Search results

"Military intelligence analyst and Harvard Law student Tyler Vigen illustrates the golden rule that "correlation does not equal causation" through hilarious graphs inspired by his viral website. Is there a correlation between Nick Cage films and swimming pool accidents? What about beef consumption and people getting struck by lightning? Absolutely not. But that hasn't stopped millions of people from going to tylervigen.com and asking, "Wait, what?" Vigen has designed software that scours enormous data sets to find unlikely statistical correlations. He began pulling the funniest ones for his website and has since gained millions of views, hundreds of thousands of likes, and tons of media coverage. Subversive and clever, Spurious Correlations is geek humor at its finest, nailing our obsession with data and conspiracy theory"--

Concept measurement & the determination of unidimensionality are usually based upon techniques that depend on the relations between the items measuring the concept or concepts: a set of items having the same relations with one another is inferred to be unidimensional. However, an additional criterion for unidimensionality is that these items also have the same relations with a set of predictors. One multivariate technique that explicitly considers these relations between predictors & items is canonical correlation-a statistical technique that 'analyzes that portion of the interitem correlation related to the set of predictors.' After analysis of a set of items with factor analysis, canonical correlation was applied, & the results demonstrated the promise of the technique for exploratory data analysis: two dimensions, 'political activism' & 'governmental moral distaste,' emerged, which were not evident under examination with factor analysis. 5 Tables, 3 Figures, Appendix. Modified HA.

"Correlation matrices (along with their unstandardized counterparts, covariance matrices) underlie much of the statistical machinery in common use today. A correlation matrix is more than a matrix filled with correlation coefficients. The value of one coefficient in the matrix puts constraints on the values of the others, and this is a major theme of the volume. The volume is written accessibly, and key points are illustrated with a wide range of lively examples, including correlations between intelligence measured at different ages through adolescence; correlations between public health expenditures, health life expectancy, adult mortality, and other country characteristics; correlations between wellbeing and state-level vital statistics; correlations between the racial composition of cities and professional sports teams; and correlations between childbearing intentions and childbearing outcomes over the reproductive life course"--

AbstractA multi‐product monopolist sells sequentially to a buyer who privately learns his valuations. Using big data, the monopolist learns the intertemporal correlation of the buyer's valuations. Perfect price discrimination is generally unattainable—even when the seller learns the correlation perfectly, has full commitment, and in the limit where the consumption good about which the buyer has ex ante private information becomes insignificant. This impossibility is due to informational externalities that require information rents for the buyer's later consumption. These rents induce upward and downward distortions, violating the generalized no distortion at the top principle of dynamic mechanism design.

We develop a trade model with correlation in productivity across countries. The model spans the full class of generalized extreme value import demand systems and implies that countries with relatively dissimilar technology gain more from trade. In the context of a multisector trade model, we provide a tractable and flexible estimation procedure for correlation based on compressing highly disaggregate sectoral data into a few latent factors related to technology classes. We estimate significant heterogeneity in correlation across sectors and countries, which leads to quantitative predictions that are significantly different from estimates of models assuming independent productivity across sectors or countries. (JEL C38, F11, F13, F14, L16, O30)