Documente Academic
Documente Profesional
Documente Cultură
Usefulness
Ease of use
Problem Practical
Theoretical
Research design
Framework
Responses
How to execute
Data collection
Analysis
Discussion
Recommendations
assume
Data coding
Spss (recommended)
Or excel
Missing vale
Suggested by hasiel
Outliers:
Bad data
Normality
Multicollinearity
Design instrument
1. Indicator reliability
a. Factor loadings should be = or greater than 0.5
b. Deleting one item will improve loading of other loadings
c. Do not delete items
d. Delete lowest one and run
e. And then delete other than run
f. If you are deleting item from a variable will affect the items of that variable
2. Internal consistency reliability:
Convergent vailidity
4. discriminant validity
falnel and locker method widely used
top score should be higher of the lower and its nearby value
cross rooting
htmt method
Conceptualized problem
Lowest loading run and again lowest loading and again run
If data proves these four test it means that it has passed 1st step
Conceptually different from eachother these Ivs should not have collinearity
VIF score should be less than 5 if scores are greater than 5 than these two IV will not be used
together
Higher order construction by adding two variables by back it with the theory and literature
Assessment of path coeficients
Hypothesis testing, hypothesis can be directional ( x positive influence on y can be positive as well as
negative) and sometimes non directional ( where you say x have relaations/impact on y) directions is
found through literature. Non directional is used where literature is no avaiale. Non directional is two
tailed t test. If directional then one tailed t test. Two tailed 1.96. spss runs two tail default in regression.
Normality of data small sample size normality problem. Pls is the non pyramatiic approach to deal with
the non normal data, follow a procedure of bootstrapping. T values says suffort in what significance.
Significance of three type, 1% 5% and 10%. P values is computed on the basis of t value
Assessment of R2
Explained as in model Iv how much variance they explain Ivs can be more in the model 0-100% both
ends hard and almost impossible 10% s required for publish. Haeralt 2017 .20 is substantial for
behavioral science.
If R2 is more or 80% then there is problem with your data, you do not have to report ajst R2
Assessment of F2
Ivs every what is Ivs relative importance over Dv no any threshold, if f2 is 0.02, 0.15 mid, 0.3 high cohem
1988
Assessment of Q2:
This above analysis was for simple excluding mediation and moderation
Statistics
Theory( literature)
Mediation:
Calculates sth called indirect effect look in the paper how theoriticaly justify
Three approach
Confidence interval
Preacher and hayez citation 2006, 2007 ,and 2008 said do bootstrapping and see confidence interval