A study sought to determine whether the ancient Chinese art of acupuncture could help infertile women become pregnant.41 One hundred sixty healthy women undergoing treatment with artificial insemination were recruited for the study. Half of the subjects were randomly assigned to receive acupuncture treatment 25 minutes before embryo transfer and again 25 minutes after the transfer. The remaining 80 subjects were instructed to lie still for 25 minutes after the embryo transfer. Results: In the acupuncture group, 34 women became pregnant. In the control group, 21 women became pregnant.
(a) Describe how the three principles of experimental design were addressed in this study.
(b) The difference in the percent of women who became pregnant in the two groups is statistically significant. Explain what this means to someone who knows little statistics.
(c) Explain why the placebo effect prevents us from concluding that acupuncture caused the difference in pregnancy rates.
a) Control- treatment was applied to some of the individuals but not the others for comparison; Random- Individuals receiving treatment were randomly assigned. This helps eliminate bias;
Replication - The experiment is repeatable for future data collection.
b) It is very unlikely that the difference was caused by chance because the study was large enough.
c) There is no group that received a placebo (without the subjects being aware that they received a placebo).
Sampling and Data
okay for this problem. We were looking at a study where they did some acupuncture. Like the overall goal of of the study was to improve fertility for women getting pregnant on, do they? We're testing to see if acupuncture improve the rate of fertility and women getting pregnant. Okay, so I'm going to restate the problem because their overall goal is to say, Hey, here's this experiment. We're doing experimental design, being a treatment and testing it out. So improvement in general simply states improve fertility. Um And so then we are plan thio include The fertility for these women was to do this treatment to part of the group and not the other part of the group. You can re read the details, but I'm going to summarize it here. So we have 160 healthy women and basically half received the acupuncture treatment. Have to not Okay, So 160 healthy women were part of this experiment and have received acupuncture after not results showed some interesting, uh, interesting results came out of it because they had 34 successes with the group that had the acupuncture and same time they had 21 successes and pregnancies without the acupuncture, we have a few follow up questions. The good design. And what does this all mean? Okay, so there's a restatement, and it's kind of wordy, but starts is a little bit wordy, but that helps us layout What we want to answer. So the question is, what are the good parts of this experimental design? Okay, so the good news about this method that we kind of highlighted out there, the good news was that a good elements of experiment experiment design has a control. Okay, so it has a control because some did not receive the treatment. And when you have control on your experiment is like, you know, more classic experiments in science class. You want to basically test it with some, but not with other. So good element of design has control. Second, uh, part of it has, um, just randomly assigned. Okay, So when they did, then they broke it out to see who received the acupuncture and who did not. There was a random method of selection. Okay, So random selection on who received it or not What the purpose does. Doing random for good experimental design it eliminates bias. So you're trying to pick and choose who gets something who doesn't. So you're having a true, better experiment. And the third piece of good experimental design is replication. Meaning is your plan good enough? You can repeat it again. So the studies look interesting. Can you do it again? And so this method is simple enough? Yeah, comprehensive enough that they could repeat it and see if this is just, uh, a little bit by chance that we had more success is, um but you kind of good experimental design you want, You want to be repeatable. So that answers part A Has those to be elements of good design Question be asked us, um, do we have Can we trust these results? Okay, so can we trust these results? So again, 116 we have 34 successes in 21 successes. Without the treatment eso the results look pretty good. Okay, So the results Yeah, it helps me trusted because the group is large enough and we did a good material design, but Okay, eso no experiment is perfect, but there seems to be a good experimental, you know, it seems to be good evidence 34 versus 21 is pretty good. So if it was less or closer, like 34 32 it feel less a little less, um confident and later statistics. We have more specific ways of really drilling down those numbers. But for this part of the chapter and part of the year they were at, it seems like the fact that they're pretty relatively large groups is enough to thio. Put some confidence in our results and find the third question is, um, but we're pretty confident confidence about about what we should. What should we do, or how can we improve this results kind of thing so this could be improved, but he should be redone. It could be improved if you have like a more true control group. And by that, we mean is if it's more true, control group. We kind of did 160 we had half receive the acupuncture and half did not so a better way of doing this. Like I said, the results were pretty good, but the results would be better and much more sound scientifically sound if those who did not way see the acupuncture thought they're receiving. It's called the placebo effect, so it's kind of weird, but you almost fake. You know, you're doing like that, that the true acupuncture, but sometimes just know, thinking that you're getting a good medication or a good drug or good acupuncture may affect how your feelings and your emotions and how you think something will work. So you wanna always have a placebo effect? Um, so it's gonna summarize that by saying we approved by adding the placebo. So if we did the study again and it half didn't have to not, you'd want the half that did not. Not only to receive it, they want them Thio receive some kind of, uh, lack of a better word pretend acupuncture. So that way it takes out the mentality that year. Everyone thinks they're getting something that is helping them out. So interesting question, interesting context. But the big idea is we wrapped up the we now we talked about what is good experimental design and the results. The results are big enough. We can kind of trust them, but would be even better if we had more of a troop of placebo