Chapter 8: Analyzing Quantitative Data —

Quantitative research deals with:

 Measurements
 Numerical values
 Statistical interpretation

Researchers use statistics to:

 Summarize data
 Discover patterns
 Make predictions
 Compare results
 Determine significance of findings

The chapter emphasizes that:

“Numbers are meaningless unless we can find the patterns beneath them.”

Statistics help researchers answer:

“What do the data mean?”

EXPLORING AND ORGANIZING A

DATA SET
Before doing any statistical calculations, researchers should:

 Observe data carefully

 Think critically
 Look for hidden patterns
 Organize data meaningfully

The chapter strongly stresses that:

Careful observation of data is extremely important.

“How researchers organize data affects the meaning
revealed by those data.”
Time-series study
Meaning:

 Data are collected repeatedly over time.

Such studies often reveal clear visual patterns.

USING COMPUTER SPREADSHEETS

Organizing large amounts of data manually is:

 Difficult
 Slow
 Time-consuming

Modern researchers use spreadsheet software like:

 Microsoft Excel
 LibreOffice
 Spread32

Advantages of Electronic Spreadsheets

Once data are entered:

 Calculations become automatic

 Changes update automatically
 Data can be reorganized easily

Major Spreadsheet Functions

1. Sorting
Data can be rearranged:

 Highest to lowest
  Alphabetically
 By age
 By category etc.

Example:
Math scores can be sorted from youngest to oldest students.

2. Recoding
Researchers can transform old data into new categories.

Example:

Age Group
7–9 Group 1
10–12 Group 2
13–15 Group 3

3. Formulas
Spreadsheets can calculate:

 Mean
 Totals
 Percentages
 Standard deviation

Many formulas are already built into the software.

4. Graphing
Software automatically creates:

 Line graphs
 Pie charts
 Bar graphs

Researchers can customize:

 Axes
 Labels
  Legends

5. “What If” Analysis

Researchers can test multiple possibilities quickly.

Examples:

 Compare males vs females

 Compare different age groups
 Remove one subgroup and reanalyze

This makes experimentation easier.

Researchers must choose statistics according to:

 Nature of data
 Research purpose
 Scale of measurement
 Distribution pattern

Looking at data in only one way gives an

incomplete understanding.
Therefore researchers use multiple statistical techniques.

FUNCTIONS OF STATISTICS
Statistics have TWO major functions:

1. Descriptive Statistics
These describe and summarize data.
They explain:

 Center of data
 Spread of data
 Relationships among variables

Examples:

 Mean
 Median
 Mode
 Standard deviation

2. Inferential Statistics
These help researchers:

 Draw conclusions about populations

 Use sample data to make predictions
 Test hypotheses

Researchers study small samples and generalize findings to larger populations.

Example of Inferential Statistics

An immigration officer meets only some Egyptians.

From this small sample, the officer forms ideas about Egyptians in general.

Similarly:
Researchers study samples to infer characteristics of entire populations.

Statistics as Estimates of Population

Parameters

Parameter
A parameter is:

A characteristic of an entire population.

Examples:

 Population mean
 Population standard deviation

Statistic
A statistic is:

A value calculated from a sample.

Researchers use sample statistics to estimate population parameters.

Statistical Symbols
Characteristic Population Symbol Sample Symbol
Mean μ M or X̄
Standard deviation σ s or SD
Proportion P p
Total number N n

Single-Group Data
A study produces single-group data when information is collected from only one group.

Example
 Test scores of one classroom
 Heights of students in one school
 Blood pressure readings of one group of patients
In such studies, researchers usually analyze:

 Average performance
 Variability
 Trends within the same group

Multi-Group Data
A study produces multi-group data when researchers compare two or more groups.

Example
 Boys vs girls
 Experimental group vs control group
 Private school students vs government school students

Researchers mainly examine:

 Differences between groups

 Similarities between groups
 Effects of treatments or conditions

Different statistical techniques are required for single-group and multi-group data.

Because:

 Single-group studies focus on describing one group.

 Multi-group studies focus on comparing groups.

Continuous vs Discrete Variables

A variable is any characteristic that can change or vary.

Example:

 Age
 Height
 Number of students
 Marks

Variables are mainly of two types:

1. Continuous Variable
A continuous variable can take infinite possible values within a range.

You can measure it very precisely.

 Infinite possible values

 Values can lie between two numbers
 Can contain decimals/fractions

Examples
Age
A person can be:

 18 years
 18.5 years
 18.75 years
 18 years 2 months etc.

So age is continuous.

Height
Examples:

 160 cm
 160.2 cm
 160.25 cm

Infinite possibilities exist between two heights.

Weight,temp

2. Discrete Variable
A discrete variable has only specific separate values.

No values exist between them.

  Countable values
 No fractions usually
 Separate distinct categories
 Usually counted, not measured

Examples
Number of Students
Possible values:

 20 students
 21 students
 22 students

Not possible:

 20.5 students

Grade Levels
 9th
 10th
 11th
 12th

No “9.5th class.”

Nominal, Ordinal, Interval, and Ratio

Scales
These are scales of measurement.

They determine:

 Type of data
 Statistical methods usable
 Mathematical operations possible
Nominal Scale
Numbers represent categories only.

Examples:

 Male = 1
 Female = 2

Here, 2 is not greater than 1. The numbers are only labels.

Other examples:

 Religion
 Political affiliation
 Blood group

Suitable statistic

 Mode

2. Ordinal Scale
Ordinal data show order or ranking.

Example:

 Class rank
 Competition positions

If a student has Rank 1 and another has Rank 2, we know the first student performed better.
However, we do not know how much better

Suitable statistics

 Median
 Percentiles

3. Interval Scale
Interval data have equal intervals between values.

Example:
  IQ scores
 Temperature in Celsius

Difference between:

 10°C and 20°C

is equal to
 20°C and 30°C

But zero is not a true zero.

0°C does not mean absence of temperature.

Features

 Order exists
 Equal intervals exist
 No true zero point

Suitable statistics

 Mean
 Standard deviation
 Correlation

4. Ratio Scale
Ratio data are similar to interval data but also have a true zero point.

Example:

 Height
 Weight
 Income

If income = ₹0, it means no income.

Ratios are meaningful:

 40 kg is twice 20 kg.

Features

 Order exists
 Equal intervals exist
 True zero exists
 Most advanced scale
Suitable statistics

All statistical techniques can be used.

Normal Distribution (Bell Curve)

Many natural and human characteristics follow this pattern.

Examples:

 Height
 IQ
 Crop production
 Test scores
 Biological traits

The graph looks bell-shaped and symmetrical.

Characteristics of Normal Distribution

1. Symmetrical Shape
 Left side mirrors right side.
 Both halves are equal.

This means:

 Data are evenly distributed around the center.

2. Highest Point at Center

Most values occur near the middle.

Very high or very low values occur less frequently.

Thus:

 Average values are most common.

3. Mean = Median = Mode
In a perfect normal curve:

 Mean
 Median
 Mode

all occur at the same central point.

Percentages in Normal Distribution

The chapter explains how population percentages are distributed around the mean using
standard deviations.

Standard deviation helps determine:

 How spread out data are

 How close scores are to the average

Smaller SD:
→ Scores tightly clustered

Larger SD:
→ Scores widely spread

Between Mean and 1 Standard Deviation

About:

34.1%
lies on EACH side.

So together:

68.2%
of data lies within 1 standard deviation of the mean.

Between 1 SD and 2 SD
About:

13.6%
lies on each side.

Beyond 2 SD
Only:

2.3%
lies at each extreme end.

This means:

 Extremely high or low values are rare.

NON-NORMAL DISTRIBUTIONS
Not all data follow a perfect bell curve.

Some distributions are distorted.

SKEWED DISTRIBUTIONS

Skewness means:

One side of the distribution stretches farther than the other.

The stretched side is called: Tail

Positive Skew
In positive skew:

 Tail extends toward the right

 Peak lies toward left

Usually caused by:

 A few extremely high scores

Example:

 Income distribution

Most people earn average salaries.

Few earn extremely high salaries.

Negative Skew
In negative skew:

 Tail extends toward left

 Peak lies toward right

Caused by:

 A few extremely low scores

KURTOSIS
Kurtosis refers to:

How peaked or flat a distribution is.

Leptokurtic Distribution
  Very peaked
 Tall center
 Data tightly clustered

Meaning:
Most values are close to mean.

Platykurtic Distribution
 Flat and broad
 Data widely spread

Meaning:
Greater variability exists.

PERCENTILE RANKS
Percentile ranks are often used in:

 Achievement tests
 Aptitude tests

Meaning of Percentile Rank

A percentile tells:

How many people scored below a person.

Example:
90th percentile means:

 Person performed better than 90% of people.

Formula for Percentile Rank

Percentile ranks are ordinal data.

PARAMETRIC VS NONPARAMETRIC
STATISTICS
Statistical methods are divided into two categories:

1. Parametric statistics
2. Nonparametric statistics

PARAMETRIC STATISTICS
Parametric statistics assume certain things about data.

Two major assumptions:

Assumption 1
Data should use:

 Interval scale OR
 Ratio scale

Assumption 2
Data should follow:

Normal distribution
If assumptions are violated:
  Results may become inaccurate.

Examples of Parametric Statistics

 t-test
 ANOVA
 Pearson correlation

NONPARAMETRIC STATISTICS
Nonparametric statistics:

 Do NOT require normal distribution

 Can handle ordinal data
 Can handle skewed distributions

Useful when assumptions fail.

Why Not Use Nonparametric All the Time?

Because:

Parametric statistics are more powerful and sophisticated.

DESCRIPTIVE STATISTICS
THREE MAJOR MEASURES
1. Measures of central tendency
2. Measures of variability
3. Correlation measures

MEASURES OF CENTRAL TENDENCY

Central tendency means:
The center point around which data cluster.
It represents the “typical” value.

Three Main Measures

1. Mode
2. Median
3. Mean

Each has different uses.

MODE
Definition:

Most frequently occurring value.

Example
Data:

3, 4, 6, 7, 7, 9, 9, 9, 9, 10, 11

Mode = 9

because:

 9 appears most often.

Features of Mode
Advantages
 Simple
  Useful for nominal data

Disadvantages
 May not represent true center
 Unstable across samples

MEDIAN
Definition:

Middle value in ordered data.

Finding Median
Arrange data:

 Lowest to highest

Then:

 Find middle score

If Number of Scores is Odd

Median =

 Exact middle score

If Number of Scores is Even

Median =

 Average of two middle scores

Uses of Median
Median is best for:

 Ordinal data
 Highly skewed data

Why Median Works Better in Skewed Data

Extreme values affect mean heavily.

Median ignores extremes.

Example:

3, 4, 5, 5, 6, 9, 15, 17, 125

Mean of Above Data

Mean becomes:
21

But most scores are actually near:

5–9

So mean becomes misleading.

Median of Same Data

Median = 6

This better represents center.

MEAN (Arithmetic Average)

Definition:
Arithmetic average of all scores.

Formula for Mean

Important Feature of Mean

Mean uses:

Every score in the dataset.

Limitation of Mean
Sensitive to:

Extreme scores (outliers)

GEOMETRIC MEAN
Used for:

Growth phenomena
Examples:

 Population growth
 Biological growth
 Compound interest
Formula for Geometric Mean

Growth Curves
Stages:

1. Slow beginning
2. Rapid acceleration
3. Leveling off

Researchers must choose statistics according to:

 Data characteristics
 Distribution shape
 Measurement scale

NOT according to personal preference.

MEASURES OF VARIABILITY
Variability means:

How spread out the data are.

It tells us:

 Whether scores are close together

OR
 Widely scattered

Two datasets may have the same mean but very different variability.
Importance of Variability
Measures of central tendency alone are insufficient.

Example:

Two classes may both have:

 Mean = 75

But:

Class A

Scores:
74, 75, 76

Class B

Scores:
40, 60, 75, 90, 110

Same mean, but very different spread.

Thus:

Researchers must measure

dispersion/spread also.

THE RANGE
The simplest variability measure is:

Range
Definition:
Difference between highest and lowest score.
Formula for Range

Example Showing Problem with Range

Family sizes:

1, 3, 3, 3, 4, 4, 5, 5, 6, 15

Range:

15−1=1415-1=1415−1=14

But most families actually have:

3–6 children.

The extreme value 15 distorts interpretation.

INTERQUARTILE RANGE (IQR)

To avoid extreme-score problems, researchers use:

Interquartile Range

Quartiles
Data are divided into:

Four equal parts

Quartile 1 (Q1)
25% of scores lie below it.
Quartile 2 (Q2)
Middle point.

Equivalent to:

Median
50% above and 50% below.

Quartile 3 (Q3)
75% of scores lie below it.

Formula for IQR

Meaning of IQR
Shows spread of:

Middle 50% of data

Average Deviation
Represents:
Average distance of scores from mean

STANDARD DEVIATION
The most important variability measure.

Definition:

Average spread of scores around the mean.

Variance
Variance =

Average of squared deviations

Standard Deviation
Standard deviation =

Square root of variance

This returns the value to original measurement units.

Small SD:
→ Scores clustered closely
Large SD:
→ Scores spread widely

Importance of Standard Deviation

Used in:

 Inferential statistics
 Hypothesis testing
 Normal distribution
 Correlation
 Regression

It is one of the most important statistics.

VISUAL INTERPRETATION OF
VARIABILITY

Low Variability
When scores cluster tightly around mean:

 Data are more similar

 More homogeneous

High Variability
When scores spread widely:

 Data become more diverse

 Less representative of average

Z SCORES (STANDARD SCORES)

A z score tells:

How far a score lies from the mean in SD

units.

Formula for z Score

Where:

 X = individual score
 M = mean
 s = standard deviation

Positive z Score
Means:

 Score is above mean

Example:
z = +2
→ score is 2 SD above mean.

Negative z Score
Means:

 Score below mean

Example:
z = -1
→ score is 1 SD below mean.

z=0
Means:

 Score exactly equals mean.

IQ Score
The book gives IQ scores as an example of:

Interval Scale data

IQ (Intelligence Quotient) scores are assumed to have:

 Equal intervals between values.

Example:

IQ Scores

85

95

105

115

The difference between each pair is 10, so the intervals are equal.

This means:

 Difference between 85 and 95

is considered equal to
 Difference between 105 and 115

Important Point About IQ Scores

The book explains that:

Zero is not a true zero in IQ scores.

Stanine
Stanine means:

“Standard Nine”
It is a method of reporting test scores by dividing performance into 9 categories.

Stanines help simplify interpretation of scores.

Features of Stanine
 Scores range from 1 to 9
 Middle stanine = 5
 Based on normal distribution
 Used in educational and psychological testing

CORRELATION
Correlation studies relationships between variables.

“When one variable changes, does another

change too?”

Examples
 Height and weight
 Study time and marks
 Temperature and ice cream sales
Positive Correlation
As one variable increases:

 Other also increases.

Example:

 Study time ↑
 Marks ↑

Negative Correlation
As one variable increases:

 Other decreases.

Example:

 Stress ↑
 Sleep ↓

No Correlation
Variables unrelated.

Change in one does not affect other.

Correlation Coefficient
Represented by:

r
Range:
Interpretation of r
Value Meaning

+1 Perfect positive correlation

0 No correlation

-1 Perfect negative correlation

Important Limitation of Correlation

The book gives an extremely important warning:

Correlation statistics are based on the

assumption that the relationship between
variables is linear.

Linear Relationship
A linear relationship means:

Relationship forms a straight line.

As one variable changes:

 Another changes at constant rate.

Problem with Nonlinear Relationships

Some relationships are:

Curved instead of straight.

In such cases:

 Correlation may fail to detect the relationship properly.

Example from Book: BMI and Anxiety

The chapter gives an example using:

 Body Mass Index (BMI)

 Anxiety level

Researchers found:

 Anxiety was highest in people who were:

o Very underweight
o Very overweight
 Anxiety was lowest in people with:
o Average BMI

This forms:

U-shaped relationship
not a straight-line relationship.

Why Correlation Fails Here

A regular correlation coefficient assumes:

Linear relationship
But BMI and anxiety show:

Nonlinear relationship
Therefore ordinary correlation may not properly represent the association.

Scatter Plots
Correlations are often shown using:

Scatter diagrams
Each point represents:

 One participant’s scores on two variables.

Strong Positive Correlation

Points form upward pattern.

Strong Negative Correlation

Points form downward pattern.

Weak Correlation
Points scattered randomly.

IMPORTANT WARNING
The chapter emphasizes:

Correlation does NOT prove causation.

Just because two variables relate does not mean one causes the other.

Example
Ice cream sales and drowning deaths may both increase in summer.

But:
  Ice cream does not cause drowning.

Summer temperature affects both.

Coefficient of Determination
The chapter introduces:

Coefficient of determination
Represented by:

r²
how much one variable can be explained or
predicted by another variable.

Example

Meaning:
64% variability is shared.

 Larger r2→ stronger relationship

 Smaller r2→ weaker relationship

INFERENTIAL STATISTICS
Inferential statistics allow researchers to:

 Draw conclusions about large populations

 Use only small samples for analysis
 Estimate population characteristics
 Test hypotheses statistically

Two Main Functions

1. Estimating population parameters

2. Testing statistically based hypotheses

ESTIMATING POPULATION
PARAMETERS
Researchers usually study only a sample, not the whole population.

Example: Connecting-Rod Pins

A company manufactures metal pins for machinery.

Problem:

 Some pins may be faulty.

 Researcher Jan selects a random sample.

She wants to estimate:

1. Average pin diameter

2. Variation in diameter
3. Proportion of acceptable pins

This is estimation of population parameters using sample data.

Important Assumption
Inferential statistics assume that:

The sample is randomly selected and representative of the population.

SAMPLING DISTRIBUTION OF THE

MEAN
When many random samples are taken:

 each sample has its own mean

 those sample means form a distribution

This is called the sampling distribution of the mean.

Key idea:

Even if individual scores vary a lot,

sample means vary less.

STANDARD ERROR OF THE MEAN

The standard deviation for the distribution of sample means is called

Standard Error of the Mean (σM)

Formula:

Where:

 σ = population standard deviation

 N = sample size

Meaning of Standard Error

 Small standard error → sample mean close to population mean
  Large standard error → less accurate estimation

Increasing sample size decreases standard error.

Estimated Standard Error

Usually we do not know the population standard deviation

So we estimate it using the sample standard deviation

The formula becomes:

Where:

 S = sample standard deviation

 N = sample size

POINT ESTIMATES vs INTERVAL

ESTIMATES
A. Point Estimate
Single value used to estimate population parameter

example:

Sample average diameter = 0.712

So Jan says:

“I think the population mean is 0.712.”

This ONE number is called:

Point Estimate
This becomes estimate of population mean.

Problem:

 Exact estimate is unlikely to be perfectly accurate.

B. Interval Estimate
Gives a range within which population parameter probably lies.

Example:

“The true population mean is probably between 0.708 and 0.716.”

Confidence Interval
This interval usually has a confidence level.

Example:

95% Confidence Interval

Means:

Researcher is 95% confident that the true population mean lies inside this range.

Example:

0.708 to 0.716

Understanding Confidence Intervals

If sample means form normal distribution:

About:

 68% lie within 1 standard error

 95% lie within 2 standard errors
Thus researcher can estimate likely population range.

HYPOTHESIS TESTING
Second major function of inferential statistics.

TWO TYPES OF HYPOTHESES

1. Research Hypothesis
A prediction or educated guess made by researcher.

Example:

 “Teaching method A improves scores.”

Purpose:

 Guides research

2. Statistical Hypothesis
Used during statistical testing.

Usually a:

Null Hypothesis (H₀) there is no effect, no difference, or no

relationship between variables.

Null hypothesis states:

Any observed difference occurred only due to chance.

Example:

 Two group means differ only because of random variation.

TESTING THE NULL HYPOTHESIS
Researchers compare:

 actual observed results

with
 results expected by chance

If probability of occurring by chance is very small,

the null hypothesis is rejected.

Statistical Significance
Researchers calculate probability that results happened by chance.

If probability is very small:

 Result is considered statistically significant.

Significance Level (Alpha, α)

Common alpha levels:

Alpha Level Meaning

.05 5% chance result occurred randomly

.01 1% chance

.001 0.1% chance

If:
p < .05

→ Reject null hypothesis.

Example
If probability of result due to chance is:
 1 in 1000

then:

 likely NOT due to chance

 some real factor caused the difference

TYPE I AND TYPE II ERRORS

Type I Error (Alpha Error)
we believe

there is realRejecting null hypothesis when it is actually true.

effect or diff

Meaning:

 Concluding effect exists when it really does not.

Example:
Researcher says medication works when it actually doesn’t.

False postive

Probability of Type I error = α

Type II Error (Beta Error)

Failing to reject null hypothesis when it is false.

Meaning:

 Concluding no effect exists when it actually does.

False negative

Example:
Medication actually works, but researcher concludes it doesn’t.

Trade-off Between Errors

Reducing one error often increases the other.
Lower α:

 fewer Type I errors

 more Type II errors

Higher α:

 more Type I errors

 fewer Type II errors

POWER OF A STATISTICAL TEST

Power = ability to correctly reject false null hypothesis

If something REAL is happening,

a powerful test can detect it..

Ways to increase power:

1. Use Larger Sample Size

Larger samples:

 Reduce sampling error

 Produce more accurate estimates

2. Improve Validity and Reliability

Better measurements:

 Reduce error
 Increase likelihood of detecting true effects

3. Use Parametric Statistics When Possible

Parametric tests are generally more powerful.

Nonparametric tests:

 Require fewer assumptions

 But are less powerful
.

RESEARCH HYPOTHESIS vs NULL

HYPOTHESIS
Often they are opposites.

Example:

Research hypothesis:

Groups are different

Null hypothesis:

Groups are the same

Researchers indirectly support research hypothesis by rejecting null hypothesis.

IMPORTANT POINT
Statistics alone are not enough.

Researchers must:

 interpret findings
 explain meanings
 connect results to research problem

EXAMPLES OF INFERENTIAL
STATISTICAL PROCEDURES

PARAMETRIC STATISTICS
Used for interval/ratio data and normal distributions.
1. Student’s t-test
Purpose:

 Compare two means

Types:

 Independent samples t-test

 Dependent samples t-test

2. ANOVA (Analysis of Variance)

Purpose:

 Compare 3 or more means

Uses F-ratio.

If significant:

 follow-up post hoc tests needed

3. ANCOVA
Analysis of covariance.

Purpose:

 Compare means while controlling another variable (covariate)

More statistically powerful than ANOVA.

4. Regression
Predicts dependent variable using independent variable(s).

Types:
  Simple regression
 Multiple regression

Important:
Correlation ≠ causation.

5. Factor Analysis
Finds clusters of related variables called factors..

Used to identify underlying factors/themes.

7. Structural Equation Modeling (SEM)

Advanced statistical technique.

Examines:

 Complex relationships among variables

 Direct and indirect effects

Can include:

 Mediating variables
 Moderating variables

NONPARAMETRIC STATISTICS
Used when:

 assumptions for parametric tests are violated

 data are ordinal or nominal

1. Mann–Whitney U Test
Compares medians of two groups.

Nonparametric equivalent of independent t-test.

2. Kruskal–Wallis Test
Compares medians of 3+ groups.

Equivalent of ANOVA.

3. Wilcoxon Signed-Rank Test

Compares correlated ordinal variables.

Equivalent of dependent t-test.

4. Chi-Square Test (χ²)

Tests how closely observed frequencies match expected frequencies.

Applicable for:

 nominal
 ordinal
 interval
 ratio data

5. Odds Ratio
Measures relationship between two dichotomous variables.

Example:

 smoking vs heart disease

6. Fisher’s Exact Test

Used for:
  very small samples (n < 30)

Alternative to chi-square/t-test for small samples.

META-ANALYSIS
Meta-analysis = analysis of analyses.

Researchers combine results from many previous studies.

Useful when:

 many studies already exist on same topic.

STEPS IN META-ANALYSIS
1. Search for Relevant Studies
Systematic search using:

 journals
 databases
 keywords

2. Select Appropriate Studies

Include only studies matching:

 treatment
 variables
 population
 methods

3. Convert Results to Common Statistical

Index
Different studies may use:

 t-tests
 ANOVA
 regression

Researcher converts findings into common index.

Usually:

Effect Size (ES)

Effect Size
Effect size tells:

How strong or large the effect is.

Example
Medicine may:

 Reduce stress slightly

 Reduce stress greatly

Effect size measures this magnitude.

IMPORTANCE OF META-ANALYSIS
Helps researchers:

 summarize large bodies of research

 identify overall trends
 resolve conflicting findings

But:

 mathematically complex
 requires strong statistical knowledge
USING STATISTICAL SOFTWARE
PACKAGES
Examples:

 SPSS
 SAS
 Minitab
 SYSTAT
 Statistica

Advantages of Statistical Software

1. Wide Range of Statistics
Can perform:

 advanced analyses
 handle large datasets
 manage missing data

2. User-Friendly
Results displayed in tables and graphs.

3. Assumption Testing
Software checks:

 skewness
 kurtosis
 normality
4. Graphics
Creates:

 charts
 graphs
 tables

IMPORTANT WARNING
Software cannot think for the researcher.

Researchers must:

 choose correct statistical tests

 interpret results properly