What are the key elements of the research process that you need to know in order to critique research presented in this book?
The field of psychology relies on a scientific approach, and becoming familiar with how psychological research is conducted can help you be a more critical consumer of it. Moreover, as we just discussed, the history of using the scientific method to study girls and women has not always been bias-free. One goal of this book is to help you develop skills in uncovering hidden gender biases and appreciating the complexities associated with conducting psychological research.
One complexity involves how we define what it means to be a woman. You may like to rely on biological definitions. However, as we’ll explore in Chapter 4, not everyone with a vagina, for example, considers themself a woman, and not all people who consider themselves women have a vagina. Furthermore, not all people who consider themselves women act or dress in ways that one might expect a woman to act. Researchers have clarified that it’s possible to refer to someone’s sex or gender, and that these are not necessarily the same thing. The APA publication manual instructs authors to use sex for biology and anatomy and gender for attitudes and roles associated with a biological sex (American Psychological Association [APA], 2009), yet there’s still a tendency to confuse sex and gender (Smith, Johnston-Robledo, McHugh & Chrisler, 2010). It’s not always clear, for example, when a researcher is referring to biology, psychology, or both, when studying girls and women. Furthermore, as we’ll discuss in Chapter 4, the idea that there are only two sexes and two genders is an oversimplification of reality.
Because defining woman isn’t always clear, research about sex and gender can be controversial. Problems come up when observations are interpreted through an oversimplified lens, which, especially when conveyed in the popular press, can be demeaning to everyone. For this reason, feminist psychologists have called for a more socially responsible approach to psychological research, one that doesn’t shy away from a complicated exploration of sex and gender (Fine & Roberts, 1999).
Doing Research
Because we’re constantly surrounded by sex and gender, ideas for research can come from everyday interactions. For example, let’s say you’re planning a day at the beach during a family vacation, and you notice a distinct difference between two of your cousins. Your 6-year-old cousin excitedly puts on her bathing suit and grabs her Frisbee. Your 13-year-old cousin, in contrast, spends over an hour trying on different bathing suits, standing in front of the mirror, and evaluating her body. Although there are probably lots of explanations for this, you think their ages might explain their different behaviors. This is a theory, or a proposed explanation for why certain things occur. Initially, theories can come from personal observations or from previously established concepts. They are then refined through the process of collecting data and drawing conclusions based on that data. Let’s explore how.
Image Description
Both wear two-piece swimsuits but the girl who is about nine years old has a more modest suit compared to the skimpy bikini on the girl who is about fifteen years old.
Do you think these girls experienced getting ready to go to the beach the same way? Do you think they have different levels of concern about their appearance? What would be involved in doing scientific research to answer these types of questions?
First, a theory will lead to a scientific question that can produce a testable hypothesis, or prediction. In our example, your hypothesis might be: “As a girl ages, her body anxiety increases.” Each of the italicized words or phrases in the hypothesis is a variable, and variables can be measured in many different ways. For example, to measure body anxiety you might use direct observation by recording behaviors that reflect body anxiety—such as by noting how many times a girl examines her body in a mirror. As another option, you might use a survey to ask participants to report their beliefs, attitudes, and behaviors related to body anxiety. Although the conceptual definition of body anxiety is the same, the measurement, or operationalization, of it can vary quite a bit. And data can change depending on how the researcher decides to operationalize a variable. For example, the 13-year-old may not want to admit she’s anxious, but as a researcher you might observe her frequently checking herself in the mirror. This measurement decision could affect the results, since anxiety might be lower if operationalized through self-report and higher if operationalized through observation.
Returning to our example, the research process requires expanding your investigation beyond your cousins and collecting data from a larger sample of girls of different ages. A common method is to recruit girls you already know, referred to as a convenience sample. You might ask fellow classmates or post a message about your study on Twitter. Of course, this might mean that all the people in your sample share similar characteristics because they all have something in common: their relationship to you! It’s more challenging to recruit participants you don’t know, and for this reason, it might be necessary to offer an incentive for participating in your study. This type of recruitment takes longer and requires more resources—like money to pay for the incentives.
Once data are collected, more often than not you’ll use statistics to make interpretations about the data. Most psychological research relies on quantitative methods, an approach that attempts to represent variables with numbers. For example, body anxiety could be represented numerically as the score on a self-report measure so that higher numbers signify higher levels of body anxiety. Let’s say that a group of 6-year-olds averages a score of 15 and a group of 13-year-olds averages a score of 25. When psychologists use statistics to examine this difference, they’re interested in evaluating statistical significance. Statistical significance suggests that if body anxiety was not actually related to age, finding a difference this large (15 vs. 25) by chance would be unusual. How unusual this would be is often indicated as probability, or a p-value, and results are typically considered statistically significant if p < .05. This level of statistical significance means that there’s a probability (p) of less than 5% that the results of the study are due to chance.
If you were to find statistically significant differences in body anxiety among girls of different ages, you could have confidence in assuming that development is related to body anxiety. If the results were not statistically significant, however (i.e., p > .05), you might have to revise your original theory: Maybe body anxiety is related to general levels of anxiety, rather than age. In other words, maybe your 13-year-old cousin is just a more anxious person than your 6-year-old cousin. With this revised hypothesis, the scientific process can begin again. However, this time, you might reconsider your study’s design, as design decisions can also play a role in non-significant findings.
When researchers don’t find statistical significance, it can be frustrating because they have to start the process over again and typically don’t submit the original findings to an academic journal for publication. In general, academic journals only publish studies that show statistically significant findings. This means that most published psychology research reflects a preference for finding differences, rather than similarities, between groups—including gender differences, a topic we’ll focus on in Chapter 3.
When researchers do find statistical significance, however, they can write up the findings in a report and submit it to an academic journal for peer review. During the peer-review process, experts in the field review the research and determine if it meets a rigorous standard for publication. If the reviewers and the journal editor decide that the research is of high value, it will be published. Such research then has the potential to influence people, policies, and social structures.
In our example, you might submit your report to a journal that focuses on gender, such as Psychology of Women Quarterly or Sex Roles. By doing this, you know that the reviewers are other scientists with expertise in gender. Once the research is published, a media outlet might feature the study in a magazine or on a morning talk show. The study might also influence curriculum decisions in middle schools, high schools, or after-school programs.
Experiments versus Correlational Research
What are the key differences between experimental and correlational designs?
If you have already taken a statistics or research methods course, you may have noticed that the above example isn’t an experiment. Instead, it would be considered a correlational design. In a correlational design, relationships between variables are examined—in this case, the relationship between age and body anxiety. One of the most important tenets of psychology is that correlation does not imply causation. In other words, even if variables are related, it doesn’t mean that one caused the other. For example, if umbrella sales go up when grass grows, it doesn’t mean that selling umbrellas causes grass to grow or that growing grass causes increased umbrella sales. In this case, they’re both caused by another variable: rain. However, in an experiment, the researcher manipulates a variable in order to investigate whether changes in that variable (the independent variable) cause a change in another variable (the dependent variable).
A critical component of experiments is random assignment—the process of assigning participants to conditions in a way that guarantees all participants have an equal chance of being in any group. Returning to our example, you couldn’t randomly assign people to different age conditions because people already have an age. In other words, you couldn’t randomly assign people to the 6-year-old condition or the 13-year-old condition because the people in those groups already exist as 6-year-olds or 13-year-olds. Therefore, even if you were to find a difference between 6-year-olds and 13-year-olds, you couldn’t conclusively say that age causes the change in body anxiety. You could, however, design an experiment that explores another variable that might affect the development of body anxiety. For example, you could randomly assign girls to two conditions. In the manipulated condition, sometimes referred to as the experimental group, you could tell the girls that boys are watching them. In the control condition, the group you’re comparing the effects of the manipulation against, you could tell the girls to simply enjoy one another’s company. If you were to find that body anxiety is statistically significantly higher in the manipulated condition, then you could confidently state that the anticipation of being watched by boys caused the increase in the girls’ body anxiety. Because all participants had an equal chance of being in either condition, you likely eliminated other factors that might have influenced the outcome. However, it’s important to note that many variables cannot be randomly assigned, including sex and gender. Therefore, the ability to use psychological research to make causal statements about the role of sex or gender is limited.
try it for yourself
In what ways have your views on any topic been influenced by reports about scientific findings? How often have you questioned the results of a scientific study presented in the news or through social media? Have you noticed bias in the way science is reported? Why or why not? Ask three people if they would question a scientific finding presented in the popular press. Now find a news story summarizing a research study, locate the original research report, and compare the two. Is the summary in the news story accurate? Is it complete? What are the differences between the way the journalist framed the results and the way the scientists did?
