Lecture 3 Correlational Research Why Do We Use Correlational Research What happens when you cannot do experimental research cause and effect you want to find out what causes something to happen for ethical or practical you cannot manipulate the independent variable or you cannot control the confounding variable reasons You do correlational research Definition of Correlational Research Correlational research does not give us a cause and effect relationship it gives us a correlational relationship When two things are correlated changes in one are typically accompanied by changes in the other the two things change together Positive and Negative Correlation When one variable increases and the other variable also increases this is said to be a positive correlation between the two variables Positive and negative do not have any connotation of good or bad in this instance Height and weight are good examples of a positive correlation as you grow taller you tend to gain weight When one variable increases and the other variable decreases this is said to be a negative correlation GPA and absence from class are good examples of a negative correlation as absence in class goes up GPA tends to go down Example Universities want to find a measure for success in potential students The most powerful predictor of performance in a class is attendance in the class Statistical Measures of Correlation Correlation is a statistical measure between two variables So the strength of the correlation connection between the two variables ranges between 1 and 1 A correlation of 1 means that two variables are perfectly positively correlated if A goes up B goes up if A goes down B goes down by the same amount Correlations of 1 are very very rare A correlation of 1 means that the two variables are perfectly negatively correlated if A goes up B goes down or vise versa by the same amount A correlation of 0 means that the variables are not associated connected at all Changes in one are not accompanied by any predictable pattern of change in the other the two variables are random with respect to each other 0 8 and 0 8 mean exactly the same thing in terms of strength of the association 1 When Do We Use Correlational Research Disadvantages of Correlational Research Observational studies where data is being gathered by observing people in their natural environment and cataloguing their actions Animal research in the wild living with the animals Child research observing children There are always bias issues in the sense of the people who are doing the cataloguing are so ignorant of how to do what they re doing that the data is worthless Case studies are observations of one subject often used in clinical psychology A single case study as far as justifying a conclusion is pretty much worthless People try to combine case studies to see if different patterns emerge over a number of case studies Surveys and tests polls surveys are structured questionnaires that are used to gather information about peoples attitudes and behaviors that are not observable You cannot act on any data gathered in a psychological study Well done correlational research is good however the best correlational research is not as good scientifically as the best experimental research Why From experimental research you can draw justified causal conclusions you cannot do this in correlational research because of the way the data is gathered and because you cannot control the independent and confounding variables In correlational research you can draw conclusions about correlations between variables taken on its own evidence of a correlation does not provide evidence of a causal connection In correlational research evidence of a causal connection is equally strong in either direction If A is correlated to be the claim that B causes A is equally as strong as the claim that A causes B it can go either way On the basis of a correlation alone you cannot state with an certainty that there IS a causal relationship nor can you state if there is a causal relationship which direction it goes in Example In any given year it is usually the case that over 90 bicycle riders who die were not wearing a helmet Not wearing a helmet doesn t cause you to die dieing doesn t cause you to not wear a helmet With all correlational data you must be careful to look for other things that might be involved One other thing correlated in this example is skill riding a bicycle there is a strong positive correlation between not wearing a helmet and being clueless on the bike Because there is a positive correlation between skill and not wearing a helmet the correlation between death and not wearing a helmet cannot be taken at face value This is generally referred to as the third variable problem 2 Third Variable Problem This is a reason why you can never ever draw causal claims based on correlational data the nature of the data and the analysis that is does in correlational research does not rule out third variables that could be the causal factor you cannot eliminate these factors using correlational design you can eliminate them using experimental design Example foot size and vocabulary size are positively correlated within the child population Having a big vocabulary obviously does not cause your feet to grow you don t store the words in your feet and having big feet does not cause your vocabulary grow There is a positive correlation here but no one in his or her right mind would try to make a causal claim So what s going on Age Little kids have little feet and don t know a lot of words adults have bigger feet than kids and know more words There is a third variable age Pseudo science people claim the mantle of science but the conclusions they reach are not based on sound scientific tests not about what you are studying it s about how you are doing it General Note 3
View Full Document
Unlocking...