English 字幕

← Independent Samples - Intro to Inferential Statistics


Showing Revision 5 created 05/25/2016 by Udacity Robot.

  1. Welcome to Lesson 11. In Lesson 10, you learned about dependent samples or
  2. repeated measures. Just to refresh your memory, that could be where we give the
  3. same person two different conditions to see how they react to each one. Maybe a
  4. control and then a treatment, or maybe two types of treatments. Or, this could
  5. be longitudinal, where we measure some variable at some point in time, and then
  6. again at another and see if the variable changes. Or this could be a pre-test
  7. and post-test. What was the measurement of a variable before and after a
  8. treatment? These are just a few situations in which we would use dependent
  9. samples. This type of research design is really useful because it controls for
  10. indivdual differences. In other words, if we gave someone some kind of
  11. treatment, then those same individual differences will be present the next time
  12. we give that same treatment. That way we can see how two different treatments
  13. play out under the same conditions. Because there's controls for individual
  14. differeneces, we could then use fewer subjects. And this is more cost
  15. effective, less time consuming, and generally less expensive. However, there
  16. are also a few disadvantages. One of which, is carry-over effects. For example,
  17. let's say we have this new method of teaching math. You want to know if it's
  18. going to be effective. If you use the same group of students to test this new
  19. teaching method, inevitably, they're going to be better at math the second time
  20. around. So, if the first time we teach them one way, and then the second lesson
  21. we teach them a different way, they'll already be better at math from learning
  22. it the first time. Then we don't know if the results after the second treatment
  23. are due to the fact that it was effective or due to the fact that they've
  24. learned math before. That's just one example. The second measurement can be
  25. affected by the first treatment. And the order in which we give the treatment
  26. might influence the results. For example, say we want to test two types of
  27. pills. What if the first pill has some kind of interaction with the second
  28. pill? And so, by taking it in that order they affect the results. Therefore, in
  29. this lesson, you're going to learn about independent samples. Whereas,
  30. dependent samples deals with within subject designs, independent samples deals
  31. with between subject designs. In this case, the advantages of dependent samples
  32. are the disadvantages of independent samples. And the disadvantages of
  33. dependent samples are the advantages of independent samples. Does that make
  34. sense? With independent samples, we need more subjects because we need to
  35. randomize the two groups taking the two treatments. We need a larger end to
  36. control for individual differences as best as possible. That means it's more
  37. time consuming and generally more expensive. But then the advantages of
  38. independent samples are that we don't have carry over effects. Therefore, we
  39. can give one treatment to one group, another treatment to another group, and
  40. not worry about one treatment effecting the other, because each person or each
  41. subject only gets one treatment. With independent samples, we can do an
  42. experimental test where we give treatments to the subjects. Or observational,
  43. where we simply observe characteristics of two different populations, and then
  44. compare them. Everything is exactly the same. The Null and Alternative
  45. Hypotheses, the t-statistic, and the way we make our statistical decision.