The Practice of Social Research

advertisement
Foundations of Sociological Inquiry
Statistical Analysis
Today’s Objectives





Why use Statistics?
Descriptive Statistics
Inferential Statistics
Multivariate Techniques
Questions?
The formula Y = f(X) tells us that
1.
2.
3.
4.
5.
X is the dependent
variable.
Y is the dependent
variable.
f is the dependent
variable.
need to know what Y, f,
and X represent to
determine the
dependent variable.
None of these choices
is correct.
62%
17%
12%
4%
1
2
3
4%
4
5
Why Use Statistics?
Statistics enable us to construct simplified
representations of a complex social world.
Why Use Statistics?
Statistics enable us to construct simplified
representations of a complex social world.





Begin with a sociological question
Identify data to answer the question (collect,
observe, record)
Analyze data (statistics help)
Present your findings (statistics help)
Situate your findings in relation to what we think we
already know (statistics help)
Recommended Salary for Job Candidates:
$4000 $70,000 $40,000 $80,000 $120,000 $135,000 $70,000 $50,000 $67,000.00
$500,000 $50,000 $75,000.00 $60,000 $150,000 $20,000 $50000.00 $70,000 $80,000
$62,000 $200,000 95,000 $75000 $70,000 $80000 $75,000 $45,000 a year $100,000
$250,000 $65,000.00 $45000.00 $75,000 $88,000 $80,000.00 $150,000 $55,000
$130,000 $60,000 $78,000 $150,000 $50,000 70000 $45,000-60,000 $80,000 $75,000
$55000 $40,000 95,000 $80,000 $30,000.00 $80000 $30000 $70,000 $50,000 $50,000
$65000 $80,000 $? $80,000 $50000 $50000 (I have no idea how much Marketing
Executive gets paid usually) 150,000 $74,000 $60,000 $60,000 $65,00 $80,000 $65,000
$90,000 $70,000 $90,000 $80,000 $45000 $45000 $35000 $100,000 $85,000 $50,000
$60000 80000 $85,000 $58000 $60000 $70,000 $80,000 $70,000 $40000 $70,000
$80,000 $60,000 $200,000 $80,000 $50000 $60,000 - $75,000 $80,000 $60,000 $45,000
$50,000 $90,000 $30,000 $60,000 50000 $200,000.00 $40000.00 $60000 $50,000
$75,000 $60000 $180000 $120,000 $80000 $55,000 $50,000 85000 $145,000 $ $85,000
$55,000 $70000 $75, 000 $60,000 60000 $ 10,000 $100000 $65000 $85,000 $80,000
$60,000 $ 70,000 $80,000 $75,000.00 $100,000 $50000.00 $70,000 $95,000 $92,000
$70,000 $50,000 $68,000 $80,000 $40,000 $30,000 $50,000 $60,000 $40,000 $80,000
$65,000 $i dont know $90,000 $60,000 $70,000 $80,000 $65000 $70,000 $ $100,000
$72000 $70,000 $50,000 $110,000.000 $80000 $18,000 $110,000 $200,000 $100,000
$80000
Descriptive Statistics (summary)

Statistical computations describing either the
characteristics of a sample or the relationship among
variables in a sample




Data reduction
Measures of association
Regression analysis
Other forms of multivariate analysis
Recommended Salary for Job Candidates
Mean
Std. Dev
N
Male
Respondents
$77,277
$37,904
96
Female
Respondents
$78,837
$56,897
65
Source: Data were collected from students enrolled in
Sociology 300.
Recommended Salary for Job Candidates
Difference in Means

Is the difference in mean salary recommended by
men and women statistically significant?
Difference in Means

Is the difference in mean salary recommended by
men and women statistically significant?

Conduct a t-test
t = 0.20, df = 154, p-value = .84
95 percent confidence interval (-13392, 16512)
Difference in Means

Is the difference in mean salary recommended by
men and women statistically significant?

Conduct a t-test
t = 0.20, df = 154, p-value = .84
95 percent confidence interval (-13392, 16512)

We should not reject the null hypothesis that the true
difference in means is equal to zero
Recommended Salary for Job Candidates
Multivariate Analysis

Is the difference in mean salary recommended by men
and women statistically significant, controlling for parental
status of applicant?
Multivariate Analysis


Is the difference in mean salary recommended by men
and women statistically significant, controlling for parental
status of applicant?
Conduct a regression analysis of recommended salary
Variable
Male Respondent
Parent Applicant
Intercept
Estimate
-1452
-14018
85846
t-value
-0.18
-1.77
13.18
P-value
0.86
0.08+
<.001***
Multivariate Analysis


Is the difference in mean salary recommended by men
and women statistically significant, controlling for parental
status of applicant?
Conduct a regression analysis of recommended salary
Variable
Male Respondent
Parent Applicant
Intercept

Estimate
-1452
-14018
85846
t-value
-0.18
-1.77
13.18
P-value
0.86
0.08+
<.001***
We should not reject the null hypothesis that the true
difference in recommended salaries, controlling for
parental status of applicant, is equal to zero
Inferential Statistics

The body of statistical computations relevant to
making inferences from findings based on sample
observations to some larger population.


Sampling error
Non-sampling error
_____ indicate the likelihood that the
relationship observed between variables in a
sample can be attributed to sampling error
only.
66%
1.
2.
3.
4.
Ex-post facto
hypothesizing
Tests of statistical
significance
Disconfirmation
Disambiguation
19%
11%
4%
1
2
3
4
Statistical Significance

Statistical Significance is a general term referring to
the likelihood that the relationship observed in a
sample could be attributed to sampling error alone.
Statistical Significance

Statistical Significance is a general term referring to
the likelihood that the relationship observed in a
sample could be attributed to sampling error alone.

Tests of Statistical Significance are a class of
statistical computations that indicate the likelihood
that the relationship observed between variables in a
sample can be attributed to sampling error alone.
Statistical Significance

Statistical Significance is a general term referring to
the likelihood that the relationship observed in a
sample could be attributed to sampling error alone.

Tests of Statistical Significance are a class of
statistical computations that indicate the likelihood
that the relationship observed between variables in a
sample can be attributed to sampling error alone.

Level of Significance, in the context of tests of
statistical significance, the degree of likelihood that
an observed, empirical relationship could be
attributed to sampling error.
_____ are statistical measures used for
making inferences from findings based on
sample observations to a larger population.
1.
2.
3.
4.
Descriptive
statistics
Inferential
statistics
Both of the above
Neither of the
above
59%
27%
11%
3%
1
2
3
4
A statistical significance level of .05 means that
1.
2.
3.
4.
5.
the probability that a relationship
as strong as the observed one
can be attributed to sampling
error alone is 5 percent.
we can be 5 percent sure that
the relationship is real and not
due to sampling error.
there is an .05 percent chance
that a relationship as strong as
the observed one can be
attributed to sampling error.
the difference we observed in
the table is 5 percent different.
there is a 5 percent standard
error in the observations.
38%
31%
19%
9%
4%
1
2
3
4
5
Questions?
Download