Unit Zero: Science Practices Flashcards

Cognitive Biases and Mental Shortcuts

Cognitive biases are characterized as mental glitches or shortcuts that frequently distort human thinking, serving as the conceptual equivalent of optical illusions within the realm of ideas. One primary bias is confirmation bias, which involves the tendency to actively search for information that aligns with pre-existing beliefs while disregarding contradictory evidence. A common example is an individual who believes social media is inherently harmful and consequently only acknowledges news reports that highlight its negative aspects. Hindsight bias occurs after an event has concluded, leading individuals to believe that they had predicted the outcome all along, such as a person claiming that it was obvious a specific couple would separate only after the breakup has occurred. Overconfidence bias involves the belief that one possesses more knowledge or accuracy than is objectively true, exemplified by a student assuming they will achieve a perfect score on a test despite having engaged in minimal study. The tendency for perceiving patterns in random events involves the human brain's instinctual drive to find order in chaotic or arbitrary data. This can manifest as seeing human faces in cloud formations or believing that wearing a lucky shirt contributed to a sports team's victory. These biases persist because the human brain is evolved to seek meaning and detect patterns, a trait that helped ancestors survive but can now lead to erroneous conclusions like believing in non-existent lucky streaks.

Ethical Red Flags and the Misuse of Psychology

Psychology must not be weaponized to support discrimination, stereotypes, or pseudoscience. It is considered profoundly dangerous and unethical to utilize biased or flawed research to claim that one race or gender is naturally smarter or more aggressive than another. Ethical psychological practice requires that findings are not distorted to marginalize specific groups or promote harmful social agendas.

Research Design Architectures: Experimental, Quasi-Experimental, and Non-Experimental

Psychological research is categorized into three primary methodologies based on the manipulation of variables and the establishment of causality. Experimental methods involve the direct manipulation of an independent variable to determine its effect on a dependent variable, aiming to identify causal relationships. These studies typically utilize random assignment to control and experimental groups, and are characterized by high internal validity and reproducibility. An example would be testing how caffeine affects memory by strictly controlling caffeine intake among different groups. Quasi-experimental methods also involve variable manipulation but lack the component of random assignment. Researchers often use this method when true experiments are impossible, such as comparing reading programs across existing classrooms or studying the effects of public policy in different cities. While they can suggest causality, they carry less certainty due to potential confounding variables within pre-existing groups. Non-experimental methods involve observation or measurement without the manipulation of variables to explore descriptions and correlations. These include surveys, case studies, and naturalistic observations. These methods cannot establish causation but are highly useful for studying rare phenomena or behaviors in real-world settings with high external validity.

Specific Non-Experimental Research Methodologies

Surveys are a method of collecting self-reported data using questionnaires or forms to gather information on attitudes and opinions from large groups efficiently. However, they are susceptible to social desirability bias and wording effects, where leading questions like "How dangerous do you think violent video games are?" can produce poor data. Case studies provide an in-depth analysis of a single individual or a small group, such as studying a unique patient with amnesia. While they offer rich, detailed insights into rare conditions, their findings are not generalizable to the broader population. Interviews involve structured or unstructured verbal questioning to deepen understanding and clarify responses, though they are subject to interviewer bias and are time-intensive. Naturalistic observation involves watching subjects in their natural environments without interference, which preserves authentic behavior but prevents the determination of cause and effect due to a lack of variable control. Correlational research examines the relationship between two variables, such as GPA and sleep hours, allowing for prediction and the use of large data sets without proving causation. Finally, meta-analysis involves the statistical review and combination of results from multiple studies on a single topic, such as analyzing $30$ different studies on the effectiveness of therapy for anxiety. This increases statistical power but can be affected by publication bias and inconsistencies in the quality of the original studies.

Components of Experimental Design: Variables, Controls, and Blinding

In a standard experimental design, the experimental group receives the treatment, while the control group does not. A placebo, which is a fake treatment like a silent or static-filled headphone track, is used to control for the psychological expectation of improvement. The independent variable ($IV$) is the factor the researcher manipulates, while the dependent variable ($DV$) is the measured outcome. Confounding variables are outside factors that systematically vary with the $IV$ and may influence the $DV$, such as a plant growth study where the experimental group receives fertilizer but also happens to receive more sunlight. Demographic variables include characteristics of the participants such as age and gender. To maintain scientific integrity, researchers use operational definitions to define variables in specific and measurable terms, such as defining "sleep quality" as the number of nightly awakenings. This allows for precise replication by other scientists. Blinding procedures are also vital; single-blind studies ensure participants do not know their group assignment to reduce bias, while double-blind studies ensure that neither the participants nor the researchers know who is in which group, specifically targeting the reduction of experimenter bias.

Sampling Dynamics and the Mechanics of Generalizability

The population represents the entire group a researcher wishes to study, while the sample is the smaller subset actually examined. Representativeness is achieved through random sampling, where every individual in a population has an equal chance of being selected. This reduces bias and is a requirement for generalizing findings to the broader population. Sampling bias occurs when certain members of the population are more likely to be included than others, such as convenience sampling where researchers only select participants within easy reach, like students at their own school. Generalizability refers to the extent to which findings from a sample apply to the whole population. Evaluating this involves checking if the sample was representative, if random sampling was utilized, and if there was diversity in age, race, and geography. For instance, a memory study using only $25$ Ivy League students lacks generalizability because the sample is too specific to represent all students.

Data Interpretation and Measurement in Psychological Science

Psychological measurement uses quantitative tools, which produce numeric data like test scores on a Likert scale of $1$ to $5$, and qualitative tools, which provide descriptive data from open-ended interviews. A correlation coefficient ($r$) measures the relationship between two variables, ranging from $-1$ to $+1$. A value of $r = +1$ indicates a perfect positive correlation, $r = -1$ indicates a perfect negative correlation, and $r = 0$ indicates no relationship. It is an essential tenet of psychology that correlation does not equal causation. This is due to the directionality problem, where it is unclear which variable causes the other, and the third variable problem, where an external factor affects both variables simultaneously. For example, if a study finds an association between outdoor temperature and the number of football concussions, it demonstrates a correlation but does not prove that temperature directly causes head injuries.

Statistical Foundations: Measures of Central Tendency and Skewed Distributions

Measures of central tendency help summarize large data sets. The mean is the average, the median is the middle value in an ordered set, and the mode is the value that occurs most frequently. In a normal distribution, represented by a symmetrical bell-shaped curve, the mean, median, and mode are all located at the center and are identical. Percentile rank is used to compare individual scores, such as being in the $90 ext{th}$ percentile, meaning one scored better than 95 ext{%} of the group. Skewed distributions occur when the curve leans; a positive skew (right-skewed) has a tail going to the right, indicating most scores are low with a few high outliers, resulting in ext{Mean} > ext{Median}. A negative skew (left-skewed) has a tail going to the left, indicating most scores are high with a few low outliers, resulting in ext{Mean} < ext{Median}. A bimodal distribution displays two separate peaks on a graph, suggesting the presence of two distinct subgroups or common score ranges within the population.

Measures of Variation and the Properties of the Normal Distribution

Measures of variation include the range, the difference between the highest and lowest values, and standard deviation ($SD$), which measures how spread out scores are from the mean. A low $SD$ suggests scores are clustered together, while a high $SD$ indicates significant variability. Adding or subtracting the same number from all values leaves the $SD$ unchanged, while multiplying or dividing all values causes the $SD$ to scale accordingly. Adding an extreme outlier will increase the $SD$, and removing one will decrease it. Within a normal distribution, specific percentages of data are predictable: 68 ext{%} of scores fall within $1$ standard deviation of the mean, 95 ext{%} fall within $2$ standard deviations, and 99.7 ext{%} fall within $3$ standard deviations.

Inferential Statistics and the Evaluation of Practical Significance

Inferential statistics allow psychologists to determine if findings from a sample are applicable to the population. The $p$-value indicates statistical significance; a result is considered significant if the $p$-value is less than or equal to $0.05$, suggesting a less than 5 ext{%} probability that the results occurred by chance. Effect size, such as Cohen’s $d$, measures the strength of the relationship. Cohen’s $d$ values of $0.2$, $0.5$, and $0.8$ represent small, medium, and large effects, respectively. A result can be statistically significant but have a small effect size, meaning it may not have real-world practical significance. Regression toward the mean describes the phenomenon where extreme scores tend to move closer to the average upon repeated testing or with more data collection.

Ethical Guidelines in Human and Animal Research

Psychological research must adhere to strict ethical standards established by the American Psychological Association ($ext{APA}$) and Institutional Review Boards ($ext{IRB}$). The key principles can be remembered using the acronym CANDID. Consent involves informing human participants of the study's purpose, risks, and their right to withdraw. Assent is required for minors along with guardian consent. Anonymity and confidentiality must be maintained to protect participant identities. No harm (Nonmaleficence) mandates that researchers minimize physical and psychological distress. Deception is only permitted if it is essential to the study, does not cause harm, and is followed by immediate debriefing. Debriefing involves explaining the true purpose of the study and providing support if needed. For animal research, the Institutional Animal Animal Care and Use Committee ($ext{IACUC}$) ensures researchers justify the use of animals, provide humane housing, and minimize pain. After a study's completion, it must undergo peer review by experts to ensure credibility and accuracy before being published and ideally replicated by other scientists.

Questions and Discussion

Q.1. One of Dr. Whitmore’s students believed cognitive abilities remained stable during the lifespan. When reading research where $60$-year-old and $80$-year-old groups outperformed younger groups on factual knowledge, the student claimed it was an obvious finding because older people accumulate more knowledge. What is this? Response: (A) Hindsight bias. Q.2. Jacob learns expressing gratitude increases well-being. He believes the outcome is obvious and everyone would intuitively recognize it. What bias is this? Response: (C) Hindsight bias. Q.3. Dr. Kebede tests if gratitude journals increase well-being. After reading journals from a gratitude group and a non-tasked journal group, he concludes gratitude works. What bias likely affects his interpretation? Response: (A) Confirmation bias. Q.4. Participants were randomly assigned to drive a simulator while talking on a cell phone or without. What research type is this? Response: (D) Experimental. Q.5. Researchers randomly assigned $50$ children to wear nose plugs and $50$ to not wear them while rating food. What does random assignment help with? Response: (C) It allows researchers to understand if differences were due to the independent variable. Q.6. In a study of $25$ students, participants misremembered the word "dacha" as "house." To run this as an experiment, what is necessary? Response: (D) Participants must be randomly assigned to groups. Q.10. A client improves during therapy due to the expectation of improvement rather than the therapy itself. This is? Response: (A) A placebo effect. Q.11. A researcher compares massed practice vs. distributed practice strategies using vocabulary quiz scores. What is the dependent variable? Response: (C) The vocabulary quiz scores. Q.12. Professor Ahad compares an $8$ a.m. class and a $1$ p.m. class on the same exam. The $1$ p.m. class scores higher. What is the problem with concluding the $8$ a.m. students are inferior? Response: (D) Time of day is a confounding variable. Q.13. Why provide an operational definition of sexual orientation? Response: (A) To help everyone understand and collect data in the same manner. Q.18. Psychologists are uneasy that 80 ext{%} of participants are from Western, educated societies. Why? Response: (D) A representative sample is essential to generalize results to the entire population. Q.22. Respondents report being healthier or less prejudiced than they truly are. What explains this? Response: (C) The social desirability bias. Q.24. Dr. Messina uses her own psychology students for research. What type of sample is this? Response: (B) Convenience sample. Q.27. A researcher finds mnemonic devices improved memory for $10$ people but shouldn't draw conclusions yet. Why? Response: (C) The researcher should expect results to regress to the mean as more data is collected. Q.32. A normal distribution has a mean of $50$ and $SD$ of $8$. Mary scored $59$. What percentage scored lower? Response: (D) 84 ext{%}. Q.36. Given a chart of $50$ students, how many earned scores higher than two standard deviations above the mean? Response: (A) $2$. Q.39. What change makes standard deviation increase? Response: (A) If the tallest person were taller. Q.42. In a study where participants wanted to quit the coin toss but were told they had to finish, which guideline was violated? Response: (B) Violation of the right to withdraw. Q.47. Jeremy wants to use third graders for a mnemonic study. Why might the IRB deny his plan regarding consent forms? Response: (C) Children can express approval, but only parents or guardians provide consent.