STATISTICAL DATA ANALYSIS USING SOFTWARE
1. User-Friendly & Menu-Driven Packages
These tools are characterized by "point-and-click" interfaces, making them ideal for
beginners, students, and social scientists who require robust results without extensive
programming.
SPSS (Statistical Product and Service Solutions)
• Best For: Social sciences (Sociology, Psychology, Demography).
• Strengths: Extremely easy to navigate; excellent for descriptive statistics, ANOVA,
and factor analysis.
• Weaknesses: Not suitable for time-series analysis or multilevel regression.
• Key Note: The oldest established program (1960s), designed for researchers to enter
data and formulate results quickly.
MINITAB
• Best For: Education and basic business research.
• Strengths: Highly compatible across platforms (Linux, Mac, PC); excellent for
teaching research methods due to clear dialog boxes.
• Weaknesses: Weak in General Linear Models (GLM) and multilevel regression.
2. Power-User & Programmable Systems
These packages offer more flexibility and handle larger datasets but require a higher technical
learning curve.
SAS (Statistical Analysis System)
• Best For: Pharmaceutical research, banking, and "enormous" datasets.
• Strengths: Can handle up to 32,768 variables; superior in mixed model and
multivariate analysis.
• Weaknesses: Difficult to learn; errors in programming are hard to debug; weak in
ordinal/multinomial logistic regression.
STATA
• Best For: Econometrics and publication-quality graphing.
• Strengths: Smart data management; handles time-series analysis well.
• Versions: Ranges from Stata/IC (standard) to Stata/MP (for multiprocessor speeds).
R & MATLAB
• Best For: Advanced applied statistics and custom model implementation.
• R: Free (open-source) with the richest library of statistical models.
• MATLAB: Superior graphics and visualization for papers.
• Commonality: Both are essentially programming languages, allowing users "creative
freedom" to implement any specific model.
STATISTICAL DATA ANALYSIS USING SOFTWARE
III. FEATURES OF STATISTICAL SOFTWARE
Statistical software has some common characteristics that make it reliable and suitable for data
analysis:
1. Data editor is in rows and columns which make it very easy to enter numeric data.
[Link] is availability of menu bar comprises drop-down menu, quick analysis as well as brief
user manual.
3. Statistical level of measurement is put into consideration in data entry
4. They follow the initial steps in research project
(a) Getting your data ready to enter into the software.
(b) Defining and labeling variable
(c) Entering data appropriately with each row containing each case and each column as
variable.
(d) Data checking and cleaning is possible.
• All data should be numeric, although it may not be all variables it is not desirable to use
letter or word (String variable) as data. This can be achieved by recoding the letter or word
(string data) into desirable numeric and labeled appropriately.
• Data exploration can be done to check for errors and other accuracy.
• The statistical level of significance for rejecting null hypothesis (Ho) is when your p-value
significance is less than 0.05.
The Fundamentals of Data Coding in SPSS
Coding is the process of assigning numerical values to represent specific responses. Because
SPSS is designed to perform mathematical calculations, it prefers numbers over text ("string"
data). For instance, instead of typing "Agree" or "Disagree" a hundred times, you assign them
values like 1 and 2.
Phase 1: Setting the Infrastructure (Variable View)
The Variable View is your blueprint. Every row here represents a single question or data
point. To code effectively, you must focus on three specific columns:
1. Name & Label: The "Name" is a short code (e.g., Job_Sat), while the "Label" is the
full description (e.g., "On a scale of 1-5, how satisfied are you with your current
role?").
2. Values: This is the most critical step for categorical data. Click the "..." button to open
the Value Labels dialog. Here, you define your code: 1 = Very Dissatisfied, 2 =
STATISTICAL DATA ANALYSIS USING SOFTWARE
Dissatisfied, and so on. This allows you to enter "1" during data entry while SPSS
remembers what that "1" actually means.
3. Measure: You must identify if your data is Nominal (labels like "Department"),
Ordinal (ranked scales like "Junior, Mid, Senior"), or Scale (continuous numbers like
"Annual Salary").
Phase 2: Data Entry and Cleaning (Data View)
Once your variables are defined, switch to Data View. Each row represents a single
participant (case). If you have set up your Value Labels, you can click the Value Labels icon
on the toolbar to toggle between the numeric code and the text description. This is a great
way to spot-check for entry errors—if a cell shows a "6" when your scale only goes to "5,"
you know there’s a typo.
Phase 3: Advanced Coding Techniques
Often, the raw data isn't ready for analysis immediately. You may need to transform it using
the Transform menu:
• Recoding into Different Variables: This is a "safety-first" technique. If you want to
collapse a 10-point scale into a 3-point scale (Low, Medium, High), use this function.
It creates a new column, preserving your original raw data in case you need to revert.
• Reverse Coding: In many surveys, some questions are phrased negatively to ensure
participants are paying attention. If a "5" on Question A means "Very Happy" but a
"5" on Question B means "Very Sad," you must reverse code Question B so that
higher numbers consistently represent the same sentiment across your dataset.
• Computing Variables: If you have five different questions measuring "Public Space
Appropriation," you can use the Compute Variable tool to calculate a mean score
across those five items, creating a single "Total Index" for your analysis.
Phase 4: Handling the "Missing"
Participants often skip questions. In the Missing column of Variable View, assign a "Discrete
Missing Value" (like 999). This tells SPSS to exclude that specific cell from calculations like
means or regressions, ensuring your results aren't skewed by arbitrary zeros or blanks.