Soc 3306a Lecture 3 Developing the Research Question
Dec 18, 2015
Soc 3306aLecture 3
Developing the Research Question
Research questions and objectives (adapted from Fig. 3.1 Blaikie 2000)
‘What’ and ‘Why’ questions
What? Descriptive questions
What types of people are involved? What is their characteristic behaviour? What are consequences of their behaviour?
Why? To find causes for or reasons why
Why do they think/act this way? Why does this behaviour have particular
consequences?
Using statistics to answer research questions Statistics are mathematical tools
used to organize, summarize, and manipulate data.
Data gathered through survey items and are the scores on the variables used in a statistical analysis. Data is simply information expressed
as numbers (quantitatively).
Statistical Applications
Two main statistical applications: Descriptive statistics Inferential statistics
Descriptive Statistics Univariate descriptive statistics
summarize the information on one variable at a time. %, mean, standard deviation, histogram
Bivariate descriptive statistics summarize the relationship (i.e. describe the strength and direction) of the relationship between two variables Gamma, correlation coefficient,
scatterplot
Descriptive Statistics
Multivariate descriptive statistics describe the relationships between three or more variables. Multiple correlation, two-way ANOVA,
multiple regression. Can describe the complex and multi-
layered relationships in the social world.
Inferential Statistics
Can use to make inferences from a random sample statistics in order to generalize the results to a population.
Can also use to test for significant differences in group means or proportions.
This is “hypothesis” testing Z or T-tests, one-way ANOVA (F), Chi-
square statistic
Variables….
Used to calculate statistics Are concepts in numerical form
that can vary in value Have traits that can change values
from case to case. Examples:
Age, Gender, Race, Social class change for different individuals in a survey study
Case
The entity from which data are gathered Can be individuals, groups,
organizations, nations etc. The “unit of analysis” In the datasets you will be using,
each case is an individual survey respondent
Reliability and Validity Validity = truthfulness of a measure
Is it measuring what the researcher thinks it is measuring?
Reliability = the consistency or dependency of a measure Does the measure consistently give
the same results? Reliability can be assessed numerically
with statistics like Cronbach’s alpha
Measurement Error Systematic: the measure (ie survey
item) is flawed or distorted and does not reflect a respondent’s true attitude or behaviour this is validity
Random: lack of agreement between repeated uses of a measure This is reliability
Precision error: related to the “level of measurement” of a variable
Important aspects of Quantitative Variables:
a) Independent vs dependent variables Antecedent (control) and Intervening
b) Continuous vs. discrete categories Mutually exclusive and exhaustive
attributes c) Levels of measurement:
nominal ordinal interval/ratio
Independent and Dependent Variables
In causal relationships: CAUSE EFFECTindependent variable X dependent variable Y
Independent variables (“causal” or “explanatory”) are those that are manipulated. Can use more than one in a multivariate analysis (X1, X2, X3….)
Dependent (“outcome” or “response”) variables are measured for variation in response to X.
Discrete and Continuous
Discrete variables are measured in units that cannot be subdivided. Example: Gender
Continuous variables are measured in a unit that can be subdivided infinitely. Example: Age
Level Of Measurement
The mathematical quality of the scores of a variable. Nominal - Scores are labels only,
they are not numbers. Ordinal - Scores have some
numerical quality and can be ranked. Interval-ratio - Scores are numbers
All mathematical operations possible. Most “precise” level of measurement
Survey items as variables
Survey items are questions used to measure demographic, behavioural, or attitudinal aspects of individuals (cases)
Each item is a variable Response categories should have
mutually exclusive and exhaustive attributes
Items can be combined to create composite measures (i.e. Index, Scale)
Using one of the datasets to answer ‘what’ and ‘why’ Look at the variables in one of the
datasets available for your research to see which variables might be suitable to answer ‘what’ and ‘why’ questions
Examine the actual questions asked and the level of measurement of answers
Which variables are of interest to you? What type of questions can you ask?
Hypothetical problem statement: (Uses variables from a 2007 CCHS dataset)
“How is overall health affected by the food choices that are made?”
Is this problem important? Why? Other questions related to this problem:
What factors determine one’s food choices? What are the consequences of the food
choices people make? What role do doctor/dentist visits play?
Other related questions What age groups or gender make
healthier food choices? What are the consequences of less
healthy food choices? What group is more likely to visit
the doctor/dentist regularly? Does ability to chew affect food
choices and consequently, overall health?
Identify specific variables in a dataset that can be used to solve the problem….
Respondent’s Age Respondent’s Gender Consumption of fruit (index) Consumption of vegetables (index) # of doctor, dentist visits Respondent has trouble chewing Respondent’s overall health status
Relations Between Variables
Hypothesis: a statement that describes the relationship between two or more variables.
Null hypothesis: What is actually tested in a statistical test
Alternate hypothesis: The research hypothesis. “Rejection” of the null builds up evidence for the research hypothesis
Testable Hypotheses Hypotheses are stated in terms of
your chosen specific variables:
1. Higher consumption of fruit and vegetables leads to better overall health (main hypothesis)
2. Ability to chew increases the likelihood of higher fruit/vegetable consumption
3. # of dentist visits is related to the ability to chew
The Causal Model(See Figure 4.7 in Gray and Guppy)
Possible model to answer research questions and test hypotheses above:
DVOverall Health Status
Main IVConsumption of Fruit and Vegetables
AV (Control)Gender
AV# of Dentist visits
Antecedent Variable (control)Age
Intervening variableDoctor visits
AVAbility to chew