Sampling Research'''

1
Sampling Research...

• Making some sense of qualitative data

2
Methods of collection

• Archives / databases
• direct observation
• survey sampling

3
Method of analysis

• Chi square (X2)
• goodness of fit
• test of independence

4
One way X2 - goodness of fit

• Expected outcomes of
• tossing a coin (5050)
• rolling a die (16.616.616.616.616.616.6)
• selecting an UG student (25252525)
• Null hypothesis observed frequencies will not
  differ from what is expected (i.e., expected
  proportion x total sample size)

5
The decisive question
Is the observed frequency significantly
different from what was expected?
6
Applying this principle to the friendliness
project...
7
X2 determines if this result is significantly
different from what was expected (i.e., equal
frequency of response).
significant X2 and p values
8
Degrees of freedom
df number of cells (C) -1
9
In this example, df 2
10
Two way X2 - Test of independence of 2
variables

• Expected outcomes of any given cell
• total in row, multiplied by total in column
• product divided by total sample size
• Null hypothesis observed frequencies will not
  differ from what is expected (see formula above)

11
Are responses independent of observers gender?
12
Is that test significant?
NO!
13
Degrees of freedom
df number of columns minus 1 (C-1) times number
of rows minus 1 (R-1)
14
In this example, df 2
15
Are responses independent of nature of contact?
16
Is that test significant?
Yes!
17
Are responses independent of gender and nature
of contact?
Eye Contact Only
18
Is that test significant?
No!
19
Are responses independent of gender and nature
of contact?
Eye and Verbal Contact
20
Is that test significant?
Not quite!
21
Discussion of results

• Are responses of subjects independent of observer
  gender?
• Are responses of subjects independent of nature
  of contact?
• Is campus friendly based on this experiment?

22
Experiment variables

• Control
• gender of investigator
• nature of contact

• Confounding
• time of day
• groups of people
• gender of subject
• gender similarity
• weather
• location on campus
• appearance of observer
• ethnic similarity
• age similarity
• classification of subject
• etc, etc, etc.

23
How else might we measure friendly?

• people helping people (e.g., volunteer groups)
• level of positive interaction in classes, food
  court, on sidewalks, etc.
• content of school newspaper
• etc.