Decision theory,linear partial information and statistical dominance

The paper deals with decision making under conditions of linear partial information, i.e. when probabilities of states of nature are not known precisely, but are subject to linear constraints. Conditions ensuring strict and weak statistical dominance of one strategy over another are derived. It is also shown that weak dominance in terms of payoffs is equivalent to weak dominance in terms of regrets. The new results are more general than those obtained by Fishburn and by Kmietowicz and Pearman for weak and strict ranking of probabilities, and include them as special cases. The new results can be employed in practical decision making in several ways.     

Long-term low-dose dehydroepiandrosterone replacement therapy in aging males with partial androgen deficiency

Dehydroepiandrosterone (DHEA) and dehydroepiandrosterone sulfate (DHEAS) age-related withdrawal is very likely to be involved in the aging process and the onset of age-related diseases, giving rise to the question of whether preventing or compensating the decline of these steroids may have endocrine and clinical benefits. The aim of the present trial was to evaluate the endocrine, neuroendocrine and clinical consequences of a long-term (1 year), low-dose (25?mg/day) replacement therapy in a group of aging men who presented the clinical characteristics of partial androgen deficiency (PADAM). Circulating DHEA, DHEAS, androstenedione, total testosterone and free testosterone, dihydrotestosterone (DHT), progesterone, 17-hydroxyprogesterone, allopregnanolone, estrone, estradiol, sex hormone binding globulin (SHBG), cortisol, follicle stimulating hormone (FSH), luteinizing hormone (LH), growth hormone (GH) and insulin-like growth factor 1 (IGF-1) levels were evaluated monthly to assess the endocrine effects of the therapy, while β-endorphin values were used as a marker of the neuroendocrine effects. A Kupperman questionnaire was performed to evaluate the subjective symptoms before and after treatment.

The results showed a great modification of the endocrine profile; with the exception of cortisol levels, which remained unchanged, DHEA, DHEAS, androstenedione, total and free testosterone, DHT, progesterone, 17-hydroxyprogesterone, estrone, estradiol, GH, IGF-1 and β-endorphin levels increased significantly with respect to baseline values, while FSH, LH and SHBG levels showed a significant decrease. The Kupperman score indicated a progressive improvement in mood, fatigue and joint pain.

In conclusion, the present study demonstrates that 25?mg/day of DHEA is able to cause significant changes in the hormonal profile and clinical symptoms and can counteract the age-related decline of endocrine and neuroendocrine functions. Restoring DHEA levels to young adult values seems to benefit the age-related decline in physiological functions but, however promising, placebo-controlled trials are required to confirm these preliminary results.     

Estimation of parameters of 1-out-of-2:g repairable system

Moment estimators of l-out-of-2:G repairable system are supplied under four sampling schemes assuming that the failure and repair time distribution of the units are exponential with unknown parameters λ, μ respectively. Information metrices of the estimators are supplied. Also it is shown that the estimators are asymptotically normally distributed in every sampling scheme.     

Estimation of parameters of 1-out-of-n:g repairable system

A single unit system supported by N-l inactive standbys and a repair facility is considered when λ,μ are unknown, λand μbeing failure and repair rates of the unit respectively. Three sampling schemes are considered to obtain moment estimators λ^?:and μ _?:when the performance of the unit have to be observed in the system only. Asymptotic variances of the estimates are supplied.     

Upper bounds for the variances of certain random variables

Jacobson (1969) proved that the variance of a unimodal distribution on the unit interval is bounded above by 1/9. This paper generalizes Jacobson's result by deriving sharp upper bounds for the variances of α-unimodal distributions and for the variances of S-shaped distributions.