SPSS
Ram Saran R
(1827318: [Link]@[Link])
Measurement- Phenomena of measuring something
Metric -Anything we derive out of measurement
Data- Two Categories
Primary Data
Secondary Data
Numeric --- a) Interval data(3) b) Ratio(4)
Non Numeric(categorical data) --- a) Ordinal(2) b) Nominal(1)
1. Nominal- Only for reference; E.g. Aadhar Card No, Locations,
Blood Group etc.
2. Ordinal- Can be ranked; E.g. Rating of faculties in CU( Likert
Scales)
3. Interval- Difference between successive intervals is equal. Have
presence of grading zero. E.g. Temperature scale
4. Ratio Scale-
Nominal and Ordinal: Weakest Measurements: Non Metric
Data: Non Parametric
Interval and Ratio: Strongest Measurement: Metric Data:
Parametric
Any data taken out of descriptive data- Stastic
SPSS Software- 20
Variable view- Shows data dictionary: Makes sense to dataset
Role- Change to Target from input, it becomes dependent
variables
Two types of Data Analysis
Supervised- when data has dependent variable[ e.g. y=g(x) ]
Unsupervised- when there is independent variable[x1,x2,x3….xn]
Data Preparation
Missingness
Outliers- Unnamed Observations- High or low number will lead to skewing
of data from mean value
Coding
Missingness – Causes: - Operator’s mistake,
Failed to collect the data
Overcoming: - Take and replace with average,
Remove the column or row
(Finding missing values in excel- select tab and press Ctrl+G- go to special and
check “blank”)
Imputation (works only on numerical values)
Imputation means to replace
Analyze- Multiple Imputation- Analyze Patterns (used only for numerical)
“N” represents no. of missing values
Go to the file Data, click Analyze- Multiple Imputation- Impute Missing
Data Values- select the data with highest no. of missing values-create new
dataset( select one more variable for reference purpose or SPSS will show
error)
We can change the number of imputation (e.g. 5 imputations means 5 data
sets will be generated). Better to use more datasets to increase accuracy.
Depend on the objectives.
(Accuracy- Errors are less)
Change Original Data to various dataset numbers to see imputed data
Save the dataset as excel file— File-Save as- Excel
Open saved excel file – apply filter to data- uncheck all blanks
Copy it to new sheet
Data-data analysis- descriptive statistics-ok- input range (select including
the range)- check labels- give output range-summary statistics- confidence
level- ok
Look at skewness values- if its positive, it means its right skewed. If the
value is close to zero, it is symmetric.