Study Guide & Take Home Quiz on Statistical Model Building and Regression

Unit I. Study Guide & Take Home Quiz # 1
Section 1: Statistical Model Building and Logistic Regression
There is a Chapter 13 Review with various types of questions on pages 334 – 338 ( Kellar & Kelvin).
Although not required as a part of this study guide and take home quiz, it is highly recommended that
you attempt the review questions.
1. Complete the following: Regression models are used extensively by researchers. They allow us to:
Identify and determine the validity of relationship that exist between dependent and independent
variables involved in the model.
2. List the various types of regression that can be used.
a. Linear regression
b. Polynomial regression
c. Logistic regression
d. Ridge regression
3. Complete the following: By multivariate we mean that the model:
Outcome of the model is predicted with more than one predictor variables
4. Multivariate regression models allow researchers to consider the effects of several independent
variables on one dependent variable of interest simultaneously. This is important for several reasons:
1. Comparing the coefficients and drawing conclusion on the effect of predictor variable from the
coefficient
2. Show the magnitude of effect of each predictor variable in the model in relation to the dependent
variables
3. Helps in showing the correlation of the predictor variable with the dependent variable
4. Gives the prediction of the dependent variable considering all the predictor effects
5. Define, describe & discuss the following::
a. Odds: These are numerical expressions of some events reflecting likelihood of occurrence of the
events. It can either be odds for or odds against. Odds for reflect the occurrence while odds against
reflect the likelihood that the event will not happen.
b. Odds ratio
This is the measure of association between an exposure and an outcome of an event. It gives the odds
representation that a particular result will be achieved provided a certain exposure in comparison to the
odds of the results in the exposure absence.
c. Risk Ratio or the Relative Risk
In the wound infection examination, the risk ratio is used to bring the comparison between two groups
by simply dividing the cumulative incidence in the exposed group by the unexposed group cumulative
incidence.
d. Odds Ratios Versus Risk Ratios
Odds ratio is a ratio involving two odds i.e. the odds of treatment group divided by odds in the control
group while risk ratio is the ratio involving two probabilities i.e. the probability of the treatment group
divided by the probability in the control group

e. Adjusted and Unadjusted Odds Ratios:
Adjusted odds ratio is the standardized simple odds ratio from two groups while unadjusted odds ratio is
the simple ratio of outcome probabilities from two groups
6, Complete the following statement: Logistic regression models are used when the:
When there is need to predict the categorical dependent variable from continuous /categorical
independent variables, when the assumption of linearity in the normal regression is violated by
categorical dependent variable.
7. Complete the following: Logistic regression models ask the following question:
Can the given set of predictors correctly predict the outcome categories? Are all the predictors having
relative importance?
8. What type of data is needed for logistic regression?
Quantitative data
9. Discuss how logistic regression works.
Like the linear classifiers, logistic classifiers apply the calculated logits or scores in the prediction of the
targeted classes. The input is classified to class 1when the output is formally greater than 0.5, class 2 will
be classified when the output is closed to zero and is less than or equal to 0.5 which can be achieved by
maximizing likelihood using the maximum likelihood estimate (MLE).
10. What kind of information is obtained from logistic regression models:
Odds ratio are obtained from logistic regression using more than one explanatory variable. The
association of variables are analyzed together to avoid confusing effects.
11. Statistical Model Building and Logistic Regression are advanced statistical procedures and require
additional course work to fully understand and calculate.
True _true____ False _____
Section 2: Linear Regression
There is a Chapter 14 Review with various types of questions on pages 367 - 370 ( Kellar & Kelvin).
Although not required as a part of this study guide and take home quiz, it is highly recommended that
you attempt the review questions.
1. Provide a narrative overview of what is linear regression:
It is the comparative analysis that is used to compare dependent and independent variables and to
determine significant predictor variable and the magnitude of their effects on the model. As well, linear
regression explains relationship between outcome variable and predictor variables.
2. Give an example of a research question addressed by linear regression:
Is the body weight having an influence in the blood cholesterol level? This is a forecasting an effect
question.
3. What types of data are required to use/compute a Linear Regression Model?
Continuous Quantitative data
4. What assumptions must be met in order to use Linear Regression?
There should be linear relationship between dependent and independent variable and the data should
not have outliers. The data need to be normally distributed and no multicollinearity exists in the data
and lastly, there should be little or no effect of autocorrelation.
5. In a short narrative explain simple regression:
This is the type of regression that allows the study and summary of the relationship between two
quantitative continuous variables.
6. There are two questions that we want to answer with a linear regression model what are they?
The forecasting an effect questions and the trend forecasting questions.

7. Discuss each of the approaches to selecting variables for inclusion in multiple linear regression
models.
Forward selection: Starts from completely empty equation by plugging in predictor variables one at a
time starting with the ones having the highest correlation with the outcome variable.
Backward elimination: This is the reverse process. All the involved independent variables to be used in
the equation are entered into the equations first and having them deleted one at a time if they do not
have effect to the regression equation.
Stepwise selection: This involves analysis at every step in the determination of the contribution of
predictor variables previously entered into the equation. The effect of the previously added predictor
variable is easily understood in the equation.
Block-wise selection: this is a type of forward selection that is arrived at in blocks or sets. Based on
psychometric consideration, predictors are grouped into blocks and have each block applied differently
and ignoring other predictor variables.
8. Linear Regression is an advanced statistical procedure and requires additional course work to fully
understand and calculate.
True _____ False _False____
Section 3: Exploratory Factor Analysis
There is a Chapter 15 Review with various types of questions on pages 396 – 398 ( Kellar & Kelvin).
Although not required as a part of this study guide and take home quiz, it is highly recommended that
you attempt the review questions.
1. Provide a short overview of Factor Analysis:
This is the technique applied in the reduction of the large number of variables into lower number of
factors. The maximum common variance is extracted from all variables by factor analysis and putting
them into a common score.
2. In the research literature of nursing and other health care professions, factor analysis is most often
used as part of the instrument development process. Discuss this use of factor analysis in the
instrument development process:
Factor analysis is used in the validation of the instruments in the measure of the scale used in the
instrument and also measuring the reliability of the instrument by using the Cronbach’s alpha
3. Complete the following: The building of theory is a principal purpose of research, and factor analysis
may support such eff9orts in a variety of ways:
By reducing the data that provide the least variance account
By showing the validity and variability the instrument used in the research
By incorporating the principal component analysis in the analysis when the KMO < 0.6
4. Discuss how factor analysis can be used for data reduction:
Through factor analysis, the principal component analysis is conducted that help in the reduction of data
by providing the effective dimensionalities of data by considering the variability of data.
5. For Factor Analysis what type of data is required?
Qualitative and quantitative data
6. Exploratory Factor Analysis is an advanced statistical procedure and requires additional course work
to fully understand and calculate.
True _true____ False _____
Section 4: Path Analysis