SOC202H1 Lecture Notes - Multiple Choice, Standard Score, Confidence Interval
This preview shows half of the first page. to view the full 2 pages of the document.
SOC202H1S Qualitative Analysis in Social Sciences
February 7 2012 – Week 5
- Today we will talk about confidence interval
10 things you should know about C.I.
1. Confidence intervals are all about range of possible values of a parameter
Whether or not if it is high or not, it is not based on entire population.
It is sample-based. What if it’s wrong?
The quality of sample you have is essential
It is based on theoretical idea of the sample
If you can just come back to visualizing the concrete it will provide the
estimate of the population
You have to translate it into a theoretical sample distribution. What is
the possibility that you got Gwen Stefani? Very unlikely.
2. CIs are based on the notion that we use sample stats to estimate the value of
a population parameter
Not a distribution of scores on the first test
The sample size matters and that is why sample size has influence on
3. Calculation for a CI is as follows (refer to the slide)
Instead of memorizing formula the calculation is point estimate
The size of error is essential
How far am I off from the point?
To calculate the error term you need standard deviation.
The sample is really critical in this
Correlation between education and income is related. It doesn’t make
sense to you now, it will later on after the first test.
4. Obtain the standard error
Go back to the idea of spread was small, like 7,8,7,8,7,8, it is very little
spread. However, if it is 2,7,4,0,8, sample deviation is huge so more
Taking what you have and comparing to hypothetical
5. We need to obtain the error term
6. What is the critical Z score and how to obtain it
1.96 is the number you should remember
Low probability of scooping up a sample
The larger the standard error, the questionable the answer.
What is the true population mean
o The standard error size is an indicator of how many
variability is there in population
7. What happens to the size of width of the CI when the size of sample standard
Sample-to-sample estimates may be worse
You're Reading a Preview
Unlock to view full version