not only provides evidence of the positive experiences of participants in the Integrated Expert System, but helpful suggestions in making the other Systems more appealing to future participants. It is hoped these data and interpretations will be valued and utilized for improving Expert systems for behavior change in the future. Acknowledgements I would to acknowledge the significant support provided for this research by the National Cancer Institutes' (NCI) Research Supplements for Underrepresented Minorities (PA-01-079). I would also like to acknowledge the support of Wayne F. Velicer, Ph.D. for providing numerous opportunities to pursue my research interests; Mark L. Robbins, Ph.D. , my major professor for his mentorship in my academic and career developments; Ginette G. Ferszt , Ph.D. for her guidance in qualitative methods and the Cancer Prevention Research Centers' Survey Center staff for their accommodations in this research. Finally , I would to acknowledge the unconditional support of my family and dedicated colleagues throughout this journey.


List of
Since there is limited data in the literature describing participants' experiences with these Expert Systems ( e.g. what made it easy or difficult to utilize, what aspects of the experience contributed to a sense of satisfaction). An in-depth examination using qualitative methods to obtain a better understanding of participant experiences was warranted. This examination can ultimately provide valuable information regarding the future use of Expert Systems.

Cancer Prevention
According to the Centers for Disease Control and Prevention, the top ten cancer sites include: prostate, female breast , lung and bronchus, colon and rectum , urinary bladder, non-Hodgkin lymphoma , melanomas of the skin , kidney and renal pelvis and ovary (2005). The majority of these cancers can be prevented (Shibuya, et al., 2002) by minimizing unhealthy behaviors such as smoking, eating fatty foods, leading a sedentary lifestyle and access UV exposure.
The field of cancer prevention has progressed from individual (i.e. one-on-one physician interventions, family interventions) to population-based interventions (i.e . community interventions, research). This evolution was based in part on evidence indicating behaviors that increase risk (i.e. socio-cultural , economic and environmental) for diseases are not merely individual , but exist in entire populations (Bernstein et al., 2002;Pienta & Esper , 1993;Ward, et al., 2004) . Therefore, population -based preventions efforts require that interventions reach significant percentages of populations at risk (Fendrick et al., 1999;Janz et al., 2003 ;Prochaska , et al, 2005).
Improvements in a variety of technologies have made the dissemination of population-based interventions more feasible. Population cancer prevention strategies now include , but are not limited to, telephone, mail and computer-based technologies.
Computer-based technologies allow providers to administer assessments and interventions while maintaining high fidelity to theory or content , which is often difficult for health care providers to deliver with consistency and accuracy. In addition , these types of interventions can be accessed by individuals from practically anywhere they have internet access. Research suggests population -based interventions have been helpful in reducing rates of cancer overall , but disparities still exist between Whites and ethnic minorities (Edwards, et al., 2005).
Health Disparities. As population -based cancer prevention programs were made more readily available in the last twenty years, there was an increase in attention to health disparitie s. According to Braveman (2006) , a health disparity is defined as: ... a particular type of difference in health or in the most important influences on health that could potentially be shaped by policies, it is a difference in which disadvantaged social groups systematically experience worse health or greater risks than more advantaged groups.
2 Disparities in cancers between Whites and Blacks include differences in: risks of getting cancers, prolonged period prior to treatments , stage of diseases at diagnosis and poorer survival rates with Blacks suffering considerably more than Whites (Wong, et al., 2009) even when all intuitive demographic variables are equivalent (i.e. education , age, gender). There are several hypotheses as to why these disparities exist including: style of interaction of the provider, utilization of services, satisfaction with previou s services, access to services, cost of services, and perceived trust in health care provider (Alesina, A & Ferrera E.L., 2000;Corbie-Smith et al., 2002;Doescher , M.P. et al., 2000 ;Hulka, et al., 1975;Johnson & Nies , 2005;Mutchler , J.E. & Burr, J.A., 1991;Richman et al., 2007}. It is important to continue to examine the sources of health disparities , especially where the largest gaps exist.

Expert Systems
In recent years, remarkable improvements have been made in the use of computer technologies, some of which are Expert Systems. Expert Systems are sophisticated computer programs that mimic reasoning and problem solving of human experts with more consistency and accuracy. The decision making of the Expert System utilizes a combination of empirical data and a theoretical framework for intervention purposes . The Expert System is just as effective at providing an intervention and sometimes more so than a human expert. Expert System research has revealed successful tailored interventions for individuals whom endorse a variety of at risk behaviors including : smoking , diet, exercise , diabetes management , UV protection , alcohol consumption , condom use, and mammography screening (Evers et. al, 2006;Johnson et al., 2006 ;Jones , H. et al, 2003;McCabe , S.E., 2006;Prochaska et.al. , 2001;Velicer et. al., 1993;Velicer & Prochaska, 1999).
Expert Systems for behavior change can operate in variety of ways. In general, Expert Systems include some type of assessment and feedback to the individual that is guided by decision rules codified in a computer program. An individual is assessed for a particular behavior such as smoking. This assessment can include (but is not limited to) frequency of behavior, context of behavior, and willingness to change behavior. This information is analyzed based on a theoretical framework (i.e. Transtheoretical Model).
The information is then reprioritized (i.e. which behavior is easiest to change or which behavior will have the largest impact on overall health) and an intervention with feedback is delivered to the individual. This feedback can be tailored both theoretically and empirically and can also include suggestions or ideas for the individual to modify the identified behavior(s). Expert Systems interventions for behavior change are provided through a variety of mediums including print materials, automated feedback via telephone , and internet interventions via computer.
Expert System interventions can potentially reach more individuals than healthcare professionals, are comparatively cost effective, and can work with a variety of behaviors simultaneously . Due to the promising benefits of Expert Systems , there has been an increase of research in the area.

Transtheoretical Model (]TM) . Expert Systems for behavior change at the Cancer
Prevention Research Center are based on the theoretical framework of the Transtheoretical Model (Prochaska , et al, 1992;Prochaska, et al, 1994). The core construct of the Model is Stages of Change. There are five Stages of Change in this model that categorize an individual 's readiness to change including: Precontemplation-no intention to change, Contemplation-thinking about change, Preparation-intentions to change, Action-actively engaged in change and Maintenance-maintaining change for a prolonged period of time with relapse prevention. Outcome variables in Expert Systems for behavior change include: decisional balance (pros and cons of change), self-efficacy ( confidence to change), temptations and behavior specific concerns. Expert System intervention s then attempt to encourage participant s' use of change Processes that will lead to progression through the stages toward successful behavior change and maintenance of that change.

Project HEALTH: Computerized Population Programs for Three Cancer Risks1
Project HEALTH was a population-based cancer prevention program administered from the Cancer Prevention Research Center at the University of Rhode Island. The major objectives of Project HEALTH were to implement and assess Expert System interventions on populations of individuals with multiple risk factors for cancer.
Project HEAL TH provided one of three Expert System intervention s (i.e. Telecommunications, Modular and a newly developed Integrated intervention) for participants that were at concurrent risk 2 for cancer by: smoking, having poor diets and leading sedentary lifestyles3. Participants in Project HEAL TH represented a select sample (only 10 to 12%) of the total United States population as they engaged in three known behaviors that put individuals at risk for cancer. Because this total sample is small, Expert System was different in the delivery method , organization and format of the feedback that was provided . 4 The control group did not receive an intervention and therefore will not be discu ssed here.

6
Participants in Telecommunications group were invited to call an automated system to receive interventions on the three targeted behaviors . The feedback provided to participants during the call included : their stages of change, pros and cons of behavior change, strategies for change (i.e. overcoming temptations , getting support, making a commitment) and a summary for each unhealthy behavior. On average, each call took 20 minutes to complete. Participants could call into the system on a weekly basis and receive additional information on the targeted behaviors. The system was designed so that the overall feedback was not able to be delivered to a participant in one telephone call ( on average, it took 5 to 6 telephone calls to deliver the completed feedback) . The perceived benefit of this Expert System is that the intervention simulated a human conversation, which may appeal to the general participant.
The Modular Expert System consisted of printed reports that were mailed to participants and provided tailored feedback on each of the three unhealthy behaviors .
The printed feedback included detailed information on: their stages of change , pros and cons of behavior change, strategies for change (i.e. overcoming temptations, getting support , making a commitment) and a summary for each unhealthy behavior. A total of 9 reports (3 mailed packets including feedback on 3 behaviors) were delivered to participants over the course of 12 months. expectations, experience) and that the response occurred after the service or product was consumed. Hence , this newer theory has face validity in that incorporates the commonalities of most theories of satisfaction.
While most theories contain these components , the way in which satisfaction is defined depends on the field of interest. For instance, in the marketing field definitions of satisfaction are often product-oriented . For example, satisfaction is the sense that the product met an individual's needs. In the social sciences, more emphasis is placed on the experiences of the consumer in the definition. Thus, satisfaction is the positive experience an individual had while using a service. While the theoretical underpinnings are similar, definitions remain divergent, which alludes to the complicated nature of assessments to study satisfaction.
Literature suggests at least two overarching components of satisfaction. Some researchers have found that satisfaction it directly linked to outcome measures (Ennew, et al., 1999;Spreng, et al, 1996). Outcome measures can vary also depending on the field of research. These can include , but are not limited to: stages of change, timeliness of service , if the expected product was provided. Other researchers found that satisfaction is directly correlated with the perception of value/quality placed on the service or product (F om ell et al., 1992;Hallowell, 1996). It appears both of these ideas are valuable components to consider when examining satisfaction.
Given the inconsistencies in definitions, how reliable are satisfaction data? Sitzia ( 1999) found of the 195 studies that examined satisfaction, the majority displayed little evidence of reliability or validity . Avis et al. (1995) argues the construct of satisfaction is not grounded in the values and experiences of the consumer. This dismissal creates an imbalance of power as consumers are considered less in planning and evaluation of services. Concluding, not only are the definitions of satisfaction inconsistent , but measures to assess satisfaction can often be weak as they fail to incorporate the experiences or values of the consumer.
Given the limitations, satisfaction is an important concept to examine as it is often used in a variety of ways such as for professionals to guide programs or for consumers to choose products . It is important to understand the concept from a consumer perspective.
In considering components of satisfaction, McKinley et al. (1997) facilitated focus groups to develop a patient satisfaction questionnaire. These researchers discovered the following aspects: interpersonal interactions , quality of care , outcome of care and access all important in overall satisfaction. This attempt to develop a measure based in qualitative methodologies provided some confirmation of the commonalities found in theories of satisfaction and an example of the depth in components of satisfaction. Meuter et al., (2000) studied satisfaction with technology-based service encounters and found several factors leading to satisfactory evaluations by customers. They identified the main satisfaction is "a complex system" that incorporates both the experience and perceptions of consumers.
This researcher adheres to the integrated definition of satisfaction as suggested by Giese & Cote (2000). Therefore, satisfaction is defined as an evaluation of services based on the experience and perception s of an individual. With a clear definition , developing a measure to assess satisfaction is warranted. Because the components of satisfaction were not clearly understood for participants utilizing Project HEAL TH Expert Systems (as they had not been thoroughly assessed prior to this research), qualitative methodologies were employed in an attempt to understand participants' experience and ultimately understand their definition of satisfaction.

Methodology
The primary aims of this research were to examine the experiences and levels of satisfaction among participants in three Expert Systems. The secondary aims were to investigate any differences in levels of satisfaction among differing demographics (e.g. gender and race). The information gathered from this research will hopefully be used in the future to improve the three Expert Systems.

Research Questions
The

Mixed Method Research Design
In order to answer the research questions, a mixed method design was selected.
This type of design contains elements of qualitative and quantitative approaches (Tashakkori & Teddlie, 1998). While a mixed method design was chosen , this researcher still maintained a qualitative stance in designing the interview guide, coding, analyzing and interpreting the data. Lincoln & Denzin (2000), well-known qualitative researchers , indicate there are several activities that define the qualitative research process. These include how a researcher approaches the world, a framework (theory, ontology) that specifies a set of questions (epistemology) which then influences the approaches a researcher uses to answer these questions (methodology, analysis). The overarching term that encompasses ontology, epistemology and methodology is paradigm (Lincoln & Denzin, 2000). There has been, and continues to be, a great deal of debate regarding the relative importance of maintaining congruence between a researcher ' s ontology, epistemology and methodology in qualitative and quantitative research. Other scholars, often referred to as pragmatists, assert that qualitative and quantitative research are compatible (Brewer & Hunter , 1989;Datta, 1994;Howe, 1988;Tashakkori & Teddlie , 1998) and can be used to complement each other in research studies. According to pragmatists, both approaches are useful and the decision to use one or both are based on what will work best to answer the research questions. Pragmatists believe that it is the research questions that are most important, not the researcher's worldview , and "pragmatism is the best paradigm for justifying the use of mixed methods (Tashakkori & Teddlie , 1998).
Pragmatism served as the philosophical orientation for this research. Purposive sampling techniques were used to select a sample that could best address the phenomena being studied . Since the primary aim was to elicit information related to the experiences of participants (i.e. How did you experience Project HEAL TH? including expectations, satisfaction , likes, style) semi-structured telephone based interviews were determined to be the best method for data collection. The interview questions were based on the literature review and this researcher's values and interests . The types of questions asked by this researcher were primarily ' hows' as opposed to ' whys.' Analysis of the data was conducted using both qualitative and quantitative approaches. This researcher gathered information about the participants' experiences of Expert Systems and then made generalizations about each group.
In qualitative research, establishing trustworthiness is imperative to ensure the best quality of research and to minimize limitations. The researcher bears the responsibility to demonstrate the findings of an inquiry are credible. Lincoln & Guba (1985& 1994 developed criteria and strategies that can be combined to address trustworthiness. A number of these strategies were used in this study including prolonged engagement, reflexive journaling , informal member checks, dependability audit and triangulation.
This researcher was engaged with the data over the course of one year. The interviews were conducted and by this single researcher over the course of six months.
Transcription of the interviews over the course of four months, development and refinement of codes over the course of one year allowed this researcher to be immersed in the data (prolonged engagement). The participants were informed that all of their opinions were valued. Throughout the interviews, responses given by participants were validated by the researcher (member checking). Memos were kept throughout the study and recorded methodological decisions which were discussed with faculty (reflexive journaling). Informal consultation with faculty throughout the research process, allowed for a dependability audit which examined all of the decisions made from the beginning of the research study to the analysis and interpretation of results. Lastly, the quantitative analyses were used to validate the qualitative data specifically related to participant satisfaction ratings (triangulation).
When conducting a study that is purely qualitative or incorporates a significant qualitative component , it is important to discuss any qualities of the researcher that could possibly influence the research process. This researcher is a Black female who has been interested in the topic of health disparities for several years, specifically between Whites and Blacks. This researchers race seemed to be helpful in developing the interview protocol as cultural factors like style of interactions were thoroughly considered. Race , however, did not seem to be a concern when conducting the interviews. This may have been due to the fact the interviews were conducted over the telephone ; the results may have been different if the interviews were conducted in person. It is clear race influenced the conception of the research , but it likely did not influence participant's experiences when conducting the interviews. Over 260 individuals were contacted via telephone to request their participation in this research over the course of a six month period. A total of 58 interviews were conducted, which constitutes 22% of the individuals contacted. There were a variety of 6 Sample size is relevant to statistical power in quantitative research, but has less relevance in qualitative research (Camic, et al, 2003). This researcher believed the sample of 30 would provide adequate representation of the experiences of participants. 7 The terms Caucasian and White have been used interchangeably in the literature. The term White has been used more often in research in the last few years, therefore , I will be using this term for the remainder of this document. 8 The terms African American and Black have also been used interchangeably in the literature. I believe the term Black is more descriptive than African American when describing individuals of the African Diaspora. Therefore , the term Black will be used throughout the remainder of this document.

Participants
dispositions which prevented the remaining 200 individuals from participation additional interviews which were incomplete for various reasons (i.e. desire to discontinue and lack of recall regarding feedback), but were included due to the valuable information provided.
The Black sample 14 included 4 participants, which made up less than 7% of this research sample. These participants were all women, 3 of whom were in the Modular group and 1 was in the Telecommunications group. Their age range was from 47 to 64, with a mean age of 54.

Procedures
All procedures of this research were approved by the Internal Review Board at the University of Rhode Island.
Confidentiality. Participants signed a consent form as a part of Project HEAL TH (see Appendix A). The Principal Investigator , Wayne F. Velicer, Ph.D. indicated this form provided consent to this research as well. The consent states: "You will be asked to participate in several telephone surveys during the next two years. The number of telephone surveys will depend on the group you are assigned to ... " This research was considered a part of conducting one of these telephone surveys.
This researcher complied with procedures of confidentiality on the consent form, which states: All data will be coded with a number and will be scored on password-protected computers, separated from you name. Only authorized researcher will have access to any identifying information. There will be no reports remaining that identify you as an individual project participant. Information linking to you name will not be released to anyone outside the research group.
As a part of the research team , this researcher was able to access the name and contact information of an individual via computer system, but no paper records were kept 14 A total of 56 Black participants and 1502 White participants were enrolled or had completed Project HEAL TH at the time this research was being conducted. It is possible the sample size for Black participants was too small and the proposed goal of recruiting 3 to 5 participants was not realistic for this research .
including this identifying information. After data were collected, access to this information ceased.
When participants were contacted, they were informed of: their confidentially, their decision to not participate or quit at anytime and provided with a contact number where they could receive additional information about their rights or to file a complaint (see Appendixes A and D). In addition, they were notified their participation may not directly benefit them, but their participation would provide valuable information in designing future programs , which may benefit others. They were further informed the purpose ofthis recorded call was to examine their experience and satisfaction (see

Appendixes C and D).
Project HEAL TH participants were recruited using a random digit dial telephone methodology to contact individuals across the United States. Due to the wide area in which participants resided, it was deemed most appropriate to conduct semi-structured interviews via telephone.
A list of approximately 450 identification numbers with the three grouping variables (i.e . group assignment, level of participation, and race) were provided by the Survey Center. A purposive sampling technique was utilized in hopes of obtaining an equivalent number of participants in each of the grouping variables. As a part of the research team, this researcher was able to access the Survey Center computer system to get participant contact information. Since the sample was from various parts of the United States, careful consideration was given to time of day and time zones were confirmed using an area code finder prior to calling.
Over 260 individuals from this list were contacted via telephone and asked to participate. Participants were informed of the purpose of this research and aspects of confidentia lity. They were provided with an opportunity to ask questions and/or express concerns (see Appendixes C and D). If individuals were agreeable at this point, they were asked permission to conduct the interview at that time and the recording was started . If individuals were agreeable, but were not available at that time, they were asked for an appropriate time to complete the interview and called at that time.

Data Collection and Measures
Demographic Data. Archival data from the Survey Center provided demographic information including gender, age and race. Group assignment, level of participation and stages of change were also provided by the Survey Center. No additional information was requested from participants. According to Patton (1987), there are six basic types of interview questions which include: experience/behavior; opinion/belief ; feelings; knowledge; sensory; demographics. This interview guide contained each type of question with the exception of demographics questions (as explained above) . Experience or behavior questions 20 include descriptions of experiences , behaviors, actions , and activities. The majority of questions in the interview guide consisted of experience and behavior questions since this was the primary aim of the research. Opinion/belief questions provided this researcher with an understanding of the cognitive and interpretative processes of the individuals , which was important in how they viewed the Systems . Feeling questions were specifically directed at the participants ' level of satisfaction because this represents their emotional response to their experiences and thoughts. Emotional responses are thought to be an important aspect of satisfaction and participation (Liljander & Strandvik , 1997).
Knowledge questions assessed familiarity with the Systems and gave this researcher a better sense whether the participants grasped the information in a manner consistent with original Investigators intentions. Lastly , sensory questions allowed the individual an opportunity to describe the stimuli, in this case, the type of Expert System to which they were exposed. This provided a sense of the experience of interacting with each of the Systems. The time frame of questions addressed past, present and the future behaviors, thoughts and feelings; specifically, past experiences with the System, current impressions of the system , and future health behaviors. Recommendations from Depth Interviewing (Patton , 1987) were followed with regard to the sequencing of questions. Noncontroversial present behaviors were asked first, then interpretations, opinions and feelings about behaviors. The last section of the interview was devoted to future -oriented behaviors.
Opportunities were allotted for potential follow-up questions to elicit more information or clarify the information presented by the participant. Some follow-up questions were detail oriented and others were clarification probes. For instance, "I want to make sure I understood what you said correctly. What I got from that was .. . Is that correct?" There was also time allotted for probing questions. For instance, "Tell me more about that" or "Would you be more specific ." A sample interview is provided in the appendix for review (see Appendix F).
The interview was designed to take approximately 10 minutes to administer.
However, variations in length of interviews were expected based on the nature of the participant (e.g. talkative versus quiet). Interviews for this research varied in length from 5 minutes to 55 minutes. The interviews were conducted in English , which was the primary language of the majority of participants from Project HEAL TH.

Analytical Procedures
Qualitative . Interviews were recorded electronically using resources provided by the Survey Center at the Cancer Prevention Research Center (CPRC) . The Survey Center has private acoustic paneled enclosed workspa ces and desks. Each workspace has a headset telephone for hands-free telephone interviewing. The interviews were recorded electronically using the UBS Blast system (Version 1.94) and transferred to a Universal Serial Bus (USB) Flash Drive , where they were stored in a locked cabinet on the CPRC premises.
Each interview was transcribed verbatim by this researcher. Each transcription of the interview was thoroughly examined for formal and informal identifiers (e.g. name, place of employment, place of residence) , which were removed to protect the privacy of participants. The transcription process took approximately four month s to complete.
The data were analyzed using a qualitative analysis and management program (NVivo 7). NVivo 7 allowed this researcher to search and assess relationships of text 22 with the ability to mark specific items for analyses. These data were organized using the process of coding or with nodes. A code or node is a collection of references about a specific theme, place, person or other area of interest (NVivo 7 Manual, 2006). This researcher coded complete sentences, paragraphs and larger sections of the interviews to provide a context. A tree node or a catalog of nodes was then used to organize the data for easy comparisons (see Figure 1). As a result of the tree node, this researcher was able to organize the data with overarching themes as well as specific parts of that theme.
While rare, some participants' had multiple nodes within the same overarching theme.
For example, one participant indicated they participated in the program to help others as well as to improve his health .
The coding process was completed over the course of one year and involved two distinct phases. First, this researcher organized the data by coding according to the interview questions (i.e. Tell me how the program met/did not meet your expectations?).
This type of coding is closely related to topic coding, in that there are preconceived topics and data is coded according to those . As the coding based on the 18 questions progressed, other codes began to emerge from the data (i.e. helping others , suggestions) and these were coded as well. During the second phase, the participants' group was utilized as an overarching code. At this point, the coding closely resembled descriptive coding which identifies the individuals or groups. This second phase was helpful in the process of comparing group experiences and satisfaction. At various points throughout the coding process, this researcher informally consulted with colleagues, however, no one was directly involved in the coding in NVivo.

23
Quantitative . Quantitative data were analyzed using Statistical Package for the Social Sciences (SPSS) , Versionl6 .0. SPSS is an advanced mathematical and statistical software program used for analyses. These data, that provide support to the qualitative data, are presented throughout the results section and referred to in the discussion section.

Themes
The major topical headings from the interview guide served as a framework in examining themes (Reasons for Participation, Expectations , Likes, Style, Experience 15 , Reaction to Feedback, Trust, Satisfaction and Suggestions). Therefore the results section was separated by these themes. Within each of these overarching themes, frequencies of codes and appropriate exemplars of participant endorsements are provided. In addition, quantitative analyses are included under the satisfaction theme.

Reasons to Participate
Participants were asked their reasons for participating in Project HEAL TH. In general, participants Reasons for Participation (see Figure 2) fell into one of four areas: an interest in changing their behavior (n=16), helping others or research (n= 19), general interest in the topics (n=9) or no reasons (n=23). A number of individuals recalled a desire to change their unhealthy behaviors and described their overall health: "Well , I think it was primarily because I was, you know I am getting on in years and I was concerned for my health (R16)." Others were interested in changing a specific behavior, for instance: "Um, I was thinking about quitting smoking (R57) " and "Uh, it gave me a chance of the dieting and ways of quitting to smoke (R49)." Other participants were motivated to help others : "I thought it would be a good learning experience and also turn around and be able to help other people (R52). " A number of individuals reported a general interest in the program: "I just thought it was an interesting study and you know there was always a chance of bettering yourself (R36) ." Still others indicated they had no reasons or could not recall reasons of why they participated.

Expectation s
Participants were asked about their expectations with Project HEALTH and how those expectations were or were not met. Responses were coded into 3 categories (see I: Before you began to participate in the program , tell me some things you expected to happen? For instance , some people say they expected to be smokefree or to have a better diet or exercise program. Of the 17 participants that expected behavior change, 11 indicated they changed one, two or all three behaviors (i.e. eating habits, exercise habits, decrease in smoking).
Of the participant s that stated they expected change, some (n=6) reported no behavior changes . Of the participants that did not report changes, over half (n=4) indicated Project HEAL TH did provide helpful information . For instance, a Modular group participant stated: I think it met my expectations ju st fine. I probably expected a little more of myself that I was actually going to do but that was also because I was hoping to be having another child during that time and that didn 't happen yet and we decided to wait. So, I was expecting myself to get healthier than I actually did, but um .. . still with just the new knowledge and having all the new literature to look through and you know , keep tabs on myself. It's something I can still utilize in the future and something I will be aware of (R36).
One participant in the Telecommunication s group acknowledged her lack of participation as an explanation for her expectations of behavior change not being met: I: Can you tell me how the program met or did not meet your expectations with regards to these?
R48: Um, it didn't, because I didn 't participate enough I think. You know what I mean?
I: Right, and you were invited to call the automated system. And were you able to do that at all.
R48: I don't think I ever did, no.
Participants that fell into the category of 'no expectations' or ' basic project expectations ' (n=38) reported expectations like : receiving telephone calls, receiving feedback , etc. The following statement is an example of basic project expectati ons in the Modular group.
Rl 6: .. . no, I had no expectations other than to participate and then get some feedback as to how I compare to other people in the country , I guess.
27 Several participants (n=3) alluded to the desire to help others through their participation. For example: "I just kind of one of those things , it was cancer research and that's always a good thing so I am always one to help out (R14)."

Likes
Participants were asked what they liked about the program and if these 'likes' had an influence on their level of participation . The category of likes (see Figure 4) was coded across the 3 groups into 5 codes including: feedback (n=23), helping others (n=5), interactions with staff (n= l0), the telephone surveys (n=8), none (n=l0) and 1 participant in the Integrated group indicated she enjoyed participating because it "it was easy (R50). " Several participants in the Modular (n=14) and Integrated (n=5) groups described their enjoyment of the feedback reports. For instance: "The report you would get after every telephone call. Because just in case you didn't remember all the questions, you got to read and see if you made any progress or not (R45)" and " ... when I did receive the progress reports , that was nice actually to just to be able to see on paper where you are starting from and where you may be and um ... then the little calls, the check-ins (R36)." Only a few people (n=4)  A couple of people in each group indicated their interaction with staff (especially during the telephone survey) was pleasant and were motivated to continue participating.
For example: "The people that called were real nice. And ifI didn 't have time to talk, they would call me back later (R24);" "Nothing in particular , except that you guys were polite and always very forthright in everything that you said and asked. That's why I kept going with it (R32); " "The people were real courteous (R35)" and "Well , I liked the fact that they weren ' t calling you every week (both laugh). And they were very polite and patient (R43)." A couple of participants stated their desire to help others made their participation worthwhil e: "Well, I felt like I was contributing to something (R37) ."

Style
Because the nature of feedback differed for each group , the style was examined separately for each group. However, the data revealed an overlap in codes.
In the Telecommunications group, codes (see Figure 5) were divided into 5 categories , including: confusing questions (n=2); irrelevant (n=2); repetitive question s (n=2); time consuming (n=2) and no difficulties (n= l 7). It should be noted when responding to the style question , the majority of participants referred to the telephone survey as opposed to the feedback or interactions with staff .
In the Modular group , participant respon ses fell into one of 7 codes (see Figure 5), including: confusing questions (n=5); infrequent contacts (n=2); irrelevant (n=2); repetitive questions (n=3); time consuming (n=7); too much information (n= l) and no difficulties (n= 18). Similar to the Telecommunication s group, the majority of participant s referred to the telephone survey when discussing their perception of style. Two participants felt an increase in contact would have helped their behavior change: "If it was more contact than 3 months apart or however often it was. You know , I might have been might is the key word , have been a little more diligent (ROS)." In addition , another participant indicated he received the telephone surveys, but not the written feedback : I guess to be honest I probably had a chance to take part in something that was pretty good, but I never realized it because I didn ' t have any of the materials to know what it was that I was suppo sed to be trying to correct or curve. I mean obviously eating and smoking habits , that type of thing. But, I ... I don ' t know. I guess without knowing what the materials were , I don't know how much more benefit I could have gotten out of that ... (R38) .
On the opposing end, this participant indicated she felt there was too much information in the printed feedback. She says: R02: After the first one, then I saw what it was, I looked at it and thought "Oh okay I'll get to this to this later. " And that' s basically how I did it.
I: Okay , so because there was so much information there , it was hard to go through a second and third time.
In the Integrated group , the data was divided into 4 codes (see Figure 5) including : confusing questions (n=8); repetitive questions (n= l); infrequent contacts (n= I) and no difficulties (n=8). When responding to this question, the majority of participants referred to the telephone survey as oppo sed to the feedback or interactions with staff. Below is an example of a participants' description of confusing questions. There were several ways in which trust (see Figure 6) was examined including: if participants would recommend the program to others , if the participants would participate again in this project or a similar one and how believable was the information provided by the program. Of the 51 participants that answered the question regarding the recommendation of Project HEAL TH, 42 stated they would recommend the program to others (i.e. family member or friend). Forty-five participants stated they would participate in this or a similar program again and 6 indicated they would not. Of the 6 individuals that stated they would not participate again, 2 also stated they would not recommend the program to a friend.
Of the 16 participants that responded to this question in the Telecommunications group, 12 indicated they would participate in a similar program again and 13 would recommend the program to family or friends. Because I see family and friends who are vastly overweight and they don ' t have any physical limitation s and they don't seem to be doing anything about it. But I think when they realize where their health score is or where they are healthy wise , I believe they would take into consideration the fact that you need to do something and you need to do it now. And a lot of the family members that I see they are much younger than I am. So if they start now , just think how healthy they would be in a year or two (R09) .
Five participants indicated they would not recommend the program for various reasons , including: don't remember the project to recommend it (R02); friends are too old to participate (R04); time consuming (R15) ; and one stated he does not (as a general rule) recommend anything to others (R26).
All of the participants in the Integrated group (n= I0) indicated they would participate in a similar program again and would recommend the program to others .
When asked why she would recommend the program to others, R07 said: It did give you some ideas about what roads to take to change , you know your eating habits , exercise , and I am trying to think what other questions were on there . Eating habits, exercise ... Oh, cancer prevention. Um, it just made you aware.
The majority of participants in the Telecommunications group did not doubt the truth of feedback statements (n=4), but several mentioned the difficulty in implementing some of the suggestions due to their environment. For instance: I am going to group with that and with the smoke thing, it's hard to take up a buddy that doesn't want to smoke and stick to that, cause I don't know him. Find somebody that doesn't want to smoke and hang out together. I suppose I could do that, but I keep running into people that smoke. I don' t live in a smoke-free environment , so that was a tough one (RS l ).
The majority of participants in the Modular group also did not doubt the truth of feedback statements (n= l5) and similarly mentioned the difficulty in implementing some of the suggestions due to their environment. For instance: "Well some of the things they would say, like with smoking or eating did I find it harder ifl was at parties or around people. I mean that didn't really apply to me (RI 5)." And: Well , I am sure all the statements are true and they were all for my good and benefits. I just, in fact this [indicates where he lives] where I am living now , we have this clubhouse over there with an exercise room and a swimming pool outside. And I've been there twice. I just , I just can't get into it (R44).
In the Integrated group several individuals responded to this question (n=8) and the responses were split, as in half doubted the statements and the other half did not. One participant describes how she slowly learned to trust the statements. She says: Well, I would say in the beginning, yeah. Again, it still goes down to the, you still find excuses, you find all kinds ofreasons why that's not right ... or whatever. Even like with the smoking , for the longest time I was like "I gotta quit, I gotta quit." But, you know what, I've just got too much stress and it's all excuses . It's not real fact, it's not really keeping you. I don't have time to exercise, you only need five or ten minutes. It may take longer that way, but that's all you really need. Nobody can say they don't have or can't find five or ten minutes. You know, the isometrics now, you can do things while you are sitting at your desk at work. I mean , I just don't ... it's just excuses , that's what it is (R33).
Another participant described his doubt in the statement about finding a friend to exercise with. He says: "It's highly unlikely that I would find somebody to work out with and stuff like that. But, then again, I never always had to have that anyway. I was able to do either. But, yes, they are beneficial (R32)." Of the 4 Black participants, 2 participants (R03-TLC & RI 7-M) indicated they did not receive feedback, so they were unable to comment on the trustworthiness of the statements. Three of these participants indicated they would recommend the program to others and would participant in a similar program in the future. The other participant was not asked these questions in part because she had not accessing her feedback through the 38 TLC automated system . The number of Black participants was too small to make comparisons of trust to White participants.

Satisfaction
Participants were asked two questions related to their satisfaction with Project HEAL TH. They were asked to describe their satisfaction in their own words (see Tables   4, 5 and 6) and to rate their overall satisfaction on a scale of 1 to 10 (see Tables 1, 2, and 3). Fifty-one participants responded to this question with a range in scores of 1 tol0 [M=7.63, SD=2.08, Mode=8]. Of the 5 participants with no scores, 4 did not complete the interview and one participant felt she was unable to answer this question stating: "Uh, not applicable. I wasn ' t dissatisfied and I wasn't satisfied, it was just like it didn't matter.
I don't know how to answer that on a one to ten (R28)." Satisfaction ratings were examined within each of the 3 groups. Due to the unequal and low numbers of participants in each group , a statistical comparison of satisfaction scores consisting of all 3 groups was not appropriate . Instead, these were examined independently in each group.  Table 5) as found in this response: I guess I could say that I was very surprised that it was as long as it was, I was expecting it more to be a month or something. It was really cool that it was a long progressive thing . participants seemed the most satisfied (see Table 6), however, it fact this was the smallest group should be taken into consideration . One participant remarked: I am very satisfied with it. I am even more satisfied with it because it made me really look at myself. You know, I am able to run around with my son more and able to play with him more, you know a lot more energy and hopefully if I keep up with it you know, it will be even more so. So, I think it really just brought full circle and right up to the surface how important it is to um ... at least try to stay healthy in the world of fast everything. Just if for nothing else, you know obviously for yourself, but if for nothing else just for our kids so we can at least give them a good basis and they can learn from positive role models. And they may not make the same crumby choices that got us all chunky and you know, unhappy as a lot of people are because unfortunately the heavier you get you know you just get unhappy and miserable with yourself . And you know, that's why I think these kinds of programs can be helpful and they are beneficial to everybody whether they take it at the time or whether they learn after the fact (R36).
Of the 4 Black participants, the satisfaction ratings were 1, 8, 9 and 10. The score of 1 was given by a woman in the TLC group whom was unable to access her feedback through the automated system . The 3 remaining participants were in the Modular group.
The number of Black participants was too small to make comparisons of satisfaction to White participants.

40
The rating of satisfaction was then examined in relation to the overall level of participation. An independent-sample s t-test was conducted to compare the satisfaction scores for Low (n=32) and High (n=24)  There was a small, positive correlation between satisfaction and exercise stage variables  Within the Modular group, suggestions included: making the printed material s shorter (e.g. bullet-points or highlights) (n=2), adding telephone check-ups to the project (n= l), providing an overall report for the group (n=2), possibly adding an on-line survey (n=l) and minimizing the time from telephon e survey to receiving the printed materials (n=l). This participant talked about having a shorter feedback report : No, but I did see they did have helpful hints and guides or whatever to help you to get to places I guess, to get to a place to quit smoking. But, I didn't happen to read them so ... cause there was a lot of other stuff in there with it and so it was like reading a letter instead of. . . I'd rather just read a list and go real quick , you know(R02) ?
This participant talked about the timing of when she received the feedback materials : "Uh, you know when they called for the questionnaire. It seems like it was a long time before I got the feedback. And then a couple times I went ' Oh, yeah I remember that! ' It ju st seemed like it was a long period in between (R24)."

Discussion
The aim of this research was to qualitatively examine the experience and satisfaction of participants in three Expert Systems. To gain a broad perspective, a variety of participants were recruited using a purposive sampling technique. The themes found in this research (i.e. Reasons to Participate , Expectations , Likes, Style, Reaction to feedback, Trust , Satisfaction and Suggestions) were not surprisingly similar to the underlying factors considered when developing the interview guide . However, this researcher did not identify the codes within each theme prior to conducting the research; rather these codes emerged from the data.

Themes
In general , Reasons to Participate were consistent across the three groups .
Participants were generally interested in the topics , changing their behaviors or displayed a sense of altruism. While these codes may appear simplistic, these results provide important information for future investigators . These data confirm the underlying goals of Project HEALTH: to aid individuals in changing unhealthy behaviors and to add to the knowledge base of multiple behavior change. Future participants could ask, why should I participate and the response would be: "In research like this, individuals have participated in hopes of changing their unhealthy behaviors and to help others in similar struggles." When we look at the theme of Expectations , several individuals expected to change their behaviors and only a couple expected to help others. Participants who expected behavior change did not necessarily change more or less than those with no expectations. The conclusion of this theme suggests that expecting behavior change does not guarantee change. As with any change , the desire or expectation is not enough; there 46 must be environmental support , interpersonal support and internal motivation. More than once a participant remarked on the difficulty of quitting smoking while their friends or family members still smoked (i.e. lack of interpersonal and environmental support).
Due to the length of time from individuals consenting to participate in Project Style of interaction has been cited in the literature as a contributing factor in health disparities. However, this research revealed no differences in the perceived Style of Project HEALTH between Black and White participants. This researcher believes that there were too few Black participants in this satisfaction study (i.e. Black participants =4 & White participants=52) to be able to meaningfully make comparisons on the key variables by race. In addition , the phrasing of the style question was considered as a possible explanation for the lack of differences. In health disparity research , the operational definition of style consists of a face-to-face interaction between two people .
The interventions in Project HEAL TH were provided outside the context of a hospital setting or an office, therefore the style (i.e. interactions with staff) was very limited and restricted to telephone interactions . It would have been helpful to consider alternate definitions of style to make the question more sensitive. Health disparities between Whites and Blacks in this country continue to exist and we need clearer definitions of what these gaps consists of. Two ways to address this issue is to recruit more Black participants in studies such as Project HEAL TH and the current research project to clearly define the known contributing factors in health disparities.
There were an astonishing number of comments made about the telephone surveys . Some individuals enjoyed the telephone survey and utilized it as a reminder to make changes in their behaviors. But, for the majority of participants , the telephone survey was lengthy and consisted of confusing and repetitive questions. From the perspective of the Investigator , the number of questions in the survey was appropriate to gather the necessary information for staging and tailored feedback on unhealthy behaviors. From the participants' perspective , being on the telephone for lengthy periods of time was not always desirable. One participant commented he would have preferred to provide this information for Project HEALTH through a mailed survey. While this may have been convenient for the participant , the additional financial burdens and questions of retaining psychometric characteristics ( e.g. participants may not complete the survey in one sitting and over time their responses may be influenced by a host of factors) would be substantial. Some revisions in the telephone survey, however, may be helpful in retaining participant engagement. Future Expert System intervention research may need to find ways to minimize the respondent burden of the assessment process.
Given the unique characteristics of participants in Project HEAL TH, the metamessage 'You need to change,' was not a novel theme as many have heard direct exhortations to change their unhealthy behaviors from family, friends and health care professionals. It was interesting how participants responded to feedback from Project HEALTH about their behaviors compared to other sources. Some felt the face-to-face interaction with their physicians or other health care providers was most effective in addressing their unhealthy behaviors as they were being held accountable, which they felt was motivating. Others felt the detailed information provided in the Project HEALTH feedback was more effective as it provided information on the benefits of change and helpful suggestions on how to change. There were no noticeable differences between those that preferred feedback from their provider and those that preferred feedback from Project HEAL TH ( e.g. felt pressured to change, stages of change , etc.), it just appeared that individuals had their own beliefs of how change would most likely be sustained.
While receiving the same messages from multiple sources is optimal in successful behavior change, the current health care climate (where physicians may interact with patients for 15 minutes once a year) calls for more opportunities to deliver efficient tailored information as evidenced in Expert Systems.
Overall, the reaction to feedback (of those participants that received feedback) was positive, especially towards the beginning of the project. Participants noted their appreciation of Project HEALTH's assessment through the telephone surveys which provided tailored feedback. However , multiple remarks ( especially within the Modular group) were made about the lengthy and repetitive nature of the feedback over the course of the program. This researcher wondered if the nature and form of the feedback could be altered over the course of an intervention to address these concern s, especially if a participant is at the same stage of change across assessments. For instance , the Modular group could receive more frequent feedback that is shorter in length or the Telecommunications group could receive briefer feedback over the telephone. It seems these types of approaches are needed to increase participant engagement that could increase participants' use of feedback materials as well as increased calling into an automated system.
The overwhelming majority of participants in all three groups indicated they would participate in a similar program again, and would also recommend Project HEAL TH to a friend. These endorsements are exemplars of trust, which participants further attested to when asked about the believability of the feedback. The few instances ( across all groups) when participants doubted the truth or that elements of the feedback did not apply to them in Project HEALTH, were minor when compared to the overall value of the feedback. For instance , a couple of participants doubted that working out with a friend would be helpful as their friends lived a great distance away or their friends were unable to work out. These instances suggest that gathering additional information (i.e. Does the participant have friends to work out with?) for more specific tailoring would help engage more participants. So, there were instances when participants doubted the truth of a statement or an element of the feedback, but displayed trust in Project HEAL TH by their willingness to participate again or recommend the program to others.
According to the participants in this research, satisfaction grossly consisted of: style, level of trust in Project HEALTH and if initial expectations were met. Satisfaction did not appear to be clearly related to whether a participant changed their unhealthy behaviors (outcome measures in Project HEALTH). While some participants mentioned their behavior changes when describing satisfaction, none of the participants that rated satisfaction low suggested their lack of change was related to their satisfaction.

Participant Profiles
In order to compare the groups, this researcher thought it was important to have a snap shot of the average participant before developing a group picture. This researcher used the corresponding data in each group to create a profile participant for each group. A profile may provide a better sense of individuals participating in these Expert Systems.
Based on these interviews , the typical participant in the Telecommunications group was a White male in his S0's who was interested in participating to help research.
When he found out what the project had to offer, he decided this was a good opportunity to change his diet and smoking habits. He called into the automated system a couple of times, but lost his password. He found the telephone surveys helpful reminders to eat better and reduce his smoking , but soon after the call , he would resort to old habits.
Given this profile, which is a conglomerate of statements and beliefs from participants in the Telecommunications group, how would improvements be made in this type of Expert System? While this profile participant was motivated to make changes , it might be helpful to provide him with periodic reminders over the telephone . The telephone reminders could be automated about once a month. In addition , the password system could be revised to make participation more convenient. One of the significant complaints in the Telecommunications group was difficulty with the password. After a participant experienced frustration with the pas sword, they were less likely to engage in the interventions .
The typical participant in the Modular group was a White woman in her late 40 ' s.
She initially participated because it had to do with smoking cessation. She has been trying to quit smoking for some time now because she lacks the stamina to activity play with her children. However, she did not have any expectations of changing going into the program. She enjoyed the printed materials as they resembled report cards. While she was disappointed at times with her progress, she liked the fact she could reread the materials and use them as a motivator. She wished the reports were shorter or had a list of bullet-pointed ideas. The suggestions were helpful although some did not apply to her, like exercising with a friend which was not feasible. She made a few changes including walking more, which helped decrease her nicotine intake. In the case of this profile, it might be helpful to revise the feedback materials slightly. The Investigators might consider a bullet-point style, which may exclude some important information . Another suggestion would be to send the same information in smaller increments so it is not as overwhelming for participants (e.g. an overview with bullet-points , specifics about smoking behaviors , diet behaviors, and exercise suggestions).
The typical participant in the Integrated group was a White woman in her mid 50' s. She found the style of Project HEALTH helpful in supporting her efforts to improve her health overall. The questions asked during the telephone survey were similar to those of her health care provider , both of which were helpful reminders to continue making small changes. She would recommend this project to a friend because it was a good companion to other sources of support for healthy behavior change. Potential recommendations for improvement are very limited given this profile , which is likely due to the small sample size. However, based on the data available in the Integrated group , this participant profile could serve as an exemplar for using this type of Expert System.

Group Comparisons
Given these profiles and the data , how do the three Expert Systems compare?
Overall , the Integrated group seemed to have the most positive experience as participants in this group were likely to: rate their satisfaction high, experience positive behavior change and desire to recommend participation to others. Project HEAL TH Investigators hypothesized the Integrated group would do well compared to other systems because it: was based on prior knowledge of successful intervention systems and provided integrated information (as opposed to information on each behavior) on behavior change which allowed participants to work at their own pace. It is also likely this group did better due to the reduced project demand as participants in this group received the least amount of feedback reports. We could hypothesize this intervention is less prescriptive than the others , potentially offering a gentler approach.

Modular versus Telecommunications. Both Modular and Telecommunications
Expert Systems provided detailed feedback on each of the three targeted behaviors . In general , participants in the Modular had a higher level of participant and reported reading the feedback . Participants in the Telecommunications group found accessing the feedback difficult and when participants did access it, they found the feedback repetitive.
It would appear that providing information on more than one behavior would be better received by participants if it was in printed form.

Integrated versus Modular. Both the Integrated and Modular Expert Systems
provided printed feedback to participants . The difference was the Integrated Expert System provided overall information about health and the Modular system was detailed in each of the three targeted behaviors. Based on the data, it would appear presenting overall information would be more well-received as the Integrated group yielded fewer complaints about the format, displayed more satisfaction and more healthy changes in behaviors. Although, this finding represents a comparison between the groups with the largest and smallest sample sizes and should be interpreted with caution.

Limitations
There are several limitations specific to this research . First, the timing of these interviews from the completion of the 12 month telephone survey was not consistent across participants, with the interview completion ranging from one month to twelve months. There was no evidence to suggest this variability in length of time from the 12 month survey until participation in this study had an impact on participant recall.
However , it would appropriate in the future to standardize a length of time to conduct interviews to help ensure dependability.
Second, the number of participants interviewed in the research was relatively small compared to the overall project (less than 4% of the total sample). Moreover , the lack of suitable representation of Black participants in this research was likely due to an ambitious sampling technique. At the time of this research, there were a total of 56 Black participants who had participated in Project HEALTH. It is possible the sample size for Black participants was too small and the proposed goal of recruiting 3 to 5 participants per group was not realistic for this research. However, for purposes of qualitative research, the number of participants has minimal relevance compared to the data gained.
This researcher believes the information gained from the experienc es of participants was substantial. While the information gained is valuable, generalizability of these findings to other Expert Systems should be considered with caution .
Third, individuals who participated in this research may possibly have had a more po sitive view of Project HEALTH than individuals who declined to participate in this satisfaction study. This concern about a positive or negative bias is often a limitation of research that cannot be entirely ruled out as an explanation for study results. Participants were provided with ample opportunities and encouraged to discuss aspects of Project HEAL TH they liked as well as those they did not like as much. Given the number of negative remarks and constructive criticisms offered by participants, this researcher believes the participants in this sample were fairly representative of the typical participant who participated in Project HEALTH.
Lastly, a single researcher interviewed all participants, transcribed the data , coded and interpreted the data. A recommendation for strengthening future research studies includes the use of two or more coders in the analytic phase of the research, a confirmability audit to examine the product of the inquiry and the interpretations made by the researcher.

Future Directions
There are several changes this researcher would consider making to the interview guide prior to administering it again including: a revised question about experience and more sensitive questions to detect stylist difficulties.

Satisfactio n as described by Participants
It was nice ... anytime somebody called to talk they were always polite , patient... I didn 't feel like I was bothering anybody , you know and they didn 't make me feel like they was bothering me. So, it was nice.
Um, just need better follow up.
It provided information with the questions and it didn 't take very long.
Overall, it was okay. I have no problems with . .. any part of it. And 1 must not have because I kept doing it. And I have never, I have not continually done that before. I enjo yed talking to people. They were all cordia l with me, and like I say, I was impressed with their concern.
Real neutral. There was no satisfact ion or dissatisfaction , it was a no factor.

10
Well, I enjoyed the question s and it had me thinkin g.

7
Well I think overall it was pretty comprehensive.
The people were nice and the surveys were easy to understand and 20 9 l guess it would have been helpfu l if I applied myself more to it.

9
As far as the program , I think its fine.  It didn't take up a lot of time and it didn't seem like it was judgmenta l.
Like when you said "Well, a little bit of progress is better than nothing." Like, I have a lot of encouragement here, if I am going to change.
I was satisfied with it. I just didn't go along with all of it, by doing it.
... I was very satisfied with it.   You have been asked to take part in a research project described below . The researcher will explain the project to you in detail. You should feel free to ask questions . If you have any more questions later, Dr. Velicer, the person mainly responsible for this study (phone: in Rhode Island: (800) 555-2854; outside Rhode Island: (800) 7TT-3537), will discuss them wilh you. You must be at least 18 years old to be in this research project .
Description of the proje ct: The purpose of this research is to learn more about ways to help people change unhealthy behavior and to evaluate differenl ways to help people change unhealthy dietary behavior. smoking habits and sedentary lifestyle.

What will be done:
If you decide to take part in this study, here is what w ill happen. You will be asked to participate in several phone surveys during the next two years. You will be randoml y assigned (like tossing a coin) to one of four research study groups . The number of phone surveys will depend on the group you are assigned to; at a minimum you will be surveyed again at 12 and 24 months after the inttial survey. You may or may not receive materials that deal with keeping a healthy diet, quitting smoking or exercising . You may or may not be offered participation in an automated telephone counseling system called TLC that is designed to help you change your health behavior. If you are in this TLC group yo u will be required to make several (toll-free) phone calls to the system In order to get the ·information that is especially tailored to your needs. After two years. the study will be terminated and you will no longer receive survey q uestions or any materials.

Ri5k5 or d~comfort:
The only discomfort or inconvenienoe asso ciated wilh the study is that associated with the surveys and wtth the eventual effort involved in participating in one of the automated counseling programs.

Appendix A-continued: Consent Form
Expected benefits of study: Participation i n this research may help you make better decisions about your hea lth. Even if there is no direct benefit to you for taking part in this study, your honest answers will provide valuable information in designing future health education programs, which may benefit others .
Confid en tiality Your participation in this study is strictly confidentia l. All data will be coded w~h a nu mber and will be stored on password-protected compu t ers, separated from your name. Only authorized resea rchers will have access to any identifying informa1ion. There will be no reports remaining that identify you as an individua l project participant. Informa tion linking to your na me will not be released to anyone outside the research group .
Decision to quit at any time The decision to take part in th is study is up to you. You do not have to part icipate. If you decide to take part in this study, you may qu it at any time . 'M1atever you decide will not penalize you in any way . If you wish to quit , simply inform Dr . Velice r (phone: in Rhode Island : (800) 555-2854 ; outside Rhode Isla nd: (800) 777-3537) of your decision .

Rights and Complaints
If you are not satisfied w~ the way this study is performed, you may discuss your complaints with Dr. Velicer (phone: in Rhode Island : (800)  I: So, there weren ' t any parts that were unclear for you.

R44: No.
I: How has the program been helpful in changing your behavior overall?
R44: Well, really the only thing that has really changed is my smoking. I am smoke less. As far as my diet ... well, I eat a lot more fruits than I used to. I always liked fruit , but my wife never did, so we didn't have it. So, since I am by myself, I eat quite a bit. I eat out about three times a week . I try to get fish every time.
I: Okay. Tell me some ways in which the program has not been as helpful in changing your behavior ? So, give me an example of something that has either stayed the same or gotten worse .
R44: Well , they do want me to get out and exercise. I have a stationary bike and maybe every two weeks or so, I am on it. Not like I should be.
I: So, was that the same before you participated in the program. Like maybe getting on the bike a couple times a month.
R44: Yeah, that stayed the same. I didn't really do any extra exercise .
I: Oka y. I am wondering if there is anything that you would like us to do differently ? For instance, some things you would like to see more or less of that would make the program better in the future?