me myself and I lecture 6- cause and effect Flashcards
how do we know if the sample is a good representation?
We only ever have a SAMPLE. If our sample is large, we hope it will be a good representation of the true population. We never know for sure how close our sample is or isn’t to the true population. The larger and more unbiased it is, the more closely representative it is likely to be.
what is the sample average?
The sample average is known as the point estimate and it is our best guess at what the population average is given that we never know what it is.
what is the point estimate?
our best estimate (the point estimate) of the population average is the sample average.
what is a smaller sample more likely to contain?
sample error
what is the 95% confidence interval?
Light green line represents the sample average, our point estimate that most closely represents the population average.
95% confidence interval is calculated based on how much height variation there is in the sample as a fraction of the overall sample size. You don’t need to worry about the underlying maths here.
The 95% confidence interval is a range where is it reasonably likely that the true population average lies, given that we never know the true population average.
what does the 95% confidence interval describe? and what do large and small sample sizes cause?
describes the range within which we are reasonably confident the population average is located
If our sample size is large, it is likely our point estimate will be close to the true population average and the 95% confidence interval range will be short. Smaller sample – we have a greater margin of error. Larger 95% confidence interval.
what do overlapping notches suggest?
suggest there may not be a statistically significant difference between groups. We test this with a statistical model.
when do we use chi2 test and fishers exact test?
Chi2 test is not ideal here because of the small sample sizes. We should have at least 5 samples in each group for this.
We can use an alternative test that is more suitable for small sample sizes – Fisher’s Exact Test.
