DATA ANALYSIS AND MANAGEMENT

Choosing the appropriate statistical test

One of three main topics in the lecture

Dealing with scales and items

One of three main topics in the lecture

Sample demonstration using OpenEpi

One of three main topics in the lecture

Sex and hypertension

A research question example used in the scenario

Sex and systolic blood pressure

A research question example used in the scenario

100

The number of respondents in the scenario

40 males and 60 females

The sex distribution of respondents in the scenario

25 out of 40 males

The number of males diagnosed with hypertension in the scenario

35 out of 60 females

The number of females diagnosed with hypertension in the scenario

120 with an SD of 5

The mean systolic blood pressure for males in the scenario

115 with an SD of 20

The mean systolic blood pressure for females in the scenario

Millimeters of mercury (mmHg)

The unit blood pressure is measured in

Identify, define, and categorize them

Three things to do with research variables

NOIR classification

One way to categorize variables

Continuous versus discrete classification

Another way to categorize variables

Nominal and Ordinal

NOIR categories that fall under discrete or categorical variables

Interval and Ratio

NOIR categories that fall under continuous variables

Determine which statistical test to use

What variable classification helps determine

Outcome variables and predictor variables

Another concept to understand about variables

Dependent variables

Another word for outcome variables

Independent variables

Another word for predictor variables

Cause

What the predictor variable is quote unquote

Effect

What the outcome variable is quote unquote

Association or correlation

The relationship typically sought in observational studies

Association

The type of relationship the research questions in the scenario are looking for

Predictor variable

The variable assigned as Sex in the scenario, based on causality direction

Outcome variable

The variable assigned as Hypertension in the scenario, based on causality direction

Hypertension and Systolic blood pressure

The outcome variables in the scenario

Jasp

An open source statistical software mentioned as a source for flowcharts

Determine what type of statistical test to use

What a decision tree or flowchart helps determine

Sex

The predictor variable in the scenario

Hypertension and Systolic blood pressure

The outcome variables in the scenario

Categorical or Nominal

The classification of Sex as a variable

Categorical

The classification of Sex as a variable in the continuous vs categorical dichotomy

Categorical or Nominal

The classification of Hypertension as a variable in the scenario

Yes or No

The possible values for the Hypertension variable in the scenario

Numerical value

How Systolic blood pressure was measured in the scenario

Ratio

The classification of Systolic blood pressure as a variable in the scenario's measurement

Continuous

The classification of Systolic blood pressure as a variable in the continuous vs categorical dichotomy

Sex and Hypertension

The two categorical variables in the scenario

Systolic blood pressure

The continuous variable in the scenario

Hypertension

The outcome variable for the first research question

Categorical

The type of outcome variable for the first research question

One

The number of outcome variables focused on in most cases according to the lecture

One

The number of predictor variables for the first research question

Sex

The predictor variable for the first research question

Categorical

The type of predictor variable for the first research question

Pearson chi-square or likelihood ratio

The statistical test to use for the first research question

Chi-square analysis

A more common name for the Pearson chi-square test

Systolic blood pressure

The outcome variable for the second research question

Continuous numerical

The type of outcome variable for the second research question

One

The number of outcome variables for the second research question

Sex

The predictor variable for the second research question

One

The number of predictor variables for the second research question

Categorical

The type of predictor variable for the second research question

Two

The number of categories in the predictor variable (Sex) in the scenario

Males and Females

The two categories for the predictor variable (Sex) in the scenario

Did we measure each male multiple times?

A question asked to determine same or different entities

Different or independent entities

How respondents are classified if each is a separate entity

Different or independent entities

The most likely classification for entities in most research studies

Independent t-test or point biserial correlation

The statistical test for the second research question, assuming different entities

Independent t-test or t-test

The recommended test for the second research question

Parametric tests

Statistical tests used if data are normally distributed

Nonparametric alternative

Used if data values do not follow a normal or Gaussian distribution

Independent t-tests and Pearson chi-square

Two common statistical tests discussed initially

One-way ANOVA

An extension of the independent t-test

More than two categorical predictor variables

When to use one-way ANOVA

ANOVA

A test among three or more categories for a categorical predictor variable

Regression analysis

An important set of statistical tests

Multiple regression and logistic regression

Two very basic types of regression analysis

The outcome variable

What is very important in determining whether to use multiple vs logistic regression

Continuous

The type of outcome variable used with multiple regression

Categorical

The type of outcome variable used with logistic regression

Logistic regression

The regression type to use with Hypertension as the outcome in the scenario

Multiple regression

The regression type to use with Systolic blood pressure as the outcome in the scenario

Statistical significance

What t-tests, ANOVA, and Chi-square analysis provide

Strength of association

What t-tests, ANOVA, and Chi-square analysis do not tell you

Strength of association

What regression analysis (multiple/logistic) can give you an idea of

Odds ratios or coefficients

How strength of association is shown in regression models

Pearson correlation

A common test providing statistical significance and strength of association

Two continuous variables

The type of variables needed for Pearson correlation

Systolic blood pressure and age

An example of two continuous variables for Pearson correlation

Pearson coefficient (r)

What you get from Pearson correlation besides the p-value

0 to 1

The range of the Pearson r value

Perfect correlation

What a Pearson r of 1 signifies

No absolute correlation

What a Pearson r of 0 signifies

Higher the correlation

What a higher Pearson r value indicates

Deciding on data analysis plan and study design

What knowing the correct statistical test helps with

Likert scales and Likert items

Common elements in survey forms

Contains several items

A characteristic of a Likert scale

A single question

What is not a Likert scale

Response levels are arranged horizontally

A characteristic of a Likert scale

Response levels are scored with consecutive integers

A characteristic of a Likert scale

Verbal labels

What response levels in a Likert scale are anchored with

More or less evenly spaced gradations

What verbal labels in a Likert scale should connote

Bivalent and symmetrical about a neutral middle

A characteristic of verbal labels in a Likert scale

Measures attitudes in terms of level of agreement/disagreement to a target statement

A characteristic of a Likert scale

A single item in a Likert scale

The definition of a Likert item

Likert items, not Likert scales

What individual questions in a survey questionnaire are

A Likert scale

What multiple Likert items form