SESSION 13

QUANTITATIVE DATA ANALYSIS

Dr ZHENG Weijia
Offi ce ： IB941
Email ： [email protected]
Session Overview
• Quantitative data preparation
• Assessing the normality of a
distribution of data
• Basic descriptive statistics for analysing
quantitative data
• Choose & run the appropriate statistical
test to test a given hypothesis
• Interpret the findings of the tests
Planning data analysis
should start …
a) Once data is collected.
b) When designing your data
collection tool, e.g. questionnaire
c) Once you entered your data on
a data file.
Collect the right data
first!
“What does it mean if a finding is
significant, or that the ultimate in
statistical analytical techniques have
been applied, if the data collection
instrument generated invalid data at
the outset?” (Jacoby, 1978, p. 90)
Data Analysis
• A set of methods and techniques
used to obtain information and
insights from data
• Helps avoid erroneous judgements
and conclusions
Quantitative data preparation
• So, you collected your data, what’s
next?
- Questionnaire checking
- Coding and prepare a codebook
- Prepare the data file
- Enter your data
- Screen and clean the data
Questionnaire checking
• A questionnaire returned from the
field may be unacceptable for several
reasons.
- Parts of the questionnaire may be
incomplete.
- The pattern of responses may
indicate that the respondent did not
understand or follow the instructions.
- The responses show little variance.
- One or more pages are missing.
 Treatment of Unsatisfactory
Responses
• Returning to the Field
• Assigning Missing Values:
- Assigning missing values to unsatisfactory
responses, e.g., substitute a neutral value.
• Discarding unsatisfactory respondents
(casewise & pairwise deletion)
Missing responses should be kept at a minimum, and
the implications of each treatment procedure should
be seriously considered before selecting any option.
Treatment of Unsatisfactory
Responses
• Discarding Unsatisfactory Respondents
- Casewise deletion:
 Cases or respondents with any missing
responses are discarded from the analysis.
 Throwing away large amount of data,
however, is costly and may seriously affect
the results at times.
Treatment of Unsatisfactory
Responses
• Discarding Unsatisfactory Respondents
- Pairwise deletion:
 All cases, or respondents with any missing
values are not automatically discarded.
 Rather, for each calculation only the cases
or respondents with complete responses
are considered.
 Appropriate when the sample size is large,
there are few missing responses, and the
variables are not closely related.
Coding and preparing a
codebook
• Coding involves grouping and
assigning value to various responses
from the survey instrument
• A codebook is a summary of the
instructions you will use to convert
the information obtained from each
subject or case into a format that
SPSS can understand.
Coding and preparing a
codebook
• Preparing the codebook involves
deciding how you will go about:
- defining and labelling each of the
variables
- assigning numbers to each of the
possible responses.
A CODEBOOK EXCERPT
Column Variable Variable Question Coding
Number Number Name Number Instructions
1 1 ID 1 to 20 as coded
2 2 Preference 1 input the number circled.
1=Weak Preference
7=Strong Preference

3 3 Quality 2 Input the number circled.

1=Poor
7=Excellent

4 4 Variety 3 Input the number circled.

1=Poor
7=Excellent

5 5 Value 4 Input the number circled.

1=Poor
7=Excellent

6 6 Service 5 Input the number circled.

1=Poor
7=Excellent
Preparing the data file
• This involves a number of steps:
– Creating the data file.
– Entering the information obtained from
your questionnaire in a format defined
by your codebook.
– Checking the data file for errors.
– Correcting any errors.
Frequency distribution
• Reports the number of responses that
each question received.
• Organizes data into classes or groups of
values.
• Shows number of cases that fall into
each class.
• Can be illustrated simply as a number ，
or as a percentage ， or histogram.
• Can be used to check the data for errors.
FREQUENCY HISTOGRAM
Frequency distribution
• A frequency distribution for a variable
produces a table of frequency counts,
percentages, and cumulative
percentages for all the values
associated with that variable.
• Indicates the shape of the distribution
of the variable, and the extent of
missing, illegitimate and extreme
responses.
Frequency distribution
Frequency distribution
Use graphs
E.g., A pie chart shows a proportion:
What is your usual visit occasion to
Starbucks?
1. Breakfast
2. Lunch
3. Coffee breaks
4. Other
Describing distributions
• An important aspect of the
"description" of a variable is the
shape of its distribution.
• Typically, a researcher is interested in
how well the distribution can be
approximated by the normal
distribution.
Describing distributions
• There are several ways to check the
normality of a distribution:
- Check histogram with normal curve
- Check the skewness and kurtosis
statistics
Describing distributions
• Skewness measures the deviation of
the distribution from symmetry. If the
skewness is clearly different from 0,
then that distribution is not normal.
• Kurtosis measures the “peakedness”
of a distribution. If the kurtosis is
clearly different than 0, then the
distribution is either flatter or more
peaked than normal.
Symmetric vs. Skewed
Distribution
High vs. Low Kurtosis
Descriptive Statistics
• Used to summarise and describe the
data
• Descriptive statistics are normally
generated with frequency distribution
to help summarise information in the
frequency table
Descriptive Statistics
• Includes:
- Measures of location/central tendency
(mean, median and mode)
- Measures of variability/dispersion
(range, variation, and standard
deviation)
- Measures of shape (skewness and
kurtosis)
Measures of location
• The Mean –the sum total of values
divided by the number of cases.
• The Mode – the value that occurs most
frequently. It represents the highest
peak of the distribution.
What is the mean of this distribution?
1, 4, 2, 4, 5, 2, 1, 1, 7
Answer: 3
Measures of location
• The Mean –the sum total of values
divided by the number of cases.
• The Mode – the value that occurs most
frequently. It represents the highest
peak of the distribution.
What is the mode of this distribution?
1, 4, 2, 4, 5, 2, 1, 1, 7
Answer: 1
Measures of location
• The Median
- the middle value of a ascending or
descending rank ordered distribution,
exactly half of the responses are
above and below the median value.
- If the number of data points is even,
the median is estimated as the
midpoint between the two middle
values – by adding the two middle
values and dividing their sum by 2.
Measures of location
What is the median of this distribution?
1, 4, 2, 4, 5, 2, 1, 1, 7

Reorder the distribution as an

ascending rank:
1, 1, 1, 2, 2, 4, 4, 5, 7
The median is the middle value of the
above rank, i.e., 2
Measures of location
• The most appropriate measures of
location are ：
- mode for nominal data,
- median for ordinal data, and
- mean for interval or ratio data.
Measures of Variability
• The range: the interval from the
lowest to the highest value in an array
of data.
E.g.,
where the highest value is 7 and the
lowest value is 1, the range is 7 - 1 = 6.
Measures of Variability
• Variance
- The average of the squared differences
from the Mean
- The extent to which values are
dispersed.
- Increases as values differ significantly
from the mean.
Measures of Variability
• The standard deviation:
- the square root of the variance.

- a “standard” way of knowing what is

normal (i.e., values within one Standard
Deviation of the Mean), and what is
extra large or extra small (i.e., values
outside one Standard Deviation of the
Mean).
SPSS Detailed Steps:
Frequencies
Select ANALYZE on the SPSS menu bar.
Click DESCRIPTIVE STATISTICS, and select
FREQUENCIES.
Move the variable you wish to measure to the VARIABLE(S)
box.
Click STATISTICS.
Select MEAN, MEDIAN, MODE, STD. DEVIATION,
VARIANCE, and RANGE.
Click CONTINUE.
Click CHARTS.
Click HISTOGRAMS, then click CONTINUE.
Click OK.
Hypothesis testing
• Null hypothesis: the hypothesis used in
testing
• The alternative hypothesis is the logical
opposite of the null hypothesis.
• Presumed to be true until there is an
evidence that it is not.
• If the evidence is inconsistent with the
null hypothesis (H0), then the null
hypothesis is rejected and the alternative
hypothesis (Ha) is accepted.
A Broad Classification of
Hypothesis Tests
Testing for differences between
groups
Hypotheses - differences
Null hypotheses examples:
“Both brands are equally popular with
customers in China”
“Males and females are equally likely to smoke”
“The control group and the test group of
customers are equally positive about the brand
image”
“Students’ attitudes to the brand were the same
before and after viewing the advert”
t-Tests
• Used when you have two groups (e.g.
males and females, British and Chinese)
• When you wish to compare the mean
score on some continuous variable (e.g.
the mean score of buying intention)
t-Tests
There are three types:
– One sample, when you want to make a
statement about a single variable (e.g.
comparing actual sales against a set target).
– Paired samples, which are “related”, i.e., the
same people are tested on two occasions.
– Independent sample t-test is used when
there are two different groups.
Assumptions for t-tests
• Normality (the variable has a symmetric
distribution)
• Ratio or interval data
• The population variance is estimated
from the sample
Independent samples t-test
• Null (H0):
There is no difference between the two
groups (i.e. they are the same or equal).
• Alternative (Ha):
There is a difference between the two
groups.
Independent Samples t-test
on SPSS
• Analyse/
• Compare means/
• Independent sample t-test
– Choose the dependent “test” variable
– Choose the grouping variable
– Define the two groups
– Continue
– OK
Results
Interpreting Results
Levene’s Test : This tests the equality of
variances of the samples.
– If the Sig. value is larger than .05 (e.g. .07,
or .10) you should use the first line of the
table “equal variances assumed”.
– If the value is smaller than .05
(e.g. .01, .001) this means the variances for
the two groups are not the same – use the
second line “equal variances not assumed).
Interpreting Results
Sig. 2-tailed
– If the Sig. 2-tailed is equal to or less
than .05 (e.g. .03 or .001) then there is a
significant difference in the mean scores.
– If the Sig. 2-tailed value is above .05
(e.g. .06 or .10) then there is no significant
difference in the scores.
Interpreting Results
• In the example presented in the output
shown, the Sig. (2-tailed) value is .039.
• Therefore: there is a statistically
significant difference in the mean value of
restaurant excitement between males and
females.
Testing for relationship between
variables
 AN EXAMPLE:
RELATIONSHIPS
• Is the price related in any
way to the number sold?
• How can you measure it?
• How can you tell whether the
relationship is “significant”
rather than just observable?
- Use tests to check this
relationship
- Assume that there is not a
relationship – and when you
find you can’t support that,
you have to accept that there
is a relationship.
Hypothesis - Association
Null hypothesis H0:
“No relationship exists between price and
number of mobile phones sold.”
Non-directional hypothesis H1 :
“There is a relationship between the price
and number of mobile phones sold.”
Directional hypothesis:
“The number of mobile phones sold
increases with a decrease in the price.”
Correlation (relationships)
• Positive correlation: as a score on one
variable increases, the corresponding
score on the other variable does the
same.
E.g., As customer satisfaction increases –
customer loyalty increases.
• Negative correlation: when one
variable goes up, the other goes down.
E.g., As the price increases, sales decrease.
 Correlation
• Pearson correlation co-efficient (r): a
statistical measure of the strength of a
relationship between two metric variables
• The Pearson correlation coefficient formula
assumes:
– Variables measured using interval or ratio scales.
– The relationship between the two variables is
linear/straight-line.
• No “cause and effect” inferred
• Ranges between +1 and –1 (zero (0) = no
correlation)
Interpretation of the size of a
correlation
Pearson Correlation on SPSS
• Analyse/
• Correlate
• Bivariate
- Choose the variables you want to test
for association
- OK
Quiz time
In a frequency distribution, two variables
are considered at a time.

True or false?
Quiz time
The mean, mode, and median are
associated with the measures of location
of frequency distribution.

True or false?
Quiz time
A mathematical distribution whose
objective is to obtain a count of the
number of responses associated with
different values of one variable and to
express these counts in percentage
terms is called ________.
A) sampling distribution
B) random distribution
C) frequency distribution
D) systematic distribution
Quiz time
A frequency distribution for a variable
produces ________.
A) a table of frequency counts
B) percentages
C) cumulative percentages
D) all of the above
E) none of the above
Quiz time
Hypotheses related to differences in the
population means can be tested using
the t test.

True or false?
Quiz time
In the case of conducting the t-test on the
means of two samples of observations,
the samples can be ________ or
________.
A) independent; random
B) random; exclusive
C) exclusive; dependent
D) independent; paired
E) set-up; breakdown
Quiz time
A systematic relationship between two
variables in which a change in one
implies a corresponding change in the
other is called ________.
A) causation
B) analysis of variance
C) correlation
D) deviation
E) standard deviation
Quiz time
A method for handling missing
responses in which respondents with any
missing responses are discarded from
the analysis is called pairwise deletion.

True or false?
Quiz time
A book containing coding instructions
and the necessary information about
variables in the data set is called a(n)
________.
A) editing book
B) sampling record
C) codebook
D) questionnaire
E) checking book
Quiz time
Treatment of missing responses poses
problems, particularly if the proportion of
missing responses is ________.
A) between 1 percent and 3 percent
B) between 3 percent and 5 percent
C) between 5 percent and 7 percent
D) between 8 percent and 9 percent
E) more than 10 percent
Quiz time
Which of the following is NOT an
alternative for the treatment of
unsatisfactory responses?
A) return the questionnaire to the field to
get better data
B) assign missing values
C) discard unsatisfactory respondents
D) delete the question from the survey
E) All of the above are alternatives.
Quiz time
The assignment of a symbol to
represent a specific response to a
specific question, along with the data
record and column position that symbol
will occupy, is called ________.
A) editing
B) coding
C) sampling
D) pretesting
E) selecting
Quiz time
A statistic that indicates the
distribution's dispersion is called
________.
A) measures of location
B) measures of variability
C) measures of association
D) measures of relativity
E) measures of difference
Quiz time
A measure of central tendency given as
the value above which half of the values
fall and below which half of the values
fall is called the ________.
A) mean
B) mode
C) median
D) maxima
Quiz time
The mean squared deviation of all the
values from the mean is called the
________.
A) variance
B) range
C) median
D) mean
E) geometric mean