PSY 130 Lecture Notes - Lecture 17: Statistical Hypothesis Testing, Statistical Significance, European Credit Transfer And Accumulation System
Spring Semester
2018
Psych 130
Psychology
Personality Development and Effective Behavior
● Statistical significance:
o How can researchers say whether discrete group averages on their dependent
variables reflect real impact ects or only fluctuations of chance
o Luckily, statisticians have established formulas that allow us to estimate the
probability that, by chance alone, the difference between the averages would have
occurred
o There are several types of statistical assessments, each suitable for differentiated
data types and differentiated testing designs
o Several of the more common measures are a variance analysis, a chi-square test
and a coefficient of correlation
o Returning to the example of the waitress, if the two averages are erased by such a
small amount that it could have been triggered by a chance fluctuation, we
conclude that the difference did not hit a statistical significance cancelation
o Conversely, if the difference is so high that it was not induced by chance in all
probability but reflects a true difference between the two waiting types, we
conclude the difference is statistically important
o The inference in the above case will be that one type of waitressing does seem to
result in better tips than the other type
o Statistical testing does not really provide a yes or no answer to our question,
though
o What they tell us is the statistical probability that the disparity was caused by
chance between the classes
o For example, suppose we apply a statistical test to the data of the waitress, and
find that one out of every four times a difference this big will happen by chance
o How can we make out of that
o They reflect a real impact at the different averages
o To have a lot of faith in such a statement would be difficult
o We may have found a genuine difference, but there is a strong possibility that the
finding is only a fluke
o And when will we assume that we really have a difference
o Traditionally, the amount of meaningful chance psychologists use is 05
o Thus, if the difference between the scores is so high that by chance it will occur
less than 5 percent of the time, the difference is possibly true
o When reviewing studies in this book, we claim differences were identified
between classes, indicating that researchers performed appropriate statistical tests
and found a statistically significant erence of cant diff
o Yet bear in mind that these findings are not automatically in any way
"significant"
o Only minor variations can be statistically important when researchers use a large
number of participants
