You are on page 1of 112

• Recommended Books:

• Hair, Mary, Money, Samouel, Page (2016). Essentials of Business Research


Methods (3rd Ed.). Routledge.
• Hair, J. F., Black, W. C., Babin, B. J., & Anderson, R. E. . (2010). Multivariate
data analysis (7th ed.). Upper Saddle River: NJ: Prentice Hall.
• Pallant, Julie. (2011). Spss survival manaul: A step by step guide to data
analysis using spss for windows 4rth edition. Berkshire, UK: Open
University Press: McGraw-Hill Education.
• Field, (2013) Discovering Statistics Using IBM SPSS Statistics (4th Ed.). Sage
• Tabachnick, B.G., & Fidell, L.S. (2007). Using multivariate statistics (5 ed.).
Boston: Pearson Education.
• Hair, Hult, Ringle, Sarstedt. (2016) A premier on partial least squares
structural equation modeling (PLS-SEM) (2nd Ed.). Sage
• Census
• Researchers prefer to collect data from all members of a population
under investigation
Sample

• Is a relatively small subset of the population.

• It is drawn using either probability or non-


probability procedures.
Whether a probability of non-probability is used

But careful consideration of sampling


design issues is necessary in selecting the
sample.
• Probability Sampling is typically used in quantitative
research

• Selection of representative sample from the population using a random


procedure to ensure objectivity in selecting the sample.

• The findings from the sample data then can be generalized to the
population with a specified degree of accuracy.
• Non-probability sampling is typically used in
qualitative research

• Judgment is used to select the sample in qualitative research.


• Findings from the sample can be used to describe, discover, and
develop theory.
• While the findings can be used to generalize to the population, it
cannot be done with a specified degree of accuracy.
• Sample size is an important consideration in both
qualitative and quantitative research

• To generalize to the population a sufficiently large


sample is required.

• In contrast, when using a qualitative approach to


develop theory on organizational behavior a small
number of cases may be sufficient.
Sampling
• Is part of basic research process. It involves answering
following questions:

1) Should a sample or a census be used?


2) If a sample, then which sampling approach is best?
3) And, how large a sample is necessary?

Objective is to minimize sampling error


• We collect information for decision making therefore
we need to involve those people who are
knowledgeable.

• We refer to the group of knowledgeable people as Population or


universe.

• A population is thus a total of all the elements that share a common


characteristics.
• E.g people, supermarkets, hospitals
• In probability sampling researchers are able at a
minimum to calculate the error associated with a
particular sampling design and can make decisions
with this knowledge in hand.
• In non-probability sampling researchers are not able to
calculate the error but have made informed judgments
in an effort to obtain usable sample information.
Remember, a sample must be representative of the
population from which it is drawn.

Only, use of an appropriate sampling design can achieve


this objective.
The sampling process
1. Defining the target population
2. Choosing the sampling frame
3. Selecting the sampling method
4. Implementing the sampling plan
1. Defining the target population

The research objectives and scope of the study are critical in


defining the target population.

Target population is complete group of objects or elements relevant


to the research project.

Elements or objects available for selection during the sampling


process are known as sampling unit (people, households,
businesses, or any logically unit relevant to the study’s objectives).
• Element Employees with incentive pay
• Sampling unit customer service representatives and branch managers
• Extent All branch locations in Sindh province
• Time March 2011
2. Choosing the sampling frame

Provides a working definition of the target population.

Sampling frame is a comprehensive list of the elements from


which the sample is drawn.
Eg. Yellow pages listing of restaurants, a telephone directory,
company’s internal database, electronic directories.
Ideally, a sampling frame is an accurate, complete listing
of all the elements in the population targeted by the
research.
In reality, sampling frame is often flawed
in a number of ways:
1. It may not be up-to-date
2. It may include elements that do not belong to the
target population
3. It may not include elements that do belong to the
target population
4. It may have been compiled from multiple lists and
contain duplicate elements as a result of the manner in
which the list was constructed.
What’s the point!
Before drawing a sample from the sampling frame list,
the researcher must therefore confirm the list’s accuracy
irrespective of its origin.
3. Selecting sampling method
• Depends on number of theoretical and practical issues.
They include:
• Considering the nature of the study
• The objectives of the study
• And the time and budget availability
Sampling Methods
Probability Non-probability
• Simple random Convenience
• Systematic Judgment
• Stratified Snowball/Referral
• Cluster Quota
• Multistage
1) Simple random:
 is a straightforward
 Assigns each element of target population an equal probability of being
selected.
 Drawing names from a hat
 Easy (drawing names from hat) with a small population
 Large target population; other approaches are necessary

 One example of simple random sampling is random digital dialing in a


telephone survey
 Issues: nonworking numbers, refusal to answer a telephone, lack of
time/interest etc
Procedure for drawing large samples
1. Sequentially assign a unique identification number to
each element in the sampling frame
2. Use a random number generator to identify the
appropriate elements to be selected in the sample
3. Ensure that no element is selected more than once.
2) Systematic sampling

Involves selecting an initial starting point on a list, and


thereafter every nth element in the sampling frame is
selected.

Suppose, you have a list of 10,000 students attending a


particular university and you want a sample of 500
students.
• Your sampling objective is a representative cross-
section of the student body.

• To draw the sample you must determine the sample


size and then calculate the sampling interval.

• Sampling interval is the number of population


elements between each unit selected for your sample.
• In this case, the sampling interval is 20
(10,000 students / sample of 500 = 20)

7, 27,47,67 and so on.


Systematic sampling produces representative data if
executed properly.
• To work properly, the sampling interval must divide
the sampling frame into relatively homogeneous
groups, if there is a cyclic sequence to the sampling
frame instead of a random sequence, systematic
sampling will not work.
• For instance, Alphabetical listings are considered
random not cyclic.

• In contrast, if we wanted to do weekly interviews with


our customers and our interval was 7, the sample
would produce biased information, because we would
always interview on the same day of the week. To be
truly random, we must conduct interviews across at
least several different days of the week.
• Lets say, if our list of customers is 1030 arranged according
to frequency of dinning, and the first 100 names on the list
eat at our restaurant at least once a week and the remaining
930 eat at our restaurant an average of four times a year,
we would have a problem using systematic sampling.
• If the sampling interval is 10 and our sample size is 103,
then our sample may underrepresent the frequent
customers (only 10 frequent customers) and overrepresent
the less frequent customers (93 nonfrequent customers)
Thus, we must know ahead of time if there are
underlying systematic patterns in the data so we can
account for them in our sampling plan.
3) Stratified sampling
Requires the research to partition the sampling frame
into relatively homogeneous subgroups that are distinct
and nonoverlapping, called strata.
• The researcher usually does the stratification on the
basis of some predetermined criteria that may be the
result of his/her past experience, or stratification could
even be specified by the client.

• In restaurant example, the researcher may wish to


stratify customers on the basis of characteristics such as
age, marital status, family size, income level, frequency
of eating out, level of satisfaction, who selected the
restaurant, or a combination of these.
• The researcher determines the total sample size as well
as the required sample sizes for each of individual
strata.
• For example, total sample size might be 400 and the
four individual strata might each have a sample size of
100.
• The stratified sample is the composite of the samples
taken from the strata.
• Elements for the stratified sample are usually selected
either by drawing simple random or systematic
samples of the specified size from the strata of the
target population.
• With stratified sampling, elements must be selected
from all the strata of the total sample.
•Practically, stratified sample is selected
in one of two ways; Proportionately or
disproportionately.
• In Proportionately stratified sampling, the overall
sample size will be the total of all the elements from
each of the strata.
• The number of elements chosen from each stratum is
proportionate to the size of a particular stratum relative
to the overall sample size.
• So if we have a stratum that is 25 percent of the target
population, then the size of the sample for that stratum
will be 25 percent of the total sample.
• For example, if we use proportionately stratified
sampling to select a sample of males and females at IBA
with 10,000 student, and 6,000 students are females and
4,000 students are males, then the overall sample would
include 60 percent females and 40 percent males.
• In disproportionately stratified sampling the sample
elements are chosen in one of the two ways;
• One approach involves choosing the elements from
each stratum according to its relative importance.
• Relative importance is usually based on practical
considerations such as economic importance of the
various strata.
• For example, if that restaurant is located in an area
dominated by older individuals who dine out less
frequently, then sampling a high proportion of younger
customers that dine out more often would be viewed as
more important to him.
• With disproportionately stratified sampling based on
economic or other reasons, the sample size from each
stratum is determined independently without
considering the size of the stratum relative to the
overall sample size.
• The more important a particular stratum is considered,
the higher will be the proportion of the sample
elements from the stratum.
• Another approach to selecting a disproportionately
stratified sample considers the variability of the data
within each stratum.
• Elements from each stratum are selected based on the
relative variability of the elements.
• Strata with high relative variability will contribute a
higher proportion of elements to the sample. Similarly,
lower the variability of a stratum the lower will be its
proportional representation in the total sample.
• For example, assume IBA with 10,000 students has 50 percent
male students and 50 percent female students. We know that
almost all the male drink beer, and there is wide variation in beer
drinking habits, with some drinking beer every day and a very
small number not drinking beer at all.
• On the other hand, only a small proportion of the female students
drink beer and not very often (they prefer fresh juices), so there is
not much variation in their beer consumption patterns.
• In this example, we would sample a larger number of male
students in our survey so we could more accurately represent
male beer consumption patterns, Since female students do not
vary much in their beer consumption habits, the smaller sample of
females should still accurately represent their behavior.
4) Cluster sampling
• In cluster sampling, the target population is viewed as
being made up of heterogeneous groups,
• Examples of cluster are ethnic groups, companies,
households, business units, and geographic areas.
• the most frequently used type of cluster sampling is
geographic area sampling.
• For example, assume you want to interview managers of
banks in Sindh province.
• The research could obtain a list of zip codes in which banks
are located: each area is then a cluster.
• The cluster to be sampled would be randomly selected, and
then all bank managers or a random sample of managers of
banks would be interviewed in each of the selected clusters.
• This process generally works well with and produces
representative data.
Cluster sampling procedure
1. Define the cluster characteristics in a way that ensures clusters are
unambiguously identified in the target population. In this manner,
the total number of clusters in the population will be known ahead of
time.
2. Decide on how many clusters to sample.
3. Choose the cluster(s) in a random manner
4. Obtain a sampling frame for the chosen clusters
5. Decide whether to conduct a census on the chosen cluster(s) or
whether to take a probability sample from the cluster(s).
6. If a probability sample is desired, determine the total sample size. If
more than one cluster will be used, then the sample size should be
allocated appropriately. This is generally done on proportionate
sampling basis.
5) Multistage sampling
• Involves a sequence of stages.
• These stages are illustrated by the following example.
• The problem is to investigate the views of medical
practitioners in the US concerning the use of medical
software to assist in patient diagnosis. The first stage is to
select a random sample of regions in the US. The regions
are clusters. The Second stage is to select a random sample
of hospitals from the selected regions, and then either
collect information from all medical practitioners from the
chosen hospitals or a random sample from within each of
the chosen hospitals.
• Even more complex multistage sampling is possible.
Determining sample size

Efficient sample sizes can be drawn from


either large (infinite) populations or small
(finite) populations.
Sampling from a large population
Researchers often need to estimate characteristics of
large populations. To achieve this in an efficient manner,
it is necessary to determine the appropriate sample size
prior to data collection.
• Determining sample size is complex because of many
factors that need to be taken into account simultaneously.
• Challenge is to obtain acceptable balance among several of
these factors. These factors include:
• the variability of elements in the target population,
• The type of sample required,
• Time availability,
• Budget
• Required estimation precision,
• And whether the findings are to be generalized and, if so, with what degree
of confidence.
What is ‘good’ and ‘bad’ data?
• You must understand how SPSS works and how to talk to it
appropriately.
Pre-testing (Refer Hunt et al., 1982)
• What items?
• Length, layout, format, number of lines for replies, sequencing
• Individual questions, respondents hesitate

• What method?
• Personal interviews, phone, and email
• Debriefing (after) or protocol (during)?
Pre-testing (Continued……)
• Who should do?
• Best interviewers
• Who are the subjects?
• Respondents who are similar as possible
• Representative vs convenience
• How large a sample?
• Vary from 12, 20, 30 to 100
Remember, then comes pilot-
test.
A recent reflection by Joe F. Hair
•Remember, Research design has
implications for the quality of the
data collected and analysed.
• The data you enter must come from
somewhere—responses to a questionnaire,
information collected from interviews, coded
observations of actual behaviour, or objective
measurements of output or performance.
• The data are only as good as the
instrument that you used to collect
them and the research
framework that guided their collection.
• From
• Review of the literature,
• Formulation of hypotheses,
• Choice of study design,
• Selection and allocation of participants,
• Recording of observations and collection of data.

• Decisions made at each of these stages can affect the quality of


the data you have to analyse and the way you address your
research questions.
•In designing your own study I would
recommend that you take your time
working through the design process to
make it the best study that you can
produce.
• Reading a variety of texts on the topic will
help
A few tips
are here…..!
1.
• Consider what type of research design (e.g. experiment,
survey, observation) is the best way to address your research
question.

• There are advantages and disadvantages to all types of research


approaches; choose the most appropriate approach for your
particular research question. Have a good understanding of the
research that has already been conducted in your topic area.
2.
• If you choose to use an experiment, decide whether a between-
groups design (different cases in each experimental condition)
or a repeated measures design (same cases tested under all
conditions) is the more appropriate for your research question.
• There are advantages and disadvantages to each approach (see
Stangor 2006), so weigh up each approach carefully.
• In experimental studies, make sure you include enough levels
in your independent variable. Using only two levels (or groups)
means fewer participants are required, but it limits the
conclusions that you can draw.
• Is a control group necessary or desirable? Will the lack of
control group limit the conclusions that you can draw?
3.
• Always select more participants than you need, particularly if
you are using a sample of humans.

• People are notoriously unreliable—they don’t turn up when


they are supposed to, they get sick, drop out and don’t fill out
questionnaires properly! So plan accordingly. Err on the side of
pessimism rather than optimism.
4.
• In experimental studies, check that you have enough
participants in each of your groups (and try to keep them equal
when possible). With small groups, it is difficult to detect
statistically significant differences between groups (an issue of
power).
• There are calculations you can perform to determine the sample
size that you will need. See, for example, Stangor (2006), or
consult other statistical texts (we will discuss in detail later in
the course).
• Wherever possible, randomly assign participants to each of
your experimental conditions, rather than using existing
groups.
• This reduces the problem associated with non-equivalent
groups in between-groups designs.
• Also worth considering is taking additional measurements of
the groups to ensure that they don’t differ substantially from
one another.
• You may be able to statistically control for differences that you
identify (e.g. using analysis of covariance).
• Choose appropriate dependent variables that are valid and
reliable. It is a good idea to include a number of different
measures—some measures are more sensitive than others.
• Don’t put all your eggs in one basket.
• Try to anticipate the possible influence of extraneous or confounding
variables. These are variables that could provide an alternative
explanation for your results.
• Sometimes they are hard to spot when you are immersed in
designing the study yourself. Always have someone else
(supervisor, fellow researcher) check over your design before
conducting the study. Do whatever you can to control for these
potential confounding variables.
• Knowing your topic area well can also help you identify possible
confounding variables. If there are additional variables that you
cannot control, can you measure them? By measuring them, you
may be able to control for them statistically (e.g. using analysis of
covariance).
• If you are distributing a survey, pilot-test it first to ensure that
the instructions, questions and scale items are clear.
• Wherever possible, pilot-test on the same type of people who
will be used in the main study (e.g. adolescents, unemployed
youth, prison inmates). You need to ensure that your
respondents can understand the survey or questionnaire items
and respond appropriately.
• Pilot-testing should also pick up any questions or items that
may offend potential respondents.
• If you are conducting an experiment, it is a good idea to have a
full dress rehearsal and to pilot-test both the experimental
manipulation and the dependent measures you intend to use.
• If you are using equipment, make sure it works properly. If you
are using different experimenters or interviewers, make sure
they are properly trained and know what to do. If different
observers are required to rate behaviours, make sure they know
how to appropriately code what they see.
• Have a practice run and check for inter-rater reliability (i.e. how
consistent scores are from different raters). Pilot-testing of the
procedures and measures helps you identify anything that
might go wrong on the day and any additional contaminating
factors that might influence the results. Some of these you may
not be able to predict (e.g. workers doing noisy construction
work just outside the lab’s window), but try to control those
factors that you can.
CHOOSING APPROPRIATE SCALES AND
MEASURES
• Many different ways of collecting ‘data’,
• depending on the nature of your research.
• This might involve measuring output or performance on some
objective criteria, or rating behaviour according to a set of
specified criteria.
• use of scales that have been designed to ‘operationalise’ some
underlying construct or attribute that is not directly measurable
(e.g. self-esteem).
• Thousands of validated scales are available for research use
•How do you find the right
measure for research your
purpose?
• A thorough review of the literature in your topic area is the first
place to start.
• What measures have been used by other researchers in the area?
• Sometimes the actual items that make up the scales are included in
the appendix to a journal article; otherwise you may need to trace
back to the original article describing the design and validation of
the scale you are interested in.
• Some scales have been copyrighted, you need to purchase ‘official’
copies from the publisher.
• Other scales, which have been published in their entirety in journal
articles, are considered to be ‘in the public domain’.
• Acknowledgement is a must.
• In choosing appropriate scales there are two characteristics that you
need to be aware of:
• reliability and validity.
• Both of these factors can influence the quality of the data you obtain.
• When reviewing possible scales to use find out reliable and valid
scales.
• But, it is also important to pre and pilot-test them with your
intended sample.
• Sometimes scales are reliable with some groups (e.g. adults with an
English-speaking background), but are totally unreliable when used
with other groups (e.g. children from non-Englishspeaking
backgrounds).
PREPARING A QUESTIONNAIRE
• A poorly planned and designed questionnaire will not give good
data with which to address your research questions.
• In preparing a questionnaire, you must consider how you intend to
use the information; you must know what statistics you intend to
use.
• Depending on the statistical technique you have in mind, you may
need to ask the question in a particular way, or provide different
response formats.
• I would suggest that you read further on the topic if you are
designing your own study.
• A really great book for this purpose is DeVaus (2002) or Boyce
(2003).
Question types
• Most questions can be classified into two groups:
• closed or open-ended.
• A closed question involves offering respondents a number of
defined response choices.
• They are asked to mark their response using a tick, cross, circle,
etc. The choices may be a simple Yes/No, Male/Female, or may
involve a range of different choices. For example:
• Closed questions are usually quite easy to convert to the numerical
format required for SPSS.
• For example, Yes can be coded as a 1, No can be coded as a 2; Males
as 1, Females as 2.
• In the education question shown above, the number corresponding
to the response ticked by the respondent would be entered.
• For example, if the respondent ticked Undergraduate university, this
would be coded as a 5. Numbering each of the possible responses
helps with the coding process.
• For data entry purposes, decide on a convention for the numbering
(e.g. in order across the page, and then down), and stick with it
throughout the questionnaire.
• Sometimes you cannot guess all the possible responses that
respondents might make—it is therefore necessary to use open-
ended questions.
• The advantage here is that respondents have the freedom to
respond in their own way, not restricted to the choices provided
by the researcher. For example:
• Responses to open-ended questions can be summarised into a
number of different ategories for entry into SPSS.
• These categories are usually identified after looking through the
range of responses actually received from the respondents.
• Some possibilities could also be raised from an understanding
of previous research in the area.
• Each of these response categories is assigned a number (e.g.
work=1, finances=2, relationships=3), and this number is
entered into SPSS.
• Sometimes a combination of both closed and open-ended questions
works best.
• This involves providing respondents with a number of defined
responses, and also an additional category (other) that they can tick if
the response they wish to give is not listed.
• A line or two is provided so that they can write the response they
wish to give.
• This combination of closed and open-ended questions is particularly
useful in the early stages of research in an area, as it gives an
indication of whether the defined response categories adequately
cover all the responses that respondents wish to give.
Response format
• In asking respondents a question, you also need to decide on a response format.
• The type of response format you choose can have implications when you come to do your
statistical analysis.
• Some analyses (e.g. correlation) require scores that are continuous, from low through to
high, with a wide range of scores.
• If you had asked respondents to indicate their age by giving them a category to tick (e.g.
less than 30, between 31 and 50 and over 50), these data would not be suitable to use in a
correlational analysis.
• So, if you intend to explore the correlation between age and, say, self-esteem, you will
need to ensure that you ask respondents for their actual age in years.
• Be warned though, some people don’t like giving their exact age (e.g. women over 30!).
• Try to provide as wide a choice of responses to your questions
as possible.
• You can always condense things later if you need to (We will
discuss it further later).
• But, don’t just ask respondents whether they agree or disagree
with a statement—use a Likert-type scale, which can range from
strongly disagree to strongly agree:
• This type of response scale gives you a wider range of possible scores, and increases the statistical
analyses that are available to you.
• You will need to make a decision concerning the number of response steps (e.g. 1 to 6) that you use.
DeVellis (2003) has a good discussion concerning the advantages and disadvantages of different
response scales.
• Whatever type of response format you choose, you must provide clear instructions.
• Do you want your respondents to tick a box, circle a number, make a mark on a line?
• For some respondents, this may be the first questionnaire that they have completed. Don’t assume
they know how to respond appropriately.
• Give clear instructions, provide an example if appropriate, and always pilot-test on the type of
people that will make up
• your sample.
• Iron out any sources of confusion before distributing hundreds of your questionnaires.
• In designing your questions, always consider how a respondent might interpret the question and
all the possible responses a person might want to make.
• For example, you may want to know whether people smoke or not. You might ask the question:
• In trialling this questionnaire, your respondent might ask
whether you mean cigarettes, cigars or Shesha.
• Is knowing whether they smoke enough?
• Should you also find out how much they smoke (two or three
cigarettes, versus two or three packs),
• and/or how often they smoke (every day or only on social
occasions)?
•The message here is to consider each
of your questions, what information
they will give you and what
information might be missing.
Wording the questions
• There is a real art to designing clear, well-written questionnaire items.
• Although there are no clear-cut rules that can guide this process, there are
some things you can do to improve the quality of your questions, and
therefore your data. Try to avoid:
• long complex questions
• double negatives
• double-barrelled questions
• jargon or abbreviations
• culture-specific terms
• words with double meanings
• leading questions
• emotionally loaded words.
• When appropriate, you should consider including a response
category for ‘Don’t know’ or ‘Not applicable’.
• For further suggestions on writing questions, see De Vaus
(2002) and Kline (2005).
Independent & Dependent Variables
• Independent variable: A variable thought to be the cause of some
effect. Used in experimental research to denote a variable that the
experimenter manipulated.
• Dependent variable: A variable thought to be affected by changes in
the independent variable. This is the outcome variable.
• Predictor variable: A variable thought to predict an outcome variable.
Basically an independent variable.
• Outcome variable: A variable thought to change as a function of
change in a predictor variable.
From your initial observation you generate explanations, or theories, of
those observations, from which you can make predictions (hypotheses).

Scientific statements are ones that can be verified (tested) using empirical
evidence.
Levels of Measurement
• Categorical
• Binary variable: Only 2 categories (male, female)
• Nominal variable: More than two categories: young, middle age, elderly.
• Ordinal variable: The variable is ordered by some attribute, such as pain.
(each interval does not represent and EQUAL distance). Ex: RPE & Pain ratings
• Continuous
• Interval variable:
• Ratio variable:
Measurement Error

• Measurement error is the discrepancy between a variables actual value and its
measured value.
• Some variables more prone to errors than others: attitude, pain, volume of gas
expired, blood pressure, height and weight.
• Factors that can influence measurement error:
• Accuracy of instruments
• Random variation in the variable
• Adherence to sound measurement principles
Validity and Reliability
• Validity refers to whether an instrument actually measures what it
is designed to measure.
• DEXA, hydrostatic, and skinfolds can all measure the percent body fat.
DEXA has the highest validity of the three.
• Reliability refers to the consistency of the instrument.
• The easiest way to test reliability is to measure the same people
twice (test-retest reliability).
Correlational vs Experimental Research

• In correlational research we observe what goes on without directly


interfering with it.
• Ex: what is the relationship between anxiety and performance?
• In experimental research we manipulate one variable to see its
effects on another.
• Ex: What is the effect of caffeine on reaction time?
Using Descriptive Statistics

• The descriptive information gives you an idea about the average


level of satisfaction and something about the variability of scores:

• Comparative Statistics
• Suppose you have been funded by a government agency to evaluate
the operation of two charity-sponsored counseling centers and you
get following summary data:

• Average satisfaction score at the uptown location was 82.4 (based on 54


client scores)

• Average satisfaction score at the Downtown location was 78.5 (based on 55


client scores)
• Assuming that the clients are representative at each location, you
have some evidence to make a decision about which center is more
effective in terms of satisfaction scores.

• Your data suggest that the uptown location may do a better job as
far as the satisfaction score is concerned since the score for uptown
is 3.9 point higher that the score for the downtown location.

• However, what if the average scores were only 1 point apart? Or 10


points apart?
• What level of difference would it take for you to conclude that the
average score for one location was significantly higher than for the
other?

• Could the difference in scores be due to random fluctuations? If you


did the survey again during some other time period, is there a
reasonable chance that the downtown location would produce a
better score?
• These questions are addressed with a properly designed and
executed statistical analysis
• Correlations Statistics
• To learn more about your survey results, you could examine your
data in another way.

• Ignoring for a moment the location of the center, you may want to
compare the relationship between educational level of clients and
satisfaction scores.
Understanding Hypotheses testing, Power, and Sample Size

• Alternate (Ha): The population means of the two groups are different

• Null hypotheses (H0) = the population means of the two groups are
the same
• In hypotheses testing two types of errors can occur

• Type-I Error:

• In statistical hypothesis testing, a type I error is the rejection of a true null hypothesis, while a
type II error is failing to reject a false null hypothesis.

• Type-II Error
Understanding the p-Value

• The “evidence” used to reject a null hypotheses is summarized in a


probability called a p-value.