For each research finding decide whether there is evidence...

For each research finding, decide whether there is evidence of causation, correlation, or both. If it is only a correlation, name a possible lurking variable that may be the cause of the results. a. As the sales of television sets have increased, so has the number of overweight adults. Does television cause weight gain? b. A study in an elementary school found that kids with big feet were better readers than kids with small feet. Do big feet make children read better? c. The more firefighters sent to a fire, the longer it takes to put out the fire. Does sending more firefighters cause a fire to burn longer? What does it mean to say that data has a strong negative correlation? (Multiple Choice) (A) The variables in the data have no relationship at all (B) A linear model is appropriate, and the slope of that line is negative (C) There is a negative causation relationship in the data (D) One variable in the dataset always has a negative value

Transcript

00:01 This is a couple of questions that asks us to distinguish between causation and correlation.

00:08 So first let's get definitions down of these terms.

00:12 Causation refers to a causal relationship in which some event a causes some other event b.

00:21 So there's a strong arrow of causation means that if a happens, then it makes b happen.

00:29 There is usually some kind of mechanical connection where a is physically interacting with parts of the world that create b.

00:39 Whereas a correlation is a weaker connection.

00:43 It's just a connection between a and b where maybe we tend to see them together, but we don't know if a is causing b, b is causing a, or there's some lurking variable, like maybe a variable c down here.

00:58 That tends to cause both of them together.

01:04 So let's look at our three questions here, abc, and see if any of them have evidence of causation.

01:13 So a has us relate the sale of television sets with the number of overweight adults.

01:28 So those are our two variables.

01:30 We're counting up how many adults are overweight, and we're also looking at the sales statistics.

01:36 For television sets.

01:38 They tell us that these two things are related to each other, that maybe we see more sales of television sets and more percentage of overweight adults.

01:46 Are those two things causally related or just correlated? i think that it might be tempting to find a causal arrow here to say that more television sets make more adults sedentary, and that increases the chance of overweight adults.

02:04 So there is a possible connection here where you might say there is causation.

02:11 So more tv sets leads to a more sedentary lifestyle because people choose to spend their recreation watching tv instead of going outside and exercising.

02:27 And that leads to more overweight in the population.

02:31 So that's one potential causal model.

02:34 I don't really have evidence for that.

02:37 Model, though.

02:38 So honestly, if i were answering this question, i would say, well, there is a correlation between them, but i don't have any evidence of causation.

02:46 And a lurking variable that might be behind both of these things is there might be some kind of lifestyle change that leads to both.

02:56 So there might be a lifestyle change.

02:59 For instance, you could have people living in like small apartments where they're not able to get a lot of outside exercise, and that could lead them to both buy tv sets because they don't have other ways to entertain themselves and to become overweight because they are more sedentary.

03:22 Or we could even see a causation link the other direction.

03:27 So we could see that overweight adults because they have fewer, they're less able to exercise, feel comfortable outside.

03:37 Maybe there is a social stigma.

03:39 They may be more likely to buy television sets.

03:43 So that would be a causal arrow in the other direction.

03:50 I can see a potential connection that way.

03:53 So i don't feel comfortable attributing this to a causal link...