Simple Linear Regression - a measure of variation of the Yi values around their

- There is only one independent and dependent mean, Y. Simply known as total variation
variable regression sum of squares (SSR)
Independent variable is the variable that can - the explained variation of the total variation.
possibly predict the dependent variable
error sum of squares (SSE)
Correlation vs Regression - the unexplained variation of the total variation.
Scatter plot – shows relationship between two
variable (IV DV) error sum of squares (SSE)- the portion of the total
variation in the DV that is explained by variation in
Correlation – used to measure the strength of the the IV.
association (linear relationship) between two Note: 0 ≤ r 2 ≤ 1
variables.
Positive – upward Negative - downward Interpretation:
Regression analysis used to: (percentage) of the variation in (DV) is (explained or
Predict the value of DV based on the value of at least predict if above 80%) by variation in (IV)
one IV
Explain the impact of changes in an IV on DV Standard Error of Estimate
*Income and spending, wage - standard deviation of the variation of observations
DV: the variable we wish to predict or explain around the regression line is estimated.
IV: variable used to predict or explain the DV
Example: ASSUMPTIONS OF REGRESSION
Only one IV, X Linearity
Relationship between X and Y is described by a - states that the relationship between variables is
linear function linear. if it look like “isaw” then there is a violation in
Changes in Y are assumed to be related to changes the assumptions of regression
in X

TYPE OF RELATIONSHIPS

EQUATION OF SIMPLE LINEAR REGRESSION

Population:
Y = β0 + β 1 X 1
independence of errors
Relevant range of interpolation – range of
- requires that the errors are independent of one
response of the respondents must be within the
another.
range

MEASURES OF VARIATIONS
total sum of squares (SST)
 - processed by forming the variables into a structure 3. Communalities of Variables
called factors. Below 0.50 is a candidate for deletion or failing to
Normality of Errors explain the other variables.
- error values are normally distributed for any given Pseudo-factor
value of X - variables come from item-indicators or item- If it falls into on or above 0.40, but below 0.50; it
questions that describe the factor. depends to researcher’s discretion if the he will
remove it or not, (it has significant factor loading and
Basis: higher factor loading - the higher number in has theoretical support)
the group will be the name of the factor loading.
4. Scree Plot
- respondents: min. 100 observations. - 1 the eigen value should be greater than 1 to make
it significant.
- if you are going to split the samples for validation - if below 1, the factors are not considered anymore
purposes the respondents must be 200 or higher. in the analysis.

General rule: 5. Factor loading

Minimum is to have at least five times as many Factor loading Sample size needed for
observations as the number of variables to be significance
Equal Variance (Homoscedasticity) analyze. More acceptable is 10:1 (10 respondents .30 350
- the probability distribution of the errors has constant per one variable). .35 250
variance. .40 200
Interpretation: .45 150
.50 120
1. For Measures of Sampling Adequacy (MSA) .55 100
through KMO is needed to proceed factor analysis .60 85
and Bartlett’s Test of Sphericity. .65 70
.70 60
- KMO (Kaiser-Meyer Olkin Test) it should be
.75 50
greater than 0.5 for a satisfactory factor analysis to
proceed.
- look into rotated compenent matrix
0.90 or above, marvelous
0.80 or above, meritorious
0.70 or above, middling
0.60 or above, mediocre Multiple Regression Analysis
0.50 or above, miserable
Below 0.50, unacceptable Multiple regression models
- use two or more independent variables to predict
- Bartlett’s test, it should be significant (0.05 or the value of a dependent variable
lesser) to make the factor analysis reliable. Equation:
FACTOR ANALYSIS Y =b0 +b 1 X 1 +…+ bk X k
Exploratory factor analysis (EFA) 2. Total Variance Explained Predictors or explanatory variables
- only be used for if you can’t find or identify a The cumulative variance explained should be 60% or - independent variables
variable in the RRL. higher to make it valid and reliable.
Criterion or explained variable
- dependent variables The ZPRED (X) and ZRESID (Y) were measured on
the scatterplot to test the presence of a pattern in
coefficient of multiple determination graph.
- represents the proportion of the variation in Y that is - if the data show a diamond pattern or triangle
explained by the set of independent variables. pattern indicate the presence of homoscedasticity

Adjusted r-squared Sufficient number observations:

- to take into account both the number of independent - the number of parameters (# of the variables) times
variables in the model and the sample size. the number of observations (respondents per
- comparing two or more regression models that variable) or (total) is (more than or less than)
predict the same dependent variable but have a sufficient as there are (total respondents).
different number of independent variables.
Main analysis:
overall F test The coefficient of determination is (r-squared value).
- to determine whether there is a significant Its ANOVA has a F value of (#); look into SSR) which
relationship between the dependent variable and the means that the IV used together in this model
entire set of independent variables. The variables account for (percent) of the variances,
which has a good explanatory power, being close to
0.80.

Interpretation:

Collinearity test: (Coefficients Table)

If the VIF values is greater than 5, then there is a
presence of collinearity.

Normality test: (Test of Normality)

P-values of Kolmogorov- Smirnov and Shapiro-Wilk
tests

Homoscedasticity Test: