Research Design in Sonography

Learning Objective • Students will demonstrate awareness of research statistics and research

design resources for professional development

Purpose of Learning Research Statistics and Research Design • Knowledge of statistics and research design is applied to help us

become better job applicants, sonographers, critical thinkers, and decision-makers.

• Sonographers are inundated with information, so it’s important to be adept consumers of this knowledge.

• This means asking questions and considering the validity of claims before we accept them as truth.

• Statistics and research methods can help us learn how to interpret and address data and information that we encounter.

Research Design

What Research Is (and Is Not) • Research is used to inform and potentially improve professional

practice • Research helps answer professional questions • Research guides decision-making • Research contributes to the professional body of knowledge • Research informs future innovations

But… • Research does not tell what is true or correct or “right” • Research does not prove a point of view—it only gives some evidence

Getting Started: Framing a Research Design • Research topic: a broad issue or area of study that is important to

investigate • Select topics from:

• Professional experience • Industry trends • Prior research studies • Existing theories • Communications within professional networks • Collaborations with other professionals

• Research questions: to focus and establish boundaries for the research study

Create a Hypothesis • A proposition to be tested; also called the alternative hypothesis • A statement of the relationship between two variables • Links the research questions with the research design • A testable or measurable statement • Null hypothesis: a statement related to the alternative hypothesis,

which states that nothing is happening, there is no relationship, or there is no effect

• For example, if the alternative hypothesis is that an increase in study time will correlate with an increase in students’ test scores, the null hypothesis would be that there is no effect on test scores when students increase study time.

• In hypothesis testing, researchers either reject or do not reject the null hypothesis (they do not “prove” the alternate hypothesis)

Choose a Research Design • Framework for answering the research questions and testing the

hypothesis (if applicable) • Includes:

• Approach to the research study • Type of research design • Selection of study participants • Research Methods

• Data collection methods • Data collection procedures • Data analysis strategies & statistics

Defining the Research Approach • Inductive research: start with data collection from observations or

tests, identify patterns in the data, suggest a theory • Deductive research: hypothesis testing; start with theory, create

hypothesis, collect/analyze data from observation or tests, confirm or reject the null hypothesis

• Abductive research: designed to explain incomplete observations, surprising facts, or puzzles.

Collect Research Data • The research questions indicate what types of data must be collected to

answer the questions • Types of data:

• Qualitative: words, descriptions, concepts, ideas, experiences, meanings • Quantitative: numerical data • Primary data: original information collected to answer the research questions (e.g.

from surveys or interviews) • Secondary data: information collected by other researchers and repurposed for the

research questions (e.g. census data, labor statistics) • Descriptive data: data collected through observations of existing actions that the

participants are taking; no research intervention • Experimental data: data collected after systematically intervening in the

participants’ behavior or treatment

Data Analysis: Apply Statistics

Research Statistics Overview • There are two types of research statistics:

• Descriptive statistics: statistics that summarize a data set within a study • Inferential statistics: statistics that help a researcher draw conclusions or

establish probabilities about the outcomes of a study

Descriptive Statistics • Statistics that describe a data set may include:

• Number of study participants (n) • Mean (average) • Median • Mode • Range • Quartiles • Variance • Standard deviation • Visuals such as distributions, histograms, and stem-and-leaf diagrams

• For example, when case studies report on research done with human subjects, there is often a table describing the population: the overall size of the group, sizes of the study’s subgroups, and demographic characteristics

Examples of Descriptive Statistics in Action • Median home price • Mean: average high and low temperature for a given day • Median income and Range of income for sonographers as a job category • Mode: most frequently purchased category of a product in the last month in

a retail business

Advanced Descriptive Statistics • Within descriptive statistics, it’s possible to analyze data from more

than one variable through bivariate and multivariate analysis • Bivariate statistics are a type of descriptive statistics that help a researcher

describe data from 2 variables and analyze a potential relationship between two variables

• Correlation statistics (regression analysis, Pearson’s r, Spearman’s rho) • Covariance • z-scores • Scatterplots

• Multivariate analysis is a type of descriptive statistics to help a research describe data from more than 2 variables, through tools like:

• Cross-tabulations • Contingency tables

Examples of Bivariate and Multivariate Descriptive Statistics • Bivariate: correlation between height and weight in children • Multivariate: cross-tabulations in a political poll result

Inferential Statistics • Researchers can get data from a sample and draw conclusions or make

predictions about the population from which the sample is drawn • Inferential statistics are used to:

• Make estimates about a population • Test hypotheses about populations

Sampling Error & Confidence Intervals • Because the sample being studied is smaller than the population it

represents, the data collected from the sample will inherently have a sampling error.

• Sampling error is the difference between the true value for the population and the value that the researcher identifies in data from the sample

• Confidence intervals describe the range of values that are most likely true for the population

Examples of Inferential Statistics in Action • A 20 subject sample size participates in a study of a new blood

pressure drug. Inferential stats (t-test) helps the researcher make inferences about the results to the population and in comparison to existing treatments.

• 20 patients are randomly assigned as 4 sample groups to use 4 different blood pressure medications. Blood pressure is measured before and after the patient started using their assigned medication to find the mean blood pressure reduction for each medication. Inferential stats (ANOVA) are used to draw inferences about the effects of the drugs on the population and in comparison to each other.

• Political polling research firms poll a small sample of people about political issues and candidates. Using inferential statistics, researchers can determine the preferences of the whole population based on the

Key Vocabulary for Statistics and Research Methods

• Population • Sample • Statistics • Variable • Analysis • Data set • Correlation • Research question • Hypothesis • Methodology

Resources to Learn Statistics and Research Methods

• Khan Academy: Statistics and Probability • Corporate Finance Institute: Inferential Statistics • Khan Academy: Introduction to Experiment Design • Khan Academy: Types of Statistical Studies • Khan Academy: Observational Studies and Experiments • Khan Academy: Identify the Population and Sample • Khan Academy: Correlation and Causation • Khan Academy: Data Collection and Conclusions •