One-way ANOVA Introduction Two variables: 1 Categorical variable (factor/IV), 1 Quantitative variable (response/DV) Main Question: Do (the means of) the quantitative variables depend on which group (given by categorical variable) the individual is in? ANOVA looks at differences between groups. Note: We usually refer to the sub-populations or the same population but with different treatments as “groups” when doing ANOVA. Introduction At its simplest ANOVA tests the following hypotheses: H0: The means of all the groups are equal μ1 = μ2 = μ3 = μi Ha: Not all the means are equal Introduction Usefulness: – Similar to t-test – More versatile than t-test – Compare one parameter (response variable) between two or more groups Introduction Why Not Just Use t-tests? – Tedious when many groups are present – Using all data increases stability – Large number of comparisons some may appear significant by chance Introduction Examples: – ”An organization has three different branches. Turnover level differs across the three branches and management wants to know whether this may be explained by the extent to which employees are satisfied with their working environment across the branches. Fifty employees are randomly selected at each branch and given a questionnaire measuring how satisfied they currently are with the working environment”. Introduction – Researchers investigate the effects of control type on firm performance. The research question is whether a real difference exists in performance between owner- and manager-controlled firms (McKean and Kania, 1978). – Investigators want to investigate whether demographic factors (e.g. age groups, races, education level, annual income level, and employment status) and investment experience (novice, intermediate, advance) have influence on retirement planning intention. Introduction – Researchers investigate the behavior of noise traders and their impact on the market. There are three groups in the experiment (accordingly with researchers’ treatments): informed traders (who possess fundamental information), liquidity traders (who have to trade for exogenous reasons), and noise traders (who do not possess fundamental information and have no exogenous reasons to trade); (Bloomfield, O’Hara, and Saar, 2007). – Researchers investigate the impact of moods (i.e. Negative, positive, and neutral) on ethical judgment of auditors (Cianci and Bierstaket, 2009). Introduction The researchers investigate the effects of advertising models’ eye color (blue, green, and brown) in ad viewers responses to the ad (Simpson, Sturgen, and Tanguma) Introduction What can we conclude from the examples? ANOVA Assumptions There are Three basic assumptions used in ANOVA: The populations from which the samples were taken are normally distributed. Homogeneity of variance Random sampling. Notation for ANOVA •n = number of individuals all together • i = number of groups • x = mean for entire data set is Group i has • ni = # of individuals in group i • xij = value for individual j in group i • xi = mean for group i • si = standard deviation for group i How ANOVA works ANOVA measures two sources of variation in the data and compares their relative sizes • variation BETWEEN groups • for each data value look at the difference between its group mean and the overall mean ( xi - x ) 2 • variation WITHIN groups • for each data value we look at the difference between that value and the mean of its group ( xij - xi ) 2 How ANOVA works The ANOVA F-statistic is a ratio of the Between Group Variation divided to the Within Group Variation: F Between Within MSG MSE This compares the variation between groups (group means to overall mean) to the variation within groups (individual values to group means). This is what gives it the name “Analysis of Variance.” A large F is evidence against H0, since it indicates that there is more difference between groups than within groups. Note: it is easier to look at the P-value to indicate whether the H0 is rejected or not If the P-value is less than or equal to a, reject H0. If the Pvalue is greater than a, fail to reject H0. How ANOVA works Step 1: The null hypothesis is H0 : 1 2 3 • Step 2: The alternative hypothesis is H a : not all of the i are equal • Step 3: The significance level is (usually =? is set to one of the values {0.01, 0.05, 0.1} How ANOVA works Step 4: Calculate the F-statistic: F Mean Square Group MSG or Mean Square Error MSE MSG, MSE and the F-statistic are found in the ANOVA table when the analysis is run on the SPSS How ANOVA works Step 5: Find the P-value Step 6. Reject or fail to reject H0 based on the P-value. Step 7. State your conclusion. How ANOVA works Levene’s test: H0: σ12 = σ22 = σ32 = σi2 → Homogeneity of variance Ha: σ12 ≠ σ22 ≠ σ32 ≠ σi2 – Homogeneity fulfilled → Equal variances assumed. – Homogeneity rejected → Equal variances not assumed. Note: •ANOVA is still robust even when the homogeneity assumption is not fulfilled, as long as the sample sizes are roughly equal or the deviation is only of a moderate level. As a rule of thumb, if the largest std.dev < (2 x the smallest std.dev) then we need not to be concerned about this assumption. •Equal variance assumed or not assumed will affect to Post Hoc test methods (p.20) How to perform ANOVA in SPSS? One-way ANOVA – Choose Analyze > General Linear Model > Univariate – Click the DV (only one click) to highlight it and then transfer it to Dependent Variable box by clicking the corresponding arrow. – Doing a similar procedure for IV and transfer it to Fixed Factor(s) box by clicking the corresponding arrow. – After that, click the option button and check for Homogeneity of Variance. Note: SPSS uses a Levene’s test of homogeneity of variance. – Back to the former box. How to perform ANOVA in SPSS? Post Hoc Test: The results from the ANOVA do not indicate which of the three groups differ from one another. To locate the source of this difference we use a post hoc test (commonly Tukey test and the more conservative is Scheffé test; equal variance is assumed in these tests). – Click Post Hoc and check Tukey box, click Continue button. – Last, click OK button and wait a moment while SPSS analyzes the data. Note: • Tukey performs all of the pairwise comparisons between groups. • Scheffe performs simultaneous joint pairwise comparisons for all possible pairwise combinations of means. Can be used to examine all possible linear combinations of group means, not just pairwise comparisons. How to perform ANOVA in SPSS? If equal variance is not assumed, some post hoc tests could be used: – Tamhane's T2. Conservative pairwise comparisons test based on a t-test. – Dunnett's T3. Pairwise comparison test based on the Studentized maximum modulus. – Games-Howell. Pairwise comparison test that is sometimes liberal. – Dunnett's C. Pairwise comparison test based on the Studentized range. How to perform ANOVA in SPSS? One IV or Factor Is F-value significant? Yes No Are there more than 2 groups? Stop Yes No Do Post Hoc comparison Stop How to perform ANOVA in SPSS? Exercise 1: Open job satisfaction.sav An organization has three branches in three different region. Management wishes to know whether employees are satisfied with their job differs across regions. A total of 218 employees are randomly selected at the regions and given a questionnaire measuring how satisfied they currently are with their job”. How to perform ANOVA in SPSS? Does management find evidence that employees’ job satisfaction differs across regions? Which branch differs from the others? How to perform ANOVA in SPSS? This is how the data set is shown How to perform ANOVA in SPSS? How to perform ANOVA in SPSS? Transfer Satisfaction variable to dependent variable box and region variable to Fixed Factor(s) box. After that click Options How to perform ANOVA in SPSS? Check the homogeneity check-box and after that click Continue How to perform ANOVA in SPSS? Click Post Hoc… How to perform ANOVA in SPSS? 1. Transfer Location variable from factor(s) to Pos Hoc Tests for: 2. Check the Tukey Check-box 3. Click Continue How to perform ANOVA in SPSS? Click OK and wait a minute How to perform ANOVA in SPSS? The number of sample in each region Homogeneity test’s result P-value for Levene’s Test Ho: σ1 = σ2 = σ3 Ha: At least one σ is different than the others How to perform ANOVA in SPSS? Result of ANOVA P-Value for ANOVA Ho: μ1 = μ2 = μ3 Ha: At least one μ is different than the others Conclusion: There is a difference in employees’ job satisfaction across regions. How to perform ANOVA in SPSS? South region is significantly different from others How to perform ANOVA in SPSS? Exercise 2: Open Training.xlsx file. Read the instruction in the Training.xlsx file and the raw data. Open Training.sav file. Observe how we handle the raw data and convert it into three treatments in order to analysis it using ANOVA. Perform the ANOVA test using file Training.sav. Answer the questions. Report this exercise 2 in written form by the end of this course week. Test yourself What is ANOVA? Why do we use ANOVA? What are ANOVA assumptions? How to test ANOVA assumptions? What do we do when the equal variance is not fulfilled? What does it mean when the F value in ANOVA result is statistically significant? What does the post hoc test answer? References Agresti, A. (2007) Ch 12: Comparing group: Analysis of Variance (ANOVA) method, Retrieved on 26/04/2012, from http://www.stat.ufl.edu/~aa/sta6127/ch12.pdf. Bloomfield, R., O’Hara M., and Saar G. (2007) ”How Noise Trading Affects Markets: An Experimental Analysis”, Available at SSRN: http://ssrn.com/abstract=994379 or http://dx.doi.org/10.2139/ssrn.994379. Cianci, A. and Bierstaker, J. 2009. "The Effect of Client Importance and Performance Feedback on Auditors' Technical and Ethical Judgments." Managerial Auditing Journal, Vol. 24 Iss: 5, pp.455 – 474. Ghozali, I. (2005) Multivariate analysis application with SPSS, Diponegoro University Publishing, Semarang. Ho, R. (2006) Handbook of univariate and multivariate data analysis and interpretation with SPSS, Taylor & Francis Group, Boca Raton, FL. McKean, J. R., and Kania, J. J. (1978) “An Industry approach to owner-manager control and profit performance”, Journal of Business, Vol. 51 No. 2, pp. 327-342. Murray, J. (2010) Analysis of Variance – Homework and Exam, Retrieved on 27/04/2012, from http://www.murraylax.org/bus735/fall2010. Pruim, R. (nd) ANOVA: Analysis of Variance, Retrieved on 30/04/2012, from http://www.calvin.edu/~rpruim/courses/m243/F03.