BENGKEL ASAS SPSS

1
BENGKEL ASAS SPSS

• BY
• NOR IDAYU MAHAT
• FAKULTI SAINS KUANTITATIF

2
Some Revisions

• Do you familiar with those terms?
• Population vs. Sample
• Objects/respondents
• Variables vs. Constant value
• Parameter vs. Estimator
• Randomness

3
Basic Practice in Analysing Data

• Identify and determine the purpose of your study.
• Collect data.
• Perform initial analysis on the data.
• Pursue with complex analysis (if necessary).
• Documentation.

4
Initial Data Analysis

• It is a good practice to look at the data and to
  identify their main features before pursuing
  complex analyses e.g. hypothesis testing,
  modeling and multivariate analyses.
• Objectives
• To identify some anomalies.
• To determine a suitable technique that can be
  employed on the data.
• For validation purposes.
• To make better interpretation on the obtained
  results.

5
Understand Your Data

• How many objects/respondents?
• How many variables have been observed?
• Are those variables are numeric, categorical or
  mixed of both?
• Do we have groups of objects/respondents?
• Can you identified some irregular values in the
  data set?

6
Missing Value

• Definition Objects with no value in some
  variables.
• Some strategies to handle missing value
• Exclude objects with missing value.
• Replace a missing value by the mean of all
  available values for the relevant variable.
• Imputation missing values can be replaced by
  some suitable numerical entries.

7
Outlier

• Definition Values that are distinctly different
  from other values.
• Outliers may contribute to biased estimated value
  and this leads to give misleading results.
• Strategies to handle outliers
• Outliers due to recording errors should be
  corrected.
• If the values are genuine then some thought must
  be given as to whether or not they should be
  retained.

8
Handling Many Variables

• Analyse each variable separately
• Easy to perform, but often we face some
  difficulties to make general conclusion about the
  data.
• Analyse all variables simultaneously
• Involve complex procedures but we may easily
  conclude the obtained results.

9
Strategies to View Your Data

• Use statistic value.
• Familiar statistics can be used such as mean,
  variance etc. But, there are too many values!
• Plot (e.g. histogram, bar chart etc.)
• Better than value of statistic but limited to 2
  or 3 variables at one time.
• Map the data ( e.g. this can be done with
  Principal Component Analysis, Data Dimensional
  Scaling etc.).

10
Some Informative Statistics

• General statistic measures
• Sum
• Maximum value
• Minimum value
• Frequency
• Ratio

• Center value
• Mean
• Median
• Mod
• Robust mean

11
Some Informative Statistics

• Dispersion of observed values
• Range
• Variance
• Standard deviation
• Skewness
• Kurtosis

• Relationship between two variables
• Covariance
• Correlation

12
Graphical Display

• To visualise distribution of data.
• histogram bar chart box and whiskers.
• Movement of data.
• Line plot time series plots.
• To check relationship between two variables.
• Line plot scatter plot.
• To investigate distribution.
• Normal probability (P-P) plot Quantiles (Q-Q)
  plot.