Part 1: When reading a journal or newspaper article that cites statistics or presents statistical results, what are some things that you should consider to assess whether the conclusions are credible? Why? (Hint: review Chapter 1 and the other chapters we have covered in the textbook for specific criteria. The more criteria listed the better.)

PART 1 Many things can contribute to an article's credibility: knowing what data is relevant to the subject, what data we can really obtain, what approaches are better or worse for collecting the data, and more. Credible scientists consider these: * Define hypothesis. * Describe how they expect to confirm or refute the idea. * Describe the observations made and procedures taken to keep other factors constant. * Explain their findings (and certainty) about the hypothesis. * Specify future learning goals. * Define results that could discredit their findings. * Define data logic. * Define how they considered their own criticisms while evaluating their own performance. * Explain how claims were generated from references. * List any possible objections. * Utilize rational thinking (Stemwedel, 2011) For credibility of an article, the following factors are examined: 1. Sample size: If the sample is too small, the results cannot be generalised for the entire population. Larger samples are more dependable. 2. Error margins: These might estimate a study's sample size. Larger error margins require fewer sample sizes. 3. Bias in Sampling: If a study was conducted on people's ability to identify a quote from classic literature, the results would be inaccurate. Researchers sampled at their convenience, not randomly. Misleading questions, unsuitable incentives for the sample make it impossible to make firm conclusions from the data. 4. Context: Without context, the article or report's facts could be misinterpreted (Sandstrom, 2018). 5. Goal of the study: Variables affects hypothesis, techniques, and findings. Knowing the study's goal helps apply and interpret its findings. 6. Examining the researchers' backgrounds and funding sources helps validate their conclusions. Research should be conducted transparently, showing the many stages of the study to ensure objectivity, and it's best to establish if any organisation has an interest in particular research finding. An outside sponsor or funder may impair the study's objectivity. 7. In-depth interviews, self-completed questionnaires, and secondary sources. Secondary data dependability and usefulness to the research are key. 8. Inconsistencies such as contradicting statements, hasty conclusions, presuming causation when correlation exists, inappropriate extrapolation, and circular reasoning. 9. Listing study restrictions. 10. Confounding variables: Confounding arises when a two-variable relationship is caused by a third, unseen variable. Confounding occurs when an experiment's results can be attributed to unmanipulated causes. When scientists declare a correlation between two or more variables, they indicate a causal relationship. Determine a causal link between two variables. 11. Avoidance of unethical behaviour: * Don't modify and conceal facts in statistics to show only what you want to see. * Engaging in unethical data analysis from collection to interpretation. Incorrect data gathering questions might induce bias. As an example of unethical data interpretation, assume a company surveys clients and only releases good results. * Ignoring aberrant facts to prove a point. * Using wrong statistical procedures, omitting unfavourable findings, failing to incorporate all important information, inflating the number of study participants, mislabelling or misinterpreting graphs, and sorting through data without sufficient documentation. 12. Linear regression, a standard tool, only estimates slope accurately in ideal situations. In practise, correlation is rarely perfect, leading linear regression to the wrong result. * Linear regression only evaluates linear dependent and independent variable relationships. It assumes linearity. Not always. * A linear regression study determines if independent factors and dependent variable mean are correlated. * Outliers affect linear regression. Linear regression assumes unrelated data. No association exists between subject ratings. This isn't always a good idea. This doesn't make sense in space and time clustering. Instead of openly comparing two variables, authors typically conclude that they are different because one variable's effects are statistically significant and the other is not. Inaccurate assertions about a particular citation may also occur (Sullivan, 2017).