Two phenomena (things that can be observed and measured) are correlated if they move together. Children’s ages and heights are positively correlated; they increase together. Prices of fixed-coupon bonds and market interest rates are negatively correlated; when rates rise, prices fall, and vice versa.
Just because things are correlated does not mean that they are related by cause and effect. Here is a dumb example. We observe a high correlation between bicycle riders raising their arms and turning.
Physical scientists would observe the correlation between raised arms and turns. They would hypothesize any number of possible relationships. One could be direct (arm signal causes turn). They would measure everything they could (amount of traffic, time of day, age of rider, location, turning of handlebars, weight of rider, type of bicycle…) and conduct controlled laboratory experiments in which they held everything constant except one variable to see whether turns ensued. They would conclude that only one factor – handlebar turns – led to turns. Nothing else matters. They would develop a theoretical model (turning handlebars change bicycle motion consistent with Newton’s Laws.
Social scientists – and that includes appraisers – cannot conduct controlled experiments in which we hold everything else constant and change one variable at a time. Even in the bond example above, there is more going on that might be relevant: stock prices, economic conditions, regulatory changes, and so forth.
That is a critical difference: the ability to conduct controlled experiments. Because we appraisers cannot do this, we often look for correlations and wrongly conclude that they imply causation. So do many others. For an interesting example, see
http://www.carolmoore.net/articles/sunspot-cycle.html
in which the author claims that sunspot activity caused the Berlin Wall to fall in 1989. Note the “scientific explanation” at the end of the first paragraph as well as the disclaimer at the top of the page!
I raise this issue of correlation and causation because I read many academic and professional articles about financial markets in which correlation (a tool of statistical analysis) and causation are mixed together. That can be a recipe for disaster, like not signaling a turn when you are bike riding!