Longitudinal research designs in music therapy: recommendations from a study of people with dementia.

Ledger, Alison ; Baker, Felicity

Abstract

Longitudinal research studies in music the-ap), are not frequently reported within the music therapy literature despite the need for these Within the discipline. This article aims to encourage music therapists to undertake longitudinal research, by providing information on aspects of this type of research design. The authors discuss some advantages of longitudinal research approaches, and present the challenges expected to be faced by those conducting longitudinal research studies. Reflections on the authors' experiences in investigating long-term effects of music therapy, on people with dementia are described, along with recommendations as to how future researchers might address or accommodate some of these challenges.

Key words: music therapy, dementia, longitudinal research, long term effects.

Introduction

In searching for published studies to support the research design of a study undertaken by the first author (1) (hereafter "principal investigator"), there was a shortage of music therapy studies which adopted longitudinal research designs. While longitudinal research has been undertaken in other related medical and education disciplines (Kazdin, 2003; Portney & Watkins, 2000), no documentation of such research in music therapy could be found. The authors felt that such studies should be encouraged, resulting in the development of a longitudinal research study with people with dementia. The impetus of this article evolved through discussions and reflections upon the problems encountered in undertaking this "long term" longitudinal research project. The authors felt this topic was worthy of sharing with others who might consider implementing a long-term project such as this. The following provides (a) an overview of the advantages and challenges of longitudinal research and its application in other related disciplines, (b) an outline of the authors' views on the potential benefits and applications for longitudinal research, (c) descriptions of the obstacles the principal investigator encountered in collecting the data and, finally, (d) recommendations to assist researchers in making decisions about whether to adopt longitudinal research designs.

What is Longitudinal Research?

Longitudinal research involves repeatedly testing pre-defined measures within the same group of people over a long period of time (Kantowitz, Roediger, & Elmes, 2001; Uncles, 1988). The main focus of such methods is the time-ordered study of processes. Here, one tries to discover "how" and "why" a phenomenon exists, examining aspects of both constancy and change over time (Baltes & Nesselroade, 1978). The concept of "over time" is ambiguous when music therapy researchers and clinicians are looking at an effect of treatment. Time can be considered to be over the period of a session, or over a specified program, which might be over weeks, months, and years. In the case of longitudinal research, "longitudinal" is understood as ranging from several months to many years (Leedy & Ormond, 2001). Kazdin (2003) pointed out that the most corm-non time frames adopted in longitudinal research range from one to a few years. Saldaha (2003) stated, "Longitudinal means a, lonnnnnnng time" (P. 1).

Longitudinal research has been considered fundamental to understanding any organism's development (Uncles, 1988). It is usual for repeated time-ordered observations of an individual, or individuals, to be implemented with the goal being to identify processes and causes of intra-individual and inter-individual patterns of change (Baltes & Nesselroade, 1978). In the most traditional method of longitudinal design, known as balanced, all participants are assessed for an equal number of occasions and the timing of these assessments is also identical across all participants (Singer & Willet, 2003). This is distinct from cross-sectional research in which people of different ages or stages of illness are compared at the same point of time (Leedy & Ormond, 2001). For example, a cross-sectional study might compare a group of 5 year olds with a group of 6 year olds in 2005, whereas a balanced longitudinal study would test 5 year olds in 2005 and retest those same children in 2006 when they had turned 6 years of age.

As the above definition lays emphasis on study length as the distinguishing feature of longitudinal research, this indicates that longitudinal studies could fit within quantitative or qualitative (Saldana, 2003) research approaches. This article discusses the advantages and challenges of adopting a longitudinal design within a quantitative method.

Advantages of Longitudinal Research

There are many advantages to adopting longitudinal research designs to measure the effects of treatments. As longitudinal research allows researchers to observe and measure changes over time in treatment and non-treatment groups, researchers can more convincingly explain outcomes as the result of treatment and not some other factor. This is not true of other research designs, in which the direction of the relationship between one characteristic and another may not be clear (Kazdin, 2003). For example, if a study found that participants in music therapy were more expressive than others who did not participate in music therapy, does this mean that music therapy brought about an increase in self-expression? Or were those who sought music therapy treatment more expressive in the first place? A longitudinal research design could actually observe a change in the degree of self-expression over time in both the group of participants who received music therapy and the group of participants who did not.

Assessing participants repeatedly can also verify the accuracy of the initial findings (French, 1993), and does not depend on participants' memory, which may lead to recall bias (Cohen, Manion, & Morrison, 2000). In this way, a longitudinal research method would be a more reliable way of evaluating emotional responses to therapy than relying on people to remember how they felt in the past.

Another advantage of longitudinal research is that participants' personal characteristics remain constant, setting them apart from cross-sectional studies. Cross-sectional studies are subject to cohort effects, in that different participants may be subject to different influences that may affect the results of the research. As an illustration, someone who is 70 in 2005 would have experienced different long-term environmental conditions to someone who is 20 in 2005 (Leedy & Ormond, 2001).

Observing the same people over time is valuable in analysing change at the individual level (Cohen et al., 2000). Longitudinal research can identify individual growth curves and behaviour patterns, and an individual's status can be interpreted in terms of his/her own past growth and unique set of genetic and environmental influences (Burns, 1997). The wealth of time also enables the researcher to examine milestones (e.g., bereavements, hospital admissions) as they arise (Cohen et al., 2000). On a practical level, the time-span of longitudinal studies allows the researcher to fill in gaps in the data later, if an important detail has been omitted during one testing phase (French, 1993).

Longitudinal Research in Related Disciplines

Longitudinal studies have led to increased knowledge in fields related to music therapy, such as medicine, psychology, education, and sociology. These designs have been employed to observe changes that occur with ageing (Portney & Watkins, 2000), and intellectual growth during child development (Leedy & Onnond, 2001; Portney & Watkins, 2000). Birth-cohort studies, which track changes in a large group of people born in the same year, have particularly attracted public interest. The UK National Child Development Study is an example of a birth-cohort study, which has followed the characteristics of people since their birth in one week during March 1958 (Burns, 1997; Cohen et al., 2000). Longitudinal psychology researchers have also given us information regarding the musical development of children from infancy to 12 years of age (Davidson, 1994) (2).

In addition, longitudinal research has been used to investigate the progression of a disease state or to clarify the natural history of a disease (Portney & Watkins, 2000). Other longitudinal studies have identified changing needs of particular groups of people or organisations (Cohen et al., 2000; French, 1993). Longitudinal research has provided convincing evidence regarding causal relationships (Gomm, Needham, & Bullman, 2000), or monitored the impact of an event (e.g., natural disaster, war or injury) on people's health (Jolley, 1999; Kazdin, 2003). In epidemiological studies, these designs have been adopted to generate and test theories about protective, risk, or causal factors for various illnesses. These studies have led to major advances in the understanding of life-threatening illnesses, such as AIDs, heart disease, and various forms of cancer (Kazdin, 2003).

The use of longitudinal studies reflects a change in focus from immediate outcomes to the longer-term well being of people. Longitudinal research designs have evaluated the long-term benefits (or adverse effects) of various therapies by comparing people receiving treatment to people who are not undergoing therapy over time (Crombie & Davies, 1996). In the field of education, longitudinal studies have evaluated the efficacy of various instructional programs (Keeves, 1997). For example, Baloche (1994) assessed the merit of specially designed music education materials in increasing levels of creativity and cooperation amongst fourth-grade students. Those students who undertook the music curriculum showed improved creativity and cooperation at the end of the school year, as compared with students who did not undertake the music curriculum.

Potential for Longitudinal Research in Music Therapy

Existing music therapy literature indicates that more rigorous longitudinal testing could contribute to the knowledge currently available. Music therapy is defined as "a. process, something that takes place over time and involves growth, change, and development" (Schmidt Peters, 2000, p. 3). Music therapy is "evolutionary rather than momentary, sequential rather than singular and gradual rather than sudden" (Brusica, 1998, p. 33). Longitudinal research is therefore ideal for capturing some of the processes that occur in music therapy.

Studies have shown that music therapy can assist people to develop skills over time. A recent example is Register's (2001) evaluation of the effect of an early intervention music curriculum on four-year olds. The children showed significant improvements in prewriting and print concepts over the course of a school year, and particularly when music sessions were designed with specific academic measures in mind. Similarly, Wager (2000) reported on the gains made by an adult with autism over four years of music therapy. These included improvements in awareness, interaction, and fine and gross motor skills.

The potential for implementing and evaluating long-term effects of treatment using longitudinal designs is further supported by the emerging findings that music therapy may produce cumulative effects over time. In a study of 10 people who had cerebral vascular accidents or traumatic brain injuries, it was observed that those who attended more music therapy sessions showed greater improvement in social interaction capacities (Wheeler, Shiflett, & Nayak, 2003). Similarly, participants with Alzheimer's disease sat significantly longer with an increasing number of music therapy sessions and as they became increasingly familiar with the therapist (Groene, 1993). The effects of music therapy were significantly greater for cancer patients who received a greater number of sessions when compared with those who received fewer sessions (Hilliard, 2003). While participants became increasingly ill, their quality of life was enhanced over time. This was not true of a control group of people not receiving music therapy, whose quality of life scores decreased with declining physical health (Hilliard, 2003).

In some settings, the goal of music therapy is not to improve but to maintain a person's social, cognitive, communication, or physical skills over time. The understanding that music therapy may maintain skills of people with Alzheimer's disease is supported by reports of long-terra clinical work (Clair & Bernstein, 1990x; Clair & Bernstein, 1990b; Hanser & Clair, 1995). The case study of "Mr O" is a good example (Clair, 1991). Mr O maintained participation in music despite 3 years of marked physical and cognitive deterioration. Within music therapy, he maintained the ability to interact successfully and appropriately with others. Longitudinal studies involving large subject samples and control groups could further support assertions that music therapy develops, increases, or maintains people's skills over time.

Aside from the outcomes of treatment, longitudinal research can also focus on clinical issues such as the effects of different treatment approaches employed over time. For example, Rolvsjord (2001) presented a case study in which didactic music therapy with a seventeen-year-old girl developed into more psychotherapeutic work over the course of a year and a half. In a similar way, Glyn (2003) described the change from an exclusively music-based approach to a more verbal one, in four and a half years of music therapy with a forensic psychiatric patient. In addition, longitudinal research could track the effects of outside influences, such as relocation (Atkinson, 2003), separation from a loved one (Lipe, 1991), or change in medication (Wager, 2000) on a person's progress in music therapy. These examples indicate that longitudinal research could bring to light evolving music therapy approaches that are necessary for successful outcomes.

The importance of consumer views in today's climate of greater consumerism and freedom of information has been emphasised (Moss, 2003). Health care recipients are increasingly being described as customers, and customer satisfaction levels are now frequently used to plan, prioritise, and deliver better health services. Longitudinal research methods could be used to follow-up whether clients were satisfied with their music therapy treatment, and Moss (2003) has already begun to use research for these purposes. She administered a feedback questionnaire to older adults with physical disabilities or frailty who had received music therapy in the prior six months. Results from this questionnaire led to recommendations for future projects. Maack & Nolan (1999) followed-up 25 former Guided Imagery and Music clients, who had completed their last session 2-78 months prior to the study. The researchers sought to determine whether the clients believed a change had occurred in GIM therapy and whether gains were maintained afterwards. Questionnaire responses showed that clients felt satisfied with their GIM treatment, and were able to foster additional change or maintain the positive effects independently after termination of therapy. Like Moss (2003), Maack & Nolan's (1999) results revealed areas for further research.

A further study by Wheeler (2003) addressed music therapists' perceptions of a symposium that they had attended. Wheeler (2003) sought input from participants in The First International Symposium on Music Therapy Training 23 years after the meeting. From letters, e-mails and phone calls, Wheeler learned the impact of the symposium on participants and the development of music therapy in their countries. Participants' responses also highlighted needs for sharing and translating literature from other countries, as well as respecting the diverse approaches existing worldwide. Valuing the opinions of others, whether they are clients or other music therapists, and consulting them about the development of music therapy is potentially empowering for all involved. Longitudinal research employing questionnaires with established psychometric properties could assist in this regard.

An Application of Longitudinal Research Designs for People with Middle-Stage Dementia

The idea for the principal investigator's study emerged from her music therapy work with people with dementia who displayed severe agitated behaviours. These behaviours included distressing behaviours such as calling out, insulting others, wandering, or clinging to people. While her music therapy sessions with people who have dementia was designed to promote appropriate social interaction, she wanted to create something more preventative or empowering for these people. An examination of the music therapy literature showed that while several music therapy studies have demonstrated the immediate effects of music in reducing existing agitated behaviours, few have explored whether music therapy can have more lasting effects on people with dementia. As reported lengths of Alzheimer's disease are as long as 21 years (American Psychiatric Association, 2000), and music therapists often develop relationships with people with dementia over several years, longitudinal research with this population was considered warranted and well overdue.

While the focus of this paper is not to discuss this research project (3), it is necessary to provide a short description in order to orient the reader to the discussions of the problems encountered along the research journey. The purpose of the research was to investigate whether long-term group music therapy programs, early in the dementia progression, can reduce the onset of agitated behaviours shown by people with Alzheimer's disease. As music therapy assists participants to sustain feelings of identity, achievement, control, belonging, and purpose (Rued, 1998), it was predicted that group sessions would also lead to reductions in agitation.

Thirty people attending weekly group music therapy were followed for one year and compared with 30 people who were not receiving music therapy. People who were rated as mild to moderate on the Reisberg Global Deterioration Scale (Reisberg, Ferris, de Leon, & Crook, 1982) were of particular interest. People within this stage of disease progression are favoured for clinical trials due to their relatively high degree of change within a year (Gauthier, Poirer, & Gray, 1996). In Australia, people in the mild to moderate stages are more likely to be found in community settings, but for ease of recruiting and follow-up the primary investigator was studying residents of 13 various nursing homes. Measures of participants' agitation were being collected every 3 months over 1 year, using the Cohen-Mansfield Agitation Inventory (Cohen-Mansfield, Marx, & Rosenthal, 1989). This is a tool for which validity and reliability have been established (Foli & Shah, 2000), and one that has been used in previous music therapy and agitation studies (Nugent, 2002). It measures the frequency of 29 agitated behaviours during the previous 2 weeks. Scores were gained through questioning a caregiver and were summarised according to the type and severity of behaviour described (physical aggressive, physical non-aggressive, verbal aggressive, or verbal non-aggressive). It was hypothesised that music therapy participants would manifest fewer, less frequent, or less severe agitated behaviours over time, than those not receiving music therapy.

Challenges of Longitudinal Research

Whilst investigation at different points in time gives the longitudinal research design its strength, the use of different time points gives rise to multiple theoretical and practical problems (Keeves, 1997). Researchers are often forced to make compromises and longitudinal data may become less reliable.

Firstly, longitudinal research requires extended obligation to a single project. Once a longitudinal study is begun its design cannot be changed according to changes in thought and practice, which occur over time (Portney & Watkins, 2000). This can be frustrating if one discovers a "better" way of doing something two months into a 12-month project. Such experiences were evident during this study as the results of new studies on dementia emerged and their implications on the current study became clear. For example, the initial understanding that agitation worsens systematically over time with the progressive decline of other symptoms of Alzheimer's disease was found to be inaccurate. Agitation is more episodic in nature and behaviours are known to fluctuate over time (Eustace et al., 2002). Such knowledge would suggest that studying agitation long-term may be complex and that it would be easier to show significant effects of music therapy in maintaining other areas of functioning (e.g., cognitive skills, communication skills). However, once the research had commenced, change in the design was not possible.

On a general level, longitudinal research approaches imply collecting data on the same subjects over a period of time and therefore demand that researchers be highly accurate and scrupulous when collecting their data (Magnusson, 1991). A mistake made (either by the researcher, the research assistant, or other members directly involved in the project) can seldom be satisfactorily remedied at a later stage. It is rarely possible to "go back" and collect the data if it were not collected, incorrectly collected, or incomplete at the scheduled assessment time. This became apparent in the current study, when a research assistant collected data for control participants at incorrect time-points (i.e., four rounds of data within 7 months, rather than 5 rounds of data within 12 months). This was only discovered when the research assistant submitted several rounds of data. As a result, two further agitation measurements were required and the investigators had to wait 3 months longer than expected to gain final data.

Since both the participants and support staff are asked for a long-term commitment of their time, it is even harder to recruit for longitudinal studies than for other types of research. As it is known that participants who volunteer for research are not necessarily representative of a population as a whole, it is likely that this bias is even more accentuated in the sample of participants volunteering for longitudinal research (French, 1993). Furthermore, few major investigations of this type are able to allocate participants randomly to experimental and control groups. Instead, convenience samples are chosen as a more practical alternative (Keeves, 1997) and this means that results have limited generalisation to a population as a whole. In order to gain a sufficient number of participants to determine treatment significance in this study, 1 1 music therapists were approached to request assistance in carrying out the music therapy treatment condition. Of these 11, only one was able to commit to the year long study. As a result, most of the experimental participants were chosen from nursing homes that were convenient for the principal investigator to visit each week. The recruitment phase was by far the most difficult period in the research. As data was collected from nursing homes where music therapy services had never been previously provided, it was considered an invasion of privacy for the researchers to access residents' medical records before consent for research participation had been gained. The researchers therefore needed to rely on nursing home staff to choose appropriate residents for the research. Unfortunately, the sample recruited for participation did not f t the selection criteria precisely and this was only evident after data collection had commenced. Due to pressure to complete the study within a given time frame, these participants who did not fit selection criteria were included in the study. The study results will therefore have limited generalisability. It is envisaged that similar time constraints would be part of studies funded by various research grants and other researchers should anticipate these problems when planning a time line for study completion and when preparing their budget for the grant applications.

Another obstacle in longitudinal research is participant attrition, which is again more accentuated when participants need to commit for long periods of time. In the case of research involving older adults, participant deaths are a likely cause of attrition. Of the 60 participants in the study, 14 died during the 12-month data collection period. Interestingly, 11 died in the control group, but only three died amongst those receiving music therapy. This poses many problems when attempting to analyse and interpret the findings. Further, participant attrition weakens the internal validity of the research, in that those participants who drop out may not be typical of the participants as a whole (French, 1993). In the initial phase of the study, a couple of participants chose not to attend music therapy sessions and were therefore dropped from the study. One may question whether these participants were actually more agitated or sicker than those who continued, and whether this upset the balance between the experimental and the control groups. Researchers need to consider that in undertaking longitudinal research with people who have dementia, as many as 25% of the initial sample may be lost through natural attrition. Thus the sample may need to be substantially bigger than the number required to meet the power of the study to anticipate such significant attrition.

Nevertheless, participant attrition, particularly due to participant deaths, raises problems when deciding how to treat their missing data. Recent trends in statistics include retaining participants' data up to the point when they dropped out and then imputing scores for the remaining data in an ad hoc way (e.g., using the participant's mean score or another participant's score), or creating remaining data points based on the person's position on the distribution prior to leaving the study (Cohen-Mansfield, & Werner, 1999; Newman, 2003). Regardless of which approach is chosen, it is important to consult a mental health researcher experienced in researching people with dementia prior to making such a decision.

When conducting a longitudinal study, researchers must also consider that their results may be subject to history effects. In other words, there may be something special about the time of research, which makes the results specific to the group being studied (Kazdin, 2003). This leads to the question of whether the results of the current study can be generalised to people with dementia residing in nursing homes in 20 years time or whether they are only relevant for people who have experienced similar histories to those in the study. For example, future nursing home residents may manifest agitation differently to current residents who have experienced hardships such as World War II, the Depression, and limited access to technologies. There is also evidence that nursing home residents' levels of dependency and confusion are increasing with time (Darton, Netten, & Forder, 2003). Furthermore, there will undoubtedly be advances in medical treatments for agitation and changes in ways in which health care is delivered to older people with Alzheimer's disease in the future.

The repeated research procedures in longitudinal research may have an effect on participants' behaviours. Participants may become practised in various tests or repeated interviewing may lead to sensitisation to things that had previously gone unnoticed (Cohen et al., 2000; French, 1993). It is possible within the current study that nurses completing the Cohen-Mansfield Agitation Inventory became practised at recognising behaviours as the research progressed, endorsing the presence of behaviours which they did not detect during the earlier stages of the study. This may give the impression that agitated behaviours have worsened when this may not be the case.

Due to the sheer length of time involved in longitudinal research, it is impossible to control for the wide range of variables that may influence the results in that time (French, 1993; Kazdin, 2003; Portney & Watkins, 2000). There are many outside factors that could affect a person's agitation levels during one year; an illness, hospitalisation or change in medication, a move into a dementia specific unit, or death of a friend or spouse. These potentially confounding variables will impact on the reliability of the findings if music therapy was found to positively (or negatively) influence changes in agitation.

On a more practical level, the advantages of longitudinal research come at the price of time and money (Jolley, 1999). Longitudinal designs are not ideal when immediate answers are needed (Kazdin, 2003). They are slow to produce results and there have certainly been times when the investigators' interest in this project waned. This was particularly the case early on in the research, when interpretable results seemed so far way.

Longitudinal studies are difficult to undertake when researchers are often subject to pressure to complete research and carry out multiple projects to secure promotion and future funding (French, 1993). As alluded to earlier, the restriction of completing the research within the time frame of a Master of Philosophy degree necessitated many compromises. These compromises weakened the power of the study.

Longitudinal research requires long-term commitment of funds and resources (Portney & Watkins, 2000), which can be difficult to secure in the first place. Replacement of research assistants is often necessary as people move, retire, change jobs, or pursue other interests (French, 1993). Although one music therapist volunteered to collect data for the control group at her facility, her task became difficult when she became ill and her hours of work at that facility were reduced. Further, the high staff turn over in the aged care industry meant that the principal investigator needed to continuously re-educate staff about music therapy, research protocols, and the assistance needed to collect the correct data. Gaining consistent input from staff is difficult when high turnover is "almost endemic" amongst Australian aged care workers (Bird, Llewellyn-Jones, Smithers, & Korten, 2002, p. 12).

Due to the number of testing phases required, longitudinal research can be expensive to conduct. Furthermore, due to inflation, costs are likely to increase year to year.

Other Recommendations for Future Longitudinal Research Designs

When considering undertaking a longitudinal study; it is important to consider influences that may be beyond the researchers' control. Regulation of the study environment and people's actions for an entire year are usually not possible. Implementing a-longitudinal research design is a difficult and challenging undertaking so it is recommend that researchers engage assistance and support from others. With large sample sizes comes the need for research assistants who understand what is required, are reliable, and who are committed to the project. Most important of all, when considering conducting a longitudinal study, secure financial back up is necessary as longitudinal studies are costly.

Conclusions

Insufficient longitudinal studies exist in music therapy and the profession stands to benefit from an increase in research that adopts this method. Longitudinal studies could capture progress or maintenance in music therapy, demonstrate the evolution of clinical approaches, empower clients by gaining their views on past music therapy treatment, or establish the changing needs of music therapy students and practitioners over time. The longitudinal design presents the researcher with some challenges though. These include commitment, participant attrition, practice effects, confounding variables, and cost. However, these obstacles can be overcome provided one is informed, organised, supported, willing to compromise, and able to study a large number of participants. It is hoped that others have been inspired to undertake a longitudinal research journey through reading this article.

References

American Psychiatric Association. (2000). Diagnostic and statistical manual of mental disorders. DSM-IV-TR. Washington, D.C.: Author.

Atkinson, C. (2003). The longest goodbye--a case study. British Journal of Music Therapy, 17,90-96.

Baloche, L. (1994). Creativity and cooperation in the elementary music classroom. Journal of Creative Behavior, 28, 255-265.

Baltes, P. B., & Nesselroade, J. R. (1978). History and rationale of longitudinal research. In J. R. Nesselroade & P. B. Baltes (Eds.), Longitudinal research in the study of behavior and development. (pp. 1-39). New York: Academic Press.

Bird, M., Llewellyn-Jones, R., Smithers, H., & Korten, A. (2002). Psychosocial approaches to challenging behaviour in dementia: A controlled trial. Canberra: Commonwealth of Australia.

Bruscia, K. (1991). Case studies in music therapy. Philadelphia: Barcelona.

Bruscia, K. E. (1998). Defining music therapy (2nd ed.). Philadelphia: Barcelona.

Burns, R. B. (1997). Introduction to research methods (3rd ed.). Melbourne: Addison Wesley Longman.

Clair, A. A., & Bernstein, B. (1990a). A comparison of singing, vibrotactile and nonvibrotactile instrumental playing responses in severely regressed persons with dementia of the Alzheimer's type. Journal of Music Therapy, 27, 119-125.

Clair, A. A., & Bernstein, B. (I 990b). A preliminary study of music therapy programming for severely regressed persons with Alzheimer's-type dementia. The Journal of Applied Gerontology, 9, 299-311.

Cohen, L., Manion, L., & Morrison, K. (2000). Research methods in education (5th ed.). London: Routledge Falmer.

Cohen-Mansfield, J., Marx, M. S., & Rosenthal, A. S. (1989). A description of agitation in a nursing home. Journal of Gerontology: Medical Sciences, 44(3), M77-84.

Cohen-Mansfield, J., & Werner, P. (1999). Longitudinal predictors of non-aggressive agitated behaviors in the elderly. International Journal of Geriatric Psychiatry, 14, 831-844.

Crombie, l. K., & Davies, H. T. O. (1996). Research in health care: Design, conduct and interpretation of health services research. Chichester: John Wiley & Sons.

Davidson, L. (1994). Song singing by young and old: A developmental approach to music. In R. Aiello & J. A. Sloboda (Eds.), Musical perceptions (pp. 99-130). London: Oxford University Press.

Darton, R., Netten, A., & Forder, J. (2003). The cost implications of the changing population and characteristics of care homes. International Jozn-nal of Geriatric Psychiatry, 18, 236-243.

Eustace, A., Coen, R., Walsh, C., Cunningham, C. J., Walsh, J. B., Coakley, D., & Lawlor, B. A. (2002). A longitudinal evaluation of behavioural and psychological symptoms of probable Alzheimer's disease. International Journal of Geriatric Psychiatry, 17, 968-973.

Foli, S., & Shah, A. (2000). Measurement of behavioural disturbance, non-cognitive symptoms and quality of life. In J. O'Brien, D. Ames, & A. Burns (Eds.), Dementia (2nd ed., pp. 87-100). London: Arnold.

French, S. (1993). Practical research: A guide for therapists. Oxford: Butterworth-Heinemann.

Gauthier, S., Poirier, J., & Gray, J. (1996). Effects on decline or deterioration. In R. Becker, & E. Giacobini (Eds.), Alzheimer's disease. From molecular biology to therapy (pp. 381-385). Boston: Birkhauser.

Glyn, J. (2003). 'New York mining disaster'. British Journal of Music Therapy, 17, 97-103.

Gomm, R., Needham, G., & Bullman. A. (Eds.). (2000). Evaluating research in health and social care. London: Sage.

Groene, R. W. (1993). Effectiveness of music therapy 1:1 intervention with individuals having senile dementia of the Alzheimer's type. Journal of Music Therapy, 30, 138-157.

Hanser, S. B., & Clair, A. A. (1995). Retrieving the losses of Alzheimer's disease for patients and care-givers with the aid of music. In T. Wigram, B. Saperston, & R. West (Eds.), The art and science of music therapy: A handbook (pp. 342-360). Switzerland: Harwood Academic.

Hilliard, R. E. (2003). The effects of music therapy on the quality and length of life of people diagnosed with terminal cancer. Journal of Music Therapy, 40, 113-137.

Jolley, D. (1999). Epidemiological research. In V. Minichiello, G. Sullivan, K. Greenwood, & R. Axford (Eds.), Handbook for research methods in health sciences (pp. 201-229). Sydney: Addison-Wesley.

Kazdin, A. E. (2003). Research design in clinical psychology (4th ed.). Boston: Allyn and Bacon.

Kantowitz, B. H., Roediger, H. L., & Elmes, D. G. (2001). Experimental psychology: Understanding psychological research (7th ed.). Belmont: Wadsworth.

Keeves, J. P. (1997). Longitudinal research methods. In J. P. Keeves (Ed.), Educational research, methodology, and measurement: An international handbook (2nd ed., pp. 138-149). Oxford: Elsevier Science.

Leedy, P. D., & Ormond, J. E. (2001). Practical research planning and design (7th ed.). Upper Saddle River, New Jersey: Merrill Prentice Hall.

Lipe, A. W. (1991). Using music therapy to enhance the quality of life in a client with Alzheimer's dementia: A case study. Music Therapy Perspectives, 9, 102-105.

Maack, C., & Nolan, P. (1999). The effects of guided imagery and music therapy on reported change in normal adults. Journal of Music Therapy, 36,39-55.

Magnusson, D. (1991). Preface. In D. Magnusson, L. R. Bergman, G. Rudinger, & B. Tdrestad (Eds.), Problems and methods in longitudinal research: Stability and change. (p. xv). Cambridge, NY: Cambridge University Press.

Moss, H. (2003). Service evaluation: Music therapy and medicine for the elderly. British Journal of Music Therapy, 17, 76-89.

Newman, D. A. (2003). Longitudinal modeling with randomly and systematically missing data: A simulation of ad hoc, maximum likelihood, and multiple imputation techniques. Organizational Research Methods, 6,328-62.

Nugent, N. (2002). Agitated behaviours in Alzheimer's disease and related disorders: Music and music therapy research. Australian Journal of Music Therapy, 13, 38-50.

Portney, L. G., & Watkins, M. P. (2000). Foundations of clinical research: Applications to practice (2nd ed.). Upper Saddle River, New Jersey: Prentice Hall Health.

Register, D. (2001). The effects of an early intervention music curriculum on prereading/writing. Journal of Music Therapy, 38, 239-248.

Reisberg, B., Ferris, S. H., de Leon, M. J., & Crook, T. (1982). The global deterioration scale for assessment of primary degenerative dementia. American Journal of Psychiatre, 139,1136-1139.

Rolvsjord, R. (2001). Sophie learns to play her songs of tears. Nordic Journal of Music Therapy, 10, 77-85.

Ruud, E. (1998). Music, health, and quality of life. In Music therapy: Improvisation, communication and culture (pp. 49-67). Philadelphia: Barcelona.

Saldana, J. (2003). Longitudinal qualitative research: Analyzing change through time. Walnut Creek,CA: Alta Mira Press.

Schmidt Peters, J. (2000). Music therapy an introduction (2nd ed.). Springfield, Illinois: Charles C. Thomas.

Singer, J. D., & Willet, J. B. (2003). Applied longitudinal data analysis: Modelling change and event occurrence. Oxford: Oxford University Press.

Uncles, M. D. (1988). Longitudinal data analysis: Methods and applications. London: Pion.

Wager, K. M. (2000). The effects of music therapy upon an adult male with autism and mental retardation: A four-year case study. Music Therapy Perspectives, 18, 131-140.

Wheeler, B. L. (2003). First international symposium on music therapy training: A retrospective examination. Nordic Journal of Music Therapy, 12(1), 54-66.

Wheeler, B. L., Shiflett, S. C., & Nayak, S. (2003). Effects of number of sessions and group or individual music therapy on the mood and behavior of people who have had strokes or traumatic brain injuries. Nordic Journal of Music Therapy, 12, 139-151.

Alison Ledger BMus(Mus Thy)(Hons), RMT

Sionna Academy of Music and Dance

Irish World Music Centre, University of Limerick

Felicity Baker PhD, NMT, RMT

School of Music, The University of Queensland

(1) The first author was the principal investigator and directly involved in conducting the clinical work associated with the study referred to in this article. She completed this research as pant of a Master of Philosophy. The second author was the co-investigator and supervisor to the project.

(2) Davidson (1994) provided an overview of research that addresses developmental changes in musical ability (song singing in particular), and advocated the examination of individuals over time rather than groups of people in different age groups.

(3) The findings of the research have not been finalised and will be published elsewhere at a later date.