Module 1
This is a single, concatenated file, suitable for printing or saving as a PDF for offline viewing. Please note that some animations or images may not work.

Module 1: Introduction and Methods in Psychology
Tuesday, May 14 – Sunday, May 19

Required Reading/Viewing:

Principles of Psychology, Chapters 1–2 & Appendix
(Pages 1-6; 29-59; 581-604)

Review APA (American Psychological Association) Ethical Principles (page 1–middle of page 4; section 8–8.09 (middle of page 10–page 11) [Available via apa.org and the Course eReserves.]

Module 1 online content

Discussions:	
Discussion 0: Introduce Yourselves

Module 1 Discussion:

Initial responses due Thursday, May 16, 9:00 AM ET
Two peer response due Sunday, May 19, 9:00 AM ET
Leader response due Tuesday, May 21, 9:00 AM ET
Assignments:	None
Assessments:	
Syllabus Quiz due Thursday, May 16, 5:00 PM ET

Live Classrooms (Faculty):	
There is no Live Classroom this Unit

Activity:	Complete Module 1 Review and Reflect, due Monday, May 20, 11:59 PM ET
Module 1A: A Model for Understanding Psychology


Learning Objectives

By the end of the session students will be able to:

Describe a multi-level framework for considering phenomenon in the field of psychology and give an example.
Describe types of studies in psychology and outline their uses, strengths, and weaknesses.
Describe the difference between correlational (non-experimental) and experimental psychological research and gives examples of each.
Describe the research process – from idea to peer review.
Describe the three principles governing research ethics. Give examples of each.
 

Introduction to the Model for Understanding Psychology

Remember back to the course introductory video that was on the course home page, and see if you can respond to the question below.

What do you think of when you hear the word “psychology”? 

cartoon style cat lying on a couch next to a psychiatrist
I always like this cartoon, as I think it represents a common perception of psychology—a therapeutic endeavor where we seek to understand how early experiences caused folks to have problems (often, how it’s all your parents’ fault!!). Although humorous, it’s a pretty poor idea of what psychology is.

Instead, psychology covers many things, but, at its heart, it is a science of behavior. It shares many principles with other sciences. Psychological science can be basic—when we try to understand how and why individuals behave the way they do. This is really knowledge for the sake of knowledge, and that’s important. For example, we might want to understand how human memory works, what factors are associated with healthy child development, how individuals develop anxiety problems. Psychological science can also be applied—when we try to use that basic knowledge to change (hopefully improve) human behavior. For example, we may want to use what we know about human memory to improve student performance; we may want to use what we know about healthy child development to improve early childhood education or parenting; we may want to use what we know about how anxiety develops to better treat and prevent problems with anxiety.

One of the things that makes psychology so challenging as a field is its tremendous breadth. Psychologists are interested in how the brain works, how social situations impact individual performance, how early experiences influence the later development of mental health problems, and lots more. The sheer range of potential questions is overwhelming. This is what can be referred to as the CONTENT (what we know). This also means that we have a broad range of strategies and methods that are part of psychological research. This is what can be referred to as the PROCESS (how we come to know it). Throughout this course we will focus on some of these strategies and methods.

But we need to start with a framework—a framework for thinking about complex problems in psychology. The model that you see below is an old but still influential one—Uri Bronfenbrenner published this model in 1979. What does it show? It illustrates how we have to consider the individual in context. In his model the individual is in the middle, and they are embedded within other systems. The family, schools, neighborhoods, political systems, social structures all surround this individual. Now Bronfenbrenner, as a sociologist, was really most interested in these larger systems.

Bronfenbrenner Model of Ecological Systems Theory
Bronfenbrenner Model of Ecological Systems TheorySource: Adapted from Hchokr at English Wikipedia
As psychologists, our focus is on the individual. Still, we must think of this individual in context. So, in next illustration, I include an adaptation of this model that focuses on individual functioning—this will be our framework throughout the semester. Each of these circles is a “level of analysis”; to understand complex ideas in psychology we often need to understand them at each of these levels:


Examples of the Model for Understanding Psychology

I’d like to illustrate how we can think about complex topics in psychology using this kind of general framework. So, I’m going to use two topics to illustrate: youth aggression and anorexia nervosa.

Let’s talk about youth aggression

We need a little background. Aggression is a common problem. What’s the time of life when we experience the highest level of individual aggression?  Many of you probably guessed the teenage years, but actually it’s those toddler years! Toddlers are cute, but they can kick, hit, thrown things, bite! Thank goodness it gets better pretty quickly . . ., and they are small enough that the adults around can typically manage it. Aggression in teens can lead to many problems, like delinquency and criminal incarceration. The age when a youth begins to show significant aggression begins is often important in predicting future aggressive behavior. Kids who have high levels of early aggressive behavior (elementary school and younger) seem to be a different, and potentially more worrisome group, than those who have high levels of aggressive behavior beginning in the teen years (which is usually more short-term and associated with peer influences). We can also define aggression in various ways—physical, verbal, and/or relational. Physical aggression is pretty clear—hitting or assaulting others are examples. In verbal aggression we insult, demean or threaten others. In relational aggression, relationships are manipulated—isolating, shunning, spreading rumors, destroying reputations. Aggression can take place in person or online; indeed, online or cyberaggression is associated with negative health impacts on youth victims.


So, let’s use our framework to understand youth aggression:

biological level
1. The biological level
Wow, we could consider many things, and here are just three. First, human aggression, and that of many other mammals, is generally higher in males then females, particularly after puberty. Second, we also know that the male sex hormone testosterone is associated with higher aggression. Administration of testosterone in mice leads to increases in aggression. Third, we also know that levels of serotonin—a neurotransmitter in the brain—can be associated with higher aggression. So we know the biological level is important.

psychological level
2. The psychological level.
Let’s back up a little. Those who study youth aggression distinguish between two additional types: Reactive and Proactive. Reactive aggression what you might imagine—folks who are being attacked often respond in kind. When one child hits another child and the second child hits back, that second child is engaging in reactive aggression. You think someone has insulted you, maybe you insult back. Reactive aggression is emotional and is in retaliation to perceived slights or aggression from another individual. Proactive aggression, also called instrumental aggression, is when an individual uses aggression to get something they want or achieve some other end. Proactive aggression is goal oriented. When a child twists another child’s arm behind their back to force them to give up their lunch money—that’s proactive aggression. Which one seems a little more concerning to you? (I hope you said proactive). Although, proactive aggression is also used in war situations and in some sports situations—understandable.

We know that the psychological processes involved in these two types of aggression are different. Individuals who are prone to reactive aggression often have heightened “threat perception”—they expect others to be aggressive toward them. You could say “their antennae” are up; they expect to be attacked (physically, verbally, or relationally), and sometimes when we expect to see something we see it even if it’s not there! Let’s say you have a line of 3rd graders all jostling to see something and one child bumps into another. The child that gets bumped can think “that child bumped me because they got bumped by another classmate” or “that child bumped me by mistake” or “that child bumped me on purpose”. The child who makes that last assumption—the “on purpose” assumption—is more likely to respond with reactive aggression —maybe bumping or hitting the other child back in response.

The situation is different for proactive aggression. Remember in proactive aggression, the perpetrator uses aggression to get something. Research has demonstrated that children who tend to use proactive aggression evaluate problem solutions a little differently than children who don’t tend to use proactive aggression. In one study children were all presented problems and potential solutions. Proactively aggressive kids tended to evaluate aggressive solutions more favorably than non-aggressive kids. So proactive aggression seems to be a strategy.

This distinction between reactive and proactive aggression also applies to individuals who engage in domestic violence or intimate-partner abuse. In domestic violence an intimate partner, say a husband, is aggressive toward the other intimate partner, say his wife. Some spouses do this because they feel they are being “attacked”, insulted, or demeaned by their spouse and respond with aggression. In this case they are engaging in reactive aggression. For others, they use aggression to get what they want in the relationship—bullying or abusing their partner. These folks are engaging in proactive aggression. This is not to say that one type of aggression is “better”; but by understanding the psychological level we can understand more about how and why aggression occurs. Maybe, we can also be more effective in STOPPING aggression.

basic social interactions
3. The family level
We know that individuals develop their thought processes and emotions within the context of close relationships. Interestingly, aggression tends to run in families. Children who are raised in physically abusive families are more likely than those raised in non-abusive families to go on to either be abusive or be in abusive relationships. You could say that we learn aggression at our parents’ knees. One way this happens is through modeling—we see how our parents solve problems and go on to solve problems in a similar fashion, and sometimes that’s an abusive fashion. Another way this happens is through shaping our expectations—we may learn to expect to be attacked or demeaned, or we may learn at home that aggression can work for us. That is to say, these experiences at the family level shape the psychological level. Finally, we know that parental monitoring is important—high aggression and other delinquent behaviors are less likely when parents know what their kids are up to. We hear a lot about “helicopter parenting” these days (when parents are overly involved), but lack of parental involvement can be a real problem too! Family involvement and parental monitoring are important.

larger social group level
4. The larger group level

Some groups consider aggression to be a valid way to solve problems. Think about the mafia or violent street gangs. In this groups violence is seen as a legitimate way to solve problems and get one’s needs met. This larger group level of analysis can impact the psychological level—when your group says it is ok you begin to believe that as well. Of course, the larger group level can also impact the family level by influencing the ways in which parents relate to their children and model behavior.

A particularly interesting line of research has examined “deviancy training” among young delinquent men, where they reinforce and “one up” one another for their aggressive behavior. You can also see this happening for sexual aggression—with one young man boasting about his sexual conquests, leading to more and more outlandish boasts and “legitimizing” more egregious, harassing, aggressive (and possibly illegal) behavior.

culturl level
5. The cultural level
In some cultures, violence is more normative than in other cultures. If you look at violent crime rates you see startling differences across countries. Indeed, some cultures glorify violence and aggression. In the USA we frequently see movies where violent solutions are dramatized, heroes use violence to solve problems, and athletic events glorify aggression. We believe in “letting him [our adversary] know who is the boss”—dominance is a big deal in American culture. Other cultures emphasize getting along, playing your role in the larger, very highly valued group.

An interesting example of a powerful cultural factor impacting aggression is what is known as a “Culture of Honor”. In this system men (almost always men) feel that threats to their honor or reputation (i.e, culturally unacceptable behavior of a family member, insults, threats, etc) must be answered with violence to re-establish that honor or reputation. For example, in some cultures female chastity is highly valued, and a woman having sexual relations outside of marriage (whether consensual or not) is a stain on the family’s honor. In this cultural framework, killing one’s own female family member may be considered appropriate aggression to restore the family’s honor.

Indeed, there are important cultural factors that impact aggression.
Another example 

What do you know about anorexia nervosa?

Anorexia nervosa is a mental health problem that impacts about 0.6% of individuals. Symptoms include self-starvation, severe weight loss and low body weight. It tends to emerge during the teenage years and to predominantly impact young women. Although not the most prevalent mental health disorder, Anorexia Nervosa has the highest mortality rate of any mental health problem—about 20% of those with a diagnosis die as a result of it. Even when not fatal, it can have terrible consequences. It’s been said that it impacts the “three B’s"— bones, brains, and babies. Let’s focus on bones.

Bones

When young people are teenagers their bones are growing and strengthening, with calcium increasing bone health. Individuals with Anorexia Nervosa aren’t taking in enough calcium and may end up with brittle bones—osteoporosis can emerge very early in individuals with a history of Anorexia Nervosa. 

Brains

Young people are also undergoing tremendous growth in their brains, and fat in the diet is important in this process, as it promotes myelination of axons in the cells of the brain. We will talk more about this process when we learn about the brain. If individuals do not have enough fat in their diets, brain development may not be optimal. 

Babies

In order for young women to menstruate, they have to have a certain percentage of body fat. One common complication of Anorexia Nervosa is a lack of menstruation or what we call amenorrhea; this can have a long-term impact on young women’s fertility. Overall, AN can have long-term health impacts even when it gets better or resolves. 

So how can we use our model to understand Anorexia Nervosa? Again, the framework:

biological level
1. The biological level

The hypothalamus is a small region deep within the brain that controls some aspects of our physical functioning and our behavior.

Experiments with rats demonstrate that damaging a certain part of the hypothalamus can lead to a rat overeating to a point of severe obesity—an enormous rat! Damaging a different part of the hypothalamus gives you a rat that starves itself. The hypothalamus seems to be very important in the control of eating behavior.

Studies using brain scanning (MRI) suggest that other areas of the brain may be involved as well, including the dorsolateral prefrontal context, parietal cortex and anterior cingulate gyrus. You won’t need to know the particular regions, but I do want you to recognize that this level of analysis (biological level) is important.

psychological level
2. The psychological level

Self-Perceptions are crucial as well. One thing we’ve learned is that people with anorexia nervosa tend to see themselves, not as they are, but in a distorted way. You and I may look at the person and think “what an extremely thin person!” whereas they may look in the mirror and think “I’m still so heavy!” They may focus on one part of the body that seems like a problem to them— a thigh perhaps—and focus almost exclusively on that.

Personality plays a role too. Several personality characteristics may predispose someone to develop AN, particularly perfectionism. Individuals who are perfectionistic have very high standards for themselves and little room for mistakes or errors. Internalization of a thin ideal—when folks have thoroughly accepted the idea that thin is beautiful – also places a person at increased risk for developing anorexia nervosa. Understanding this psychological level has really helped in the development of better treatments for anorexia nervosa.

basic social interactions
3. The family level

Although not always, anorexia nervosa seems to happen more in families where parents have very high expectations for success—this is a disorder that is more common in higher socio-economic groups. Also, anorexia appears to occur more when parents emphasize appearance as an important value—perhaps this is where the internalization of the thin ideal begins. Again, this level of analysis impacts other levels, including the psychological level.

larger social group level
4. The larger group level

Certain groups are more likely to develop anorexia nervosa. In general, rates of anorexia nervosa are higher in groups where appearance (and especially a thin appearance is valued)—models, dancers, gymnasts. Gay men appear more likely than non-gay men to develop eating disorders. Group values about appearance and its importance may be internalized into one’s own value system—part of the psychological level. Again, one level can shape others.

culturl level
5. The cultural level

In our culture, appearance is something that is more highly valued in women (as compared to men), and youthful appearance (especially for women) is particularly valued. We see this in the media all the time.

As an interesting case study and an example of the importance of cultural factors, rates of eating disorders like anorexia nervosa were exceedingly rare in Fiji prior to 1995, and behavior like self-induced vomiting and binge eating were low. When researchers returned in 1998 following the introduction of Western television shows, they found greatly increased rates of these behaviors in a sample of girls, and the majority considered themselves “too fat”. This speaks to the important role of media in shaping self-perceptions and potentially risk of eating disorders like anorexia nervosa. These cultural variables also impact parenting. So the levels of analysis interact with one another—one influencing the other.

Important Considerations about Culture

Psychology has been criticized for NOT considering culture adequately. Many of the studies that are conducted do not represent the broader population. Indeed, psychology (among other disciplines) tends to focus on what are known as WEIRD populations—Western, Educated, Industrialized, Rich, and Democratic. Thus, it is challenging to generalize to different populations. In the United States of America, research has tended to focus on White middle- and upper-class populations; lower-income and people of color are not well-represented in our research studies. There has been more effort made in recent years to include more diverse populations in research AND to consider their perspectives and needs. We have a long way to go as a discipline in this regard.

Module 1A Summary

To summarize, I’ve presented a model today in which complex psychological phenomenon must be understood at a variety of levels. I’ve also provided several examples. In your opinion, what is the best level at which to understand psychological phenomena? What’s the right level of analysis? Well, I would say that it depends on your goals.

Give this some thought before you show the answer.

Let’s say you want to develop a program to reduce bullying in elementary schools. You want to figure out what factors influence school bullying. Which is the best level? 
	a. Biological
	b. Psychological
	c. Basic Social Interactions
	d. Larger Social Group
	e. Culture
Show Answer
Let’s say you want to develop a program to help students to develop better study habits. You need to understand how study strategies influence memory for course material. Which is the best level?
	a. Biological
	b. Psychological
	c. Basic Social Interactions
	d. Larger Social Group
	e. Culture
Show Answer
Do You Remember?

See what you can remember from the previous material by matching the terms to their definitions.


Module 1B: Methods in Psychology

The science of psychology relies on research to uncover knowledge. This portion of the module will focus on the methods of research used in psychology, but we will also return to this important topic throughout the course. Let’s focus on some basic strategies or designs.

1. Naturalistic Observation

One can systematically observe the real world. You can learn a whole lot by systematically observing the real world. Perhaps some of you know who Jane Goodall is. She studied chimpanzees in Africa, and by observing their behavior in the real world, she learned a lot that has guided our understanding of primates. Naturalistic observation has been used in psychology to help us form theories and develop hypotheses for future research. As another example, let’s say I want to understand more about factors impacting children’s aggression. I could observe them on playgrounds and carefully track all their aggressive behavior—hitting, kicking, spitting, name calling, all that. I could observe and deduce the situations that tend to lead to this behavior, the characteristics of those individuals who were aggressive, and the consequences of their actions. I can learn a great deal from this careful naturalistic observation.

Advantages	Disadvantages	Best Used
It’s real life	Observer bias! The fact that someone is watching can change behavior. For example, do you think that children may behave differently when they know they’re being observed? It’s also hard to control the many variables that are there as well	Naturalistic observation is often best used when first learning about a particular phenomenon and we need to develop hypotheses. For example, this method led to much greater understanding of child development and enabled us to conduct more specific research.
2. Case Study

Here we chose one specific case and study it in depth. It could be a particular individual, or it could be a classroom or work setting or another “case”. Perhaps I want to learn more about geniuses—maybe I could do a case study of the late, great astrophysicist Stephen Hawking. If I want to learn how to best teach English as a second language, I could find a very successful classroom where children learn very effectively and quickly become fluent in English. This would be a case study of a specific instance (of language learning) or situation. The case study has a long and storied history in psychology, and we’ve learned much through this approach. For example, by looking at individuals with very specific, unusual brain injuries, we’ve learned a lot about how the brain works.

Advantages	Disadvantages	Best Used
We can measure many things, study in depth, collect lots of data about our particular case. This is a major advantage of the case study to be able to look at things in depth	Maybe it doesn’t generalize! How do we know if it is typical of others or unique to that situation?	
The best use of case studies is often for unusual and hard to study phenomenon. There are case studies out there that have looked at the impact of very severe child abuse on language and emotional development. The impact of this abuse was, of course, documented after the abuse was discovered and stopped.

These cases have taught us important things about development. These cases are horrible, thankfully uncommon (although even one case is too common), and certainly not something we can (or would want) to study in more controlled ways.

On the other hand, what if we wanted to study the impact of dyslexia (a reading disability) on students’ emotional adjustment? Dyslexia is not rare and not difficult to study; a case study would be a poor choice here. We would probably be better off identifying a larger sample of kids with dyslexia and follow them over time.

3. Survey

We can ask questions of large groups of participants. Examples include the US Census or the Monitoring the Future Survey, which has been used since 1975 to examine teens’ behaviors, beliefs, and attitudes.

Advantages	Disadvantages	Best Used
We can collect LOTS of information and examine relationships within that data. For example, maybe we want to look at how religious affiliation is related to particular attitudes.	
We always have to be careful about response bias! Not everyone responds to a survey, and we want to make sure that our sample is representative! How many of you have gotten surveys in the mail and not completed them? Maybe some people are more likely to respond to surveys than others. So are our survey results really representative?

When you are trying to evaluate the validity of survey results you always want to look at the rate of response. If only 20% of folks who were surveyed responded, can you really trust it? Also, we have to rely on folks' honesty in responding to questions, and for some things (like questions about sexual activity or drug use) this may not be something we can rely on. So we have to be careful about how we rely on surveys.

Surveys are best used in situations where the research wants to gather information on characteristics, attitudes, opinion or experiences of a large population of people. For example, surveys are often used to learn about people's religious affiliations, political ideas, or health behaviors.

Correlational vs. Experimental Research

In our field we need to make an important distinction between correlational and experimental studies.

What is a Correlational Study?

Here I collect data and look at associations between variables.  This can be a terrific way to do research, but it has its limitations.

Let’s start with an example. I want to study the impact of exposure to aggressive video games on kids' aggressive behavior. For the purposes of discussion, let’s say I want to conduct my study among 3rd graders. Even within a correlational study design, I can do this in a variety of ways. One way I could do it is to use a survey of parents to determine how much time kids spend on which video games; better yet, maybe I could install an app on their computers and phones that will track their videogame time and activity—that would be a good way, eh? (Of course, we would need parental permission and child assent, because, after all, we need to be ethical in our research.)

I could then also collect data on aggressive behavior, using observation and/or getting parent and teacher reports. I then analyze my data and, lo and behold, I find that they are related. More exposure to violent/aggressive video games is associated with more aggressive behavior.

Can I conclude that exposure to violent video games caused the 3rd graders to be more aggressive? If you said “no”, congratulations you are correct! It could be that the causal direction goes the other way, maybe more aggressive kids seek out more aggressive video games. It could also be a totally different thing that causes both—something we call the "third variable". Maybe parents who don’t monitor their kids as much as they should end up with both more aggressive kids and kids who play more violent video games; maybe parents who have more permissive attitudes, or even positive attitudes toward aggression, have kids who both play more violent video games and engage in more aggressive behavior. There are lots of potential causes here. The important point is that we can’t determine it—a correlational study cannot tell me be about causality.

Remember, Correlation doesn't equal causation
What is an experimental study?

Say I want to actually determine causation. Well, now I need an experimental study. Remember my topic is the relationship between exposure to violent video games and aggressive behavior. To do an experimental study I will need to assign participants to particular conditions.

The key things here are:

experimental control. I “manipulate” the independent variable, which in this case is exposure to violent videogames.
random assignment to conditions. Half of my sample of kids get assigned to the “violent” videogame condition, half get assigned to the “non-violent” videogames condition, and assignment is by chance (that is, random; you can use the “flip of a coin” to decide who gets assigned to each condition).
Let’s say I am going to assign 100 3rd graders to one of two conditions that I, as the experimenter, have preselected.

Fifty of the third graders will be assigned to play two hours a day of a non-violent video game, maybe something boring like Tetris.
The other 50 of my third graders will be assigned a violent video game, let’s say Grand Theft Auto—hardly a game appropriate for third graders. Still, I’m the experimenter and I’m deciding what my study will be.
Obviously parents have to consent to have their children participate in the study. So the kids play these games for two hours a day for one month, and then I evaluate their aggressive behavior.

Like before, I might evaluate it using teacher reports, parent reports, and a behavioral observation. If I find at the end of my study that those who play the violent video game, Grand Theft Auto, show more aggressive behavior during the observations and their parents and teachers report more aggressive behavior, I can say that my study shows that exposure to these violent video games can cause youth to show more aggression. In this kind of EXPERIMENTAL study, I can speak about causation.

This is not to say that I’ve explained all the reasons that youth can show aggressive behavior, and there may be some youth in my study who show more aggressive behavior than others, but I can say that exposure to the violent video game caused them to show greater increases in aggressive behavior then did exposure to a non-violent video game.

What's the difference between the two kinds of studies?

The big difference between correlational and experimental research is experimenter control. In the first instance I have just measured behavior and looked at its association with other behavior. In the second instance I have control as an experimenter over the experimental conditions, and I randomly assign participants to these conditions. By using random assignment, I can control for differences naturally occurring in my participants. That is to say some of my kids may be high in aggression naturally, but by using random assignment I maximize the likelihood that some of those children will be in each of my violence exposure groups.

It’s not whether I use sophisticated technology or techniques that makes research experimental; rather, it is experimenter control. 
What Do We Study?

In conducting research in psychology, we are often interested in the big ideas out there.  These are called constructs—things like altruism, empathy, self-interest, intelligence. These are big ideas and not always easy to measure. We need to pick measures that really “get at” what it is we want to study. More specific than the constructs are the variables we use in a particular study; to operationalize (or include operational definitions) is to define our variables for the purpose of a particular study. 

Constructs and Variables

So, going back to something I discussed earlier. Say I want to understand more about youth aggression and how it comes about. Aggression is the construct I’m interested in. So how do I measure this? What variables do I use? Do I use self-report? Do you think youth or their parents might not disclose everything? Most of us want to look good and might alter our reports to look good—also known as the social desirability bias. We might do the same for our kids. Do you think teachers always know who the aggressive kids are? On the one hand, teachers see a lot of kids, so they are pretty good at knowing who is and isn’t extreme in some way. On the other hand, some of the most aggressive kids might be pretty good at hiding it from teachers, possibly limiting it to when teachers are not looking. 

None of these methods of measuring is foolproof. For my study, I’m going to include the following variables (with operational definitions in parentheses): youth self-reports (Youth Self Report Form), parent report and teacher reports (Child Behavior Checklist), and peer impressions (Peer Nominations). 

One of the things I’m careful to do here is to use standardized measures that have been shown to work well in past research and to include reports from a variety of sources (kids, parents, teachers)—none of which is perfect by itself. I could also design a task where youth can choose between potential problem solutions and add up their aggression scores, with more aggressive solutions scored higher. As you can see, there are number of ways to operationalize our aggression variables; some are better than others, and none is perfect.

Who Do We Study? Populations and Samples

We also need to carefully consider who we want to study—what is our population? In the study we previously discussed looking at youth aggression and exposure to violent video games, the population of interest was all children in the 3rd grade. After all, maybe I’d like to be able to use my research to develop programs to intervene with 3rd graders to reduce aggression.

Let’s think of another example. Let’s say, I want to develop methods for helping people with Type 1 diabetes more effectively manage their medical condition. The population here would be all people with Type I diabetes. When I finish my research, I’d like to be able to generalize my findings to the population of people with Type I diabetes. 

Now I can’t study everyone in that population, so I need to select a sample of that population—a smaller number of people that I can study. I want my sample to be a good representation of the population. The best sample is one that’s randomly selected. 


Pretend it’s like a lottery: I have the name of every person in the USA with Type I diabetes and I can randomly select 100 of them for my study. That would be amazing—too bad it’s totally unrealistic—there is not such list and how would I, practically, be able to study people from all 50 states? (I’d need a big travel budget) So I need to find another way.

Let’s say then that in order to find people with Type I Diabetes I try to get referrals for all patients who were hospitalized for diabetes complications in the last year at Joslin Diabetes Center here in Boston. 

1. Is this a good strategy?
	a. Yes
	b. No
Show Answer
Here's a better idea. What if I work for a health maintenance organization? I can go into the records of all the hundreds of thousands of members and identify all those who are taking medications commonly prescribed for Type I Diabetes, and my organization can send them letters inviting them to participate in the study. This is a pretty good strategy.

It’s not perfect; those with no health insurance won’t be represented; those who don’t want to be in a study won’t be represented. Still, it’s pretty good and maybe the best I can get. All in all, we want to consider generalizability always in choosing how to obtain our sample. That is, will my findings generalize to the larger population of folks I want to study? Sometimes we stratify our sample. To make sure my findings generalize to both men and women, I include an equal number of both men and women in my sample. In this case I’m stratifying by gender.

How Do I Summarize My Data? Descriptive Statistics

Well let’s try another example. Perhaps I’m interested in starting a clothing and shoe boutique for BU undergraduates, and I want to know something about their sizes and their preferences. Maybe I do a survey to obtain this information. I collect data on their gender, height, shoe size, number of shoes, and wardrobes (number of ties,  t-shirts, pairs of jeans), and preferences (how much they like the colors red and green. What now? What do I do with my data? 

First, I want to be able to just look at my data. I can graph my data—with potential values on the x-axis and the number of cases on the y-axis. If you look at Figure 1, you can see that I have graphed shoe size for a small sample of BU undergraduates. This graph is the distribution of my data on shoe size. If you examine it you will see that 2 people wore a size 6, 5 wore a size 7, and so on.

graph of shoe size for a sample of 20 individuals.
Figure 1
 In Figure 2, I did the same thing for how much the students liked the color green,

graph showing how much the students liked the color green
Figure 2
and in Figure 3, I examine the distribution for student height. 

graph showing distribution for student height
Figure 3
Shaqille O'Neal size 22 shoe
Example of an outlier—Shaquille O’Neal's size 22 shoe
Taken by David on Flickr: Some rights reserved.
Examining the distribution of your data is helpful in identifying odd cases; for example, the old basketball star Shaquille O’Neal wore a size 22 shoe. He’s what we would call an “outlier”—he just doesn’t fit into our distribution of scores. 

The Normal Curve

An interesting phenomenon is the normal curve. When you get a large number of cases, the distribution starts to take on the “bell” shape. There are more cases toward the middle and fewer in the extremes—typical of a large number of phenomenon and very useful in psychology. As examples, height is a good example. There are a few people who are very tall and a few people who are very short, but many more people fall in the middle. That’s the typical bell curve. Now think about intelligence. There are a lot of folks who are around the middle, and a lot fewer folks who have extreme scores on intelligence, either very high or very low. 

example of bell shaped curve
Source: By Inglesenargentina (talk) (Uploads) - Own work, Public Domain, Link
Measures of Central Tendency

Secondly, I want a number to tell me what the “typical” subject looked like. These numbers (or statistics in this case) are known as “measures of central tendency”, and there are three of them. 

Measures of Central Tendency
Mode is the most common score in the group. 

Median divides the distribution in half with the remaining 50% of scores above and 50% of scores below it. 

Mean is the arithmetic average. You add up all the scores and divide by the total number of scores. You can probably compute an average in your sleep.

These are used in different circumstances. For categories (say for instance gender) we tend to use the mode. We might say “the modal (or typical) introductory psychology student is female”. We might use the median for data that doesn’t really fit the bill curve very well. We often use the median to describe the typical home price in a community. Let’s say $500,000 is the median; this is useful because there may be homes that are much, much more expensive, but the $500,000 is much more typical.

So in this case the median gives us a pretty good idea of what is typical. Most often in psychology we use the mean. That’s because many phenomena take on characteristics of the normal curve, and this allows us to use a number of statistical procedures that we’ve found useful.

Optionally, and for another look at measures of central tendency you can view the video below.


Measure of Central Tendency
More Descriptive Statistics

Commonly Used Measures of Variability

Thirdly, I need a number that will tell me how much the different participants in my research differ from one another. Are they all similar on a particular characteristic? Or are they very different? To find this out, I use a measure of variability.

Take a look at this next graph, as it gives you an example of what I mean by variability.

Examples of Variability
examples of variability
Let’s ignore the green line for the time being, and focus instead on the blue, yellow, and red lines, as each of these represent different distributions. Interestingly for each of these, the mean is exactly the same. The blue distribution has little variability, and scores for most folks are fairly similar to one another. The red line is a different distribution, and there’s a wider range of scores. The yellow distribution is very wide. So although the typical score for each of these distributions is the same the amount of variability is quite different.

Here are three measures of variability that are often used: the range, variance, and the standard deviation. Let me tell you a little about each.

Range

The range is the lowest score subtracted from the highest score. So let’s say in my data that the largest shoe size is a 13 and the smallest shoe size is a 6; so the range here would be 7 (13-6). The range tells me a little bit, but it can be influenced a lot by an extreme score. Add Shaquille O’Neal to the class and suddenly the range goes up to 16 (22-6).

Variance

The variance is a little tricky to explain, and I’ve included an example for you to delve into. What the variance does is it takes each score and subtract the mean from that score. Then each of those resulting scores is squared; the squared scores are all added up; and we divide by the number of scores. So what you end up with is the average of the squared deviations from the mean. Now if that sounds confusing read it again. You will not be the only one confused. This is a tricky concept. Although the variance can be handy for some statistical techniques, it’s a little hard to interpret. How do we figure out what it means?

Standard Deviation

One of the ways that was developed to help interpret the variance is to convert it into the standard deviation. The standard deviation is the square root of the variance. To calculate the variance you squared all the deviations from the mean and took the average, so now you just take the square root of that number. It’s a little easier to understand the idea of the standard deviation.

Now, just to drive you a little crazy I’m going to ask you to learn a little bit about the relationship between the standard deviation and the normal curve. Look at the image below, as it shows you how the standard deviation is related to the normal curve.


How to Calculate Standard Deviation
 

Attributes of the Normal DistributionAttributes of the Normal Distribution
What it tells you is that if you go from the mean of the distribution to one standard deviation above the mean, you will have included 34.1% of all scores. If you include one standard deviation above and below the mean you have included 68.2% of all scores. If you go out even further and include two standard deviations on either side of the mean, that is two standard deviations above the mean and two standard deviations below the mean, you will have included another 26.4% of the scores in your distribution (13.2% between one and two standard deviations above the mean, and 13.2% between one and two standard deviations below the mean). So that means that you have included approximately 96% of all scores within two standard deviations on either side of the mean. I would never ask you to do math on a test, but I do want you to remember these particular percentages. These numbers remain the same across all distributions, and knowing them can help you compare scores. As an example, most intelligence (IQ) tests have a mean of 100 and a standard deviation of 15. So here’s a test for you!

2. What is the score for one standard deviation above the mean on the IQ test?
	a. 130
	b. 115
	c. 85
	d. 110
	e. 70
Show Answer
3. What is the score for one standard deviation below the mean on the IQ test?
	a. 130
	b. 115
	c. 85
	d. 110
	e. 70
Show Answer
3. What percentage of people have a score between 85 and 115 on an IQ test?
	a. 44%
	b. 96%
	c. 100%
	d. 68%
	e. 25%
Show Answer
So back to our data on the clothing habits and preferences of undergraduates! In the next two tables, I have calculated and show you the mode, median, mean, range, variance, and standard deviation for some data I collected from my PS 101 students on height, shoe size, and how much they liked the color green on a scale from 1 to 7.

Research Methods: Shoe Size
Central Tendency	Variability
Mode = 5

Median = 4

Mean = 3.94

Range = 7 (6–13)

Variance = 3.13

Standard Deviation = 1.77

Research Methods: Color Green
Central Tendency	Variability
Mode = 5

Median = 4

Mean = 3.94

Range = 6 (1 – 7)

Variance = 3.85

Standard Deviation = 1.96

Just another word about outliers. We always need to be careful to check our data. Imagine I decide to look at the data on how many shoes students own. I could calculate the mean, median, variance, and standard deviation here. Now, I just had someone add the class, and her name is Imelda Marcos. Do any of you know who she was? Interestingly, she was the wife of the Philippine president Ferdinand Marcos. They were excessively rich (although the Philippine people were certainly not) and it was said that she owned over 3000 pairs of shoes.

I realize that she is dead, but for the sake of argument let’s say that she really wanted to take PS 101 at Boston University.  Imagine that Imelda just added my class, and I’ve included her data in my distribution. Oh no, what impact does that have? Well, the mean variance and standard deviation would be affected in a pretty striking way, because Imelda “carries more than her weight” in shoes. Again, she’s what we would call an outlier, and the mean and the standard deviation can be seriously impacted by her extreme score.

Do Variables Go Together? Understanding Correlation

Now that I’ve looked at my data, I have a sense of what the distributions of the variables are, what the typical participant looks like, and how much variability there is in my sample. Next I want to look at how some of these variables might be related to one another. Look at this next graph.

Scatterplot: Association of Shoe Size and Height
In this slide I have graphed students' heights on the y-axis against their shoe size on the x-axis. Each of the dots in this graph represents one participant. You can figure out his or her shoe size by going down from the dot to the shoe size on the x-axis and you can figure out his or her height by going from the dot to that position on the y-axis.

 
You might think that shoe size and height would be associated, and you would be right. The dots seem to form a pattern, with many of them falling in an area from the left bottom of the graph to the right top of the graph. That is to say, as height goes up so goes shoe size. But the next graph is a bit different.

Scatterplot: Association of Shoe Color and Liking GreenScatterplot: Association of Shoe Color and Height
Here I have graphed my subjects' shoe size on the y-axis and how much they like the color green on the x-axis. Perhaps not surprisingly, there really is no pattern here. But who would expect shoe size to be related to how much one likes the color green?

Correlation and Scatterplots

When we talk about correlation we are talking about the degree of association between two variables—and the scatterplot is basically a visual representation of correlation. By looking at the pattern we can get some idea of the degree to which the two variables are correlated.

The correlation coefficient is a statistical indicator of the degree of association.

 A few important points about the correlation coefficient
It ranges from +1.0 to -1.0
0 means no association
A positive correlation means that the scores go up and down together.
For example, the more hours you study the higher your grade; the more sodas you drink the more weight you gain.
A negative correlation means that as one score goes up the other goes down.
Some people make the mistake in thinking that a negative correlation means things are not related, and this is not the case. You can have a strong negative correlation and it means that the two things are strongly related. For example, the more hours you study the fewer mistakes you make on an exam and on your homework.

Here’s another good example. Let’s say I want to design a test that will indicate how good you are at golf. I think my test is a pretty good one, and I think that a high score will predict how good you are when you go out to play a round of golf. In case you are not a golfer, a good golfer has a low score, not a high score, when playing the golf course. If you have a high score on my golf proficiency test (suggesting you are a good player) then you should have a low score on the course, and that’s a negative correlation.

Remember, if my test is a good one I will be able to predict your score.

Perfect correlation occurs rarely, if ever, in nature. No two things are perfectly related. If I know something about your height, I have a general sense of your shoe size, but I wouldn’t know it for sure. There is still variability.
Watch this for a further explanation of correlation and causation.


Correlation CAN Imply Causation! | Statistics Misconceptions
How Do I Compare Groups? T-Tests

Let’s talk about another statistical strategy. What if I want to compare two groups to see if they are different on a particular variable? Students T-test is a strategy that I can use when one of my variables is nominal or categorical and one of them is continuous (that is it can have a range of values).

In this case I can compare men and women (gender is a categorical variable) on both height and shoe size (both are continuous, as they can take a range of values). In this case my hypothesis is that men will have bigger shoes and will be taller than women.

Here I’m comparing the distribution of scores for men to the distribution of scores for women. Of course, there is variability for both men and women—some have larger feet and are taller, some have smaller feet and are shorter.

The Students T-test compares the variability within each of the distributions, the one for men and the one for women, to the variability between men and women. I don’t want you to worry about how to do this test, I just want you to understand what it does. The Students T-test allows me to determine whether there is greater variability between genders then within genders. Look at the results below.


I’ve listed the mean and the standard deviation for shoe size and height for both men and women and I calculated a t-test for both height and shoe size. What I want you to focus on is something called the P value. The P value in the case of height it says P less than .001, and in the case of shoe size it says P less than .005. Let’s focus on height for minute. What the P value means is that if I conducted this study a thousand times, I would only find a difference this big by chance—one time by chance. It’s very, very, unlikely that such a big difference in height would simply be found by chance.

Let’s focus on shoe size for minute. What the P value means here is that if I conducted this study a thousand times, I would only find a difference in shoe size and this big by chance five times. Again, this is very, very unlikely to be found just by chance. On the other hand, it’s very likely that men and women actually do differ in height and shoe size—and that it is a real and replicable finding.

This is what we mean by statistical significance. Statistical significance means that it is very unlikely that such a finding is a result of chance. Now to say that a finding is statistically significant is not to say that it is important or interesting, only that it is very unlikely to be a chance finding. Honestly, the idea that men and women differ in their shoe size is really neither interesting nor important (unless you are in the shoe manufacturing business), but it’s very unlikely to be due to chance. So I want you to understand the idea of statistical significance as it will be very important going forward in your psychology studies.

Do You Remember?

See if you can identify the methods of research used in psychology and the differences between correlational and experimental studies by completing the exercises below.


How Do I Ask More Complicated Questions? Advanced Statistics

All in all, there are many statistical strategies for examining data and asking sophisticated and complicated questions. The ideas of correlation and the Students T-test are just the tip of the iceberg.

Structural equation modeling

In fact, there are many new strategies that allow us to examine complex data. Two of them I’ll just mention briefly. First, structural equation modeling uses the idea of correlation to simultaneously look at many relationships at the same time. Often times we want to examine complex questions that involve many variables at once. Structural equation modeling can help us to do that.

For example, we may want to look at how adjustment to college is related to the distance from home, relationships with family members, social supports available in the college setting, and the course of study. We could use structural equation modeling to look at this.

Longitudinal modeling

Second, longitudinal modeling allows us to examine change over time. One of the main questions we often ask is how people change over time as a function of experience. As one example, we may want to look at how children learn to read, and what factors might predict success in reading. As another example, we might want to look at how individuals respond to certain kinds of treatments for anxiety or other mental health challenges. By using longitudinal modeling, we can address these questions.

The best statistical approach

You might ask what’s the best statistical approach? Well, it always depends on what the question is you want to ask! This is where in psychology, as in so many other sciences, it’s essential to have hypotheses.

The Big Picture: Research Process in Psychology

The image below illustrates the process of doing research in the field of psychology.

Steps in the Writing Process
In the first step, you have a topic that you’re interested in and you review what is known about that topic. You learn what others have found in their research and figure out where the gaps are. In the second step, you develop hypotheses, or predictions, that guide your research. In the third step, you select a research method and your participants. Then you collect your data. In the fourth step, you analyze your data and you either accept or reject your hypotheses. Were your predictions good? Or did they not work out as you planned? You might do some follow-up analyses to figure out more about why. In the fifth step, you submit your findings for what we call peer review. You write up your paper, select a journal that you’d like to publish it in, and submit to the editorial board. The editor sends it out for review to experts in your field who know a lot about what it is you’re studying. This is a great way to get feedback on your work. If you’ve done your study well and your findings are interesting, they get published for many others to read. Those findings become part of the larger research literature. Then you start the process again. This is how the science works. As researchers we all contribute to the larger body of knowledge.

I hope this lecture gives you a sense of some of the basic methodology in the field of psychology and provides a good springboard as we move forward to the many topics that we will be discussing throughout this course.

Review and Reflect