2021 AP Statistics Free Response #6 (First Administration)

[@alfadoesstuff.4256]

For part D the confounding variable would be the number of games won. The confounding variable represents a true cause-and-effect relationship which may lead you to mistakenly believe that there is a cause-and-effect relationship between other variables. In this instance, the number of games won is the true relationship, which makes it look like year and type of stadium also have a cause-and-effect relationship with average attendance. This was a difficult question.

[@AllenTsaoSTEMCoach]

There's no way to know what is the true causal relationship. An experiment was not performed so these are all observational data. Any of them could reasonably be called a cause. Without an experiment, causation cannot be determined.

[@alfadoesstuff.4256]

@@AllenTsaoSTEMCoachWe have to determine, essentially, which of the three variables represents the strongest relationship. Year can't be the relationship because viewership should have increased for the old stadium as well. While it may seem intuitive that the relationship is actually between stadium type, graph II clearly shows that viewership increased for both stadium types. Thus, number of games won is the confounding variable.

[@alfadoesstuff.4256]

@@AllenTsaoSTEMCoach Although I think that it is reasonable to assume the college board will accept reasonable explanations for the other variables if they are well-worded.

[@AllenTsaoSTEMCoach]

Yeah I think it'll be scored based on how well the response explains the concept of confounding variables in the context of this situation.

[@alfadoesstuff.4256]

@@AllenTsaoSTEMCoach Also just to clarify, we can, and often do, infer causation from observational studies. For instance, there have been many studies linking smoking and higher rates of cancer, or texting and driving with higher rates of crashes. There are obviously certain conditions that have to be met, and a thorough examination of potential confounding variables, but some of the most significant products of statistics have relied on observation rather than random assignment of treatments.

[@mayasato856]

For part d, would it make sense to put the stadium is a confounding variable because likely the new stadium is larger and can fit more attendees than the old one and that's why the attendance has grown? We will never know if they could have won more games in the old stadium with the attendance they have in the new stadium because that many attendees just couldn't fit.

[@AllenTsaoSTEMCoach]

There's not one right answer. Your explanation would be a reasonable one for why stadium could be the confounding variable.

[@mayasato856]

@@AllenTsaoSTEMCoach Thanks!!

[@sarahlevin6699]

Hi! For part a, are there specific things you have to talk about when describing the spread? I only talked about range, not IQR. Would that still be ok and get me all the points?

[@AllenTsaoSTEMCoach]

As long as you discussed some aspect of spread, you should get credit for that. Range is perfectly acceptable to discuss for spread.

[@sarahlevin6699]

@@AllenTsaoSTEMCoach Thank you!

[@teddytaylor3373]

For part d I said something along the lines of “as there is not a strong correlation between year and attendance (for the old stadium which would make it hard to say there’s a strong relationship of attention and year overall, regardless of stadium), there is no statistical evidence that shows year affects attendance. Next, the line of best fit for the old stadium in the wins vs. attendance has approximately the same slope, meaning that the stadium does not affect the attendance. This only leaves the number of wins. The time that the new stadium didn’t win as many games has an attendance similar to the times the old stadium won an equal amount of games. The final graph also shows a positive relationship between games won and attendance

[@AllenTsaoSTEMCoach]

You do need to answer the question though, and it wasn't clear to me you did. What is the confounding variable in your explanation? You aren't supposed to try to "prove" which variable is definitely causing. While your arguments are OK for why you think number of wins is the most likely actual cause, the question is to discuss the nature of a confounding variable in this situation. So your answer would probably need to explicitly state something along the lines of "While there is a correlation between the stadium and attendance, the number of wins could be a confounding variable because the team has been winning more games in recent years. Thus, we can't necessarily conclude the new stadium is the cause of increased attendance."

[@brownvengeance5023]

since the graph is in thousands, do we have to say 25,000 instead of 25 in our answers

[@AllenTsaoSTEMCoach]

Yes you should use the units indicated in the graph.

[@interstellar0001]

For part D, I said year could’ve been a confounding because of cultural/economic shifts. Would that be fine given I said this in context?

[@AllenTsaoSTEMCoach]

Yeah I think that's a good way to describe "year" as the confounding variable.

[@interstellar0001]

@@AllenTsaoSTEMCoach Thank you!

[@kristiankieffer9284]

For part D I stated that the increasing popularity of the sport could be a confounding variable. Would that also be a correct answer?

[@AllenTsaoSTEMCoach]

I think they only wanted you to consider one of the 3 variables listed. So while that could be a confounding variable, I don't think that's what they were looking for.

[@aidan7364]

Could you say in part b that the new stadium has a non linear correlation because of possible influential points? (of course giving coordinates)

[@AllenTsaoSTEMCoach]

I feel like that would be OK to say. As long as you still stated a pretty strong positive association too.

[@zeitgeist1762]

For part (d) could I say something to the effect of: "Year (time) could be a possible confounding variable due to its influence on the population of the schools, technology during the time, and the type of stadium that was being used"?

[@AllenTsaoSTEMCoach]

Yes I think so!