ChapterPDF Available

Discovering Yourself in Psychology Ch 2

Authors:

Abstract

Discovering Yourself in Psychology is a work in progress. It is a free textbook suitable to introduce readers to psychology, the academic discipline. Chapter 2 discusses methods used in psychology research.
1
All Rights Reserved @ 2017 Shelia M. Kennison
2 CONDUCTING PSYCHOLOGICAL RESEARCH
Choosing a Research Design
The True Experiment
Quasi Experiments
Correlational Studies
Naturalistic Observations and Case Studies
Collecting Data
Constructing a Sample
Measuring Psychological Variables
Statistical Analysis of Data
Descriptive Statistics
Inferential Statistics
Statistical Significance
Practical Significance
Protecting Human Subjects
History of Research Ethics
Examples of Unethical Research
Research Misconduct
Professional Codes of Ethics
2
All Rights Reserved @ 2017 Shelia M. Kennison
Student Learning Objectives
Discuss how well the different research methodologies enable researchers to
establish a causal relationship between variables.
Describe how a researcher conducts a true experiment.
Discuss how psychologists obtain a representative sample from a research
population.
Discuss the importance of validity and reliability in the measurement of variables.
Describe how researchers evaluate the central tendency of a sample.
Describe how psychologists measure the variability observed from samples.
Discuss the extent to which outliers influence large and small samples.
Discuss the circumstances in which a researcher would make Type I and Type II
errors.
Discuss the significance of the Nuremberg trials and the Belmont Report in the
development of contemporary ethical principles for the conduct of research.
Discuss how the process of informed consent serves to ensure the ethical treatment
of human research participants.
Discuss the code of ethics that psychologists follow. What professional activities are
covered in the code?
3
All Rights Reserved @ 2017 Shelia M. Kennison
Psychologists use the scientific method when conducting research. Without a
doubt, using the scientific method to investigate psychological phenomena can be
challenging. As in other disciplines, researchers in psychology must develop a research
question, select an appropriate research design, plan appropriate measurement of the
phenomenon of interest, statistically analyze the data, and then interpret the results.
Because psychology research routinely involves the testing of human research
participants, researcher must ensure that human research participants are treated in an
ethical manner at all times. They must also ensure that research data are kept
confidential while the study is occurring as well as when the study is over and the data
are being analyzed. In this chapter, you will learn more about psychology research is
done.
Choosing a Research Design
Every research project begins with an idea or hypothesis about why some human
characteristic or behavior might occur. The next important step the research processing
is to choose a research design. The researcher must consider the fact that the different
types of research designs differ in their internal validity or how well the design can
establish the cause of the observed outcome. Beginning students of psychology are
often surprised to learn that there is only one type of research design that can establish
causation. If a researcher wants to be 100 percent sure that one variable causes a
change in another variable, then there is no doubt about what methodology should be
used. Using any other methodology may provide evidence that a relationship exists
between the two variables, but one cannot conclude that one variable causes the other.
4
All Rights Reserved @ 2017 Shelia M. Kennison
The True Experiment
What is the one and only methodology that can prove causation? It is the true
experiment. The true experiment, when conducted in a careful way, achieves the
highest level of internal validity. In a true experiment, a researcher manipulates one or
more variables in order to determine whether one or more other variables are affected.
Any variable that is manipulated in an experiment is called an independent variable
(IV). Any variable that is considered the “outcome variable” or is expected to be
changed by the IV is called a dependent variable (DV). Experiments can have one or
more IVs and DVs. In order for true experiments to have the highest level of internal
validity, the researcher must exert control over the research process, so that the only
difference among the different versions of the manipulation is the manipulation made by
the experimenter. Experiments on animal subjects often achieve high levels of internal
validity, because the experimenter can easily control every aspect of the animal’s
environment, such as the temperature of the room, the layout of the case, the amount of
food and water, and the amount of light received each day.
True experiments must also involve random assignment. Each member of the
research sample should have an equal chance of being assigned to any one of the
versions of the IV. Random assignment prevents the possibility that the groups will start
out different before the IV manipulation is applied. An easy technique to ensure that
participants are randomly assigned to the different groups or conditions is to put all the
participants’ names in a hat and randomly select participants for the first group, second
group, third group, etc. If there are only two experimental conditions, one might
determine a particular participant’s random assignment with a flip of a coin.
5
All Rights Reserved @ 2017 Shelia M. Kennison
Experiments involving human research participants are more challenging to
control or ensure that the other aspect of the situation that is varying is the IV
manipulation. Researchers cannot control every aspect of participants’ environments
as they can in experiments involving animal subjects. Nevertheless, true experiments
involving human research participants are routinely conducted. A familiar example is
the randomized clinical trial (RCT). RCTs are frequently used to test the
effectiveness of new procedures, drugs, or non-drug therapies to improve functioning.
In a RCT, the effect of a drug or new procedure may be compared to a placebo, a pill
or procedure that does not contain the experimental substance or procedure. Steps are
taken to ensure that the placebo looks similar in all possible ways to the experimental
drug or procedure. The use of the placebo is to prevent participants from knowing
whether they are receiving the experimental drug, because one may experience
physical benefits from merely believing that an experimental drug or procedure has
been taken or experienced. This has been called the placebo effect. Often, research
personnel who monitor the participants’ health outcomes also are not permitted to know
which participants are receiving the new drug or procedure. This type of design is a
double-blind design. This type of design prevents the research results from being
influenced by experimenter bias or the researcher’s expectation for a particular type of
outcome.
Quasi Experiments
There are circumstances when a researcher cannot test a research hypothesis in
a true experiment, such as when the variable of interest cannot be randomly assigned
to human research participants. For example, if a researcher is interested in
6
All Rights Reserved @ 2017 Shelia M. Kennison
determining whether men and women perform differently on a particular task, it is not
possible for the researcher to flip a coin to determine which participants will be assigned
to the male condition and which participants will be assigned to the female condition.
Variables that cannot be randomly assigned to participants are called subject
variables. Subject variables are frequently used in psychological research. Studies
that compare performance for different types of people (e.g., age, smoking status,
political affiliation, etc.) are called quasi experiments. Quasi experiments cannot
achieve the highest levels of internal validity, because it is always possible that the
subject variable that was used to create the groups is not the only difference that exists
between the groups. The groups might differ in other ways as well. One or more of
these other differences between the groups might contribute any difference observed
between the groups at the end of the study.
Correlational Studies
Often, a researcher who is interested in understanding more about subject
variables will use a correlational design, in which no variable is manipulated or
compared; rather, the researcher measures two or more variables from a group of
individuals in order to learn whether there is any systematic relationship between any of
the pairs of variables. Correlational studies generally have low levels of internal validity,
because the researcher cannot conclude that the relationship that is observed involves
one variable causing a change in the other. For example, one might observe that the
number of snow cones sold in a day is related to the number of violent assaults
recorded at the police station in the same town. One might first leap to the conclusion
that eating a snow cone puts one in the mood to get into a fight. However, one might
7
All Rights Reserved @ 2017 Shelia M. Kennison
then think that it is possible that the experience of fighting leads one to crave a snow
cone. Neither of these interpretations may be true. It is also quite plausible that snow
cone sales and violent assaults are not directly related at all. A third variable may be
related to them both. Researchers refer to this as the third variable problem. In this
case, temperature could be the third variable. As the temperature rises, some people in
the town like to treat themselves to a snow cone, and other people end up having
disputes with people and some of those get physical. For studies using a correlational
design, one should never conclude that there is a causal relationship between variables.
Despite the low internal validity of correlational studies, media reports of such
studies often imply that a causal link between variables exists. In 1996, a study showed
that regular coffee drinkers were less likely to commit suicide than those who never
drank coffee (Kawachi, Willett, Colditz, Stampfer, & Speizer, 1996). One might assume
that the study results provide evidence that there is some substance in coffee that
serves to protect one from the state of mind that leads to suicide. However, it is just as
plausible that some unidentified variable is related to whether people drink coffee and is
also related to people’s likelihood of committing suicide. For example, it is possible that
the overall health of participants was related to coffee drinking. Many individuals who
regularly take medications for various health problems may be advised by their
physicians to avoid coffee and other foods containing caffeine. Unfortunately, when
listening to media reports of research, one must always be cautious. The journalists
who are reporting the results of studies are likely not to have taken a course in
introductory psychology and in psychological research methods.
8
All Rights Reserved @ 2017 Shelia M. Kennison
Naturalistic Observations and Case Studies
Although correlational studies cannot provide convincing evidence that one
variable causes another, they can be an important first step in a researcher’s long-term
program of research. If a correlational study provides some evidence that variables are
related, then the researcher may conduct a follow-up study, using a different research
design. By conducting multiple studies on the same topic, a research may obtain a
more detailed picture of the phenomenon. There are two other types of research
designs that are routinely used as when researchers are just beginning to study a topic.
These research designs are the naturalistic observation and case study. Both types
of designs have very low levels of internal validity. In a naturalistic observation, one
observes behavior as it occurs naturally in life. The advantage of this type of research
is that the researcher can get a picture of behavior as it occurs in real world settings. A
disadvantage of this type of research is that there is a lack of control over the setting.
As a consequence, the researcher cannot be certain about the actual causes of the
observed behaviors. In a case study, researchers study a single participant or event in
order to gain insight into a phenomenon or process that occurs in others. There is also
a lack of control in case studies. A researcher documents that important elements of
the person and the situation and attempts to identify possible causes of outcomes;
however, no definitive causal conclusion can be drawn.
For some research hypotheses, a researcher may never be able to conduct a
study in which a causal relationship between variables can be established. Sometimes,
a true experiment can never be done. This is most often true when one is investigating
the harmful effects of one or more variables on human research participants. Consider
9
All Rights Reserved @ 2017 Shelia M. Kennison
the numerous studies that have demonstrated that cigarette smoking in humans is
correlated with lung cancer rates and rates of breathing-related diseases (Oreskes &
Conway, 2010). None of them have proven with 100 percent certainty that cigarette
smoking causes lung disease in people, technically. This is because, there has never
been a true experiment conducted involving human research participants. Why has no
true experiment been conducted? Conducting such an experiment would not be ethical.
Although, it would be relatively straightforward to recruit a group of research participants
for the experiment, offer to pay them well for their time, and then randomly assign half of
the sample to smoke a pack a day for a year and randomly assign the other half not to
smoke at all for year. After a year, you call them back to the laboratory for a complete
physical workup. I hope it is obvious why this type of experiment should never be done.
It would be unethical to expose someone to a manipulation that is expected to cause
them harm.
Constructing a Sample
Regardless of the research design that is used, it is usually the case that the
researcher’s ultimate goal is to use the study results to make a statement about what is
likely to occur in the future other people in other places, particularly for people who are
similar to those who were tested in the study. For example, if a researcher finds that
taking a new antidepressant causes 45 out of 50 people given the drug to experience
an improvement in mood, then it is reasonable to estimate that 90 percent of other
people given the drug may also an improvement of mood. The term external validity
refers to how well a study’s results can predict outcomes to other groups. Generally
10
All Rights Reserved @ 2017 Shelia M. Kennison
when conducting a research study, one strives for the study to achieve the highest
possible level of external validity.
In order to determine whether the external validity in a study is high or low, one
must consider the extent to which those tested in the study are similar to the individuals
whose outcomes the researcher wants to predict. The research population is the
group of individuals who share one or more characteristics of interest to a researcher
and whose outcomes the researcher is interested in predicting. The research sample
is a subset of a research population from which data are collected. Studies that have
the highest level of external validity use representative samples, which are samples in
which the researcher has recruited a small group of people who accurately reflect the
characteristics of the entire research population in proper proportion.
Studies that have the lowest level of external validity use biased samples or
samples that do not accurately reflect the characteristics of the research population.
For example, if a researcher is interested in a population of male and female college
students and the sample includes only female students, then the sample results might
not accurately predict outcomes for all college students. If a researcher is interested in
a population that includes adults ranging in age from 18 to 60 and the sample includes
only individuals between the ages of 18 and 22, then the results might not generalize to
individuals who are over 22. Samples can be biased for many reasons. Whenever a
sample is biased, the researcher is likely unable to fix it; there is no easy way to
transform a biased sample into a representative sample. If researchers report the
results of the sample, they must be careful to provide ample detail about the creation
11
All Rights Reserved @ 2017 Shelia M. Kennison
and composition of the sample, the nature of the bias, and how the interpretation of the
results is limited by the sample’s bias.
The best strategy for obtaining a representative samples by obtaining a simple
random sample. A simple random sample is achieved when each person in the
research population has an equal chance of being selected for the research sample.
This can be achieved by placing the names of all the members of the population in a
hat, shaking vigorously, and then selecting at random those that will be include in the
sample. Unfortunately, constructing a simple random sample is rarely possible in
studies involving human research participants, because it is typically impossible to
identify all the members of the research population by name. In order to ensure that
each member of the research population has an equal chance to be selected for the
sample, one would have to throw all their names into the hat. For large populations, not
only may one having trouble finding a hat that is large enough, but more importantly,
one may have trouble obtaining a list of names of everyone in the research population.
If one cannot put all the names in the hat, one cannot ensure that every member of the
research population has an equal chance of being selected for the research sample.
In psychology research, the most frequently used type of sample is the
convenience sample, which involves recruiting and testing participants who meet an
eligibility criterion. Participants are recruited from locations convenient to the
researcher. Participants may be recruited through flyers posted in public locations,
face-to-face appeals, as well as other forms of announcements on television, radio, or
the Internet. The participants must possess the characteristics of interest to the
researcher, which would be used to define the research population. In convenience
12
All Rights Reserved @ 2017 Shelia M. Kennison
samples, these characteristics serve as the eligibility criteria for recruitment into the
study. The convenience sample is the most commonly used sample in research. They
are used in even the most rigorous experiments involving human research participants.
For example, RCTs that are conducted by the National Institutes of Health utilize
convenience samples. Announcements of current trials are listed at
www.clinicaltrials.gov (NIH, 2011). Nevertheless, whenever convenience samples are
used, there is the possibility of obtaining a biased sample. When researchers use
convenience samples, they must be careful to evaluate the representativeness of the
sample and discuss the possibility of bias in the interpretation of the results.
There are some research designs whose external validity is always low. For
example, in a case study, a researcher may focus on an individual person. Case
studies are commonly used to study the functioning of individuals with illnesses,
disorders, or injuries. If a researcher can document that a particular therapy or
treatment produced a positive outcome for an individual, there is the implication that
others with a similar illness, disorder, or injury may also benefit from the therapy or
treatment. In order to show that the results of the case study do generalize to other,
similar individuals, a researcher may carry out a series of case studies. Many therapies
or treatments that start out being tested in case studies ultimately come to be tested in a
RCT involving a large sample of participants. In a case study, it is impossible for the
researcher to determine whether the results are indicative of what could be observed in
others or whether the results reflect idiosyncrasies of the individual.
13
All Rights Reserved @ 2017 Shelia M. Kennison
Measuring Psychological Variables
After a researcher selects the research design and decides how the sample will
be constructed, the next step in the research process is to decide the best way to
measure the variables of interest. Researchers use the term operational definition to
refer to the detailed description of how a variable is measured a study. Researchers
typically include details of their operational definitions in the methods sections of their
research reports. Doing so facilitates the replication of research, which is the
repeating of research studies for the purposes of determining whether similar results
can be obtained. Research results that cannot be replicated may be inaccurate. Only
after multiple replications can one be certain that a particular result is accurate and
reflects the true state of affairs in the world.. It is often said that science is self-
correcting, because it involves the multiple replication of important research findings
and the continual refining of research explanations. If a result or an interpretation of a
result is in error, future replications of the study are likely to improve upon the previous
error.
For many psychological variables, there are ways to measure them. For
example, when a researcher wants to assess the success of an individual, one might
use financial indicators of success, such as one’s annual income, or psychological
aspects of success, such as one’s life satisfaction, or even others. The beginning
researcher should consider how the variables of interest have been studied in previous
studies. For topics that have been studied many times before, it is likely that
researchers have developed some tried and true operational definitions. By using
14
All Rights Reserved @ 2017 Shelia M. Kennison
operational definitions that have been used in prior research, researchers also will find
that is it easier to compare their results with those obtained by others.
In most research involving human research participants, at least some of the
information collected from participant is reported in either a verbal or written form. Such
self-report procedures are referred to as surveys. One type of survey is a
questionnaire, which is a list of questions to which individuals respond in writing. It is
common for questions to be formulated so that participants report a response level
using a scale. For example, one might be asked to rate their satisfaction with life on a
scale from 1-to-7 with 1 corresponding to not satisfied at all and 7 corresponding to
extremely satisfied. Such questions are called Likert-type scales for the psychologist
Rensis Likert (Likert, 1932).
In addition to surveys, psychologists use a wide variety of other techniques to
measure psychological variables. In some studies, researchers might measure how
quickly a participant can respond to a stimulus, such as word or picture displayed on a
computer screen. In such studies, a participant may be asked to make a judgment by
pressing particular key on a keyboard. The time that the participants take to press the
key is recorded. The amount of time that a task takes to perform can be useful
information to a researcher who is attempting to understand what steps of processing
are occurring when one makes a judgment. Psychology researchers might also
measure physiological responses, such as heart rate, skin temperature or moisture as
well as electrical activity produced by muscle movements or brain activities.
When choosing how to measure the variables in a study, researchers must be
careful to make sure that the measures both valid and reliable. A measure’s validity
15
All Rights Reserved @ 2017 Shelia M. Kennison
refers to how well the operational definition of a variable accurately captures the
concept of interest. For example, imagine that a researcher is interested in measuring
the anxiety level of students in a class. The researcher finds a pretty short
questionnaire that was developed by researchers at Oxford University in English. The
questionnaire asks a series of questions about satisfaction with life, school, family
relationships, and various other aspects of daily life. Using the questionnaire as a
measure of anxiety would not be valid. Satisfaction and anxiety are different concepts.
The reliability of a measure refers to the extent to which similar results can be obtained
when the measure is carried out multiple times. In some instances, a researcher may
rely on human coders to make judgments about events or objects. It is important that
that different human coders use the same criteria when providing ratings. Interrater
reliability refers to how consistent multiple raters are in judging the same events or
objects. For example, raters may be asked rate how long an infant looks at an object
during a testing session. Each rater would complete their judgments individually. Then,
the raters’ judgments would be compared. Interrater reliability could be computed by
counting the number of times the raters agreed and dividing the number by the number
of times that they could have agreed.
Researchers who gather responses from participants or make observations of
behaviors must keep in mind that participants sometimes change their responses or
behavior because they are aware that they are being studied. The term social
desirability bias refers to the tendency of human research participants to respond or to
behave in a way that they perceive to be consistent with social norms. A person may
not want to report that they drink beer every day, starting early in the morning, even if it
16
All Rights Reserved @ 2017 Shelia M. Kennison
is true. They may not want to report that they cheat on their tax returns or dislike their
bosses. In other circumstances, participants may guess the purpose of the study and
may respond in a way that they perceive would be helpful to the researcher. A person
may guess that a taste test is being conducted by a particular company and then
provide very high ratings each time they are asked to sample on the company’s
products. Despite the participants good intentions, this type of responding prevents the
researcher from obtaining an accurate picture of the phenomenon. Researchers try to
prevent participants from guessing the purpose of the study. Demand characteristics
refer to any aspect of the research procedure that might enable the participant to guess
the purpose of the study. Researchers may carry out piloting testing or practice
sessions to run through the research procedures after which they make improvements
to the procedures before actually conducting the study.
Statistical Analysis of Data
One may think that most of the hard work in the research project is over after a
researcher finds a research hypothesis, selects the research design, decides how the
sample will be constructed, operationally defines the variables. Actually, only then is
the important work beginning. The next step in the research process is to analyze and
to interpret the results. Researchers use statistics to organize and to summarize
numerical data. Researchers have two types of statistics at their disposal. Descriptive
statistics involve using only the data from the sample to make statements about the
sample itself. The data are not used to make predictions. When researchers do want
17
All Rights Reserved @ 2017 Shelia M. Kennison
to use the results to make predictions, as is the case with most psychological research,
they use inferential statistics.
Descriptive Statistics
When describing samples, one is typically most interested in the central
tendency or the score or observation that is most typical of the entire sample. For
example, imagine that you have a summer job at the local YMCA. Your boss would like
you to collect some data about how many different activities the children and parents in
the community would like the YMCA to offer. You devise a short survey in which you
ask people to list make a recommendation. You gather your survey responses to show
your boss. Your boss asks you “What did you find?” As you begin to describe all the
different responses that you received, the boss stops you and says, “What activity do
most people want?” The boss is asking you to report the central tendency. There are
three common measures of central tendency. The mode is the most frequently
observed response. The mode is useful for both data measured on a numerical scale
as well as data that are not numeric, such as responses to the question “What is your
favorite activity at the YMCA?” In fact, the mode is the only measure of central
tendency that can be used with non-numerical data. For numerical data, one can also
use the median, which is the middlemost score or the mean, which is the arithmetic
average.
When determining the central tendency of a sample of numerical data, it is useful
to make a graphical representation of the distribution or all of the observations in the
dataset. On the horizontal or x-axis, one displays the different types of responses. On
18
All Rights Reserved @ 2017 Shelia M. Kennison
the vertical or y-axis, one displays the frequency or how many responses were
observed. Sometimes, the graphs are displayed with bars representing the frequency
of each response category. Sometimes, only a smooth curve is used to indicate the
frequency of each category. For the bar graph, the highest bar corresponds to the
mode. For the smoothed curve graph, the response category that is below the highest
part of the curve is the mode. When the mode, median, and mean are the same, then
determining the central tendency of the dataset is easy. This occurs when the data
follow a perfect normal distribution or bell curve. A normal distribution is symmetrical
and has an equal number of scores below and above the median, mean, and mode. In
contrast, a skewed distribution is not symmetrical. There are either more scores
above the median, which occurs in a negatively skewed distribution, or more scores
below the median, which occurs in a positively skewed distribution. A distribution is
skewed because there are a relatively small number of scores that are unusual in
comparison with the rest of the dataset. Such scores are called outliers. For skewed
distributions, determining the central tendency is less straightforward, as it is usually the
case that the mode, median, and mean will differ. The median is recommended as the
best measure of central tendency for any skewed distribution, because is affected less
by the presence of outliers than is the mean.
When describing a dataset, one also pays attention to the sample’s variability,
which refers to how much the observations in a sample are different from one another.
A useful measure of variability is the range of a distribution, which refers to the size of
the difference between the highest score in the distribution and the lowest score in the
distribution. When a distribution has a larger range than another, the distribution has
19
All Rights Reserved @ 2017 Shelia M. Kennison
higher variability. Most often, researchers assess the variability of a distribution using
the standard deviation, which reflects how spread out the sample is relative to the
sample mean. When computing the standard deviation, one measures how far each
score is from the mean and then finds that average. Samples with low standard
deviations have scores that are more similar to the mean that samples with higher
standard deviations.
Researchers who want to describe the relationships among variables in a sample
can use descriptive statistics to quantify the relationship that exists between variables.
The most commonly used statistic is Pearson’s r (Gravettner & Wallnau, 2009). The
statistic captures the degree to two variables share a linear relationship. If two variables
are not related in a linear way, the value of r will be 0. When the two variables are
perfectly related in a linear way, the value of r is either +1.00 or -1.00. When two
variables are perfectly correlated and their values increase and decrease together, they
have an r of +1.00. When there is a perfect correlation and the values of one variable
increases as the values of the other variable decreases, the value of r is -1.00.
Correlations with r values close to ±1.00 are stronger than correlations with values close
to 0.
When two variables are correlated, it is possible to think of the relationship as
indicating how much of the variability observed in one of the two variables is explained
by the other variable. Consider the relationship that has been found to exist between
happiness and income. Research has shown that the value for r = .18 (Hagerty, 2000).
One might wonder what percentage of the variability in happiness is explained by
income. One can compute this using the r value, squaring it, and multiplying by 100.
20
All Rights Reserved @ 2017 Shelia M. Kennison
According to this formula, only 3.2% of happiness is explained by income, leaving
96.8% of happiness level unexplained. The take home point here is that the stronger
the correlation, the more one variable can account for variance observed in the other
variable. For variables that are strongly correlated, such as those with an r = +.80,
there is 64% in one variable that can be explained by the other and 35% of the variance
unexplained. Variables that have an r = +.95, 90% of the variance can be explained,
and 10% remains unexplained. Still, one must always be mindful that the relationship
cannot be assumed to be causal.
Inferential Statistics
When researchers use data from a research sample to make a prediction about a
research population, they use inferential statistics. Work conducted by mathematicians
over 300 years ago laid the foundation for the inferential statistics that researchers use
today. The mathematicians from the distant and not-so-distant past worked hard to
uncover some remarkable regularities that occur when samples are drawn from a
population. At the heart of this work is the concept of probability, which refers to the
likelihood that an event will occur. Probability is most easily explained in terms of
random events, such as flipping a coin, rolling dice, or drawing a card from a well-
shuffled deck. One can estimate the likelihood of an event by divided the number of
possible times that event could be observed by the total number of possible outcomes.
So, when a coin is flipped, the probability of a heads is one heads in every two flips or
1/2. The probability of observing a tails is also ½. If a single six-sided die is rolled, there
are six possible outcomes (i.e., 1, 2, 3, 4, 5, or 6). So the chance of observing any one
of these outcomes on a roll is 1/6. The chance of rolling an even number is 3/6,
21
All Rights Reserved @ 2017 Shelia M. Kennison
because there is there are 3 ways in which that event could be observed. The chance
of rolling a 4 or a 5 is 2/6.
This simple rule of probability can also be used to predict what selections are
likely when sampling. Imagine that a teacher has a class of 40 students, 15 of them
male and 25 of them female. If she places all their names in a hat to draw out one a
random, she has a 15/20 chance of drawing out the name of a male student and a
25/20 chance of drawing out the name of a female student. Now imagine that you know
a little more about the class, specifically that it contains 10 freshmen, 10 sophomores,
10 juniors and 10 seniors. Of the freshmen, 1 is a male and 9 are female. Of the
sophomores, 4 are male and 6 are female. Of the juniors, all are male, and of the
seniors, all are female. What is the probability of drawing out a name that is a junior?
That would be 10/40 or ¼. What is the probability of drawing out a name that is male?
That would still be 15/40? What is the probability of drawing out one name at random
that is a freshman male? That would be 1/40.
Relying on the regularity of the laws of probability, statisticians have
demonstrated that large samples are always more representative of the population from
which they are drawn than are small samples. This fact is referred to as the law of
large numbers. Small samples have a greater chance of being influenced by outliers
than large samples. On each draw from the population, there is a greater chance of
typical examples of the population than unusual examples in the population. When one
draws from the population many times as happens when the sample is large, there are
more opportunities for typical scores to be observed than when one draws just a few
22
All Rights Reserved @ 2017 Shelia M. Kennison
times from the population. Small samples are more vulnerable to outliers have a big
impact on the sample characteristics, such as the mean and standard deviation.
Mathematicians have been intrigued by probability and sampling since the
1600s. In early 1700s, the mathematician Abraham de Moivre (1667-1754) discovered
that there is a remarkable regularity that occurs when one samples many, many times
from a population. When one keeps track of the samples, one observes something
quite amazing the distribution of sample characteristics (i.e., means, median,
standard deviations) always approximates a normal distribution. This is true even when
the population that is being sampled from is not itself a normal distribution (Gravettner &
Wallnau, 2009). This discovery is called the Central Limit Theorem. Without the
central limit theorem, scientists would not have the ability that they have to use the
results from their research samples to make predictions about research populations.
Research into the regularity of samples further found that the sample size mattered.
For very large samples, such as samples that are infinitely large, then the distribution
would be a perfect normal distribution. For smaller sample sizes, the distribution was
found to deviate systematically from a perfect normal distribution as the sample size
decreased. These demonstrations allowed statisticians in the 20th century to develop
the inferential statistics that we have today, such as the t-test and analysis of variance
(ANOVA).
Prior to the discovery of the central limit theory, there was another discovery
whose existence makes possible modern day inferential statistics. This discovery is
that normal distributions are special. Most measurements taken of the physical word,
including human characteristics and measurements of human behavior, tend to follow a
23
All Rights Reserved @ 2017 Shelia M. Kennison
normal distribution. Furthermore, it happens to be the case that all normal distributions
can be described by the 68-95-99.7 rule. In all normal distributions, one will observe
that 68% of the entire dataset will be within one standard deviation of the mean of the
distribution, which means that 34% of the dataset will lie between the mean and one
standard deviation above the mean and 34% of the dataset will lie between the mean
and one standard deviation below the mean. Further, 95% of the entire dataset will be
within two standard deviations of the mean of the distribution, which means that 47.5%
of the dataset will lie between the mean and two standard deviations above the mean
and 47.5% of the dataset will lie between the mean and two standard deviations below
the mean. Go out three standard deviations away from the mean and you will always
find 99.7% of the entire distribution.
Consider men and women’s heights in the United States. Research has shown
that measurements of height follow a normal distribution. The average height for U.S.
men is 69.5 inches and the standard deviation is 3 inches, and the average height for
U.S. women is 64 inches with a standard deviation of 2.5 inches (National Health
Statistics Reports, 2008). Having information about the mean and standard deviation
allows us to extrapolate that 68% of men in the U.S. are between 65.5 and 72.5 inches
tall, 95% are between 63.5 and 75.5 inches tall, and 99.7% are between 60.5 and 78.5
inches tall. Sixty-eight percent of women in the U.S. are between 61.5 and 66.5 inches
tall, 95% are between 59 and 69 inches tall, and 99.7% are between 56.5 and 71.5
inches tall. With the foundational knowledge of the central limit theory and the regularity
of the normal distribution, modern day researchers can collect a reasonably large
sample from a population whose characteristics are unknown and be about 95% sure
24
All Rights Reserved @ 2017 Shelia M. Kennison
that their sample mean is within two standard deviations of the true mean of the
population. This range of values that contains the true population mean is a
confidence interval. For a confidence level of 68%, the range would be the sample
mean plus or minus one standard deviation. For a confidence level of 99.7%, the range
would be the sample mean plus or minus three standard deviations.
Statistical Significance
A phrase that might already be familiar is the phrase “Is it significant?” This is
what researchers really want to know. For example, in an experiment with an
experimental condition and a placebo control condition, the researcher wants to know
whether any observed difference between the two groups is significant. The
mathematics and statistical knowledge that enables researchers to estimate unknown
population values from values observed in a sample also enables researchers to carry
out a procedure known as significance testing and to determine whether an observed
result is significant or is likely to reflect a true difference that could be observed again
and again in future experiments.
The significant testing procedure usually strikes the beginning student as rather
unintuitive. The researcher always begins by formulating a null hypothesis, which
states that the there is no difference of the type that the researcher believes might
occur. In our example, there is no difference for the IV measurement for those who took
the experimental drug and those who took the placebo. Then the researcher states the
alternative hypothesis, which states that there is a difference of the type that the
researcher believes might occur. There is a difference between the experimental and
25
All Rights Reserved @ 2017 Shelia M. Kennison
placebo groups. Researchers must approach significance testing in this way because
the statistical tools available do not enable researchers to determine whether the
alternative hypothesis is true; rather, researchers can only determine whether it is likely
or unlikely for the null hypothesis to be true. Only when the probability is very low that
the null hypothesis is true can one decide to reject the null hypothesis in favor of the
alternative hypothesis.
Researchers can choose the level of certainty that they use when evaluating the
null hypothesis. The term alpha level is used to refer this criterion. The most
commonly used alpha level is .05, which means that in order for the null hypothesis to
be rejected, the observed sample must be among the most extreme 5% of samples that
are possible in the population when no manipulation has been applied or no effect
would be expected. Alpha levels of .01 and .001 are also routinely used. For these
alpha levels, in order for the null hypothesis to be rejected, the observed sample must
be so unusual that they are among the most extreme 1% or .1% samples in the natural
variation of that population when no manipulation has been applied or no effect would
be expected. For an observed sample, a researcher can estimate the likelihood of that
such a sample occurs in the population in which no manipulation has been applied or no
effect is expected. This estimate likelihood is called the probability value or p-value.
Consequently, when p-values are lower than .05, .01, or .001, the null hypothesis can
be rejected.
Whether the researcher rejects the null hypothesis or accepts it, the researcher
may be incorrect. Two types of errors are routinely possible when one engages in
hypothesis testing Type I errors and Type II errors. When a Type I error is made, the
26
All Rights Reserved @ 2017 Shelia M. Kennison
researcher concludes that a real difference was observed in data, but the difference is
does not exist. The result would be unlikely to be observed in a replication. When a
Type II error is made, the researcher concludes that no difference was observed in the
data, but the difference does exist. The likelihood of making a Type I error is directly
related the alpha level. Type I errors are more likely to be discovered in future research
than Type II errors, because research journals are more likely to publish significant
results. When a Type II error occurs, a non significant result is observed. The
likelihood of a Type II error is directly related to the statistical power of a study, which
refers the study’s ability to detect a difference or effect, assuming that it exists. Studies
that have been carried out with careful methods and with adequate numbers of
participants have higher statistical power than studies carried out carelessly with small
samples. Studies with adequate statistical power are less likely to result in Type II
errors than those with inadequate statistical power. Researchers rely on power
calculations to determine the number of participants that are needed in a sample to
observe an effect.
Practical Significance
Researchers are concerned not only with the statistical significance of their data,
but also with the practical significance of their data. The researcher should consider the
extent to which the result that was observed in a sample is likely to translate into a
meaning difference in everyday life. When evaluating the practical significance of any
research result, one wants to know what does the observed difference mean to me in
my life or in my work. Unfortunately, reports of research studies may not do a very
27
All Rights Reserved @ 2017 Shelia M. Kennison
good job of explaining the practical implications of the results. We all must keep in mind
that a statistically significant difference in a research study does not always translate
into an important difference that one can experience in everyday life. The term
ecological validity is used to describe the extent to which a research finding reflects a
real life situation.
A numerical way for researchers to discuss the practical implications of their
results is to discuss the extent to which the statistical difference that was observed
involves a large, medium or small effect size. Consider an example of a study that
tested the effectiveness of a drug to reduce anxiety. It is possible for a researcher to
observe a statistically significant reduction in anxiety due to the drug when the reduction
is relatively small. Consumers may want to know if the reduction in anxiety that is
expected from the drug worth the possibly high cost. Researchers calculate the size of
their effects using a statistic known as Cohen’s d (Cohen, 1992). Cohen’s d is equal to
the observed difference divided by the standard deviation. Large effect sizes have
values for d of .80 and greater. Medium effect sizes have values for d around .50.
Small effect sizes have values for d of .20 and smaller. In some cases, researchers may
investigate the effect sizes observed across similar studies in order to investigate the
range of circumstances in which the effect is observed and the typical effect size. Such
an investigation is called a meta-analysis.
Frequently, studies in which different groups of people are compared find
statistically significant differences in the performances of the groups. One fact that
seems to get lost in the reporting of the results is that when a difference is observed
between the average performances of two groups of people, it is not the case that every
28
All Rights Reserved @ 2017 Shelia M. Kennison
member of one group will always out score every member of the other group. Consider
the differences observed between men and women’s heights. When one examines that
tails of the distributions, one finds that among the shortest people, there are more
women than men and among the tallest people, there are more men than women.
However, there are many women who are as tall as or taller than the average man.
Conversely, there are many men who are as short as or shorter than the average
woman. When one knows only there is a difference between two groups, it is not the
whole story.
Consider the differences that have been found between the math performance
for men and women (Halpern, 2000). On standardized tests, the average performance
has been found to be higher for men than for men. In the case of math performance
and the difference between men and women, the distributions of performance are
extremely overlapping. There are many women who perform equal to or above the
average performance of men. Conversely, there are many men who perform equal to
or below the average performance of women. Further, when one examines that tails of
the distributions, one finds the lower tails identical, suggesting that there are
comparable percentages of men and women who perform very poorly in math. The
upper tails differ, indicating that the extremely small percentage of the population of
men who are math geniuses is somewhat larger than the percentage of the population
of women who are math geniuses.
When evaluating the practical significance of group differences, one must always
consider the entire distribution of scores, rather than just the group means. When one
does this, one finds that the variability that occurring within each group is far greater
29
All Rights Reserved @ 2017 Shelia M. Kennison
than the variability that exists between the two groups. Stated in another way, in math
performance, women are far more different from each other and men are far more
different from each other than the two groups are. Because within group differences are
always bigger than between group differences, one should be extremely cautious about
predicting an individual’s likely performance, based on the existence of a performance
difference involving groups. If one is parenting a son or daughter, one cannot ever be
sure whether the child will be similar to most other boys or girls on the measured
dimension or whether the child will be an outlier for his or her group.
Our discussion of the practical implications of research reporting group
differences shines a light on a fact that easily goes unnoticed in most discussions of
psychological research. It is important to remember that there is no methodology in
existence that can predict with accuracy any outcome for any single individual.
Statistical procedures provide a useful framework for researchers to make predictions
about outcomes for large groups of individuals, particularly when using data obtained
from large samples. For example, imagine that researchers have investigated how well
new drug reduces improves memory. It was found that a sample of 100 people who
received the drug remembered 20 percent more on a memory test than a sample of 100
people who received a placebo. The researcher predicts that on average, those taking
the drug in the future will remember 20 percent more than they do normally when not
taking the drug. However, it will be the case that some may experience smaller
improvements in memory and some may experience larger improvements in memory.
The average improvement is expected to be 20%. Someone who takes the drug and
30
All Rights Reserved @ 2017 Shelia M. Kennison
does not experience any improvement is possible. The event would be unexpected, but
an example of an outlier.
Protecting Human Research Participants
An essential part of the research process is ensuring that the research is carried
out in an ethical manner. In the United States, research funded by the government is
regulated by Part 46 of Title 45 Code of Federal Regulations or 45 CFR 46. This
regulation is referred to as the Common Rule. One of the requirements specified by
the common rule is that institutions receiving federal research grants involving human
research participants establish Institutional Review Boards (IRBs) that review all
research on ethical grounds prior to the research’s implementation. IRBs are composed
of scientists and non-scientists as well as members of the community in which the
institution is located. The members of the IRB are charged with the task of ensuring
that research participants are not harmed through their participation in research and that
they are ethically treated at every stage of the research process.
History of Research Ethics
The modern procedures involved in the ethical conduct of research have evolved
over the last half century. Following World War II, the public became aware of the need
for governments to ensure that scientific research is ethically conducted. In 1945 and
1946, the Nuremburg trials documented the Nazis experimentation on prisoners (Tusa
& Tusa, 2010). After the trials, the tribunal produced the Nuremburg Code, which
provides guidelines for the ethical treatment of research participants. These guidelines
are displayed in Table 2.1. Among the most infamous of the Nazi doctors was Josef
Mengele (1911-1979), also known as the angel of death. He performed numerous
31
All Rights Reserved @ 2017 Shelia M. Kennison
Table 2.1 Nuremberg Code (1947)
1. The voluntary consent of the human subject is absolutely essential.
2. The experiment should be such as to yield fruitful results for the
good of society, unprocurable by other methods or means of study, and
not random and unnecessary in nature.
3. The experiment should be so designed and based on the results of
animal experimentation and a knowledge of the natural history of the
disease or other problem under study that the anticipated results will
justify the performance of the experiment.
4. The experiment should be so conducted as to avoid all unnecessary
physical and mental suffering and injury.
5. No experiment should be conducted where there is an a priori
reason to believe that death or disabling injury will occur; except,
perhaps, in those experiments where the experimental physicians also
serve as subjects.
6. The degree of risk to be taken should never exceed that determined
by the humanitarian importance of the problem to be solved by the
experiment.
7. Proper preparations should be made and adequate facilities
provided to protect the experimental subject against even remote
possibilities of injury, disability, or death.
8. The experiment should be conducted only by scientifically qualified
persons. The highest degree of skill and care should be required
through all stages of the experiment of those who conduct or engage in
the experiment.
9. During the course of the experiment the human subject should be at
liberty to bring the experiment to an end if he has reached the physical
or mental state where continuation of the experiment seems to him to
be impossible.
10. During the course of the experiment the scientist in charge must be
prepared to terminate the experiment at any stage, if he has probably
cause to believe, in the exercise of the good faith, superior skill and
careful judgment required of him that a continuation of the experiment
is likely to result in injury, disability, or death to the experimental
subject.
32
All Rights Reserved @ 2017 Shelia M. Kennison
horrific surgeries and procedures on prisoners at Auschwitz. He focused particularly on
twins, pregnant women, and individuals with physical deformities (Ware & Posner,
1986). The dissection of living people without anesthesia was common. Mengele was
not among the 23 Nazi doctors who were placed on trial at Nuremberg. He died in
Paraguay in 1979. His identity was not confirmed by DNA analysis until 1992.
In 1964, the World Medical Association published the Declaration of Helsinki,
which was an ethics code for medical research. In 1974, the Congress of the United
States created a National Commission for the Protection of Human Subjects in
Biomedical and Behavioral Research. The committee met in Eskridge, Maryland at the
Belmont Conference Center, and their report was called the Belmont Report. The
report set forth three core principles governing the ethical treatment of human research
participants: a) respect for persons; b) beneficence; and c) justice. Each of these
principles is today reflected in the procedures that researchers are required to follow
when conducting research.
Respect for persons refers to “protecting the autonomy of all people and treating
them with courtesy and respect.” Participants should be volunteers. Their participation
should not be forced or coerced in any way. In order for participants to exercise their
right to volunteer, there must be a process of informed consent, during which the
participant is made aware of the nature of the research and the procedures that will be
performed in the research study. In research involving research participants who are
younger than 18 years of age, parental consent must be obtained. When parental
consent is obtained, the researcher must then formally invite the child to participate and
provide information about the nature of the study in a process that is referred to as
33
All Rights Reserved @ 2017 Shelia M. Kennison
assent. All research participants, whether they are adults or children, should be
allowed to withdraw from the research study at anytime without penalty. In studies
involving deception, researchers must provide debriefing, during which participants are
informed that deception was used.
Beneficence refers to “the philosophy of do no harm while maximizing benefits
for the research project and minimizing risks to the research subjects.” Harms may be
physical as well as non-physical, such as psychological stress. Researchers should
strive to maximum benefits of the research while at the same time, minimizing any
physical or psychological harm. In circumstances in which there is no clear benefit
expected from a research study, the research study would typically not be approved by
an IRB. Participants’ time and energy should not be wasted for research that does not
stand to produce some actual benefit to society.
Justice refers to ensuring reasonable, non-exploitative, and well-considered
procedures are administered fairly the fair distribution of costs and benefits to
potential research participants and equally.” Participants who experience the burden
of the research procedures should also benefit from the products of the research. For
example, imagine that a community of individuals living near marshes is recruited to
participate in a new study to test a vaccine for malaria. Malaria is a disease transmitted
through the bite of a mosquito (Packard, 2007). It is a particularly deadly disease, killing
100 million people per year and making twice as many sick. Those who survive it,
suffer lifelong problems. Marshes are one of the favorite breeding grounds for
mosquitoes. The study is carried out and found to be effective in preventing malaria.
The researchers conducting the study leave the community and work toward bring the
34
All Rights Reserved @ 2017 Shelia M. Kennison
drug to the market. If it is the case that those who participated in the study will not
themselves have an opportunity to benefit from the fruits of the research, then one could
say that they endured the cost of the research as a participant but did not benefit. This
would violate the principle of justice. If those who participate in research are ensured
that they can also one day benefit from the future results of the study, should any
benefit occur, then the principle of justice would be satisfied.
Examples of Unethical Research
Modern day research ethics has been greatly informed by past examples of
unethical research. One of the most shocking examples of unethical research was
carried out by the United States government from 1940-1970. It has come to be known
as the Tuskegee Syphilis Study (Jones, 1981). The Department of Health and Human
Services directed the study in Tuskegee, Alabama from 1932 to 1972. Syphilis is a
contagious, sexually-transmitted disease caused by bacteria. If left untreated, syphilis
can cause blindness, brain damage, and death. Syphilis can cause miscarriages in
women and premature births of infants. Children born with syphilis may have
deformities, developmental delays, and seizures. In 1932, the United States Public
Health Service enrolled approximately 600 African-American men living in Macon
County, Alabama in the study. Of the 600, 399 had been diagnosed with syphilis and
were monitored for the purposes of determine the progression of the disease; 201 did
not have the disease. The 399 men who had syphilis were not told that they had it.
They were never told that they were enrolled in a research study. By 1947, penicillin
was known to be the cure for syphilis. The men were not given the treatment.
Furthermore, steps were taken to prevent the men from seeking medical treatment
35
All Rights Reserved @ 2017 Shelia M. Kennison
elsewhere, where they might have been informed about their disease and also cured of
it. When the study was halted in 1972, only 74 of the 399 men with syphilis were still
alive. Twenty-28 men had died of syphilis. One hundred others died of complications
directly related to syphilis. Forty of the men had wives who had also contracted the
disease. Nineteen children are known to have been born with syphilis. In 1997, the
United States government formally apologized to the survivors and the families of those
harmed in the study. The film and play Miss Evers Boys (1997) tells the story of
Tuskegee by focusing on Miss Evers a nurse who was hired to recruit men for the
study.
Unfortunately, the Tuskegee Syphilis study is not the only example of unethical
research conducted with human research participants in the 20th century. In the
decades following the invention of the atomic bomb, countless men, women, and
children living in the United States and Canada were exposed to radiation for research
purposes without their knowledge or consent. These studies have been referred to as
the human radiation experiments (Welsome, 1999; Jones, 2005). In 1994, then
President Bill Clinton called for an investigation into these studies. In 1995, the final
report was presented to Congress and is now available on the World Wide Web
(Department of Defense, 2011). The report was approximately 1000 pages long and
detailed thousands of experiments carried out between 1944 and 1994, including the
injection of radioactive substances into infants and pregnant women, the placement of
radioactive subjects in the daily meals of children who were housed in schools for those
with intellectual disabilities, and the exposure of members of the military as well as
incarcerated prisoners to high levels of radiation.
36
All Rights Reserved @ 2017 Shelia M. Kennison
Research Misconduct
The maltreatment of research participants is just one way in which researchers
may breach codes of research ethics. Anytime researchers violate the ethical code of
conduct, they are described as engaging in research misconduct. A familiar example
of research misconduct is plagiarism, which occurs when an author uses the verbatim
words of another with acknowledgement and proper citation. Another example of
research misconduct is data fabrication, which occurs when a researcher states that
data were collected when they were not; rather, the researcher fakes the data. A recent
example of data fabrication is the research reported by Dr. Andrew Wakefield (born
1957), a surgeon and medical researcher, who proposed a link between childhood
vaccines and autism (Godlee, Smith, & Marcovitch, 2011). After many years of
controversy surrounding the vaccines caused autism, it was found that the data
supporting the claim had been fabricated. A second example of data fabrication comes
from the early days of psychology. Cyril Burt (1883-1971) published many studies on
the inheritance of intelligence, from studies involving identical twins (Fletcher, 1991).
The fraud was only discovered after his death.
There are penalties for those who engage in research misconduct. Individuals
may lose their positions, depending on their employment contracts. Those who
plagiarize and profit from the work may be sued for damages by the individual whose
work was stolen. Those who commit research misconduct while working on a federally
funded research project can be barred from receiving future research grants for either a
number of years or for life (Altman & Hernon, 1997). Typically, one would be barred
37
All Rights Reserved @ 2017 Shelia M. Kennison
from receiving future research grants only after being found guilty of research
misconduct more than once.
Professional Codes of Ethics
The American Psychological Association has a detailed code of ethics. The APA
code covers both the ethical issues involving research and professional practice. APA
formed its first committee on the ethical standards for psychologists in 1947 (Hobbs,
1948). The first ethics code was published in 1953 and has been revised nine times,
most recently in 2009 (APA, 2011). The code provides guidelines for every aspect of a
psychologist’s professional life. For psychologists involve in delivering treatments in a
clinical setting, the ethics code cautious against having dual relationships with clients.
Psychologists should refrain from having any type of personal relationship with a client
away from the clinic. The current code also provides recommendations regarding the
treatment of human research participants as well as animal subjects. The code
emphasizes the responsibility that researchers have in ensuring that the results of the
research are not used by a third party to bring harm to others. The code also states that
researchers have an obligation to share their research data with others.
The APA Ethics Committee is receives and reviews allegations of unethical
conduct by psychologists (APA, 2011b). In the event that the committee finds evidence
of unethical conduct, there are a number of different sanctions that can be issued. The
psychologist in question may be reprimanded if it was found that there was a breach of
ethics, but that no one was harmed and the professional was also not negatively
affected. A censure can be issued if there was some level of harm to an individual and
to the profession. Typically, the level of harm involved in a censure is not extreme. In
38
All Rights Reserved @ 2017 Shelia M. Kennison
cases of extreme harm to an individual or to the profession, an expulsion is issued. In
cases of expulsion, one loses membership to the APA. It may be the case that when an
expulsion from APA has occurred, the offending member has already lost their license
to practice psychology at the state level and may have lost their membership privileges
at the state level. The most frequently cited reason for expulsion involves dual
relationships (Phelan, 2007).
Key Terms
1. 68-95-99.7 Rule
2. Alpha Level
3. Alternative
Hypothesis
4. Bell Curve
5. Belmont Report
6. Beneficence
7. Biased Sample
8. Case Study
9. Central Limit
Theorem
10. Central Tendency
11. Common Rule
12. Confidence Interval
13. Control
14. Convenience
Sampling
15. Correlational
Design
16. Data Fabrication
17. Debriefing
18. Demand
Characteristics
19. Dependent Variable
20. Descriptive
Statistics
21. Distribution
22. Double-Blind
Design
23. Ecological Validity
24. Effect Size
25. Experimenter Bias
26. External Validity
27. Human Radiation
Experiments
28. Independent
Variable
29. Inferential Statistics
30. Informed Consent
31. Institutional Review
Board
32. Internal Validity
33. Interrater Reliability
34. Justice
39
All Rights Reserved @ 2017 Shelia M. Kennison
35. Law Of Large
Numbers
36. Likert-Type Scale
37. Mean
38. Median
39. Meta-Analysis
40. Mode
41. Naturalistic
Observation
42. Negatively Skewed
43. Null Hypothesis
44. Nuremburg Code
45. Normal Distribution
46. Operational
Definition
47. Outlier
48. Pilot Testing
49. Probability
50. P-Value
51. Placebo Effect
52. Plagiarism
53. Positively Skewed
54. Power
55. Questionnaire
56. Randomized
Clinical Trial (RCT)
57. Random
Assignment
58. Range
59. Reliability
60. Replication
61. Representative
Sample
62. Research
Misconduct
63. Research
Population
64. Research Sample
65. Respect For
Persons
66. Significance Testing
67. Simple Random
Sample
68. Skewed Distribution
69. Social Desirability
Bias
70. Standard Deviation
71. Statistics
72. Subject Variable
73. Survey
74. Quasi Experiment
75. Third Variable
Problem
76. True Experiment
77. Tuskegee Syphilis
Study
78. Type I Error
79. Type Ii Error
80. Validity
81. Variability
Review Questions
40
All Rights Reserved @ 2017 Shelia M. Kennison
1. What is internal validity? What type of methodology has the highest level internal
validity?
2. Compare and contrast a representative sample and a biased sample? Which type of
sample is the best to use for empirical research?
3. What is the difference between descriptive statistics and inferential statistics?
4. Discuss the internal validity and external validity of the methodology referred to as the
naturalistic observation?
5. Discuss the internal validity and external validity of the case study methodology?
6. What are the three most commonly used measures of central tendency?
7. Explain the differences between a skewed distribution and a normal distribution.
8. What is the central limit theorem? How do researchers use information about the
central limit theorem to evaluate samples?
9. What is the 68-96-99.7 rule? Explain how one can use it to estimate what percentage
of cases in a population will have a characteristic of interest?
10. Contrast the circumstances in which a researcher would make Type I and Type II
errors.
11. What is the relationship between a study’s statistical power and sample size?
12. What is the relationship between a study’s statistical power and the chance of a Type II
error occurring?
13. What is meant by the statement “science is self-correcting”?
14. Discuss the extent to which researchers can predict the outcomes of large populations
and the outcomes of individuals.
41
All Rights Reserved @ 2017 Shelia M. Kennison
15. What is the Nuremberg Code? Identify three out of the 10 principles included in the
code.
16. What are the three principles set forth in the Belmont Report?
17. What is informed consent and why is it used in research involving human research
participants?
18. What is debriefing and when is it used in research involving human research
participants?
19. What was the Tuskegee syphilis study? What is its significance in the history of
research involving human research participants?
20. What were the human radiation experiments? What modern principle of research ethics
did they often violate?
... Therefore, the across the board moderating effect of gender on persisting co-sleeping maladjustment effects in favour of females, was highly striking and perplexing. The findings' consistency in statistical significance as well as in ecological validity (Kennison, 2017) suggested strongly that males and females approached persistent co-sleeping practice differently. To explain the gender specific findings, we directed our attention to literature insights into the process of gender development differences. ...
Article
A critical co‐sleeping literature review revealed individualistic and dyadic guided approaches taken insofar, ridden by conflicting results. Thereby, we situated our approach beyond the individual and dyad area where we developed anew a systemic co‐sleeping paradigm, resulting in theoretical and preliminary empirical findings. Initial cross‐gender analyses associated significantly co‐sleeping with Bowen Family Systems Theory's cornerstone constructs. However, once the moderating effect of gender was examined, significance disappeared across the board for females yet persisted for males. Specifically, male‐children time‐persistent co‐sleeping was associated negatively with differentiation and positively with chronic anxiety and other hypothesized maladjustment effects (guilty feelings and abandonment feelings if moved away from parents). Effects drew attention to Bowen's systemic construct of intergenerational emotional fusion. Guided by the empirical associations, we focused on gender development differences literature. We suggest that triangulation processes dynamically embed co‐sleeping within the family systems paradigm, with the embedment appearing to be significantly gendered.
ResearchGate has not been able to resolve any references for this publication.