Use of interactive voice response (IVR) technology in health research with children

This article reports on the feasibility of using interactive voice response (IVR) technology to obtain daily reports of attitudes toward alcohol and tobacco use among children 9-13 years of age. Two studies were conducted. The first was an investigation of the use of IVR technology to obtain daily data from a sample of primary school children over a period of 8 weeks. The second was an extension of the research to a large sample of primary and secondary school children in urban and rural areas who provided daily data over a 4-week period. Retention and compliance rates comparable to those obtained with adults were evident in both studies, supporting the feasibility of this technology with children. The results are discussed in relation to the benefits of this methodology for health research, particularly for studies of sensitive topics conducted with children and adolescents.

There has been a marked increase in the number of daily process or "experience sampling" studies in the field of health research (Leigh, 2000; Tennen, Affleck, Armeli, & Carney, 2000). These studies have addressed issues such as alcohol and drug use (Searles, Helzer, & Walter, 2000; Searles, Perrine, Mundt, & Helzer, 1995), eating habits (Bardone, Krahn, Goodman, & Searles, 2000), chronic pain and mood (Feldman, Downey, & Schaffer-Neitz, 1999), and sexual behaviors (Gillmore et al., 2001). Recent developments in computer technology have resulted in increased use of automated data collection methods, such as providing participants with palm-top personal computers for data entry, and computerassisted interviewing. The latter method includes online surveys (Kiesler & Sproull, 1986), computer-assisted self-interviewing (CASI; Turner et al., 1998), and computerautomated telephone interviewing using interactive voice response (IVR) technology (Bardone et al., 2000; Mundt, Perrine, Searles, & Walter, 1995). Online surveys and CASI require access to a personal computer, whereas telephone interviewing using IVR requires only access to a telephone. The advantages of IVR technology in facilitating reliable daily data collection point to its potential value in health research with children. However, according to a recent review of the use of IVR in 54 studies published between 1989 and 2000, there have been no applications of IVR involving children (Corkrey & Parkinson, 2002b). The aim of the present research was to evaluate the feasibility of using IVR technology in investigations of children's daily attitudes toward alcohol and tobacco use.

One-off methods of data collection, such as completion of a questionnaire or interview, usually involve retrospective estimates from participants of the typical occurrence of an event or experience. These methods may be less appropriate for research questions relating to events or experiences that could be affected by recall biases or errors. In particular, such problems are more likely to happen when events or experiences occur infrequently or are to be recalled by individuals who may be more vulnerable to recall biases and errors, such as children, whose cognitive abilities are still developing.

Methods of daily data collection can be interval contingent (e.g., having participants complete self-reports on a regular basis, as by having a daily diary), signal contingent (e.g., prompting participants to report data on multiple occasions, as in response to an electronic paging device), or event contingent (e.g., having participants complete reports when a target event occurs (Stone, Kessler, & Haythornthwaite, 1991; Wheeler & Reis, 1991). There are also a number of different modes of data collection within the daily process approach. Traditional modes include diaries, checklists, and telephone interviews (Leigh, 2000). Naturally, these methods involve a considerable amount of time and effort for training interviewers, conducting interviews, and performing data entry. More recently, modes of data collection combining telephone input with computer-assisted prompting and recording have offered several advantages. For example, participants perceive reporting to a computer as providing greater confidentiality and privacy than either pen-and-paper or live interviewing methods (Booth-Kewley, Edwards, & Rosenfeld, 1992; Mundtetal., 1995; Turner et al., 1998). This perception may partly explain why data obtained using computer-assisted methods usually include fewer incomplete or skipped items than do diary methods (Morrison, Leigh, & Gillmore, 1999; Turner et al., 1998). Moreover, with computer-automated systems, any missing data can be automatically flagged and quickly followed up either by an automated system feature (e.g., a reminder prompt to fill in the data missed on the previous day before proceeding to a current day's survey), or by a reminder call from a research assistant. Other advantages are that survey administration is completely standardized and that data entry, maintenance, and storage are automatic (Mundt et al., 1995; Turner et al., 1998). It has also been suggested that computer-assisted methods may result in more accurate reporting of socially undesirable behaviors, such as drug use and sexual risk-taking (Kobak et al., 1997; Tourangeau & Smith, 1996; Turner et al., 1998), although the evidence to support this claim is still limited (cf. Knapp & Kirk, 2003; Leigh, 2000). One important disadvantage of traditional daily diaries is that they can be completed in a single sitting for data covering several days-for instance, as many as 70% of the participants in some studies have admitted to faking compliance (e.g., Litt, Cooney, & Morse, 1998). In contrast, compliance via IVR cannot be faked, because the date and time of data entry are recorded by the system.

Telephone interviewing using IVR technology has particular benefits in comparison with other forms of computer-assisted data collection, primarily because of its ease of access for participants. Because access to a telephone is all that is required, completion of a report is relatively convenient and flexible for the participant (Mundt et al., 1995), and individuals from geographically dispersed and remote areas can readily participate, thereby increasing the representativeness of samples. Moreover, as noted by Turner et al. (1998), the verbal nature of administering survey questions via telephone allows for the inclusion of participants who have low levels of reading ability, including children. Finally, because interview questions presented via IVR cannot be overheard and responses are provided by unobtrusively pressing the keys of a touch-tone telephone, children are likely to perceive a high level of confidentiality; hence, their answers should be less liable to response bias.

Daily Designs Using Computer-Assisted Methods With Children

There are no published studies reporting the use of IVR technology with children (Corkrey & Parkinson, 2002b), and indeed, the use of computer-assisted daily designs with children is relatively rare. For instance, Sadeh, Raviv, and Gruber (2000) monitored children's sleep patterns using activity monitors ("actigraphs") and daily reports for a brief period of 5 nights. The activity monitor was a wristwatch-type device that did not require any input from the child, and thus was not affected by compliance problems once a participant had agreed to wear it. However, the daily self-report measures in that study were of the traditional pen-and-paper type, and compliance rates for these were not reported. Others have conducted research using the experience sampling method with children and adolescents 9-15 years of age (Csikszentmihalyi & Larson, 1987; Larson, 1989). These studies involved a data collection period of 1 week in which participants completed self-report forms in response to pager signals. Larson (1989) reported good compliance, with 80% complete self-reports obtained in response to the pager signals (with pager malfunction contributing to some of the nonresponses). The validity of the data was supported by a high level of consistency with collateral reports from parents. However, as discussed earlier, in some contexts this method may be vulnerable to faked compliance.

In sum, the use of computer-assisted methods in daily designs with children is still in its infancy, and to the extent that daily designs have been used, they have tended to be limited to brief periods of 1 week or less. In contrast, for many health-related attitudes and behaviors, (e.g., those concerning substance use, dietary patterns, and medication compliance), it is often desirable to obtain an accurate estimate of the behavior and associated events over an extended period of time, such as several weeks or months. Although extended daily sampling periods have been applied successfully with adults, the feasibility of obtaining reliable daily data from children over long time intervals has not been established.

The Present Studies

Several methodological issues associated with daily data collection over extended periods of time are particularly pertinent if applied to samples involving children. One issue is the potential for measurement reactivity-that is, the possibility that the process of daily self-monitoring changes the behavior under investigation. Indeed, this potential is a key assumption of cognitive-behavioral therapies that incorporate self-monitoring programs, and it is a possible threat to the validity of daily reports. However, there have been few empirical investigations of reactivity, and those that do exist have produced mixed results. In her review of daily design studies of alcohol use, Leigh (2000) reported that some studies have shown an increase in measurement reactivity but that others have not.

Researchers who have compared mean rates of the behaviors under investigation at the beginning of the study with those at the end have reported few significant differences (Csikszentmihalyi & Larson, 1987; Larson, 1989). Affleck, Zautra, Tennen, and Armeli (1999) argued that reactivity effects' are less likely to occur if several events or experiences are under investigation and participants are unable to review their daily recordings. Nonetheless, measurement reactivity remains an important issue for research in which potentially harmful behaviors such as substance use are the object of study. Especially with respect to children, the daily monitoring of substance-related attitudes, thoughts, and behaviors might raise concern among parents that their children's daily exposure to such topics might increase their intention (and hence, their risk) of engaging in these behaviors. Accordingly, one important goal of the present investigation was to examine whether children who report daily their attitudes toward alcohol and tobacco use would show an increase in their intention to use these substances after a monitoring interval of several weeks.

In addition to examining the issue of reactivity, we evaluated the feasibility of the use of IVR technology with this special population with respect to retention and compliance rates. Leigh (2000) noted that many studies using daily reports of alcohol-related behaviors in adults have not reported attrition rates. Where drop-out rates have been reported, they averaged from 10% to 20%. Retention rates in the few IVR studies assessing daily alcoholrelated behaviors in adults were generally high, with rates of 97% after 28 days (Perrine, Mundt, Searles, & Lester, 1995), 95% after 84 days (Bardone et al., 2000), and 93% after 112 days (Searles et al, 1995), but a drop to 60% after 366 days (Searles et al., 2000). It must be noted that the sample size in these studies was fairly small, varying between 30 and 55 participants. Also, the financial rewards offered for regular participation were considerable, ranging from U.S.$162 to $4,313. Whether reliable participation can be maintained over long periods without providing such large incentives is not known. Hence, a second goal of the present studies was to investigate whether good retention rates using IVR could be achieved with more modest incentives and with a large, less selective sample. The third goal was to examine whether the excellent rates of compliance (82% to 99%) with the demanding daily reporting regimen of IVR studies previously obtained with adults could be replicated with children.

In sum, this article describes an investigation of the use of IVR technology to obtain daily reports of attitudes toward alcohol and tobacco use from children aged 9-13 years. Two studies are described: an investigation with a sample of primary school children (n = 74) who participated over a period of 8 weeks, and an extension of this research to a larger, more representative sample of primary and secondary students (n = 605) who participated over a period of 4 weeks. The latter study was part of a longitudinal investigation of factors influencing the adoption of alcohol and tobacco use among urban and rural youth, the substantive results of which will be reported elsewhere. This article is primarily concerned with the methodological considerations of the implementation of IVR technology with children. Specifically, we addressed the following questions: (1) Can children be recruited to participate in studies involving such high demands on their commitment in terms of time and effort, and are there systematic differences between consenters and nonconsenters? (2) Will the retention and compliance rates for children be comparable to those obtained with adults using the same methodology, and are there systematic differences between completers and noncompleters? (3) Do children exposed to daily queslions about alcohol and tobacco use show increased intention to use these substances at the end of the survey interval? and (4) What incentives, training, and ongoing help are appropriate and necessary for this age group to obtain mastery of the technology and to maintain motivation for providing reliable and valid information over an extended period of time?

STUDY 1

Method

Participants. Children (53% girls) were recruited from two state-government schools in the metropolitan area of a large Australian city; the schools were situated in a low- to middle-class and a middle-class socioeconomic area, respectively. Of 153 parents contacted by letter, 74 (48.4%) gave consent for their children to participate. The mean ages of the children, who were in Grades 5, 6, and 7, were 9.9 (SD = 0.43), 10.9 (SD = 0.41), and 11.9 (SD = 0.35) years, respectively.

Apparatus. IVR technology combines telephone service with computer-administered questionnaires. Our system has the capacity to process 11 phone calls simultaneously, with one additional line dedicated to handling "line busy" conditions. The system is interfaced with local area network stations for data maintenance, backup, and analysis. The participants called a toll-free number and, after providing participant and password identification, followed a simple script to answer questions by pressing numbers on the keypad of a touch-tone telephone.

Measures. Daily self-report questions covered seven domains with respect to the use of alcohol, tobacco, and a control substance (cola): ( 1 ) exposure, which assessed the relative influence of others who used any of these substances; (2) desire to approach/desire to avoid, which assessed independently the desires to consume and to avoid a particular substance, thereby providing a measure of ambivalence; (3) intention or subjective probability of consuming a particular substance in the near future; (4) resistance/efficacy, which assessed how difficult it would be to say "no" if someone offered a particular substance the next day; (5) prevention, which assessed the relative influence of others in promoting abstinence or moderation; (6) active prevention, which assessed the participants' own actions to influence others toward abstinence or moderation; and (7) active exposure, which assessed the participants' own actions to influence others toward adoption or increase of use of a particular substance.

For most questions, the participants responded by pressing a number on the telephone keypad. Questions pertaining to exposure, prevention, active prevention, and active exposure were in a yes/no format ( 1 = yes, 2 = no). Questions from the desire to approach/ avoid, intention, and resistance sections used a nine-point rating scale indicating, for example, how much the participant wanted to consume a cola drink ( 1 = not at all, 9 - very much). Some questions were open-ended and required a recorded response.

Each section began with instructions concerning the question content and type of response required. Each question within a section had a common stem, so that as they became more practiced, the participants could respond before listening to the entire question. In addition, the participants could press the star (*) key to skip the instructions and advance to the specific question. The participants could change their response to the preceding question by using the hash (#) key. At the end of the survey, the participants were provided with an opportunity to redo the survey if they were not happy with their answers. To enhance the interactive nature of the survey, the children were also given the opportunity to leave a joke in the message option as part of their daily call, with the best ones being selected and rerecorded by the system manager to be played on the following day. In addition, to personalize the routine for the children, at the time of their first login, they assigned themselves a secret code name that they recorded, which was subsequently used by the automated system to address them during calls (e.g., "Welcome Superman") after they had entered a secret PIN.

Procedure. Participation in the project consisted of two components: First, the participants attended a questionnaire and training session approximately 2 h in duration. During this session, the participants completed a set of questionnaires and received instruction on the IVR protocol, as well as a detailed explanation of how the point system rewarded regular participation (see the Compensation schedule subsection below). This session was conducted in groups of 11 to 20 students and took place in a classroom.

For the second component, the participants followed the IVR protocol and recorded their responses daily for a period of 8 weeks (56 days). This component was completed by participants in their own time and usually took place in their homes, or in some cases from a telephone at their school or a public telephone. The telephone survey period began the day after the initial questionnaire and training session. In addition to the initial training, ongoing assistance was available through a message feature of the survey, as well as through a separate telephone help line by means of which the project manager could be contacted. Any participants who missed 2 consecutive days were contacted by research staff to remind them to call in.

Compensation schedule. The study contained a reward system by means of which the participants earned points for each daily survey completed, plus seven bonus points for each perfect week. If participants missed no more than 2 days during a 4-week period, the forfeited bonus points for those weeks containing missed days were restored. An additional 16 bonus points could be earned for reporting on a total of at least 50 out of the 56 days. All points were credited to an electronic "bank account," and participants could hear their current tally at the end of the survey every time they called. At the end of the study period, the participants received vouchers to exchange their accumulated points at a local department store for items of their choice, up to a maximum value of AU$20 (approximately U.S.$12). In addition, a movie ticket was provided at the halfway point of the 8-week data collection period.

Results

Recruitment and consent. Nearly 50% of parents gave consent for their children to participate. For ethical reasons, data on differences between consenters and nonconsenters were limited to a brief, voluntary questionnaire sent to all parents after consent forms from those parents who had given consent had been received. Completed questionnaires were received from 35 consenting and 24 nonconsenting parents.

Comparisons between consenters and nonconsenters were based on information about reasons for agreeing or not agreeing to give parental consent. The percentages of parents endorsing specified reasons are presented in Table 1. The most frequently endorsed reason for giving consent was that parents thought the topic of the research was important. Similarly, none of the parents who declined consent indicated that they refused consent because they felt the topic was not important. Instead, consenters and nonconsenters appeared to differ in the degree to which they wanted their children to be exposed to the topics of alcohol and cigarette use-nonconsenters did not want their children to be exposed to the topics. In addition, on the basis of responses to an open-ended question asking for any "other" reasons for consenting or not consenting, consenters agreed to participate because they valued community involvement and liked to provide a participatory role model for their children, whereas nonconsenters were concerned that exposure to the topics could encourage substance use and that participation would require too much time and effort.

Retention and compliance. The retention rate was high, with no attrition after the first week, 3% attrition after 4 weeks, and 8% attrition after 8 weeks. Thus, 92% of the sample completed the entire study period. Likewise, children's compliance rates were consistently high over the 8-week study period, with an average compliance rate of 91.4%. Compliance rates were equally high across all grade levels-91.5%, 90.1%, and 92.8% for Grades 5, 6, and 7, respectively.

Changes in attitudes toward alcohol and cigarette use during survey period. Daily responses regarding (1) the children's desires to drink alcohol or to smoke, (2) their desires to avoid drinking alcohol or smoking, (3) the likelihood that they would drink alcohol or smoke the next day, and (4) the children's efficacy (i.e., how difficult it would be to say "no" if they were offered alcohol or cigarettes the next day) were averaged across the first 14 days of the study period and compared with responses from days 29 to 42.1 Figure 1A shows that after daily, extended exposure to alcohol- and smoking-related questions, children did not report greater inclinations to take up alcohol use or smoking. The only significant differences between early and late periods of participation were that during the late period, children reported less chance of drinking either alcohol or a cola drink than they had reported during the early period.

Summary of Study 1

The feasibility of using IVR methodology with a preadolescent sample was clearly demonstrated. The attrition rate was low, and the compliance rates were very high. These results are particularly impressive given the 2-month duration of the survey period. Moreover, daily exposure to alcohol- and smoking-related survey questions during this period did not increase children's self-reported inclination to take up alcohol use or smoking. Two limitations of our first study were that the sample was relatively small and consisted only of primary school students. To test the reliability of the children's response patterns obtained via IVR technology, we aimed in Study 2 to replicate the results from Study 1, using a larger and more representative sample of children and young adolescents.

STUDY 2

Method

Participants. Children (56% girls) were recruited from stategovernment schools following procedures similar to those in Study 1. The consent rate was 33%, resulting in a sample of 662 participants from 12 primary and 5 secondary schools. Fourteen schools were located in the metropolitan area, whereas 2 primary schools and 1 secondary school were located in rural areas. The mean ages of the children, who were from Grades 5, 6, 7, and 8, were 10.27 (SD = 0.58), 11.02 (SD = 0.42), 12.08 (SD = 0.41), and 13.07 (SD = 0.39) years, respectively.

Apparatus and Procedure. The apparatus and procedure were the same as in Study 1, except that the length of the telephone survey period was 4 rather than 8 weeks.

Measures. Daily self-report questions were identical to those used in Study 1. In addition, baseline questionnaires included an assessment of personal experience with alcohol and cigarette use. Frequency of drinking was assessed with an 8-point scale ranging from never to almost every day. Quantity of drinking was assessed with a 5-point scale using descriptors appropriate for this age group ( 1 = just a little, enough to get a taste; 2 = more than a little, but not enough to feel "tipsy"; 3 = enough to feel "tipsy," but not enough to feel drunk; 4 = enough to feel drunk, but not enough to get sick or pass out; 5 = enough to get sick or pass out). Experience with smoking was assessed with three yes/no questions asking whether the children had smoked in the last week, month, and year, respectively.

Results

Retention and compliance. Of the 662 children who began the study, 605 (91.4%) were retained for the duration of the daily IVR survey component. Of the 57 noncompleters, 6 (10.5%) were unable to participate because they did not have a touch-tone telephone at home or their telephone did not have the facility to make outgoing calls. The majority of the remaining 51 noncompleters reported that they found the project requirements too demanding or difficult. On average, noncompleters made only 3.98 daily calls, and 14 never called the survey at all. The proportion of boys was higher among the noncompleters (54%) than among the completers (44%). There was no significant difference in age between noncompleters (M = 11.79 years, SD = 1.22) and completers (M = 11.67 years, SD = 1.11) [/(654) = 0.81, n.s.], although a slightly higher percentage of the noncompleters (33%) were high school students (Grade 8) than were found among the completers (24%).

Questionnaire data were available for all 57 noncompleters. These were compared with the data from a random sample of 57 completers stratified for gender and grade. A multivariate analysis of variance (MANOVA) with frequency and quantity of alcohol consumption as the dependent variables showed no significant differences between completers and noncompleters [F(2,107) = 1.77]. As expected, reported frequency and quantity of alcohol consumption were very low. The mean frequencies were 2.40 (SD = 1.29) for completers and 2.84 (SD = 1.63) for noncompleters [univariate F(1,108) = 3.36, n.s.], which represent scores between once or twice in my life and 3 or 4 times in my life. The mean quantities were 0.95 (SD = 0.29) for completers and 1.20 (SD = 0.91) for noncompleters [univariate F(I5IOS) = 2.48, n.s.], which represent scores of just a little, enough to get a taste. Similarly, fewer than 2% of completers and fewer than 6% of noncompleters reported tobacco use in the previous week. Although reported tobacco use in the past year was slightly higher among noncompleters ( 17%) than among completers (7%), this difference was not significant [χ^sup 2^(1) = 1.66, n.s.].

A total of 15,256 daily responses were obtained from the 605 participants who completed the telephone survey component. The average number of daily surveys completed per participant was 25.22 (SD = 4.82) out of 28 days, yielding an overall compliance rate of 90%. Over half (54%) of the participants provided data for each of the 28 days, and 85% provided data for at least three fourths of the study period. Compliance was higher among girls (93.50%) than boys (85.62%). Compliance declined throughout the 4 weeks of the survey period-compliance rates were 96.39%, 92.59%, 88.15%, and 83.12% in Weeks 1,2, 3, and 4, respectively. There were no differences in compliance across the school grade levels. The average duration of the telephone survey calls was 5 min (M = 301.72 sec, SD = 187.46).

Changes in attitudes toward alcohol and cigarette use during survey period. Similarly to Study 1, daily responses were averaged across the first 7 days of the study period and were compared with responses from Week 4 (Days 22-28). Figure 1B shows that the pattern of responses for the first and last weeks closely replicates the pattern found in Study 1. Responses for the last week did not differ markedly from responses during the first week. The few significant decreases observed were very small, were likely due to the large sample size, and in any event, were not consistent with greater self-reported inclinations to take up alcohol use or smoking.

GENERAL DISCUSSION

Results demonstrated that IVR technology can be reliably applied to daily process studies in health research with children. The retention and compliance rates were very high and are comparable to those obtained in research using this technology with adults. Importantly, there was no evidence that involvement in a longitudinal study with daily data collection had any adverse effects on children's general disinclination to engage in the potentially harmful behaviors under investigation (i.e., alcohol or cigarette use).

Recruitment and Consent

Despite the 8-week daily commitment required in Study 1, almost half of the contacted children participated. This consent rate is comparable to that for adolescents and young adults (32% to 57%) in a similar study requiring only a 4-week daily commitment (Morrison et al., 1999). One of the primary reasons parents gave for refusing consent was that involvement in the study was too much of a commitment of time and effort. This concern may account for the lower (33%) consent rate in Study 2, which involved not only committing to an intensive data collection period of 4 weeks' duration but also to a 3-year longitudinal project, with the intense 4-week daily data collection periods repeated in each year. Nonetheless, one third of contacted participants could be recruited, resulting in a large sample from a diverse range of metropolitan and rural schools.

Another reason parents gave for refusing consent was that they did not want their children to be exposed to the topics of alcohol and cigarette use. This objection highlights the difficulty of recruiting younger children to participate in research concerning sensitive health-related topics such as substance use. It was therefore encouraging that results from both studies indicated that participants did not report increased intention to use these substances after extended periods of daily exposure to the topics. If parents are made aware of this information during recruitment, it may be possible to achieve higher consent rates.

Retention and Compliance

Retention rates were high in both studies (92% and 91%) and were only slightly lower than those (93% to 97%) achieved in similar IVR studies with adults (e.g., Bardone et al., 2000; Perrine et al., 1995; Searles et al., 1995), even though the monetary incentives offered to the children (i.e., a maximum value of about U.S.$15) were very modest in comparison with financial rewards offered to the adults in the previous studies (e.g., up to U.S.$500 for a 4-week period; Perrine et al., 1995).

Analysis of attrition in Study 2 indicated that the majority of the participants who did not complete the telephone survey component found it too difficult to incorporate the survey into their daily routine. A systematic comparison of the noncompleters with a random sample of completers stratified for grade and gender revealed no significant differences with respect to self-reported frequency and quantity of alcohol consumption or proportion of individuals reporting past cigarette use. In sum, there was little evidence to suggest that noncompleters were systematically different from completers.

Compliance rates averaged just over 90%. In Study 1, compliance remained relatively constant across the 8-week period, whereas in the much larger sample of Study 2, it declined to about 83% during the last week of the daily calling period. These rates for 9- to 13-year-olds are within the same range as those (82% to 93%) achieved with adults in similar studies (e.g., Bardone et al., 2000; Searles et al., 1995). Moreover, 54% of the 605 completers in Study 2 were "perfect callers"-that is, they completed the telephone survey every day of the 4-week period. This percentage compares favorably with the rate of 14% perfect callers reported by Bardone et al. (2000) for a sample of university students.

The excellent compliance rates obtained in both studies suggest that the application of IVR technology is not only feasible in health research with children but offers unique features that appear to be particularly appealing to children. First, because of this method's flexible and convenient access from anywhere that children have access to a telephone, reliable participation is not dependent on the motivation and ability of adult family members or school personnel to facilitate the children's frequent availability for data collection. Second, the verbal presentation of questions is likely to be easy to process by children whose reading levels may vary considerably. Likewise, writing skills are not a limiting factor, because children respond by operating the touch-tone pad of the telephone, or by speaking directly into the phone if openended responses are of interest to the investigator. Clearly, today's children are likely to be savvy users of modern telecommunication tools, such as mobile phones and Web-based applications, and hence it is perhaps not surprising that the children in our studies complied at such high rates and with such ease with the IVR protocol.

Several other features of the IVR protocol are likely to have contributed to good compliance. In addition to ageappropriate training in the use of the technology, we provided ongoing support through a message feature as part of the telephone survey and a separate help line for contacting the project manager. This support was mainly used during the first few days of the survey period, until the participants had become familiar with the routine. As in IVR studies with adults, financial incentives were provided, albeit at substantially lower amounts. It was evident that hearing the current tally of bonus points after each daily survey functioned as token reinforcement in itself. That is, participants expressed enthusiasm at observing their tally increase, and they did not hesitate to call the help line at an early point during Study 2, when a temporary program glitch failed to provide accurate feedback on the current tally of bonus points. Clearly, the regular reinforcement via this automated feature, combined with the children's capacity to actively initiate interactions with the system and the research team, helped to maintain participants' motivation. At the same time, the ability of research staff to use automated messages that could easily be entered into the system on an ad hoc basis allowed for prompt responses to queries or concerns that were relevant to the entire sample. This ability greatly reduced the demand on staff resources in comparison with responding to every participant contact individually.

Limitations

The benefits of using IVR technology in research need to be weighed against the costs of software development and equipment (Leigh, 2000). The relatively high investment in cost and time associated with the initial setup of a custom-designed system appears to be one reason why the full potential of IVR applications for health-related research has yet to be realized. However, as the technology matures, the custom-designed, highly featured products currently in use are likely to evolve into lower cost products that will utilize modularized standard components interfaced with user-friendly desktop applications. Moreover, the costs of establishing IVR systems are partly offset by considerable savings in staff time for data collection and processing (Corkrey & Parkinson, 2002a).

Another limitation concerns the costs of incentives. The financial rewards offered in previous daily IVR studies with adults were substantial, limiting those studies to relatively small sample sizes. In contrast, the modest incentives offered in the present studies were sufficient to maintain high retention and compliance rates while extending the feasibility of this innovative technology for use with large samples. Whether this result would generalize to adult samples is a question for future research (cf. Searles et al., 2000).

Finally, the present studies did not include a control group that did not engage in daily reporting or that used an alternative mode of reporting (e.g., diaries). Hence, our conclusions regarding the feasibility of IVR research with children are based on replication of our findings across two independent and diverse samples and on comparison of our results with published data from comparable studies with adults.

Conclusion

The present studies were the first systematic evaluation of the feasibility of applying IVR technology to daily process research with children. Results demonstrated that children as young as 9 years readily embraced this new technology and reliably engaged in the daily data collection routine. Attrition over extended study periods as long as 8 weeks was low, and compliance with the daily response requirements was excellent. These results compare favorably with findings from similar studies using adult samples. Unlike in previous research with adults, the financial incentives offered in the present studies were modest, suggesting that investigations of larger samples are feasible with children and adolescents. In conclusion, there is no reason for further neglect of children in the application of computer-assisted methods to daily process designs. The advantages of this technology are well established, and the extended scope it affords in investigations of everyday processes affecting the health of our children opens promising avenues for further research.

[Sidebar]

(Manuscript received June 12, 2003; revision accepted for publication June 22, 2004.)

[Reference]

REFERENCES

AFFLECK, G., ZAUTRA, A., TENNEN, H., & ARMELI, S. (1999). Multilevel daily process designs for consulting and clinical psychology: A preface for the perplexed. Journal of Consulting & Clinical Psychology, 67, 746-754.

BARDONE, A. M., KRAHN, D. D., GOODMAN, B. M., & SEARLES, J. S. (2000). Using interactive voice response technology and timeline follow-back methodology in studying binge eating and drinking behavior: Different answers to different forms of the same question? Addictive Behaviors, 25, 1-11.

BOOTH-KEWLEY, S., EDWARDS, J. E., & ROSENFELD, P. (1992). Impression management, social desirability, and computer administration of attitude questionnaires: Does the computer make a difference? Journal of Applied Psychology, 77, 562-566.

CORKREY, R., & PARKINSON, L. (2002a). A comparison of four computerbased telephone interviewing methods: Getting answers to sensitive questions. Behavior Research Methods, Instruments, & Computers, 34, 354-363.

CORKREY, R., & PARKINSON, L. (2002b). Interactive voice response: Review of studies 1989-2000. Behavior Research Methods, Instruments, & Computers, 34, 342-353.

CSIKSZENTMIHALYI, M., & LARSON, R. (1987). Validity and reliability of the experience-sampling method. Journal of Nervous & Mental Disease, 175, 526-530.

FELDMAN, S. I., DOWNEY, G., & SCHAFFER-NEITZ, R. (1999). Pain, negative mood, and perceived support in chronic pain patients: A daily diary study of people with reflex sympathetic dystrophy syndrome. Journal of Consulting & Clinical Psychology, 67, 776-785.

GILLMORE, M. R., GAYLORD, J., HARTWAY, J., HOPPE, M. J., MORRISON, D. M., LEIGH, B. C., & RAINEY, D. T. (2001 ). Daily data collection of sexual and other health-related behaviors. Journal of Sex Research, 38, 35-42.

KIESLER, S., & SPROULL, L. S. (1986). Response effects in the electronic survey. Public Opinion Quarterly, 50, 402-413.

KNAPP, H., & KIRK, S. A. (2003). Using pencil and paper, Internet and touch-tone phones for self-administered surveys: Does methodology matter? Computers in Human Behavior, 19, 117-134.

KOBAK, K. A., TAYLOR, L. V. H., DOTTL, S. L., GREIST, J. H., JEFFERSON, J. W., BURROUGHS, D., MANTLE, J. M., KATZELNICK, D. J., NORTON, R., HENK, H. J., & SERLIN, R. C. (1997). A computer-administered telephone interview to identify mental disorders. Journal of the American Medical Association, 278, 905-910.

LARSON, R. (1989). Beeping children and adolescents: A method for studying time use and daily experience. Journal of Youth & Adolescence, 18, 511-530.

LEIGH, B. C. (2000). Using daily reports to measure drinking and drinking patterns. Journal of Substance Abuse, 12, 51-65.

LITT, M. D., COONEY, N. L., & MORSE, P. (1998). Ecological momentary assessment (EMA) with treated alcoholics: Methodological problems and potential solutions. Health Psychology, 17, 48-52.

MORRISON, D. M., LEIGH, B. C., & GILLMORE, M. R. (1999). Daily data collection: A comparison of three methods. Journal of Sex Research, 36,76-81.

MUNDT, J. C., PERRINE, M. W., SEARLES, J. S., & WALTER, D. (1995). An application of interactive voice response (IVR) technology to longitudinal studies of daily behavior. Behavior Research Methods, Instruments, & Computers, 27, 351-357.

PERRINE, M. W., MUNDT, J. C., SEARLES, J. S., & LESTER, L. S. (1995). Validation of daily self-reported alcohol consumption using interactive voice response (IVR) technology. Journal of Studies on Alcohol, 56, 487-490.

SADEH, A., RAVIV, A., & GRUBER, R. (2000). Sleep patterns and sleep disruptions in school-age children. Developmental Psychology, 36, 291-301.

SEARLES, J. S., HELZER, J. E., & WALTER, D. E. (2000). Comparison of drinking patterns measured by daily reports and timeline follow back. Psychology of Addictive Behaviors, 14, 277-286.

SEARLES, J. S., PERRINE, M. W., MUNDT, J. C., & HELZER, J. E. (1995). Self-report of drinking using touch-tone telephone: Extending the limits of reliable daily contact. Journal of Studies on Alcohol, 56, 375-382.

STONE, A. A., KESSLER, R. C., & HAYTHORNTHWAITE, J. A. (1991). Measuring daily events and experiences: Decisions for the researcher. Journal of Personality, 59, 575-607.

TENNEN, H., AFFLECK, G., ARMELI, S., & CARNEY, M. A. (2000). A daily process approach to coping: Linking theory, research and practice. American Psychologist, 55, 626-636.

TOURANGEAU, R., & SMITH, T. W. (1996). Asking sensitive questions: The impact of data collection mode, question format, and question context. Public Opinion Quarterly, 60, 275-304.

TURNER, C. R, Ku, L., ROGERS, S. M., LINDBERG, L. D., PLECK, J. H., & SONENSTEIN, F. L. (1998). Adolescent sexual behavior, drug use, and violence: Increased reporting with computer survey technology. Science, 280, 867-873.

WHEELER, L., & REIS, H. T. (1991). Self-recording of everyday life events: Origins, types, and uses. Journal of Personality, 59, 339-354.

[Author Affiliation]

WERNER G. K. STRITZKE, JUSTINE DANDY, KEVIN DURKIN, and STEPHEN HOUGHTON

University of Western Australia, Crawley, Western Australia

[Author Affiliation]

This research was supported by Western Australian Health Promotion Foundation (Healthway) Grants 7688 and 9354. We thank all staff and students at the participating schools. Correspondence concerning this article should be addressed to W. G. K. Stritzke, School of Psychology, University of Western Australia, 35 Stirling Highway, Crawley 6009, Australia (e-mail: werner@psy.uwa.edu.au).

Note-This article was accepted by the previous editor, Jonathan Vaughan.