Conducting a Better Statistics Study

Your Doctor says: " Drinking one glass of red wine per day regularly reduces the chances of getting heart disease." There are two ways to study this claim.

I.) Find 150 individuals who drink red wine regularly and follow them for 20 years.

II.) Find 120 individuals and have 60 of them drink 1 glass of wine per day, and the other 60 not drink any wine. Follow them for 15 years.

Which design is more trustworthy and why?

I would say that design number I is more trustworthy for a few reasons...

a.) Law of Large Numbers
b.) It would be hard to make sure that someone drank a glass of wine everyday for 15 years.

Does my reasoning seem correct? Any other ideas/reasons?

Quote:

---

Originally Posted by WartonMorton

Your Doctor says: " Drinking one glass of red wine per day regularly reduces the chances of getting heart disease." There are two ways to study this claim.

I.) Find 150 individuals who drink red wine regularly and follow them for 20 years.

II.) Find 120 individuals and have 60 of them drink 1 glass of wine per day, and the other 60 not drink any wine. Follow them for 15 years.

Which design is more trustworthy and why?

I would say that design number II is more trustworthy for a few reasons...

a.) Law of Large Numbers
b.) It would be hard to make sure that someone drank a glass of wine everyday for 15 years.

Does my reasoning seem correct? Any other ideas/reasons?

---

haha, what you said in (b) seems to be a reason against using II. and I has a larger number of individuals, so you can't just say "law of large numbers".

consider the fact that you want to make a comparison, those who do as your doctor says and those that do not. in I, it is one sided, you only study those that do.

moreover, it is vague, drinking wine "regularly" does not mean you drink 1 glass everyday, you can drink less, or more. in either case, you are not doing what the doctor said.

more than that, people who tend to be social drinkers may have habits that the regular population does not have. any of those habits may cause bias results.

now, with all this in mind, what would be your reasons for choosing II?

Quote:

---

Originally Posted by Jhevon

haha, what you said in (b) seems to be a reason against using II. and I has a larger number of individuals, so you can't just say "law of large numbers".

consider the fact that you want to make a comparison, those who do as your doctor says and those that do not. in I, it is one sided, you only study those that do.

moreover, it is vague, drinking wine "regularly" does not mean you drink 1 glass everyday, you can drink less, or more. in either case, you are not doing what the doctor said.

more than that, people who tend to be social drinkers may have habits that the regular population does not have. any of those habits may cause bias results.

now, with all this in mind, what would be your reasons for choosing II?

---

I meant to say that (I) sounded more likely because of points (a) and (b). But you do bring up a good point that (I) doesn't have a control group... the group that doesn't drink wine. So then that would make (II) a better choice because it is an actual experiment and the researchers could compare data from the wine drinkers to the non wine drinkers. Any other reasons you can think of why II would be favorable?