HLTC15 –Introduction to Quantitative and Qualitative Health Research Methodologies
Lecture 8: Basic Statistics and Quantitative Methods II –Health Survey Methods (Chapter 10)
Tuesday, October 30, 2012
Helps summarize and organize a large amount of research material using numerical, graphical, and tabular techniques
o E.g., histogram, frequency tables
Levels of Measurement
Nominal scale: classify cases into categories that have NO quantitative ordering
o Classify subjects according to their names
Ordinal scale: we can rank order cases according to the quantity/intensity of the variable expressed by each case
o Basically, it’s like nominal but there’s direction
o E.g., small, medium, large; how much bigger is medium vs. small? We don’t know.
Interval: the order and the interval difference between two points
o The presence of zero is arbitrary
Temperature: 0 degrees C = 32 degrees F; zero really doesn’t mean absence of something
Ratio: an encompass all the properties of nominal, ordinal, interval
o However, zero is not arbitrary
Income: $25, $30, $40, $45; $0 means you’re broke
o Professor once told me not to use this
Why? Can you tell the difference between 19% and 20 %. A better alternative is dot plots
Bar graph or histogram?
o Bar graph: X-axis (abscissa) discrete variables (i.e. Nominal or ordinal). Y-axis (ordinate) counts,
o Histogram: compared to bar graphs there is no space between the bars. The x-axis holds continuous data while
the y-axis can be counts, percentages, etc.
Characteristics of a Distribution
According to Saks and Allsop (2007):
o The shape – right (positively) skewed or left (negatively) skewed?
o The centre – Measure of Central Tendency (mean, median, mode).
o The spread – IQR (max, min, median, Q1, Q3), variance, standard deviation
o The existence of outlier
Frequency Tables (FT)
Gives more detailed information compared
FT tallies the number of times each value of
the variable appears in a distribution.
Using frequencies we can calculate
proportion and percentages.
Measures of Central Tendency Measures of Dispersion
Range: Difference between highest and lowest values
Inter-quartile range (IQR): Is the range for the middle 50
per cent of cases in a rank-ordered series: difference
between Q3 and Q1.
Variance and standard deviation
Bivariate Descriptive Statistics: Simple Comparison
Bivariate statistical analysis addresses the relationship between two variables.
Is there a relationship between our independent and dependent variables? Is there a correlation or association between
our variables? What is our r value?
Table 11.5 Lists numerous measures of association and correlation.
Measures of Association and Correlation
R values can illustrate the strength of a relationship and ranges from -1 to +1 (with 0 meaning no linear relationship)
Pg 214 Table 11.6 lists all the values and relative strengths
Furthermore, the + or – sign in front of your r value will signify a positive or negative relationship
Difference between Correlation and Regression
Correlation coeffic2ent (r): measures the strength of the relation between X and Y variables
Linear regression (r ) via the least-squares regression line is a prediction equation, estimating value of y for changes in x
Use of numerical techniques for making conclusions about a population based on information obtained from a random
sample drawn from a population.
Calculating the statistical significance of a sample result can be examined using a p-value approach or by calculating
confidence intervals (estimation).
Quantitative Survey Methods
We talked about sampling. What happens after sampling? We need to collect data in order to analyze this (either
qualitatively or quantitatively). Surveys can help in this data collection process.
Types of Surveys
Ad hoc surveys
o One-time surveys conducted with samples between 200-2000; usually administered face-to-face or through
o Used a lot in market research.
Cross-sectional surveys (e.g. British Social Attitudes Survey: annually administered, closed-ended, structured surveys
used to monitor changes in attitudes)
Surveys used in longitudinal studies (cohort or panel studies)
Survey methods utilize:
o Questionnaires distributed via face-to-face or telephone interviews (both are types of interview surveys).
o Structured observation
o Open-ended or semi-structured questions
o Structured record review
o Self-completed questionnaires through postal/mail communication and submission (e.g. Canadian Census)
Why use Survey Methods?
Address questions that are both descriptive and analytical.
o Who, what, where, when, how questions.
o If we can control for confounding variables and bias, we could possibly answer ‘why’.
o Can explore variations between two different groups. Before we decide to conduct survey...
We must ask some practical/important questions:
o General timeline for rese