Correlational Research & Guidelines for Reading Research 301 Lab Week 2 - Thursday Agenda for Today: 1. Turn in observational Method & Results 2. Discuss correlation & survey methods 3. Discuss research reading tips 4. Class survey & data entry using SPSS Announcements: Regarding last week’s homework (observation summary, data, & worksheet): I will hold on to that to aid me in grading the Methods & Results that you did for today. Quiz #2 is on Tuesday (6/20). You may have your article out during the quiz, as well as any notes/books, but be sure to read the article beforehand. What are correlations? We use correlational research when we cannot assign experimental groups due to ethics or practicality. For example, smoking or gender. In correlational research we look to see if one variable changes with another (covaries). We can also call this an association. Correlations Assess the relationships among naturally occurring variables with the goal of identifying predictive relationships. *Degree & Direction* THE MOST IMPORTANT RULE IN CORRELATIONAL RESEARCH!!! Correlation Is NOT Causation What statistics do we use with correlations? There are many statistics available for measuring the strength of correlations. For this class we will be using pearson product moment correlation Lower-case r represents Pearson Correlation such that… The formula: r= SP √SSxSSy _ Where SS = Σ( X – X )2 And SP = ΣXY - ΣX ΣY n= number of pairs of scores n What do values for r mean? r can range from -1 to +1. -1 means a perfect negative correlation (as X increases, Y decreases) 0 means there is no correlation (no relationship between X and Y) +1 means a perfect positive correlation (as X increases, Y increases) Negative Correlation How Drinking Predicts GPA Undergraduate GPA 4.00 3.50 3.00 2.50 2.00 1.50 1.00 0 2 4 6 Nights Drinking Per Week Positive Correlation How SAT Scores Predict Undergraduate GPA 4.00 Undergraduate 3.50 3.00 GPA 2.50 2.00 1.50 1.00 400 600 800 1000 1200 1400 1600 SAT Scores Other pictures of Scatter Plots Survey research is correlational Characteristics of Surveys include: Scope can be limited and specific, or broad and global Select a representative sample Use predetermined set of questions Sampling Careful selection allows generalizability to pop. *remember, our interest= population Biased sample: systematically differs from the characteristics of the population Selection bias: occurs when procedures used to select a sample results in over/under representation of some segment(s) of the population. Approaches to Sampling Nonprobability: does not guarantee that every element in the population has an equal chance of being included in the sample Convenience sampling Probability sampling: allows researcher to estimate the likelihood that their findings for the sample differ from those for the population Simple random or stratified random sampling Survey Methods 1) MAIL Pros: quick, convenient, no interviewer bias, good for “difficult” topics (anonymous) Cons: response bias (not everyone does it) Increase return rates by: “personal touch” Min effort is required Intrinsic interest of the topic to the respondent Identify w/ researcher or sponsor organization Survey Methods (cont’d) 2) Personal interviews Pros: more control over administering Cons: costly (time/$), interviewer bias How can you improve? Hire highly motivated, well paid & trained interviewers Give detailed instructions for possible situations Close supervision Survey Methods (cont’d) 3) Telephone interviews Pros: good for brief surveys, less dangerous, no problem of physical accessibility, or time. Cons: response bias, interviewer bias, cell phones, multiple phone households. How would you improve a telephone survey? Survey Methods (cont’d) 4) Internet surveys Pros: convenient, cheap, large and potentially diverse or underrepresented samples. Cons: selection bias, response bias, lack of control over the research environment. What do you think could be done to improve internet surveys? Survey-Research Designs 1) Cross-sectional: one or more samples drawn AT ONE TIME from the population. Describe and predict 2) Successive independent samples: same questions asked of different respondents over a TIME PERIOD. Describe changes in attitude/behavior but not how 3) Longitudinal: same respondents surveyed over time in order to look at individuals’ changes Hard to identify cause of change Threat=Attrition (drop-outs) makes the sample no longer comparable Questions Demographic variables: describe characteristics of the people who are surveyed Self-report scales assess attitudes/preferences Accuracy of questionnaires depend on careful/expert construction Reliability Consistency of the measurement If you get on a scale 3 times, do you get the same #? Increased by: 1) using many similar items on a measure 2) testing a diverse population 3) using uniform testing procedures Test-retest reliability method: give same questionnaire 2X to large sample. Alternate form test 0.8 reliability needed Validity Construct validity: extent to which it measures the theoretical construct it’s designed to measure Criterion validity: predictive v. concurrent Convergent validity: the extent to which 2 similar measures correlate with each other in the measure of a theoretical construct (p. 174-5). Discriminant validity: extent to which 2 measures do not correlate with each other in the measurement of a theoretical construct. Constructing a Questionnaire Wording should be clear & specific. Vocabulary should be simple & familiar. Questions should be short, not leading/loaded, should present all conditional info prior to key idea. Make sure to check for readability. Funnel questions: start broad specific topic Filter questions: general questions asked of respondents to see if they need to be asked additional questions (do you____? If so, how/why/when/etc?) Influence of social desirability: pressure for people to respond how they “should” believe not as they actually believe (e.g., perhaps participants will try to be politically correct). Why Correlation Causation Does Not Equal Causation Correlation: means 2 variables are related We can make predictions based on correlations We CANNOT infer cause of the relationship “Spurious” relationship: when the relation between 2 variables can be explained by a 3rd. Correlation, Causation (cont’d) It has been found that violent crime is correlated with ice cream sales (as violent crime increases, so does ice cream sales and vice versa). Why is this an important example of correlation not causation? Attitudes In thinking about peoples’ attitudes regarding smoking, littering, academics, and social honesty, you might identify the following attitudes: Against smoking cigarettes Against littering For being socially honest For being academically honest Knowing what you now know about correlation, which do you think will be related? Please take out a piece of paper. Write your name on the back. Answer as truthfully as possible! Scoring the survey Calculate your scores for attitudes towards smoking (questions 1, 7, 3,10, 5), littering (6, 2, 8, 9, 4,) academic honesty (11, 17, 13, 14, 19) and social honesty (16, 12, 18, 15, 20) Assign a 1 to SD, 2 to D, 3 to N, etc. Questions marked with R are to be reverse scored. So 5 to SD, 4 to D, 3 to N, etc. Or just reverse the Likert scale for questions on the second half. See directions. Entering Data into SPSS Enter titles of the variables into variable view For example Smoking, Littering, Academic and Social Enter values for the data in Data View How to figure out if you want to use an article for your research Skim the article – Six steps: 1. Read the abstract. If you want to continue, proceed to step 2: 2. Identify the main argument (locate the hypotheses, see whether or not they were supported and whether there are implications for future work) 3. Locate the supporting arguments (transitional words/phrases will help you find these). 4. Look for themes – If you see any recurring themes across articles, these may be important to the area of research. Keep an eye on how each article describes/addresses such themes How to figure out if you want to use an article for your research (continued) 5. If you’re still unsure as to what the scope of the paper is, then skim other paragraphs as needed 6. Decide if you want to keep this article as a potential article to read “cover to cover.” If you’ve determined that an article is worthy of reading in its entirety, then you’re ready for the next step– reading critically. Guidelines for Critically Reading Research Articles Take your time! If you rush, you will likely miss important details. Also, many articles require being read more than once. Keep a dictionary nearby – no one knows the definition to every word, and looking up unknown words will help you to better understand the material. Don’t own a dictionary? I like www.yourdictionary.com Be a “healthy skeptic” – as you read, thoroughly and carefully evaluate the claims being presented (e.g., “Is this the best hypothesis for this question?” “Are these results appropriately interpreted or is it a stretch?” “Does this idea/theory make sense?”) Write as you read – underline, circle, jot down notes or questions in the margins or on a separate piece of paper – your opinion of the article matters! You’ll learn much more if you write while you read. Also, don’t just underline/highlight. You really want to be processing the material as you read. Guidelines for Critically Reading Research Articles More questions to ask as you read: Introduction – What question is being addressed (hypotheses)? What sets this study apart from the previous research reviewed in the intro? Are any key terms defined? What answers to these authors hope to find (hypotheses)? Does the literature review provide acceptable context for the research questions? Guidelines for Critically Reading Research Articles (continued) Method – Who participated? How were the variables of interest operationalized? Is there a control group? If not, is there an explanation? Does it make sense?...Were the participants selected, sampled, randomly assigned to groups? Do they represent some larger population of interest? Large enough n? Guidelines for Critically Reading Research Articles (continued) Results – Is there evidence that any manipulations and/or measures worked – what kind of evidence is it? Were the main hypotheses supported? What were the main findings? Even if you are unfamiliar with the statistics used, can you understand the way that the results of the stats are described? Do the tables/figures help? Guidelines for Critically Reading Research Articles (continued) Discussion – In what way(s) does the study add to what we already know about the topic or question? If the study answered any questions, what were they? Did the study lead to any new questions – which? What are the practical implications of the findings? What were the study’s limitations? What studies should be conducted next? Do the authors’ implications make sense? Are they reasonable, or do they stretch the results? How could the study be improved? What changed would you make? Note: These questions are also helpful in writing and revising your own papers!