Quasi-Experimental Design Study Guide PDF

Alternatives to Experimentation: Correlational and Quasi Experimental Objective Objective Objective Understand how the Learn...

Alternatives to Experimentation: Correlational and Quasi Experimental Objective Objective Objective Understand how the Learn how causal Learn more results of these models can ca be that do techcniques nonexperimental constructed from not manipulate techniques may land correlation-based antecedent conditions: may not) be designs correlations, other interpreted. correlational-based methods, and quasi experimental How do quasi-experiments differ from actual experiments? Quasi, in Latin, means "seeming like." Quasi-experiments superficially resemble experiments, but lack their required manipulation of antecedent conditions and/or random assignment to conditions. How do quasi-experiments differ from actual experiments? They may study the effects of preexisting antecedent conditions--life events (living through 9/11) or subject characteristics (having autism)- on behavior A quasi-experiment might comparethe incidence of Alzheimer's disease in patients who used ibuprofen since age 50 and those who did not How do quasi-experiments differ from actualexperiments? In experiments, researchers randomly assign subjects to antecedent conditions that they create. An experiment might randomly assign subjects to either daily ibuprofen or aspirin use, and then measure their incidence of Alzheimer's How do quasi-experiments differ from actual experiments? We should use quasi-experiments when we cannot or should not manipulate antecedent conditions. Quasi-experiments could study the effect of spouse abuse on the frequency of child abuse. Problems with GQuasi Cannot establish cause with certainty(cannot be sure that spousal abuse caused the child abuse) Lacks internalvalidity, that is, the ability to conclude with confidence that the antecedent condition caused the observed differences in behavior However is higher in external validity, or generalizability, than lab experiments Are low in manipulation of antecedents, but high in imposition of units Correlations observ. servations Correlations Correlationsare used to establish relationships among preexisting variables Can be used to predict one set of behaviors from another, for instance, can predict college grades based on high school grades. Showsthe relationship between antecedent conditions and behavioral effects but the antecedents are preexisting, not manipulated. Low manipulation of antecedents, high imposition of units. Cannot be sure of cause Has poor internal validity but good external validity Correlations Acorrelational study is one that is designed to determine the correlation, or degree of relationship, between two traits, behaviors,or events. When two things are correlated, changes in one are associated with changes in the other In a correlational study, selected traits or behaviors of interest are measured first. Numbersor scores are recorded that represent the measured variable. Next the degree of relationship, or correlation, between the numbers is determined through statistical procedures. Correlations 13-15/53 One a correlation is known it can be used to predictions. If we know a person's score on one variable, we can make a better prediction of that person's score on another measure that is highly related or correlated to it. The higher the correlation, the more accurate the prediction will be. Correlation example Suppose a researcher wonders if there is a relationship between watching SesameStreet and Vocabulary in preschoolers.Might ask parents to list as many words that their preschoolers know and ask how frequently they watch SesameStreet. One variable is hours per week spend watching SesameStreet, other variable is the number of words the child knows. Then you would take these numbers, or data, and run a statistical procedure for all of the children in your study. Calculating Correlations One statistic that is used to calculate a correlation is called the Pearson Product Moment Correlation Coefficient or (r) When r is computed, there are three possible outcomes. The correlation can be positive, negative or no relationship Calculating Correlations The values of a correlation coefficient can only range between -1.0 and +1.0. The sign, + or -, tells us whether the relationship is positive or negative. The absolute value of r tells us the strength of relationship. -.34 or +.16 which is stronger? Calculating Correlations A Pearson correlation coefficient is used to calculate simple correlations (between two variables) and may be expressed as: n(50) = +.70, p=.001 Correlation coefficients have for properties. Linearity, sign, magnitude, and probability. Scatterplot A scatterplot can be created to demonstrate the direction of a correlation. It is often the first step in analyzing the correlation. Each dot stands for a person's scores.Each person has 2 scores Scatterplot Can draw a line through the scatterplot. These lines are called regression lines or lines of best fit.. The direction of the line demonstrates the direction of the correlation Weight Exercise vs.  Scatterplot Can drawa line through the scatterplot. These lines are called regression lines or lines of best fit. The direction of the line demonstrates the direction of the correlation Weight vs.Exercise Run Week Kilometers per eee Guess the correlation B 20-22 /53 Positive Correlation If the r value is positive, then there is a positive correlation between the variables. As one variable increases, the other increases to0. Also, as one variable decreases, the other decreases too. As hours of SesameStreet viewing increase, vocabulary increases. As hours of Sesame Street viewing decrease, vocabulary decreases The absolute value of r tells how strong the relationship is. The closer it is to 1.00, the stronger it is. Strong relationships allow for good prediction Negative Correlation If the r value is negative, then there is a negative correlation between the variables. This is called an inverse relationship. As one variable increases, the other decreases. As hours of SesameStreet viewing increase, vocabulary decreases. The absolute value of r tells how strong the relationship is. The closer it is to 1.00, the stronger it is. Strong relationships allow for good prediction. No relationship If the absolute value of correlation is close to 0, then there is no relationship between the variables. Sesame Street viewing has no effect on vocabulary. Curvilinear relationships Sometimes a correlation coefficient value can be close to zero and it would appear that there is no relationship between variables. However, there may be a curvilinear relationship demonstrated by the scatterplot. Curvilinear relationships Properties of a correlation Linearity - means how the relationship between X and Y can be plottedas a line (linear relationship) or a curve (curvilinear relationship) Sign - refers to whether the correlation coefficient is positive or negative Magnitude - is the strength of the correlation coefficient, ranging from -1 to+1 Probability - is the likelihood of obtaining correlation Properties of a correlation Linearity - means how the relationship between X and Y can be plotted as a line (linear relationship) or a curve (curvilinear relationship) Sign - refers to whether the correlation coefficient is positive or negative Magnitude - is the strength of the correlationcoefficient, ranging from -1 to +1 Probability - is the likelihood of obtaining correlation coefficient of this magnitude due to chance What does scatterplot show? Scatterplots - are a graphic displayof pairs of data points on the x and y axes. A scatterplot illustrates the linearity, sign, magnitude, and probability (indirectly) of a correlation. How do outliers affect corre 27-29/63 ? Outliersare extreme scores. They usually affect correlationsby disturbing the trends in the data. Causation  Causation Causation Correlation does not imply causation. Even a perfect correlation, if it exited,does not indicate a causal relationship. Even though a strong relationship exists between two variables, we cannot say that one cause the other. There are other possiblevariables that were not measured, that could have caused the effects. Causation Research on firmness of handshakes and positivity of first impressions found a positive correlation. However, isn't it possiblethat people who shake firmly are very extraverted and it is their extraversion that creates the good impression? Apositive correlation exists between the number of cars built and the number of airplanes built, but one doesn't cause the other. Why should we compute the coefficient of determination? Once we calculate r, we can then calculatethe coefficient of determination. The coefficient of determination (r2) estimates the amount of variability that can be explained by the predictor variable. It is an estimate of strength ifris.56 then r2 is.31 For example, Chaplin et al. (2000) showedthat handshake firmness accounted for 31% of the variability of first impression positivity. Why doesn't correlation prove causation? Since correlational studies do not create multiple levels of an independent variableand randomly assign subjects to conditions, they cannot establish causal relationships. Why doesn't correlation prove causation? There are three additional reasons that correlationscannot prove causation: (1)causal direction - in a correlation, we cannot be sure which variable is the cause and which is the effect. (2) bidirectional causation - both variables could cause the other variable (3)the third variableproblem - there could be some other variablethat is the cause that we have not measured Why doesn't correlation prove causation? Causal Direction Since correlationsare symmetrical, A could cause B just as readily as B could cause A. Does insomnia cause depression or does depression cause insomnia? Why doesn't correlation prove causation? Bidirectional causation Two variables- Insomnia and depression- may affect each other Why doesn't correlation prove causation? Third variable problem A third variable-familyconflict- may create the appearance that insomnia and depression are related to each other Linear Regression Analysis When two behaviors are strongly related, the researcher can estimate a score on one of the measured behaviors from a score on the other This technique is called linear regressionanalysis If we knew that time watching tv was correlated strongly with scores on vocabulary test, we could substitute someone's viewing time into an equation for the regression line which would give us an estimate of what that person's performance should be on the vocabulary test. Multipleregression When more than two related variablesare correlated,a multiple regression can be used. Researchers use multiple regression to predictbehavior measured by one variable based on scores on two or more other variables We could estimate vocabulary size using ageand television watching as predictor variables. 38-40/53 Quasi Experimental Designs Lacks important elements of experiments, such as manipulation of antecedents or random assignment to treatment conditions. Quasiexperimental designs don't look for relationships between variables like correlations; instead we are comparing different groups of subjects looking for differences between them, or looking for changes over time the same group of subjects. With quasi experimental designs we never know for sure what causes the effects that we see. Therefore, they are low in internal validity. Ex post facto design Ex post facto means "after the fact." A researcher examines the effects of already existing subject variables (like gender or personalitytype), but does not manipulate them. For example, women who a researcher wants to study are divorced to see if they are more pessimisticabout marriage than women who are not divorced and who are single. The divorce is the preexisting variable or subject variable. Ex post facto design Like correlations, there is no cause. So cannot say that divorce causes changes in attitudes. There could have been a third variable which caused the result. Ex post facto studiesare low in internal validity. Nonequivalent groups design Anonequivalent groups design compares the effects of treatments on preexisting groups of subjects. A researcher could install fluorescent lighting in Company A and incandescent lighting in Company B and then assess productivity. Cannot be sure that the lighting made the difference;it could be that onecompany is threatening layoffs, so workers are being more diligent. Also has low internal validity Longitudinaland Cross-sectionalappro 42-44/53 In longitudinal designs, the same group of subjects is measured at different points of time to determine the effect of time on behavior. In cross-sectionalstudies,subjects at different developmental stages (classes) are compared at the same point in time. Longitudinaland Cross-sectionalapproaches In longitudinaldesigns, the same group of subjects is measured at different points of time to determine the effect of time on behavior. In cross-sectionalstudies,subjects at different developmental stages (classes) are compared at the same point in time. Pretest/posttestdesign In pretest/posttest designs, a researcher measures behavior before and after an event. This is quasi experimental because there is no control condition. For example: Practice GRE test 1 six-week preparation course Practice GRE test 2. Which problems reduce its internal validity? There is no control group which receivesa different level of the IV (no preparation course). The results may be confounded by practice effects (also called pretest sensitization) due to less anxiety during the posttestand learning caused by review of pretest answers. Practice effects - people do better the second time they take an intelligence test, even when there is no special training in between. What is a Solomon 4-group design? This variation on a pretest/posttestdesign incl 45-47 /53 conditions: (1)a group that received the pretest,treatment and post test (2)a nonequivalent control group that received only the pretest and posttest (3)a group that received the treatment and a posttest (4)a group that only received the posttest