Journal of Behavioral Medicine

, Volume 36, Issue 4, pp 341–346 | Cite as

Placebo by proxy: the effect of parents’ beliefs on therapy for children’s temper tantrums



A placebo by proxy effect occurs when a patient’s response to therapy, assessed either objectively or subjectively, is affected by the behavior of other people who know that the patient is undergoing therapy. We recruited 58 children aged 2–5 years who reported frequent tantrums and examined the effect of a pharmacologically inert substance (flower essence) that is purported by the manufacturers to reduce temper tantrums. Tantrum frequency, tantrum severity, and parental mood were measured on 5 occasions over 8 days before treatment and on a further 5 occasions over 10 days after the start of treatment. Compared to the period before treatment, there was a continuing reduction in tantrum frequency (p = .002) and severity (p = .003) over the 8 days of placebo treatment. There were significant day-to-day correlations between parents’ mood and tantrum frequency (r = .23) and severity (r = .19). Children’s response to treatment for tantrums is associated with the beliefs and mood of the adult carer. We cannot say whether tantrum reduction was due to objective changes in child behavior, changes in parental perception, or both, but both are clinically important changes.


Placebo Children Tantrums Parent 


The placebo effect is well established—patients respond in accordance with the meaning which they perceive in the therapeutic encounter. However, patients are influenced not only by their own beliefs, but also by the behavior of others, including those delivering the therapy, their friends and relatives. The behavior of these other people is influenced by their beliefs about the patient’s treatment. The concept of placebo by proxy is comparatively recent (Grelotti & Kaptchuk, 2011). Placebo by proxy suggests that patients’ response to therapy is influenced by the behavior of others whose behavior is altered by knowing that the patient is receiving therapy. Although theoretically plausible, the concept of placebo by proxy has received little empirical investigation. Research is hampered by the difficulty of distinguishing the effect of the patient’s beliefs from the effect of others’ beliefs, as both are likely to share the same belief that the therapy is effective.

Very young children provide an opportunity for investigating the placebo by proxy effect, as young these children have immature belief systems. There is strong evidence that parental expectancies have a significant influence on both reports of child behavior and parent–child interactions. Hoover and Milich (1994) manipulated information provided to the mothers of boys who were given a sweet but sugar-free drink and measured mothers’ perceptions of behavior, physical activity of the child (using wrist and ankle actometers), and made additional observations of mother-son interactions. Mothers who were (incorrectly) informed that their sons had ingested sugar rated them as more hyperactive and less compliant than those provided with accurate information. However, wrist actometer measurements showed that these boys were, in fact, less physically active. Behavioral observations indicated that mothers in the sugar-expectancy condition exercised more control over their children, maintained greater physical closeness, and were more likely to criticise, look at, or talk to their sons than mothers in the control condition.

Additionally, researchers investigating stimulant treatments for attention deficit/hyperactivity disorder (ADHD) note that the dose–response curves for subjective measures (typically parental ratings of behavior) differ from those based on objective measures (e.g., tests of cognitive or other academic performance). Correlations between subjective and objective measures are often low, and subjective measures typically show greater improvements in situations where parents have positive expectancies for outcomes (Waschbusch et al., 2009). These findings lead to the possibility that the meaning of the child’s therapy to the parent may affect the therapeutic response of the child, either because of a change in the parent’s behavior that then leads to objective changes in the child’s behaviour, or because the parent perceives the child to behave differently.

In this study we investigated whether a placebo treatment would reduce the frequency of reported temper tantrums in very young children. Temper tantrums can be distressing for parents and are associated with health problems in young children (Scarborough et al., 2007; Potegal & Davidson, 2003). Parental behavior is known to be a factor influencing behavior problems in children, either through inappropriate reinforcement patterns (Strand, 2000) or through the effects of parental distress (McDonald et al., 2007). Because parents are implicated in the development of tantrums, it is possible that knowing that the child is receiving treatment will alter the behavior of the adult and hence influence the child. In particular, the parent’s knowledge that the child is being treated could lead to affective changes that reduce parental distress and lead to a more positive relationship with the child.

The aims of our study were (a) to investigate whether the frequency of reported temper tantrums was reduced by a pharmacologically inactive treatment and (b) whether the reduction in temper tantrums was associated with changes in parental mood.



Parents were recruited, provided informed consent and demographic information, and registered via a study website. Assessments of tantrums began immediately and continued for 8 days. During this period, flower essences and instructions were sent to by post. After the assessment on day 8, parents began flower essence treatment and assessments continued for a further 10 days. A follow-up question was sent via email. Excluding technical support, there was no human contact with parents, who provided data in response to telephone calls generated by an automated telephone system.


The study was advertised in local, and online media. Volunteers were offered a free bottle of flower essence in return for questionnaire completion. Volunteers were required to: be the primary caregiver of a child aged between 2 and 5 years; have a child who suffered from moderate to severe tantrums (defined as 1 or more tantrums lasting at least 5 min per day); and to own a mobile phone. Exclusion criteria were non-UK residence and use of flower essences within the past 6 months.

Participants were told that the researchers were investigating the effectiveness of a commercially available flower essence. Participants were told that the researchers neither endorsed nor rejected the treatment as a way of helping parents of children with temper tantrums, but participants were provided with information which was identified as coming from those who believed in the usefulness of flower essences.

Flower essence therapy

Bach flower essences are a common, over-the-counter, complementary remedy. Results of meta-analytic studies suggest that flower remedies have no specific effects, and are almost certainly pharmacologically inert (Ernst, 2010), and have been used in other placebo studies (Hyland et al., 2007; Hyland & Whalley, 2008). The essences are manufactured by floating flower blossoms in a crystal bowl filled with water, and this ‘stock’ essence is then highly diluted with brandy and water. No detectable trace of the original blossom remains, but proponents of Bach remedies suppose that they help via a spiritual mechanism as yet unknown to science. Complementary practitioners have added to the original Bach flower essence remedies and make special combinations for particular purposes; many hundreds of products are now sold purporting to treat specific conditions. In this study we used a product designed to reduce tantrums in young children made by the Green Man Essence Company. We used marketing materials from the company as the basis for information about flower essences given to parents. In this study flower essences were contextualised as a ‘new-age’ spiritual therapy: information booklets emphasised the role of the “vibrational essence of the flower”, and after each telephone call parents were told that: “Flower essences work best if, while giving them to your child, you imagine the spirit of the flower connecting you and your child to a pool of universal healing and love”. These instructions are consistent with the practice of complementary therapists and have been shown to increase the effectiveness of flower essences for spiritually-oriented patients (Hyland & Whalley, 2008).

Outcome assessments

Tantrum frequency

Parents were asked ‘How many tantrums has your child had within the past 24 h?’ Responses were made using the parent’s telephone keypad. Provided the child had experienced at least one tantrum, parents were asked about the severity and additional characteristics of the worst tantrum in the past 24 h.

Tantrum severity

Parents were asked: ‘On a scale of 1–9, how bad was the tantrum?’ Responses were made using the keypad, where 1 was ‘not bad at all’, 5 was ‘about average’, and 9 was ‘very bad’. In addition, characteristics of the worst temper tantrum in the measurement period were assessed. Single questions measured the incidence of ‘screaming’, ‘kicking and hitting’, and ‘crying’, and ratings were given from not at all (1) to a great deal (9). Responses to all four items were averaged to provide a tantrum severity score for each occasion.

Parental affect

Affect of parents was measured with an individual item in each call: “How happy are you today?” Responses were given on a scale ranging from ‘much more unhappy than normal’ (−4), ‘about normal’ (0), to ‘much happier than normal’ (4).


Parents were asked: ‘How many times did your child take the essence yesterday?’ and responded ‘not at all’ (0), ‘once’ (1), or ‘twice’ (2; two doses was the daily recommended maximum).


Ethical approval was granted by the University of Plymouth Human Participants ethical committee. After recruitment, parents provided consent via the study website and completed demographic information. Next, parents were provided with a unique identity code and asked to call an automated telephone line to complete the signup process by entering this code and the mobile telephone number on which they wished to receive assessment calls. The automated system then redialled the parent’s number to complete the first assessment call (day zero). Only those who completed the first assessment were sent flower essences.

Assessment calls continued for the period before treatment—in total 5 calls on days 0, 2, 4, 6 and 8. During this time a sample of the essence and an instruction pack was posted to parents, timed to arrive no earlier than day 8, and with a note that the essence was not to be used until day 9. Assessments continued on days 10, 12, 14, 16 and 18. For the assessment on day 10, the automated system inserted additional questions to check that (a) the essence had been received and (b) treatment had begun. If either of these conditions were not met, all further assessment calls were delayed by 1 day; the maximum delay for any parent was 3 days.

All assessments included measures of tantrum frequency and parental affect. The automatic telephone system was programmed to measure tantrum severity only if at least one tantrum was recorded. Additionally, assessments after the start of treatment (days 10–18) also included the measure of adherence, a reminder to take the essence in accordance with the instructions, and the message contextualising the essence as a spiritual therapy. After the study was completed, parents were sent an e-mail and invited to provide written comments via e-mail about their experience, and they were also asked to indicate whether their child had been told about the reason the child was being given the flower essence.


We modelled the frequency and severity of tantrums during the course of the study using a two-level random effects model (tantrum frequency was modelled as count data with the appropriate link function; Rabe-Hesketh, & Skrondal, 2008). Two slopes were estimated: the first for pre-treatment scores (≤day 8); the second for responses after treatment began (>day 8). Both models were conditioned on baseline (day 0) scores. We compared slopes before and after treatment, and tested this difference with the Wald test. Estimated means presented for each day were computed via the delta method. Additional multivariate multilevel models were used to estimate the association between parental mood and tantrums. These models provide separate estimates for correlation between mood and tantrums (a), within day-to-day reports for individual pairs of parents and children and (b), between parents, comparing means of mood and tantrum reports.


Of 104 volunteers who completed baseline assessments, 15 did not register a telephone number. Of the 89 who successfully registered a telephone number and received flower essences, 58 answered at least one call before and one call after beginning treatment and are included in the analyses. All were parents, of whom 48 (83 %) were mothers, 6 (10 %) fathers, 4 (7 %) unknown). Within this sample, 41 parents (71 %) completed at least 2 calls before and after treatment (i.e., responded to at least 4 calls), and 23 (40 %) completed at least 3 before and 3 after. The mean child age was 3.4 years (SD = 1.2 years) and there were 34 (59 %) boys, 18 (31 %) girls, and 6 (10 %) unknown.

Figure 1 shows the frequency and severity of tantrums before and after treatment. The Wald test indicated the slopes for severity and frequency were significantly different before and after treatment (for severity, chi2(1) = 8.82, p = .003; for frequency chi2(1) = 9.57, p = .002).
Fig. 1

Model estimated tantrum frequency and severity before and after treatment began (indicated by vertical dashed line)

Table 1 presents day-to-day and mean-level correlations between parental mood and tantrums. These results show that there were significant associations between parental mood and the frequency and severity of tantrums as they varied from day-to-day during the study. The variance components models indicated that day-to-day variation accounted for 54, 67, and 74 % of total variation in tantrum frequency, tantrum severity, and parental affect reports, respectively. Effect sizes from single group designs are often larger than those from group comparisons (Durlak, 2009). Nonetheless, the effect we observed was substantial. The standardised pre-post treatment difference was D = .73.
Table 1

Correlations between tantrum frequency, severity, and parental affect (results from multivariate variance-components model)




Between-individual (study mean) correlations



 Parent affect



Day-to-day correlations



 Parent affect



* p < .05, ** p < .01, *** p < .001

Thirty five parents responded to the post-study e-mail and 29 (83 %) said that they had not told their child about the effect of flower essences. Supplementary analysis showed that there was no significant difference between children who had been told or not, but that there was a trend towards a smaller reduction in tantrum frequency and severity for those who had been told. Seven parents provided written text describing their experience (shown verbatim in the “Appendix”). These show that for some parents the perceived effect of the flower essence was substantial.


We carried out naturalistic open label study to investigate the effect of a commercially available flower essence (a pharmacologically inactive substance) on perceived tantrums in children. We provided parents with commercially available information that flower essences worked through ‘vibrational energy’ but made clear we neither supported nor disputed those claims. Our results show that once the flower essence was administered, tantrum frequency and severity reduced over the subsequent 8 days. The gradual decrease in tantrums over time is consistent with placebo research in adults where the effect of flower essences purporting to improve mood leads to a gradual improvement in mood which asymptotes at about 7 days (Hyland & Whalley, 2008).

This study is the first clear demonstration of the placebo by proxy effect. The results cannot be explained by the child’s knowledge of treatment (i.e., a conventional placebo effect). Most children were not told by their parents why they were being given the flower essence, and there was a non-significant trend for those told to show a reduced response.

Our data also showed a day-to-day association between parental mood and tantrum frequency. We cannot say whether parental mood causes tantrums or vice versa, but the former is consistent with other research showing the effect of parental experience on child behavior (McDonald et al., 2007).

Inspection of the individual data and qualitative feedback (see “Appendix”) showed that the size of the placebo by proxy effect differed between parent–child dyads. Although there was no improvement in tantrums for some children, those sending written feedback reported sometimes substantial and moving accounts of the improvements in their child’s behavior and expressed a desire to continue with the treatment—whatever the cost. One parent speculated that the effects could have been due to placebo effect but that it did not matter as the important thing was to reduce tantrums. The substantial benefit reported by a minority of parents illustrates the strength of the placebo effect, but strong effects may be atypical.

Our findings should be generalised with caution. Placebo responses occur because of the meaning of the therapeutic ritual (Moerman, 2002). Our study provided parents with a therapy that was rich in meaning (“a universal pool of healing and love”), and similar to that sometimes found in complementary medicine. Other types of meaning occur in normal general practice: all placebos are not equal. Our population of parents was self-selecting. People select therapies and respond best to placebo therapies that are consistent with their values (Whalley & Hyland, 2009; Hyland, 2011). Our study population is likely to be biased towards those who respond well to this particular type of placebo.

A final limitation of our study is that our results fail to distinguish between the two potential mechanisms for the placebo by proxy effect: That there is an objective reduction in the child’s tantrums or that there is a reduction only in the parent’s perception of tantrums. Both mechanisms are consistent with previous research findings (Waschbusch et al., 2009), both are clinically important, and is possible that the two mechanisms interact—changes in parental perception then alter parental behaviour which then alters the child’s behavior. Further research with objective outcome measures will provide further insight into the way ‘other people’ influence the placebo effect.



The flower essences used in this study were provided without cost to the authors by the Green Man Essence Company. The study was unfunded.

Conflict of interest

There is no conflict of interest.


  1. Durlak, J. A. (2009). How to select, calculate, and interpret effect sizes. Journal of Pediatric Psychology, 34, 917–928.PubMedCrossRefGoogle Scholar
  2. Ernst, E. (2010). Bach flower remedies: A systematic review of randomized clinical trials. Swiss Medical Weekly, 140. doi:10.4414/smw.2010.13079
  3. Grelotti, D., & Kaptchuk, T. J. (2011). Placebo by proxy. BMJ, 343:bmj.d4345BMJ. doi:10.1136/bmj.d4345
  4. Hoover, D. W., & Milich, R. (1994). Effects of sugar ingestion expectancies on mother-child interactions. Journal of Abnormal Child Psychology, 22, 501–515.PubMedCrossRefGoogle Scholar
  5. Hyland, M. E. (2011). Motivation and placebos: Do different mechanisms occur in different contexts? Philosophical Transactions of the Royal Society B: Biological Sciences, 366, 1828–1837.CrossRefGoogle Scholar
  6. Hyland, M. E., & Whalley, B. (2008). Motivational concordance: An important mechanism in self-help therapeutic rituals involving inert (placebo) substances. Journal of Psychosomatic Research, 65, 405–413.PubMedCrossRefGoogle Scholar
  7. Hyland, M. E., Whalley, B., & Geraghty, A. W. A. (2007). Dispositional predictors of placebo responding: A motivational interpretation of flower essence and gratitude therapy. Journal of Psychosomatic Research, 62, 331–340.PubMedCrossRefGoogle Scholar
  8. McDonald, R., Jouriles, E. N., Briggs-Gowan, M. J., Rosenfield, D., & Carter, A. S. (2007). Violence toward a family member, angry adult conflict, and child adjustment difficulties: Relations in families with 1- to 3-year-old children. Journal of Family Psychology, 21, 176–184.PubMedCrossRefGoogle Scholar
  9. Moerman, D. E. (2002). Meaning, medicine, and the ‘placebo effect’. Cambridge: Cambridge University Press.CrossRefGoogle Scholar
  10. Potegal, M., & Davidson, R. J. (2003). Temper tantrums in young children: 1. Behavioral composition. Journal of Developmental and Behavioral Pediatrics, 24, 140–147.PubMedCrossRefGoogle Scholar
  11. Rabe-Hesketh, S., & Skrondal, A. (2008). Multilevel and longitudinal modeling using stata. College Station, TX: Stata Press.Google Scholar
  12. Scarborough, A. A., Hebbeler, K. M., Spiker, D., & Simeonsson, R. J. (2007). Dimensions of behavior of toddlers entering early intervention: Child and family correlates. Infant Behavior Development, 30, 466–478.PubMedCrossRefGoogle Scholar
  13. Strand, P. S. (2000). A modern behavioral perspective on child conduct disorder: Integrating behavioral momentum and matching theory. Clinical Psychology Review, 20(5), 593–615.PubMedCrossRefGoogle Scholar
  14. Waschbusch, D. A., Pelham, W. E., Jr., Waxmonsky, J., & Johnston, C. (2009). Are there placebo effects in the medication treatment of children with attention-deficit hyperactivity disorder? Journal of Developmental and Behavioral Pediatrics, 30, 158–168.PubMedCrossRefGoogle Scholar
  15. Whalley, B., & Hyland, M. E. (2009). One size does not fit all: Motivational predictors of contextual benefits of therapy. Psychology and Psychotherapy, 82, 291–303.PubMedCrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media, LLC 2012

Authors and Affiliations

  1. 1.School of PsychologyPlymouth UniversityPlymouthUK

Personalised recommendations