W4 Standard Error, Confidence interval, Hypothesis & Significant values Flashcards
What does Standard Deviation (SD) provide an indication on?
SD = It provides an indication of how well the mean represents the sample data.
What does Standard Error (SE) provide an indication of?
How do you calculate SE of mean?
SE = It provides an indication of how well the sample represents the population
- Divide the SD by the square root of the sample size
Why do we use samples?
- because we can’t measure the whole population
What do confidence intervals do?
- It’s a way of seeing how well the mean represents the true population mean
- This is another way to see how well the sample represents the population
- It is an estimated range of values which are 95% likely to include the REAL population mean
- Gives you a lower boundary & upper boundary
How to calculate a confidence interval:
Lower boundary of 95% CI = X - (1.96 x SE)
Upper boundary of 95% CI = X + (1.96 x SE)
X = mean of the sample
What is a hypothesis?
- An estimation or proposed explanation of a theory made on limited evidence.
What is correlation (r)?
- Tells us the strength & direction of a relationship/association
What is regression?
- Allows one variable to predict the other variable
How would you test to see if two means are different?
- Use a independent T-test
Or - Paired/dependent T-test
What T-test would you use if the means were from two separate groups were different?
- The independent T-test
What T-test would you use if the two means are from one group of people but at two different times (E.g. before & after)?
- Paired/dependent T-test
What’s a Null Hypothesis?
- States there will be no relationship between the variables or states there will be no difference between the means
What’s an Alternative Hypothesis?
- This states that there will be a relationship between the variables or states there will be a difference between the means
What are the 6 steps to hypothesis testing?
- Null Hypothesis
- Alternative Hypothesis
- Directional
- Non-directional - Select a level of significance (alpha level)
- Collect & summarise data
- Run statistical test
- Interpret significance of results (p-value)
What is the Alpha level (a)?
- Typically set at 5% [a = 0.05]
- This is an error rate associated with incorrectly rejecting the Null hypothesis.
This means when the Null hypothesis is in reality true we will still (incorrectly) reject the Null hypothesis 5% of the time.
What is type 1 error?
- You conclude there is a difference when in reality there isn’t (sampling error unrepresentative samples)
What is type 2 error?
- You conclude there is no difference when in fact there is (sample size too small)
What is the probability value (p-value or Sig)?
- The p-value provides a measurement of the strength of the evidence against the Null hypothesis
If you have a small (p<0.05) probability value (p-value or Sig) does that mean the observed difference was more likely due to chance or not?
- The smaller the probability value (p-value or Sig) the more likely the observed difference was not because of chance.
- Smaller p-value = not chance
Examples of Significance testing:
- Once the analysis returns with a p<0.03 we … the Null hypothesis and interpret the difference as … yet, there is still a … percent chance (p<0.03) that the results were due to chance (due to sampling variation).
Fill in the blanks.
- …we reject the Null hypothesis…
- …interpret the difference as ‘real’ (significant)…
- …yet, there is still a 3 percent chance (p<0.03) that the results were due to chance…
- What can you assume if you have these p-values?
1. p<0.10 to p<1
2. p = 0.05 to 0.10
3. p<0.05
4. p<0.01
5. p<0.001
- You can assume the Null hypothesis is correct as there’s no evidence against it (no relationship or no difference between means)
- There is weak evidence against the Null hypothesis but not enough to reject it.
(3, 4 & 5 is what you want!!) - Moderate evidence against the Null hypothesis and in favour of alternative hypothesis
- Strong evidence against the Null hypothesis and in favour of alternative
- Very strong evidence against the Null hypothesis and in favour of alternative