Wednesday, May 27, 2009

Race, Insurance, and Health Care

Do race or status of insurance affect treatment in emergency rooms?  The study "Race and Insurance Status as Risk Factors for Trauma Mortality" in Archives of Surgery (October, 2008) shows mortality rates are signicantly higher for African Americans, Hispanics, and the uninsured, compared to whites or people with insurance.  See the Science Daily article.

Friday, February 27, 2009

Men and Women and Religion

The Pew Forum has a graphic that compares responses from men and women on religious opinions and practices. It would be excellent for asking questions about statistical significance and confidence intervals. Questions are about belief in God, daily prayer, regular worship, affiliation, importance, etc. As always, one can follow links to a description of methodology, too.

Tuesday, February 24, 2009

Visiting the Dentist

The Journal of Pediatrics reports on a study that did a within-subjects comparison of the effects of the environment in a dentist's office on children's anxiety levels. The children visited the dentist so the researchers could gather baseline data on anxiety levels using a behavior checklist. During the second visit interventions included changes in overhead lighting, soothing music, a heavy vest to simulate the feeling of a hug, and vibrations in the chair. For the 35 children (16 of whom were developmentally disabled), duration of anxious behavior dropped from 3.69 minutes to 1.48 minutes in non-delayed children, and from 23.44 minutes to 9.04 minutes in the delayed children. Both drops are reported as significant.

Monday, February 9, 2009

Testosterone and Risk Taking

A University of Cambridge study of financial traders in London reveals a significant difference in testosterone levels between traders who are doing better than average and worse than average (measured against their personal averages). A paired t-test (t = 2.8, P = 0.012, two-tailed, n = 17) reveals the difference. A second paired t-test split the traders into groups where their testosterone levels at 11 a.m. (before most of the trading) was high or low. Results: a significant difference in profit and loss averages; t = 3.03, P = 0.008, two-tailed, n = 17; The paper also contains some useful graphics and correlational studies.

Friday, February 6, 2009

Color Effects Performance

Science Magazine reports on a variety of studies that show, among other things, a red computer background increases workers' accuracy and blue increases creativity.

Wednesday, January 21, 2009

Does A Cold Shoulder Make You Colder?

Researchers wanted to determine whether recalling a past experience of feeling excluded could make people feel colder, as measured by participants' estimates of the ambient room temperature. Sixty five students were randomized into groups; one group was told to recall a past experience of being excluded, the other of being included. The difference in room temperature estimates was significant, with the first group estimating the temperature to be an average of 2.6 degrees (Celsius) colder.

A second experiment found that people who were excluded in a mock computer exercise were more likely to desire warm beverages than people in a control group. An ANOVA test was used.

Source

Sunday, January 18, 2009

Too Many Significance Tests

A study published in April 2008 (link) reported that a pregnant woman's diet could influence the gender of her baby. A particular claim was that eating breakfast cereals increased the probability of having a boy. A Science Daily piece from Jan. 17, 2009 reports on a recently published explanation that indicates the earlier result is easily explained by chance.

According tot he Science Daily article, 132 food items were tested during two different time periods, for a total of 264 statistical tests. At a 0.05 level of significance, this number of tests would lead to, on average, 13 false positives (Type I error: rejecting the null hypothesis when the null hypothesis is indeed true).