What drives activity on Pinterest?

Researchers have released a new study that uses statistical data to help understand the motivations behind Pinterest activity, the roles gender plays among users and the factors that distinguish Pinterest from other popular social networking sites. Read More

Sporting events: Clear your memory to pick a winner

Predicting the winner of a sporting event with accuracy close to that of a statistical computer program could be possible with proper training, according to researchers. Read More

Computer analysis of massive clinical databases

A computer program capable of tracking more than 100 clinical variables for almost 400 people has shown it can identify various subtypes of asthma, which perhaps could lead to targeted, more effective treatments. A computational biologist led the analysis of patient data for the study. Read More

New statistical models could lead to better predictions of ocean patterns

The world's oceans cover more than 72 percent of the earth's surface, impact a major part of the carbon cycle, and contribute to variability in global climate and weather patterns. Now, researchers at the University of Missouri applied complex statistical models to increase the accuracy of ocean forecasting that influences the ways in which forecasters predict long-range events such as El NiDo and the lower levels of the ocean food chain. Read More

Researchers develop new generation visual browser of epigenome

ChroGPS is a software application that serves to facilitate the analysis and understanding of epigenetic data and to extract intelligible information, which can be downloaded free of charge in Bioconductor, a reference repository for biocomputational software. Scientists describe the uses of the program in a new article. Read More

Optical rogue waves: The storm in a test tube

Random processes in nature often underlie a so-called normal distribution that enables reliable estimation for the appearance of extreme statistical events. Meteorological systems are an exception to this rule, with extreme events appearing at a much higher rate than could be predicted from long-term observation at much lower magnitude. One such example is the appearance of unexpectedly strong storms, yet another are rare reports of waves of extreme height in the ocean, which are also known as rogue waves or monster waves. Read More

Machines learn to detect breast cancer

Software that can recognize patterns in data is commonly used by scientists and economics. Now, researchers in the US have applied similar algorithms to help them more accurately diagnose breast cancer. Read More

New strep throat risk score brings data together to improve care

A new risk measure called a "home score" could save a patient with symptoms of strep throat a trip to the doctor, according to a new paper. The score combines patients' symptoms and demographic information with data on local strep throat activity to estimate their strep risk, empowering them to seek care appropriately. Read More

Reduce unnecessary lab tests, decrease costs by modifying software

When patients undergo diagnostic lab tests as part of the inpatient admission process, they may wonder why or how physicians choose particular tests. Increasingly, medical professionals are using electronic medical systems that provide lists of lab tests from to choose. Now, researchers have studied how to modify these lists to ensure health professionals order relevant tests and omit unnecessary lab tests, which could result in better care and reduced costs. Read More

Finding the hidden zombie in your network: Statistical approach to unraveling computer botnets

How do you detect a "botnet," a network of computers infected with malware -- so-called zombies -- that allow a third party to take control of those machines? The answer may lie in a statistical tool first published in 1966 and brought into the digital age, say researchers. Read More

How people use Facebook to maintain friendships

New social networking research investigates how individuals use Facebook to maintain their friendships. Read More

Doctors often uncertain in ordering, interpreting lab tests

A survey of primary care physicians suggests they often face uncertainty in ordering and interpreting clinical laboratory tests. Physicians have developed their own strategies for ordering and interpreting lab tests, such as asking a physician colleague or specialist, consulting a text or electronic reference, or calling the laboratory. But physicians reported they would welcome better decision-support software embedded in electronic medical records and direct access to lab personnel through lab hotlines. Read More

Frequent cell phone use linked to anxiety, lower grade, reduced happiness in students

Results of the analysis showed that cell phone use by college students was negatively related to GPA and positively related to anxiety. Following this, GPA was positively related to happiness while anxiety was negatively related to happiness. Thus, for the population studied, high frequency cell phone users tended to have lower GPA, higher anxiety, and lower satisfaction with life (happiness) relative to their peers who used the cell phone less often. Read More

Better way to make sense of 'Big Data?'

Big data is everywhere, and we are constantly told that it holds the answers to almost any problem we want to solve. But simply having lots of data is not the same as understanding it. New mathematical tools are needed to extract meaning from enormous data sets. Researchers now challenge the most recent advances in this field, using a classic mathematical concept to tackle the outstanding problems in big data analysis. Read More

What's behind a #1 ranking?

Behind every "Top 100" list are a generous sprinkling of personal bias and subjective decisions. Lacking the tools to calculate how factors like median home prices and crime rates actually affect the "best places to live," the public must take experts' analysis at face value. Read More

To teach scientific reproducibility, start young

In the wake of retraction scandals and studies showing reproducibility rates as low as 10 percent for peer-reviewed articles, the scientific community has focused attention on ways to improve transparency and duplication. A team of math and statistics professors has proposed a way to address one root of that problem: teach and emphasize reproducibility to aspiring scientists, using software that makes the concept feel logical rather than cumbersome. Read More

Digital ears in the rainforest: Estimating dynamics of animal populations by using sound recordings and computing

A Finnish-Brazilian project is constructing a system that could estimate the dynamics of animal populations by using sound recordings, statistics and scientific computing. The canopy in a Brazilian rainforest is bustling with life, but nothing is visible from the ground level. The digital recorders attached to the trees, however, are picking up the noises of birds. Read More

Social media, self-esteem and suicide: Nations with more corruption demonstrate more social media, less suicide

In nations where corruption is rife, it seems that citizens these days find an escape from the everyday problems that trickle down to their lives by using online social media more than those elsewhere. Research also suggests that these two factors -- more corruption, more social networking -- also correlate with lower suicide rates. Read More

Collecting digital user data without invading privacy

The statistical evaluation of digital user data is of vital importance for analyzing trends. But it can also undermine the privacy. Computer scientists have now developed a novel cryptographic method that makes it possible to collect data and protect the privacy of the user at the same time.Read More

The performance of robust test statistics with categorical data.

The performance of robust test statistics with categorical data.

Br J Math Stat Psychol. 2013 May;66(2):201-23

Authors: Savalei V, Rhemtulla M

This paper reports on a simulation study that evaluated the performance of five structural equation model test statistics appropriate for categorical data. Both Type I error rate and power were investigated. Different model sizes, sample sizes, numbers of categories, and threshold distributions were considered. Statistics associated with both the diagonally weighted least squares (cat-DWLS) estimator and with the unweighted least squares (cat-ULS) estimator were studied. Recent research suggests that cat-ULS parameter estimates and robust standard errors slightly outperform cat-DWLS estimates and robust standard errors (Forero, Maydeu-Olivares, & Gallardo-Pujol, 2009). The findings of the present research suggest that the mean- and variance-adjusted test statistic associated with the cat-ULS estimator performs best overall. A new version of this statistic now exists that does not require a degrees-of-freedom adjustment (Asparouhov & Muthén, 2010), and this statistic is recommended. Overall, the cat-ULS estimator is recommended over cat-DWLS, particularly in small to medium sample sizes.

22568535
Read More