PSY 179 Critical Thinking in Psychology - Correlation vs. Causation - PDF

26.11.2024 PSY 179- Critical thinking in Psychology Correlation vs. Causation Today Correlation vs. causation...

26.11.2024 PSY 179- Critical thinking in Psychology Correlation vs. Causation Today Correlation vs. causation 1  26.11.2024 Correlation and causation Understanding the relationship between variables Defining correlation: Correlation measures the relationship between two variables, showing how one variable changes when another does. Causation: Direct cause-effect relationship- When changes in one variable directly cause changes in another. Requires controlled testing to prove. Correlation vs. causation: Correlation does not imply causation. A correlation simply indicates a relationship, but not the direction of influence or existence of a cause. Why it matters?: Misunderstanding correlation can lead to incorrect conclusions, affecting areas like health, education, and public policy. 2  26.11.2024 Spurious correlation: A statistical relationship that seems to exist between two variables but is caused by an external third factor. What is a third variable?: A third variable can create a false impression of causality between two variables, leading to spurious correlations. Birth control by the toaster method An example of spurious correlation The toaster method: An infamous example in Taiwan, where owning a toaster was found to correlate with lower birth rates. However, the relationship was due to socioeconomic factors. Understanding the third variable: The third variable in this toaster example was wealth. Wealthier families had more appliances and fewer children. 3  26.11.2024 Another example: Ice cream sales and drowning A correlation exists between ice cream sales and drowning rates in summer Why? This correlation is driven by the third variable, hot whether. The impact on decision making: Failing to recognize spurious correlations can lead to faulty conclusions in policy and research. Partial correlation Controlling for third variables Partial correlation: A statistical method that measures the relationship between two variables while controlling for the influence of a third variable. 4  26.11.2024 Example: Crime rates and temperature Testing the heat hypothesis The heat hypothesis: This hypothesis suggests a positive correlation between high temperatures and increased violent crime rates. What can be controlled as a third variable? Factors like time of day and neighborhood conditions, season and other activities also influence crime rates. Importance in research: Partial correlation is a powerful tool for reducing confounding factors and getting closer to true causality in correlational studies. Real-world implications: Understanding environmental factors in crime can help inform public policy and policing strategies. 5  26.11.2024 Directionality problem Establishing cause in correlation The directionality issue: In correlations, it is difficult to determine which variable causes the other, leading to ambiguity in interpretation. Example: Does high self-esteem lead to better grades, or do better grades lead to high self-esteem? Resolving directionality: Experimental designs with manipulation of variables can help in determining the direction of causality. Reversing assumptions While many assume high self-esteem leads to academic success, research suggests the reverse is often true- achievement boosts self-esteem. Research implications: This shift in understanding affects educational strategies and how we support students’ academic development. 6  26.11.2024 Selection bias: How sampling affects results What is selection bias?: Selection bias occurs when the sample used in a study is not representative of the population, leading to skewed results. Examples in education research: Private school students may outperform public school students, but this difference may result from socioeconomic factors, not the schools themselves. Controlling for SES: Studies must account for socioeconomic factors to accurately compare school types. Avoiding selection bias: Random sampling and careful controls can help mitigate the effects of selection bias in studies. The WEIRD bias The WEIRD bias: Western, Educated, Industrialized, Rich, and Democratic societies dominate psychological research. How does this bias affect the universality of psychological findings? 7  26.11.2024 What is illusory correlation?: The human brain tends to see The illusion of patterns or relationships between unrelated events, leading to false conclusions. correlation How we see patterns Why it matters: Illusory correlations can reinforce that don’t exist stereotypes and pseudoscientific beliefs, distorting our understanding of the world. 8  26.11.2024 Illusion of control Misinterpreting randomness What is the illusion of control?: People often believe they can control outcomes in random events, such as gamblers thinking they can influence a roll of the dice. Coincidence vs. control: People confuse random chance with control, leading to superstitious beliefs and irrational behaviors. Psychological impact: The illusion of control can affect decision-making in areas like gambling and business strategies. Case studies and confirmation bias Limits of case study evidence Confirmation bias: The tendency to search for, interpret, and remember information that confirms one’s pre-existing beliefs. Case studies are not causation: Relying on case studies can lead to confirmation bias and cannot establish causal relationships. The role of scientific testing: Only through systematic testing can real causal links be identified, avoiding biased interpretations. 9  26.11.2024 Pseudoscience and correlation Misleading interpretations in popular beliefs Pseudoscience refers to beliefs or practices that claim to be scientific but lack empirical support for testing. Correlation misuse: Pseudoscience often misuse correlational data to make false claims, such as astrology or alternative medicine practices. The role of critical thinking: Critical thinking helps to identify when correlational data is being misinterpreted or misused in pseudoscientific claims. Real-world example: Hormone replacement therapy (HRT) for menopause The impact of misleading correlation Initial findings: Early correlational studies suggested that HRT was linked to reduced heart disease, influencing medical recommendations. The role of selection bias: Later studies found that the correlation was driven by selection bias, with healthier women are more likely to use HRT. Consequences of misinterpretation: This mis interpretation led to widespread changes in medical advice, later revised after randomized controlled trials. 10  26.11.2024 Technological correlations Misleading interpretations in corporate settings The technology-productivity correlation: Companies often assume that the use of more advanced technology correlates directly with increased productivity. Confounding variables: The real drivers may be employee training, company culture, or other factors that affect productivity. Avoiding simplistic conclusions: Businesses must control for external factors before attributing productivity gains to technology alone. 11  26.11.2024 Correlation equals causation fallacy: One of the most frequent mistakes in interpreting statistics is assuming that Statistical correlation directly implies causation. misconceptions Ignoring third variables: Often third Common variables explain the observed correlation, such as socioeconomic fallacies in factors in educational outcomes. interpreting data Small sample size problem: Small samples can produce unreliable correlations, leading to conclusions that don’t hold in larger studies. Correlation and public perception The public often mistakenly Media amplification: Media believes that correlations outlets frequently prove causation, leading to oversimplify correlational flawed conclusions in areas findings, which can mislead like health, politics, and the public. education. 12  26.11.2024 Media often uses correlations to create sensationalist headlines… The original study in 1988 suggested that people who forced themselves to smile while looking at a cartoon found the cartoon funnier. That research was taken as gospel and the idea of “smiling to make yourself happier” spread like wildfire through psychology and decision science classes around the world. But when a number of labs tried to replicate the study in 2016, it didn’t hold up. 9 labs found a similar effect but at a much lower magnitude, and 8 labs found no effect at all, which when they combined the data came out to no significant observed effect. Ethical considerations in correlational research When experimentation is not possible: In some cases, ethical or practical constraints prevent the use of controlled experiments, leaving researchers to rely on correlations. Misinterpretation risks: Correlations can be misinterpreted, leading to flawed policy or interventions, particularly in health and education. Responsibility of researchers: Researchers must clearly communicate the limitations of correlational findings and avoid overreaching conclusions. 13  26.11.2024 Causal research: Experimental true causality Establishing true causality Random assignment: A key feature of experiments, random assignment helps eliminate third variables by distributing them equally across groups. Control groups: Using control groups allows researchers to isolate the effect of the variable being tested by comparing it to a baseline. Manipulation of variables: Researchers can actively manipulate variables to observe their direct effects on other variables. Summary and key takeaways Correlation does not mean causation: Correlations only show relationships, not causality. Misinterpreting this can lead to faulty conclusions. Third variable problem: Uncontrolled third variables can create spurious correlations, masking the real cause. Importance of experimental design: Proper experimental methods, like random assignment, are necessary to establish causality. 14