Class Notes
(810,031)

Canada
(493,925)

University of Toronto Mississauga
(23,382)

Sociology
(3,990)

SOC222H5
(93)

John Kervin
(32)

Lecture

# soc222Lec #2.docx (Lin. and Kranzler)

Unlock Document

University of Toronto Mississauga

Sociology

SOC222H5

John Kervin

Winter

Description

Tables Session # 2 ch. 2
Soc222 Lec #2
CCT: 1160, 2140, 2160
DV: Lab 1154
Levels of measurement:
- Each variable is nominal, ordinal, dichotomy, ratio
- All except ratios are categorical, because they cannot be divided into categories.
This matters because it effects what test you will use.
- Frequency distribution on quiz
- Sociologist’s ask research questions that involve 2 or more variables
o Haves causes and effects
o Causes: independent variable, effects: dependent variable
o If something has two categories we will call it nominal or ordinal (not doing
dichotomy)
SOC 222 -- MEASURING the SOCIAL WORLD
TABLES -- Session #2
WHERE WE ARE
Today’s Objectives: Know…
1. The difference between causation and correlation
2. Three criteria for causation
3. Importance of the three types of bivariate relationships
4. The two kinds of statistics and two key questions in statistics
5. Difference between a percentage and a proportion
6. How to produce a bar chart in SPSS
7. Measures of central tendency for category variables
8. How to produce a crosstabulation in SPSS
9. How to percentage tables 10. How to produce a bivariate clustered bar chart
Terms to Know
independent variable
dependent variable
covariation
descriptive statistics
inferential statistics
representative sample
frequency distribution
central tendency
mode
median
crosstab
effect size
BIVARIATE RELATIONSHIPS
- Something that looks like this : x y (x is causing y)
Dependent and Independent Variables
- The independent variable causes the dependent variable
- On the internet you can find a good job, a good job is found using the internet these
days
o Low internet use leads to poverty
o X must also come before y
o Gender is an independent variable, because it comes before, you are a girl or
boy before you are even born
Causation and Correlation - They are related, x changes, then y changes (whatever happens its systematic)
Criteria
1. Time Order
2. Covariation
- Internet varies from person to person, it is co-variation
- Co-variation is what statistics focuses on, does x vary systematically with y
o By itself co-variation is not enough to say that x causes y
3. Non-spurious
Three Types of Bivariate Relationships
DV
IV: Cat Rat
Cat Cat Cat Cat Rat
Rat Rat Cat Rat Rat
- The table shows that there are four possible types.
- Category causes ratio effects
- Rat rat
- Cat cat (no research uses this)
- Cat category independence to ratio dependent used in experiments psychology
- RatRat this is what we will focus on ratio independence to ration dependence
Importance of Each Type:
The Two Key Questions
- When can we conclude that two characteristic co-vary. - Two kinds of statistics to deal with when we want to show co-variation
Descriptive
- Is there a relationship, is it strong, what direction does it go in (is it positive; ones gets
bigger the other gets bigger. or negative)
- How strong is the relationship between x and y, is it positive or negative
Inferential
- How accurate are our population estimates.
- The accuracy of inferring to a population from a sample
- We use it to measure relationship strength between two variables, but this asks how
accurate is the strength of the relationship is we base it on the sample.
- Pg 20-21 (explanatory statistics should be ignored) Kr: pg. 31
representative sample
- Drawing samples: its not convenient to measure entire population
- Representative samples are those that capture and present the population diversity and
variation. Taking the two students from the front is not a good rep. of the entire lecture
hall.
o The people sitting in the back are different from those at the back, the ones in the
middle are different from those sitting in the ends.
- There is always a possibility that the sample is bad, and it does not capture the real
population.
- Sometimes we get inaccurate results due to a bad sample.
CATEGORY VARIABLES
- Variables categorized: ordinal, nominal, dichotomous
Frequency Distributions
- Frequency distributions are good, they tell us what the variables are, the categories ,
and the % of population in each category
Percentage and Proportion
- Percentage is the amount of a certain characteristic in regards to the total pop. - Proportion is the percentage divided by 100
SPSS Frequency Distribution
Department
Cumulative
Frequency Percent Valid Percent Percent
Valid Chem Eng 29 10.2 10.2 10.2
Elec and Comp Eng 47 16.5 16.5 26.8
East Asian Studies 39 13.7 13.7 40.5
English 20 7.0 7.0 47.5
Microbiology 63 22.2 22.2 69.7
Botany 19 6.7 6.7 76.4
Geography 11 3.9 3.9 80.3
Economics 38 13.4 13.4 93.7
Sociology 18 6.3 6.3 100.0
Total 284 100.0 100.0
The table shows:
- Valid percent measures only those individuals who answered.
- Cumulative only matters if you have an ordinal variable
- Nominal does not look at cumulative column
- Analyze, descriptive stats. Freq,
Bar Charts
SPSS Bar Charts
- For category variables best option is SPSS bar charts SPSS Bar Charts
Open data set
Click on “Graphs” on menu bar
Choose “Chart Builder” from dropdown menu, press ok
This opens the “Chart Builder” procedure box
• Top left: list of your variables
• Next to it: a white working area
• Bottom:
• A “Choose from” box. AKA “the gallery”
• On top of this box: four selection buttons
• Underneath: the five action buttons
Steps:
1. Decide which general type of graphic you want (left side of “Choose from” box)
2. Within that general type, pick the specific type (icons – will show name of type)
• Simple bar chart is on the left
3. Drag icon up to white working area
• This shows a sample of what you’ll get, with some spaces to fill in to
complete your chart
4. Drag your category variable into the X-axis area
• X-axis will show variable label
• Vertical Y-axis will show “Count”
5. Click “OK”
• The bar chart will show up in your output window. (Name of procedure
will be “GGraph”)
If you want to see percentages instead of counts:
• Open the “Element Properties” box (if it’s not already open)
• Edit Properties: Bar 1
• In “Statistics” area, under “Statistic”, pick “Percentage” on drop-down menu
• Ignore the question mark (since you’re percentaging on the total)
• Click “Apply”, then “Close” • Result: exactly the same appearance, but percentages i/o counts
- It shows the count / frequency of the student and the departments.
- How to get percentages instead of counts: element properties, click on bar 1. Under
statistic, where it says count, change to percentage.
- Shape of the bar chart is the same, just changes count to percentages
Central Tendency: Category Variables
- With any single category variable we ask: what’s the most typical case?
- Nominal: mode,
Nominal Variables • Mode: the category with the highest frequency
• That would be microbiology
Department
Cumulative

More
Less
Related notes for SOC222H5