QUANTITATIVE RESEARCH DESIGN Types of Non-Experimental Research Design:

In any research type, much more, in a 1. Descriptive – depicts an image or a picture of

quantitative research where you do a great deal an individual or a group
of abstraction and scientific or logical thinking, a 2. Comparative – states the differences or
research design is a part and parcel of your similarities between or among people, things,
study. Research design is defined as the rational objects, etc.
and coherent overall strategy that the 3. Correlative – shows the extent and direction of
researcher uses to incorporate all the vital variable relationships, that is, whether a negative
components of the research study. Hence, in or positive relationship exists between or among
order to find meaning in the overall process of them.
doing your research study, a step-by-step 4. Ex Post Facto. If the objective of the study is to
process will be helpful to you. measure a cause from a pre-existing effect, then
EXPERIMENTAL RESEARCH DESIGN Ex Post Facto research design is more appropriate
- a quantitative research design that bases its to use. In this design, the researcher has no
research method on a scientific activity called control over the variables in the research study.
experiment, in which a test or examination of a Thus, one cannot conclude that the changes
thing under a manipulated or controlled measured happen during the actual conduct of
environment is done to determine the validity or the study.
truthfulness of such thing. This design involves
two groups of subjects: the experimental group
on which the condition, treatment, or
intervention is applied and the control group
that is not given any treatment or condition.
- Following this experimental design, you
conduct two kinds of tests: pre-test for both
groups and post-test for the experimental or
treatment group to see the difference between
them based on the effects of the treatment or
condition given to the experimental group.

Types of Experimental Research Design:

 True Experimental Design – What proves
this as a true experimental design is its
random selection of participants. It is a
bias-free selection that ensures objectivity
of results. This design is the best way to
examine causal relationships.
 Quasi-experimental Design – The term
quasi (pronounced as kwahz-eye) means
partly, partially, pseudo, or almost. The
non-adherence of this research design to
random selection of participants is the
reason it got the name, quasi-experimental
research, which means a research with the
capacity to yield findings that are seemingly
or more or less true. Prone to bias.

NON-EXPERIMENTAL RESEARCH DESIGN

- a design is a quantitative research design that is
capable of giving qualitative and quantitative
data, but more on qualitative data; hence, this is
often used in the field of social sciences. Unlike
the experimental design that allows
manipulation or control of some aspects of the
research, non-experimental research design
shuns controlling variables. Instead, it involves
variables the way they naturally exist on earth.
 SAMPLING METHOD Types of Non-Probability Sampling
SAMPLING - The process by which researchers  Convenience Sampling - involves choosing
select a representative subset or part of the total respondents at the convenience of the
population that could be studied for their topic researcher. It is also called accidental or
so that they will be able to draw conclusions incidental sampling.
about the entire populations.  Purposive Sampling - involves handpicking
Types of Sampling: subjects, usually to suit very specific
 Probability Sampling - is a type of sampling intentions. This is also called judgemental
in which all the members of an entire sampling.
population have a chance of being selected.  Quota Sampling - is somewhat similar to
This is also called as scientific sampling, and stratified sampling, in that the population is
is commonly used in quantitative research. divided into strata, and the researcher
 Non-probability Sampling - disregards deliberately sets specific proportions in the
random selection of subjects. The subjects sample, whether or not the resulting
are chosen based on their availability or the proportion is reflective of the total
purpose of the study, and in some cases, on population. This is commonly done to the
the sole discretion of the researcher. This is ensure the inclusion of a particular segment
not a scientific way of selecting of the population.
respondents, and is commonly used in  Snowball Sampling - does not give a specific
qualitative research. set of samples, similar to snow expanding
widely or rolling rapidly. This is true for a
Types of Probability Sampling study involving unspecified group of people.
 Simple Random Sampling - is a method of Dealing with varied groups of people such as
choosing samples in which all the members street children, drug dependents, gang
of the population are given an equal chance members, informal settlers, street vendors,
of being selected. It is an unbiased way of and the like is possible in this kind of non-
selection, as samples are drawn by chance. probability sampling.
 Systematic Random Sampling - is a method  Consecutive Sampling - Also known as total
of selecting every nth element of a enumerative sampling, consecutive sampling
population, e.g., every fifth, eighth, ninth, is the process of conducting research
or eleventh element until the desired including all the people who meet the
sample size is reached inclusion criteria and are conveniently
 Stratified Random Sampling - is a method available , as part of the sample .
where the population is divided into  Self-selection Sampling - (also called
different strata (groups), and then the volunteer sampling) relies on participants
sampling follows. Age, gender, and who voluntarily agree to be part of your
educational qualifications are some research. This is common for samples that
possible criteria used to divide a population need people who meet specific criteria, as is
into strata. often the case for medical or psychological
 Cluster Sampling - is used in large-scale research.
studies. Clusters are representative samples  Maximum variation sampling - The idea
of the population as a whole. After the behind maximum variation sampling is to
clusters are established, a simple random look at a subject from all possible angles in
sample of the clusters is drawn and the order to achieve greater understanding. Also
members of the chosen clusters are known as heterogeneous sampling.
sampled. Sampling procedures are difficult  Homogeneous sampling, unlike maximum
and time-consuming. variation sampling, aims to achieve a sample
 Multi-stage sampling is a more complex whose units share characteristics, such as a
form of cluster sampling, in which smaller group of people that are similar in terms of
groups are successively selected from larger age, culture, or job.
populations to form the sample population  Typical case sampling - A typical case sample
used in your study. is composed of people who can be regarded
as “typical” for a community or
phenomenon. A typical case sample allows
you to develop a profile of what would
generally be agreed as being “average” or
“normal.”
 Extreme (or deviant) case sampling uses
extreme cases of a particular phenomenon
(outliers). This can mean remarkable
failures, successes, or crises, as well as any
event, organization, or individual that
appears to be the “exception to the rule.”
 Critical case sampling is used where a
single case (or a small number of cases) can
be critical or decisive in explaining the
phenomenon of interest.
 Expert sampling involves selecting a sample
based on demonstrable experience,
knowledge, or expertise of participants.
This expertise may be a good way to
compensate for a lack of observational
evidence or to gather information during
the exploratory phase of your research.
 RESEARCH INSTRUMENTS 3. Observation -
Research Instruments - are basic tools involves systematically watching and
researchers used to gather data for specific recording behaviors, events, or conditions as
research problems. They can be designed to they occur in their natural setting. It’s a
measure various aspects of subjects’ behaviors, powerful method for gathering data without
attitudes, abilities, or conditions. interference.
4. Physiological Devices -
Three ways to construct instrument: refer to tools and instruments used to
Adopt an instrument - measure and monitor various physiological
 taking an existing validated tool develop by parameters of the human body. These
other researchers and using it for your own devices are often used in medical, health,
study. and behavioral research to collect data on
Modify an existing instrument - bodily functions and responses.
 involves tweaking it to better fit your study 5. Psychological Tests -
while maintaining its validity and reliability. are standardized instruments designed to m
Create your own instrument easure various aspects of human behavior, c
 involves careful planning and design to ens ognition, emotions,and personality. They're
ure it effectively measures what you intend. crucial for obtaining quantifiable data on psy
chological traits and states.
DATA COLLECTION TECHNIQUES 6. Questionnaire -
1. Documentary Analysis - is a tool consisting of a series of questions
This is a very common method of analyzing designed to gather information from
data which rely heavily on exploring respondents. Most commonly used by the
documentations and records that exist researchers.
within a body, organization, institution, or a) Closed-ended questionnaire - or restricted-
group. choice offer respondents a fixed set of
2. Interview - choices to select from. Closed-ended
is a data collection method where the resea questions are best for collecting data on
rcher asks participants questions to gather categorical or quantitative variables.
detailed information. This can be done face b) Open-ended questionnaire, or long-form,
-to-face, over the phone, or via video calls. questions allow respondents to give answers
in their own words. Because there are no
Three Types of Interviews: restrictions on their choices, respondents
a) UNSTRUCTURED INTERVIEW - This can answer in ways that researchers may not
interview can be in the form of normal have otherwise considered.
conversations or a free-wheeling exchange 7. Experiment -
of ideas. The researcher must be skilled in used to test a causal relationship.Manipulate
conducting the interview so that he/she variables and measure their effects on
can steer the course of conversation. The others.
interviewer must be knowledgeable on 8. Ethnography -
the subject or topic of concern. used to study the culture of a community or
b) SEMI-STRUCTURED INTERVIEW - There is a organization first-hand by join and
specific set of questions, but there are also participate in a community and record your
additional probes that may be done in an observations and reflections.
open-minded or close-ended manner. The 9. Archival Research -
researcher can gather additional data from used to understand current or historical
a respondent to add depth and events, conditions or practices by accessing
significance to the findings. manuscripts, documents or records from
c) STRUCTURED INTERVIEW - The conduct of libraries, depositories or the internet.
questioning follows a particular sequence 10. Secondary data collection -
and has a well-defined content. The used to find existing datasets that have
interviewer does not ask questions that already been collected, from sources such as
are not part of the questionnaire but government agencies or research
he/she can ask the interviewee to clarify organizations.
his/her answers.
CHARACTERISTICS OF A GOOD RESEARCH 4. Content validity.
INSTRUMENT This is determined by studying the questions to
see whether they are able to elicit the necessary
Concise information. An instrument with high content
 Delivering information in a clear, brief, and validity has to meet the objectives of the
direct manner without unnecessary words. research. It is measured by subjecting the
It focuses on essential points, ensuring the instrument to an analysis by a group of experts
message is easy to understand and straight who are knowledgeable about the subject both in
to the point. theory and practice (three to five experts will
suffice). The experts assess the items of the
Sequential questionnaire and determine if the items
 Questions or items must be arranged well. It measure the variables being studied. The experts’
is recommended to arrange it from simplest criticisms will be considered in the revision of the
to the most complex. In this way, the instrument.
instrument will be more favorable to the
respondents to answer. 5. Construct validity.
This refers to whether the test corresponds to its
Valid and reliable theoretical construct. It is concerned with the
 The instrument should pass the tests of extent to which a particular measure relates to
validity and reliability to get more the other measures and its consistency with the
appropriate and accurate information. theoretically-derived hypothesis. Factor analysis,
a refined statistical procedure, is used to analyze
Easily tabulated the interrelationships of behavioral data and is
 Since you will be constructing an instrument particularly relevant to construct validity.
for quantitative research, this factor should
be considered. Hence, before crafting the 6. Criterion-related validity or equivalence test.
instruments, the researcher makes sure that This type of validity is an expression of how
the variable and research questions are scores from the test are correlated with an
established. These will be an important external criterion.
basis for making items in the research
instruments. RELIABILITY OF THE INSTRUMENT
Reliability refers to the consistency of results.
VALIDITY OF THE INSTRUMENT A reliable instrument yields the same results
Validity is the ability of an instrument to measure for individuals who take the test more than once.
what it intends to measure.
TYPES OF VALIDITY: METHODS OF RELIABILITY:
1. Face validity. Test-retest Reliability
Also known as logical validity, face validity It is achieved by giving the same test to the same
involves an analysis of whether the instrument is group of respondents twice. The consistency of
using a valid scale. The procedure calls only for the two scores will be checked.
intuitive judgment. The researcher determines
face validity by looking at the features of the Equivalent Forms Reliability
instrument. It includes the size of the font or It is established by administering two identical
typeface, spacing, size of the paper used, and tests except for wordings to the same group of
other necessary details that will not distract the respondents.
respondents while answering the questionnaire.
Internal Consistency Reliability
2. Predictive Validity. It determines how well the items measure the
When the instrument can produce results similar same construct. It is reasonable that when a
to those similar tests that will be employed in the respondent gets a high score in one item, he will
future, it has predictive validity. This is also get one in similar items. There are three
particularly useful for the aptitude test. ways to measure the internal consistency;
through the split-half coefficient, Cronbach’s
3. Concurrent Validity. alpha, and Kuder-Richardson formula.
When the instrument can predict results like
those similar tests already validated, it has
concurrent validity.
 DATA GATHERING STRATEGIES Experiment Attempts to determine a cause
DATA COLLECTION DESCRIPTION -Blind Experiment and effect relationship between
Questionnaires Questionnaires often make use of -Double Blind two or more variables.
-Paper-pencil- checklist and rating scales. It is Experiment
questionnaires usually sent by mail, email or Blind Experiment - the test
-Web-based personally given to the subjects do not know if they are
questionnaires respondents. getting the experimental
It can be sent to a large number of treatment or the placebo.
people and saves the researcher
time and money. It could be open Double Blind Experiment - neither
ended format or multiple choice the test subject nor the
format. experimenter measuring the
Allows the respondents to answer response knows to which group
in any way . the test subjects have been
A new and inevitably growing assigned (treatment or placebo).
methodology is the use of Internet
based research. The respondents Observation Observational techniques are
will answer online questions such methods by which an individual or
as the use of Survey monkey individuals gather first hand data
on programs, processes, or
Interview Interviews can be used at any behaviors being studied.
-Personal stage of the evaluation process.
Interviews Two types of interviews are used
-Telephone in evaluation research: structured The most used quantitative research technique
Interviews interviews, in which a carefully
-Focused Interviews worded questionnaire is
in educational research is the survey. In a
-Clinical Interviews administered, and in depth quantitative survey you may use a short answer
-Disguised interviews, in which the responses or dichotomous questions, multiple
Interviews interviewer does not follow a rigid choice answers, paragraph, check boxes, drop
form. Interview will supplement
the data gathered through
down, linear scale, multiple choice grid and more.
questionnaires
Personal interviews are done
when people usually respond
when asked by a person but their
answers may be influenced by the
interviewer.
Telephone interviews consuming
and less expensive and the are less
time researcher has ready access
to anyone on the planet who has a
telephone
In focused interview, the
interviewers focuses attention
upon a given experience and its
effect.
Clinical interview is concerned
with broad underlying feelings or
motivations throughout the course
of an individual's life experiences
rather than with effects of a
specific experience.
Disguised interview relates to the
degree to which the respondent is
made aware of the real research
purpose. Data collected by
structured, disguised techniques
on attitudes towards instant coffee
were compared with the similar
data collected by non-structured,
disguised techniques.
 RESEARCH QUESTIONNAIRE TYPES OF SCALING QUESTIONS
A questionnaire is a research instrument that consists Semantic Differential Scale
of a set of questions or other types of prompts that A semantic differential scale is a survey or
aims to collect information from a respondent. A questionnaire rating scale that asks people to rate a
research questionnaire is typically a mix of close-ended product, company, brand, or any 'entity' within the
questions and open-ended questions. frames of a multi-point rating option. These survey
answering options are grammatically on opposite
Types of questions in a questionnaire: adjectives at each end. For example, love-hate,
Open-Ended Questions: Open-ended questions help satisfied-unsatisfied, and likely to return-unlikely to
collect qualitative data in a questionnaire where the return with intermediate options in between.
respondent can answer in a free form with little to no
restrictions. Likert Scale
The Likert scale question is a psychometric scale where
Close-ended Questions: are questions that limit questions based on this scale are used in a survey. It is
respondents to a set of pre-determined answers. They one of the most widely used question types in a
are often used in surveys and questionnaires to collect survey. In a Likert scale survey, respondents don't
factual information. choose between 'yes/no,' there are specific choices
based on 'agreeing' or 'disagreeing' on a particular
TYPES OF CLOSE_ENDED QUESTIONS: survey question.

Dichotomous Questions: The dichotomous question is Staple Scale

generally a “yes/no” close-ended question. This Stapel scale is defined as a rating scale that is close-
question is usually used in case of the need for ended with a single adjective (unipolar), developed to
necessary validation. It is the most natural form of a gather respondent insights about a particular subject
questionnaire. or event. The survey question is comprised of an even
number of response options without a neutral point.
Multiple-Choice Questions: Multiple-choice questions
are a close-ended question type in which a respondent Rank Order Scale
has to select one (single-select multiple-choice Rank order survey questions are close-ended questions
question) or many (multi-select multiple choice that allow respondents to evaluate multiple row items
question) responses from a given list of options. The based on a specific factor. A ranking survey is a group
multiple-choice question consists of an incomplete of multiple-choice questions presented in a single-
stem (question), right answer or answers, incorrect column format. Rank order questions are used in both
answers, close alternatives, and distractors. Of course, pen and paper questionnaires as well as online
not all multiple-choice questions have all of the answer surveys. They offer accuracy of responses and detailed
types. For example, you probably won’t have the insights.
wrong or right answers if you’re looking for customer DESIGNING A QUESTIONNAIRE
opinion. 1. Identify the scope of your research.
2. Keep it simple.
Scaling Questions: These questions are based on the 3. Ask only one question at a time.
principles of the four measurement scales – nominal, 4. Be flexible with your options.
ordinal, interval, and ratio. A few of the question types 5. The open-ended or closed-ended question is a tough
that utilize these scales’ fundamental properties are choice.
rank order questions, Likert scale questions, semantic 6. It is essential to know your audience.
differential scale questions, and Stapel scale questions. 7. Choosing the right tool is essential.

Pictorial Questions: This question type is easy to use ADMINISTERING THE QUESTIONNAIRE
and encourages respondents to answer. It works Types of Questionnaires Based on Distribution
similarly to a multiple-choice question. Respondents Online Questionnaire: In this type, respondents are
are asked a question, and the answer choices are sent the questionnaire via email or other online
images. This helps respondents choose an answer mediums. This method is generally cost-effective and
quickly without over-thinking their answers, giving you time-efficient. Respondents can also answer at leisure.
more accurate data. Telephone Questionnaire: A researcher makes a
phone call to a respondent to collect responses
directly.
In-House Questionnaire: This type is used by a
researcher who visits the respondent’s home or
workplace.
Mail Questionnaire: These are starting to be obsolete
but are still being used in some market research
studies. This method involves a researcher sending a
physical data collection questionnaire request to a
respondent that can be filled in and sent back.
 PRESENTATION OF DATA TYPES OF BIVARIATE ANALYSIS:
Tabular Presentation of Data I. Descriptive Analysis
 Tables present clear and organized data. A  In the descriptive analysis, bivariate analysis
table must be clear and simple but can apply to almost all data visualizations.
complete. A good table should include the Types of visualization displays such as bar
following parts. Table number and title – charts, line charts, column charts, etc. can
these are placed above the table. The title is still be used for bivariate analysis.
usually written right after the table number.
Caption subhead –this refers to columns  One of the interesting data visualizations
and rows. Body –it contains all the data that are usually done with bivariate analysis
under each subhead. Source- it indicates if is a scatterplot. The scatterplot is a data
the data is secondary and it should be visualization in the form of points displayed
acknowledged. on the x and y axes. The x and y axes
Graphical Method of Presenting the Data represent the value of each variable. By
 A graph or chart portrays the visual using a scatterplot, we can see the pattern
presentation of data using symbols such as of the relationship between the 2 variables.
lines, dots, bars or slices. It depicts the trend The relationships that are formed can be
of a certain set of measurements or shows linear, exponential, seasonal, etc. according
comparison between two or more sets of to data conditions.
data or quantities.
Types of Graphs: II. Inferential Analysis
 A BAR GRAPH uses bars to compare  Inferential analysis is used to generalize the
categories of data. It may be drawn results obtained from a random
vertically or horizontally. A vertical bar (probability) sample back to the population
graph is best to use when comparing means from which the sample was drawn. This
or percentages between distinct categories. analysis is only required when:
The categories are measured independently  - a sample is drawn by a random procedure;
and compared with one another. A and
horizontal bar graph may contain more than  - the response rate is very high.
5 categories. A bar graph is plotted on either
the x-axis or y-axis.
 A PIE CHART is usually used to show how
parts of a whole compare to each other and
to the whole. The entire circle represents
the total and the parts are proportional to
the amount of the total they represent.

UNIVARIATE AND BIVARIATE DATA

Univariate Data - Data that involve one variable
is called univariate data. Univariate data are
often described using the measures of central
tendency (mean or average, mode, and median),
variations, or other descriptive statistics.
Bivariate Data - Data that involve two variables
are called bivariate data. The statistical
procedure used to determine and describe the
relationship between two variables is called
correlation analysis.
Bivariate Analysis - Bivariate analysis refers to
the analysis of two variables to determine
relationships between them. Bivariate analyses
are conducted to determine whether a statistical
association exists between two variables, the
degree of association if one does exist, and
whether one variable may be predicted from
another.
 APPLIED STATISTICS Guidelines suggested by Fraerkel (1994), with
Statistics is a branch of Mathematics that deals regard to the minimum number of participants
with the collection, organization, presentation, needed.
analysis, and interpretation of numerical data.  Descriptive studies - a sample with a
It is any measure computed on the basis of data minimum number of 100 is essential
obtained from a characteristics of a population  Correlation studies - a sample of at least 50
under study. is deemed necessary to establish the
existence of a relationship.
Uses of Statistics  Experimental and causal-comparative
 Describe data studies - a minimum of 30 individuals per
 Compare two or more data sets group, although sometimes experimental
 Determine if a relationship exists between studies with only 15 individuals in each
variables group can be defended if they are very
 Test hypothesis (it is a scientific guess) tightly controlled
 Make estimates about population  Studies using only 15 subjects per group
characteristics. should be replicated, however, before too
 Predict past or future behavior of data much is made of any findings that occur.
Types of Statistics Research
Sources of Misuse  Descriptive - Is the type of statistical analysis
There are two main sources of misuse of that limits generalization to the particular
statistics: group of individual observed. The data
 Evil intent on part of a dishonest researcher describe one group and the group only.
 Unintentional errors (stupidity) on part of a  Inferential - It enables researchers to make
researcher who does not know any better inferences about the population, based on a
Data Collection random sample.
 In research, statisticians use data in many
different ways. DESCRIPTIVE STATISTICS
 Data can be used to describe situations. Measures of Central Tendency
 Data can be collected in a variety of ways, Mean - The sum of all the scores divided by the
BUT if the sample data is not collected in an number of scores. Often referred to as the
appropriate way, the data may be so average.
completely useless that no amount of
statistical torturing can salvage them.

The Slovin’s Formula:

where :
X̄ - sample mean
Σx - sum of all data points
N - number of data points
where:
n = sample size
Median - A median is the middle of a distribution.
N= population size
Half the scores are above the median and half are
e = the desired margin of error which is usually
below the median.
0.05

where:

X = ordered list of values in data set

n = number of values in data set
Mode - Most frequently occurring score in a
distribution.

For group data:

INFERENTIAL STATISTICS
 Population: Group that the researcher wishes to
study.
For ungroup data:  Sample: A group of individuals selected from the
Find the highest number among the given datas. population.
 Census: Gathering data from all units of a
Central tendency is simply the location of the middle population, no sampling.
in a distribution of scores.  Inferential statistics generally require that data
come from a random sample.
Weighted Average - When the arithmetic mean of a  In a random sample each person/object/item of
set of numbers is computed, it is safe to assume that the population has an equal chance of being
all observe values are of equal weight. In situations chosen.
where the numbers are not equally important, a
weight that is proportional to its relative importance Null hypothesis: A hypothesis put forward to argue
can be assigned to calculate the weighted average. that a relationship or pattern does not exist.
The weighted average is computed by dividing the Alternative Hypothesis: Statement of what study is set
sum of the products of the products of the values and to establish.
their weights by the sum of the weights.
HYPOTHESIS ERROR
When null hypothesis is tested one must state
beforehand the tendency of making a wrong decision
brought about by accepting or nullifying a null
hypothesis.

TYPE 1 ERROR
E.g. in a trial of new Drug X, the null hypothesis
Rating Scales might be that the new Drug X is no better than the
A rating scale provides more than two options, in current Drug Y.
which the respondent can answer in neutrality over a - H0: there is no difference between Drug X and Drug
question being asked. Y.
A Type 1 error would occur if we concluded that the
1. Three-point Scale two drugs produced different effects when there was
2. Five-point Scale no difference between them.
3. Seven-point Scale
TYPE 2 ERROR
Variance Type 2 error is failing to detect an association when
 The variance is a measure of how spread out a one exists, or failing to reject the null hypothesis when
distribution is. it is actually false.
 It is the average squared deviation of the You kept the null hypothesis when you should not
observations from their mean (how the have.
observations 'vary' from the mean). If Drug X and Drug Y produced different effects, and it
 The larger the variance, the further spread out was concluded that they produce the same effects.
the data.
LEVEL OF SIGNIFICANCE (α) CORRELATION
It refers to the probability to commit a type 1 error. In When to use it?
educational research, the most commonly used  When you want to know about the association or
significance levels are 0.05 and 0.01. An alpha level of relationship between two continuous variables
0.05 implies that the probability of committing a Type  Ex) food intake and weight; drug dosage and
1 error by chance alone is 5 out of 100. blood pressure; air temperature and metabolic
T-Test rate, etc.
 Allows the comparison of the mean of 2 groups. What does it tell you?
 Compares actual difference between two means  If a linear relationship exists between two
in relation to the variation in the data (expressed variables, and how strong that relationship is
as the standard deviation of the difference What do the results look like?
between the means).  The correlation coefficient = Pearson’s r
 Paired t-tests: When comparing the MEANS of a  Ranges from -1 to +1
continuous variable in two non-independent
samples (i.e., measurements on the same people
before and after a treatment)
 Independent samples t-tests: To compare the
MEANS ofa continuous variable in TWO
independent samples (i.e., two different groups
of people)
ANOVA
 In statistics, analysis of variance (ANOVA) is a
collection of statistical models, and their
associated procedures, in which the observed
variance in a particular variable is partitioned Guide for interpreting strength
into components attributable to different sources  0 – 0.25 = Little or no relationship
of variation.  0.25 – 0.50 = Fair degree of relationship
 In its simplest form ANOVA provides a statistical  0.50 - 0.75 = Moderate degree of relationship
test of whether or not the means of several  0.75 – 1.0 = Strong relationship
groups are all equal, and therefore generalizes t-  1.0 = perfect correlation
test to more than two groups. How do you interpret it?
 Tests hypotheses that involve comparisons of  If r is positive, high values of one variable are
two or more populations associated with high values of the other
 The overall ANOVA test will indicate if a variable (both go in SAME direction - ↑↑ OR
difference exists between any of the groups ↓↓)
 However, the test will not specify which groups  Ex) Diastolic blood pressure tends to rise with
are different age, thus the two variables are positively
 Therefore, the research hypothesis will state that correlated
there are no significant difference between any  If r is negative, low values of one variable are
of the groups. associated with high values of the other
 Allows the comparison of 3 or more groups. variable (opposite direction - ↑↓ OR ↓ ↑)
 Looks at the variation within groups, then  Ex) Heart rate tends to be lower in persons who
determines how that variation would translate exercise frequently, the two variables
into variation between groups (considering correlate negatively
number of participants).  Correlation of 0 indicates NO linear relationship
 If observed differences are larger than what How do you report it?
would be expected by chance, the findings are  “Diastolic blood pressure was positively
statistically significant. correlated with age (r = .75, p < . 05).”
 When we compute between and within group
variability we partition the total variability into CHI-SQUARE
the two components. A chi square (X2) statistic is used to investigate
 Therefore: Between variability + Within whether distributions of categorical (i.e.
variability = Total variability. nominal/ordinal) variables differ from one another.
Chi square Test for Independence
 Involves observations greater than 2x2
 Same process for the Chi square test
 Indicates independence or dependence of three
or more variables…but that is all it tells
ETHICAL DILEMMAS