Choosing the Right Statistics

Most research data can be classified in one of the three basic categories.
Category 1: A single group of participants with one score per participant
Category 2: A single group of participants with two (or more) variables measured for each
participant
Category 3: Two (or more) groups of scores with each score a measurement of the same
variable

CATEGORY 1
This type of data often exists in research studies that are conducted simply to describe
individual variables as they exist naturally.
It is also possible that the data are a portion of the results from a larger study examining
several variables.
Although several variables are being measured, the intent is to look at all of them one at a
time.
The identifying feature of this type of research is is that there is no attempt to examine
relationships between different variables.

Statistics for Category 1 Data

Scales of Measurement
Numerical scores from Ratio
or Interval Scales
Ordinal Scores (Ranks and
Ordered Categories)
Nominal Scores (Named
Categories)
Descriptive Statistics
Mean
Standard Deviation
Proportions or Percentages
(to describe distribution
across categories)
Median
Proportions or Percentages
(to describe distribution
across categories)
Mode
Proportions or Percentages
(to describe distribution
across categories)
Inferential Statistics
Single sample t-test
(Use the sample mean to test
a hypothesis about the
population mean)
Chi-square test for goodness
of fit
(Use the sample data to test a
hypothesis about the
proportions in the population)
Chi-square test for goodness
of fit
(Use the sample data to test a
hypothesis about the
proportions in the population)

CATEGORY 2
The research studies is specifically intended to examine relationships between variables.
Different variables are being measured, so there are two or more scores each representing
a different variable.
There is no attempt to control or manipulate the variables; they are simply observed and
recorded as they exist naturally.
The researcher usually select pairs of variables to evaluate specific relationships.
Statistics for Category 2 Data
Number of Nature of Variables
Variables
Both variables measured
on interval or ratio scales
(numerical scores)
Both Variables measured
on ordinal scales (ranks or
ordered categories)
Numerical scores for one
Two variable and two values for
the second ( a
dichotomous variable)
Two values for both
variables (two
dichotomous variables)
Any measurement scales
but a small number of
categories for each
variable
All variables measures on
interval or ratio scales
(numerical scores)
Three
Numerical values and
dichotomous variables
Descriptive Statistics Inferential Statistics
Pearson correlation t-test
coefficient (describes the F-ratio
degree and direction of
linear relationship)
Simple linear regression Analysis of Regression
Spearman correlation t-test
(describes the degree and F-ratio
direction of monotonic
relationship)
Point Biserial correlation t-test (data can be grouped to
(describes the strength of be suitable for independent-
the relationship) measures)
Phi-coefficient (describes 2 x 2 chi-square test for
the strength of relationship) independence
Regroup the data as Chi-square test for
frequency distribution independence
matrix. The frequencies or
proportions describes the
data
Partial Correlation t-test
(describes the direction and F-ratio
degree of liner relationship
between two variables
while controlling the third)
Multiple regression Analysis of Regression
Partial correlation t-test
F-ratio
Multiple regression Analysis of Regression

CATEGORY 3
A second method for examining relationships between variables is to use the categories
of one variable to define different groups and then measure a second variable to obtain
set of scores within each group.
The first variable, defining the groups, usually falls into one of the following general
categories:
1. Participant characteristics (e. g. gender, age)
2. Time (e. .g. before and after)
3. Treatment conditions (e. g. with X vs. without X)
If the scores in one group are consistently different from the scores in another group,
then the data indicate a relationship between variables.
Another factor that differentiates data sets in this category is the distinction between
independent-measures and repeated-measures designs.
Independent-measures, also known as between-subjects design, requires a separate
group of participants for each group of scores. (e.g. comparing scores of male and
female)
 Repeated-measures design, also known as within-subjects design, obtains several
groups of scores from the same group of participants. (e.g. before/after study)

Statistics for Category 3 Data

Number of Descriptive
Scale Design Inferential Statistics
Groups Statistics
Independent
Independent-measures t-test
-measures
Two
Repeated- Repeated-measures t-test (paired
Interval/ measures Mean t-test)
ratio scales Independent Standard Deviation
Independent-measures ANOVA
-measures
Two or more
Repeated-
Repeated-measures ANOVA
measures
Independent
Mann-Whitney U test
-measures
Two
Repeated-
Wilcoxon signed rank test
Ordinal measures
Median
Data Independent
Kruskal-Wallis test
More than -measures
two Repeated-
Friedman test
measures
Ordinal/no
minal with Independent Proportion in each
Two or more Chi-square test
few -measures categories
categories

References:
Gravetter, F. J. and Wallnau, L. B. (2015). Statistics for the behavioral sciences Philippine
edition. Singapore: Cengage Learning Asia Pte Ltd

McDonald J. H. (2014). Handbook of biological statistics 3rd ed. Maryland: Sparky House
Publishing URL: http://www.biostathandbook.com/testchoice.html

StatSoft, Inc. (2013). Electronic statistics textbook. Tulsa, OK: StatSoft. URL:
http://www.statsoft.com/textbook/