No free lunch for cross validation

It is known theoretically that an algorithm cannot be good for an arbitrary prior. We show that in practical terms this also applies to the technique of ``cross validation'', which has been widely regarded as defying this general rule. Numerical examples are analysed in detail. Their implications to researches on learning algorithms are discussed.

Does child weight influence how mothers report their feeding practices?

Objectives. The present study aimed to ascertain whether parental reports of their feeding practices are associated with independent observations of these behaviours, and whether the reliability of maternal report depends upon the child's weight. Methods. A total of 56 mothers and their children ate a lunch to satiety which was videotaped and coded for maternal use of control during feeding. Mothers also completed questionnaires about their feeding practices and children were weighed and measured. Results. Maternal reports of controlling feeding practices were poorly related to independent observations of these behaviours in the laboratory. However, there was a significant interaction between child BMI z score and observed pressure to eat in predicting maternally reported pressure to eat. There was also a significant interaction between child BMI z score and observed maternal restriction with food in predicting maternally reported restriction. When decomposed, these interactions suggested that only mothers of relatively underweight children were accurate at reporting their use of pressure to eat when compared to independent observations. For mothers of relatively overweight children there was a significant negative relationship between observed and reported restriction over food. Conclusions. Overall there was poor correspondence between maternal reports and independent observations of the use of controlling feeding practices. Further research is needed to replicate these findings and to ascertain whether parents who are inaccurate at reporting their use of these feeding practices are unaware that they are using controlling feeding practices or whether they are responding in socially desirable ways to questionnaires assessing their feeding behaviour. © 2011 Informa Healthcare.

Maternal symptoms of depression are related to observations of controlling feeding practices in mothers of young children

Maternal depression can impair parenting practices and has been linked with less sensitive feeding interactions with children, but existing research is based on self-reports of feeding practices. This study examined relationships between maternal self-reported symptoms of depression with observations of mothers' child feeding practices during a mealtime. Fifty-eight mothers of 3-and 4-year-old children were video recorded eating a standardized lunch. The recording was then coded for instances of maternal controlling feeding practices and maternal vocalizations using the Family Mealtime Coding System. Mothers also provided information on current symptoms of depression and anxiety. Mothers who reported greater symptoms of depression were observed to use more verbal and physical pressure for their child to eat and to offer more incentives or conditions in exchange for their child eating. Mothers also used more vocalizations with their child about food during the observed mealtime when they had greater symptoms of depression. There was no link between symptoms of depression and observations of maternal use of restriction. Symptoms of depression are linked with observations of mothers implementing a more controlling, less sensitive feeding style with their child. Health professionals working with families in which mothers have symptoms of depression may benefit from receiving training about the possible impact of maternal depression on child-feeding practices, and mothers with symptoms of depression may benefit from guidance regarding its potential impact on their child-feeding interactions ©2013 American Psychological Association.

Do educators need educating?

Actual text: I was recently at the Spanish College of Optometry biennial conference and attended a meeting of contact lens lecturers from around Spain and Portugal. We discussed various ideas, mainly about how to share good practice and improve standards. What came to my mind was ‘is there a blueprint for training trainers?’ Well probably not but there are many things that we need to acknowledge such as the way students learn for example. Many educators themselves were taught by lecturers who would write on a blackboard or use acetate on an overhead projector, then came the 35 mm slide era followed by the Powerpoint era. More recently there is a move towards a much more integrated approach of various teaching methods. At my university our contact lens and anterior eye lectures generally follow a format where a narrated Powerpoint lecture is uploaded onto our internal virtual learning environment. This narrated version of the slides is designed to give the didactic element of the topic. The students listen to that before attending an interactive seminar on that topic. The seminar is also recorded so that students can listen to that afterwards. The seminar is designed to give additional information, such as case reports, or to clarify key points or for live demonstrations. It is a good way of doubling the contact time with the students without imposing further on an already packed formal timetable as the students can work in their own time. One problem that we noticed with this approach was that attendance can vary. If the students feel that they will gain something from the interactive seminar then they are more likely to attend – exam tips usually win them over! At the Spanish meeting the educators decided that they wanted to have regular meetings. The industry colleagues in attendance said that they were happy to help but could not necessarily give money, but they could offer meeting rooms, pay for lunch and evening meals. They even said that that they were happy to host meetings and invite other companies too (except to manufacturing plants). In the UK the British Committee of Contact Lens Educators (BUCCLE) meets for one day on three occasions in the year. The American Optometric Contact Lens Educators (AOCLE) meets annually at a three day event. Both these organisations get some help from industry. BUCCLE usually has one of its meetings at a university, one at a company training centre/manufacturing plant/national headquarters and one meeting the day before the BCLA annual conference. BUCCLE usually has its pre-BCLA meeting in conjunction with the International Association of Contact Lens Educators (IACLE). So when educators meet what would they discuss; well probably the focus should be on education rather than actual contact lens knowledge. For example sharing ideas on how to teach toric lens fitting would be better than discussing the actual topic of toric lenses itself. Most universities will have an education department with an expert who could share ideas on how to use the internet in teaching or how to structure lectures or assessments etc. In the past I have helped with similar training programmes in other countries and sharing good practice in pedagogy is always a popular topic. Anyone who is involved in education in the field of contact lenses should look at the IACLE web page and look out for the IACLE World Congress in 2015 in the days preceding the BCLA. Finally, IACLE, AOCLE and BUCCLE all exist as a result of generous educational grants from contact lens companies and anyone interested in finding out more about should refer to their respective web pages.

Teaching our children when to eat:how parental feeding practices inform the development of emotional eating-a longitudinal experimental design

Background: Emotional eating in children has been related to the consumption of energy-dense foods and obesity, but the development of emotional eating in young children is poorly understood. Objectives: We evaluated whether emotional eating can be induced in 5-7-y-old children in the laboratory and assessed whether parental use of overly controlling feeding practices at 3-5 y of age predicts a greater subsequent tendency for children to eat under conditions of mild stress at ages 5-7 y. Design: Forty-one parent-child dyads were recruited to participate in this longitudinal study, which involved parents and children being observed consuming a standard lunch, completing questionnaire measures of parental feeding practices, participating in a research procedure to induce child emotion (or a control procedure), and observing children's consumption of snack foods. Results: Children at ages 5-7 y who were exposed to a mild emotional stressor consumed significantly more calories from snack foods in the absence of hunger than did children in a control group. Parents who reported the use of more food as a reward and restriction of food for health reasons with their children at ages 3-5 y were more likely to have children who ate more under conditions of negative emotion at ages 5-7 y. Conclusions: Parents who overly control children's food intake may unintentionally teach children to rely on palatable foods to cope with negative emotions. Additional research is needed to evaluate the implications of these findings for children's food intake and weight outside of the laboratory setting. This trial was registered at clinicaltrials.gov as NCT01122290.