Research Instruments W4

and Statistical Analysis

for Quantitative
Studies (2/2)

NCM 111 / RES 1 / CRES 1 / BAC 105

 Percentile Ranks

“…percentile ranks “…In essence,

spread people percentile ranks are
evenly over the ordinal data and
number of possible must be treated as
ranks one might such…”
get…”
2
 Two Assumptions of Parametric
Statistics
■ The data reflect an interval or ratio scale.
■ The data fall in a normal distribution (e.g.,
the distribution has a central high point, and
it is not seriously skewed, leptokurtic, or
platykurtic). 3
 amount of
variability

Descriptive points/measures
of central
statistics tendency

extent of association
between/among
variables
4
 A point of central tendency is a point around
which the data revolve, a middle number around which
the data regarding a particular variable seem to hover.

In statistical language, we use the term

measures of central tendency to refer to techniques
for finding such a point. Three commonly used
measures of central tendency are the mode, the
median, and the mean, each of which has its own
characteristics and applications.

5
“The mode is the single
number or score that
occurs most frequently.”

6
 What is the mode in the
following data set?

3 4 6 7 7 9 9 9 9 10 11 11 13 13 13 15 15 21 26

Mode = 9

7
 The median is the numerical center of a set of
data; to facilitate our discussion, we will call each
piece of data in the set a “score.”

The median is the number in the very middle of

the scores, with exactly as many scores above it as
below it.

8
 What is the median in the
following data set?

69 70 72 73 75 76 77 78 79 81 83 83 84 85 86
87 89 90 91 92

Median = 82

9
“…the mean [is] the fulcrum point
for a set of data: it represents the
single point at which the two sides
of a distribution ‘balance.’”

“Mathematically, the mean is the

arithmetic average of the scores
within the data set.”
10
Formula for
Mean

11
“…the geometric mean… is computed
by multiplying all of the scores together
and then finding the Nth root of the
product.”

12
 Arithmetic Mean vs.
Geometric Mean
Attribute Arithmetic Mean Geometric Mean
The sum of all values
The nth root of the
Definition divided by the number of
product of n values.
values.
Applicable to any set of Applicable to positive
Applicability
numerical data. values only.
Can be calculated with Can also be calculated
Weighted Mean
weighted values. with weighted values.
Outliers have a significant Outliers have a lesser
Effect of Outliers
impact on the mean. impact on the mean.
Represents the average Represents the central
Interpretation
value of a set of data. tendency of a set of data.
Sum of values divided by Nth root of the product of
Calculation
the count of values. values.
13
“The configuration of the data
dictates the measure of central
tendency most appropriate for
that particular situation.”

14
“The nature of the data—the facts
of life—governs the statistical
technique, not the other way
around.”

(Leedy & Ormrod, 2016, p. 226)

15
“… we can often make reasonable predictions about
a population based on our knowledge of central
tendency…

In the broad spectrum of possibilities, we are betting

on the average as being the best single guess about
the nature of the total population.”

(Leedy & Ormrod, 2016, p. 226)

16
Using Measures of Central Tendency for Different
Types of Data

17
“… data are more similar if they cluster about the mean.
Scatter them, and they lose some of their uniformity;
they become more diverse, more heterogeneous. As
specific data points recede farther from the mean, they
lose more and more of the quality that makes them
‘average.’”

18
The simplest measure of variability is the range,
which indicates the spread of the data from lowest
to highest value:

Range = Highest score - Lowest score

19
The interquartile range is equal to
Quartile 3 (the 75th percentile point) minus
Quartile 1 (the 25th percentile point)

“… the interquartile range gives us the range

for the middle 50% of the cases in the
distribution…”
20
“When working with either ordinal data or
highly skewed data, some researchers report
what is called a five-number summary, which
consists of the lowest and highest numbers
(depicting the range) plus Quartile 1, the
median (which is also Quartile 2), and
Quartile 3.”

21
 Average Deviation

“… determine[s] how far away from

the mean each score is in the
distribution.”
22
 Standard Deviation

“… rather than taking the absolute value of the score–

mean differences, we square the differences. After we
have added the squared differences together and then
divided by the number of scores, we find the square
root of the quotient.”
23
 Variance

“… simply the standard deviation

squared.”
24
Using Measures of Variability for
Different Types of Data

25
“More useful in statistical analyses are
standard scores. Simply put, a standard score
tells us how far an individual’s performance is
from the mean with respect to standard
deviation units.”

26
 z-score

The simplest standard score is a z-score, which is

calculated by using an individual’s raw score
(which we will symbolize as X), along with the
mean and standard deviation for the entire group…

27
 Example: Mary’s
Extroversion Test Result

28
Converting a Z-Score to Another
Scale

29
 Mary’s Score as an IQ Score

Mary’s Score IQ Score in Stanine

Scale

30
 “… statistical manipulation of
the data is not, in and of itself,
research.”

31
 Limitation of Measures of Central
Tendency

“… merely described the center and

spread of the data.”

32
“The statistical process by which we
discover whether two or more variables
are in some way associated with one
another is called correlation.”

33
 correlation
coefficient for
two variables

Directio
Strength
n

34
 “The most widely used statistic for determining
correlation is the Pearson product moment
correlation, sometimes called the Pearson r. But
there are numerous other correlation statistics as
well. As is true for measures of both central tendency
and variability, the nature of the data determines the
technique that is most appropriate for calculating
correlation.”

35
36
37
 “Not all relationships take a linear
form… Always keep in mind that the
nature of the data governs the
correlational procedure that is
appropriate for those data. ”

38
“Correlation does not
necessarily indicate
causation…”

39
“… inferential statistics allow us to draw
inferences about large populations from
relatively small samples...”

1. To estimate a 2. To test statistically

population parameter based hypotheses
from a random
sample
40
 “… we use a sample to learn about
the larger population from which the
sample has been drawn…”

Inferential statistics can tell us how

closely these sample statistics
approximate parameters of the overall
population.
41
“… Statistical estimates of
population parameters are based
on the assumption that the sample
is randomly chosen and
representative of the total
population.”

42
“… Different samples—even when each one has been randomly
selected from the same population—will almost certainly yield
slightly different estimates of the overall population. The
difference between the population mean and a sample mean
constitutes an error in our estimation.”

43
“… the standard deviation
for the distribution of
sample means—is called
the standard error of the
mean…”
44
Statistically, when all of the samples are of a
particular size (n), the standard error of the
mean for any variable being measured is
related to the standard deviation for the
variable itself in the following formula:

45
46
 Point Versus Interval
Estimates

A point estimate is a single In particular, we specify a range within

statistic that is used as a which a population parameter probably lies,
and we state the probability that it actually
reasonable estimate of the
lies there. Such an interval is often called a
corresponding population confidence interval because it attaches a
parameter; for instance, we certain level of probability to the estimate—
might use a sample mean as a certain level of confidence that the
a close approximation to the estimated range includes the population
population mean. parameter.

47
“… A research hypothesis is a reasonable conjecture,
an educated guess, a theoretically or empirically based
prediction. Its purpose is a practical one: It provides a
logical framework that guides a researcher while
designing a research study and collecting data.”

“…Statistical hypothesis, usually [refers to] a null

hypothesis. A null hypothesis (often symbolized as
H0) postulates that any result observed is the
result of chance alone.”

48
This process of comparing observed data
with the results we would expect from
chance alone is called testing the null
hypothesis.

49
 statistically
significant
result
1-in-20 1-in-100
probabilit criterion
y

50
“… if we incorrectly reject the null
hypothesis—we are making a Type I
error (also called an alpha error)…”

we might conclude that a result is due

to chance when in fact it is not. In such
a circumstance, we have failed to reject
a null hypothesis that is actually false—
something known as a Type II error
(also called a beta error).
51
 Increasing the Power of a
Statistical Test

1.Use as large a sample size as is

reasonably possible.
2.Maximize the validity and reliability of your
measures.
3.Use parametric rather than nonparametric
statistics whenever possible.

52
53
54
References:

Leedy P. D. & Ormrod, J. E. (2016). Practical

research planning and design. Pearson
Publishing.

Arithmetic mean vs. geometric mean. (n.d.) This

vs. That. https://thisvsthat.io/arithmetic-mean-
vs-geometric-mean (Accessed September 30,
2024)
55