Lecture 7 – Statistical Models2 - summary page
1) Standard error
Allows you to determine how well your sample can be generalized to the population.
SE = standard deviation of sample means
a. How well your sample represents the greater population from which the sample
b. The SE is a measure of the standard deviation of sample means.
Standard error = √¿
Standarddeviationof your sample
= √ n−1
X = independent score, xbar = mean, n = number of cases in your sample.
SE = 12/3 = 4.
SE = 12/4 = 3.
By increasing sample size we decrease standard error. This is good! We want as low of standard
of error as possible. By increasing sample size we increase confidence in our research.
Therefore larger samples are good because they decrease error.
2) Confidence intervals
a. 95% is the conventional value.
i. So we are likely to be wrong 5% of the time.
b. If you took 100 samples, the mean of 95 of those samples are likely to be within
the range of the confidence interval.
c. Overlapping confidence intervals.
If they’re overlapping can’t be sure there’s a true difference. Finding the range tells us that we
can be confident that the true value is likely to be somewhere in that range. If overlap is huge we
can’t be confident about where the average is. The more people