Question

I put together a scatterplot of Ice Cream Sales v Temperature. It is a positive correlation. What are possible lurking variables?

Answer 1

I think there is no possible lurking variables, rather the scatterplot should be Temperature v Icecream sales. Because , increasing temerature indicates that summer is approaching. Ice creams , cold drinks are such products that are preferred in Summer for their cooling effects . In Winter, their sales reduce significantly as people prefer something hot/warm products. So, basically the data has a seasonal effect , and a direct causal relationship with temperature and ice cream sales can be shown, which will also be positively correlated. But yes, you can think of a categorical variable named 'season' ( having summer, winter, spring etc) as a lurking variable here or a quantitive variable "humidity" (since high temeperature > warm air > more humidity, low temperatire> cold air > less humidity) as a lurking variable, though I would like to suggest you, not at all think about the presence of a lurking variable in this particilar situation.

For a better understanding of lurking variable, plot the data of Ice cream sales Vs No. of deaths by drowning. You can see a possitive correlation there also. In this case, the lurking variable is the 'Temperature' , which causes the rise in Ice cream sales . Also people like to enjoy bathing in pool/rivers more in when the temperature is high (that is in summer). So, 'Temperature' indeed have a effect indirectly.