1. Class 3 Relationship Between Variables SKEMA Ph.D programme 2010-2011 Lionel Nesta Observatoire Français des Conjonctures Economiques Lionel.nesta@ofce.sciences-po.fr

2. Qualitative × Qualitative × Quantitative × Quantitative Which variables are we looking at ? Relationship Between Variables ANOVA

4. ANOVA: ANalysis Of VAriance  ANOVA is a generalization of Student t test  Student test applies to two categories only:  H 0 : μ 1 = μ 2  H 1 : μ 1 ≠ μ 2  ANOVA is a method to test whether group means are equal or not.  H 0 : μ 1 = μ 2 = μ 3 =... = μ n  H 1 : At least one mean differs significantly

5. ANOVA This method is called after the fact that it is based on measures of variance. The F-statistics is a ratio comparing the variance due to group differences (explained variance) with the variance due to other phenomena (unexplained variance). Higher F means more explanatory power, thus more significance of groups.

6. Revenues (in million of US $ ) Sector 1Sector 2Sector 3 Firm 118.021.534.8 Firm 218.021.534.8 Firm 318.021.534.8 Firm 418.021.534.8 Firm 518.021.534.8

7. Revenues (in million of US $ ) Sector 1Sector 2Sector 3 Firm 118.0 Firm 221.5 Firm 325.0 Firm 428.7 Firm 534.8

8. Revenues (in million of US $ ) Sector 1Sector 2Sector 3 Firm 119.623.730.8 Firm 219.428.432.9 Firm 321.928.535.3 Firm 421.231.731.8 Firm 524.637.035.7 Do sectors differ significantly in their revenues? H 0 : μ 1 = μ 2 = μ 3 =... = μ n H 1 : At least one mean differs significantly.

9. ANOVA df = (k – 1) df = n – kdf = n – 1 residual This decomposition produces Fisher’s Statistics as follows:

10. Origin of variationSSd.f.MSSF-StatProb>F SS-between379.12189.6 SS-within (residual)132.51211.0 SS-total511.61436.5417.70.0003 The result tells me that I can reject the null Hypothesis H 0 with 0.03% chances of rejecting the null Hypothesis H 0 while H 0 holds true (being wrong). I WILL TAKE THE CHANCE!!! The ANOVA decomposition on Revenues

11. Comparison of Means Using Student t with STATA We still use the same command ttest ttest var1, by(varcat) For example: ttest lnassets, by(type) ttest lnrd, by(year) ttest lnrdi, by(type) Beware! Unlike ANOVA, Student t test can only be perfomed to compare two categories.

12. ANOVA under STATA We still use the same command anova anova var1 varcat For example: anova lnassets isic anova lnrd isic anova lnrdi isic anova cours titype

13. Stata Instruction Sum of Squares F-Stat P value STATA Application: ANOVA

14. Anova Example in Published Paper

15.  Verify that US companies are larger than those from the rest of the world with an ANOVA  Are there systematic  Sectoral differences in terms of labour; R&D, sales  Write out H 0 and H 1 for each variables  Analyse  Comparer les moyennes  ANOVA à un fateur  What do you conclude at 5% level?  What do you conclude at 1% level? SPSS Application: ANOVA

16. SPSS Application: t test comparing means

18. Qualitative × Qualitative × Quantitative × Quantitative Which variables are we looking at ? Relationship Between Variables Chi-Square Independence Test

20. Introduction to Chi-Square  This part devoted to the study of whether two qualitative (categorical) variables are independent: H 0 : Independent: the two qualitative variables do not exhibit any systematic association. H 1 : Dependent: the category of one qualitative variable is associated with the category of another qualitative variable in some systematic way which departs significantly from randomness.

21. The Four Steps Towards The Test 1.Build the cross tabulation to compute observed joint frequencies 2.Compute expected joint frequencies under the assumption of independence 3.Compute the Chi-square ( χ²) distance between observed and expected joint frequencies 4.Compute the significance of the χ² distance and conclude on H 0 and H 1

22. 1. Cross Tabulation  A cross tabulation displays the joint distribution of two or more variables. They are usually referred to as a contingency tables.  A contingency table describes the distribution of two (or more) variables simultaneously. Each cell shows the number of respondents that gave a specific combination of responses, that is, each cell contains a single cross tabulation.

23. 1. Cross Tabulation  We have data on two qualitative and categorical dimensions and we wish to know whether they are related  Region (AM, ASIA, EUR)  Type of company (DBF, LDF)

24. 1. Cross Tabulation  We have data on two qualitative and categorical dimensions and we wish to know whether they are related  Region (AM, ASIA, EUR)  Type of company (DBF, LDF)

25. 1. Cross Tabulation  We have data on two qualitative and categorical dimensions and we wish to know whether they are related  Region (AM, ASIA, EUR)  Type of company (DBF, LDF)

26. 1. Cross Tabulation  Crossing Region (AM, ASIA, EUR) × Type of company (DBF, LDF)  tabulate continent type

27. 2. Expected Joint Frequencies  In order to say something on the relationship between two categorical variables, it would be nice to produce expected, also called theoretical, frequencies under the assumption of independence between the two variables.

28.  tabulate continent type, expected 2. Expected Joint Frequencies

29. 3. Computing the χ² statistics  We can now compare what we observe with what we should observe, would the two variables be independent. The larger the difference, the less independent the two variables. This difference is termed a Chi-Square distance. With a contingency table of n lines and m columns, the statistics follows a χ² distribution with ( n -1)×( m -1) degree of freedom, with the lowest expected frequency being at least 5.

30. 3. Computing the χ² statistics  tabulate continent type, expected chi2

31. 4. Conclusion on H 0 versus H 1 We reject H 0 with 0.00% chances of being wrong I will take the chance, and I tentatively conclude that the type of companies and the regional origins are not independent. Using our appreciative knowledge on biotechnology, it makes sense: biotechnology was first born in the USA, with European companies following and Asian (i.e. Japanese) companies being mainly large pharmaceutical companies. Most DBFs are found in the US, then in Europe. This is less true now.

32.  Analyse  Statistiques descriptives  Tableaux Croisés  Cellule  Observé & Théorique 2. SPSS : Expected Joint Frequencies

33.  Analyse  Statistiques descriptives  Tableaux Croisés  Statistique  Chi-deux 3. SPSS : Computing the χ² statistics

34. Qualitative × Qualitative × Quantitative × Quantitative Which variables are we looking at ? Relationship Between Variables Correlations

36. Introduction to Correlations  This part is devoted to the study of whether – and the extent to which – two or more quantitative variables are related:  Positively correlated: the values of one variable “varying somewhat in step” with the values of another variable  Negatively correlated: the values of one continuous variable “varying somewhat in opposite step” with the values of another variable  Not correlated: the values of one continuous variable “varying randomly” when the values of another variable vary.

37. Scatter Plot of R&D and Patents (log)

39.  The Pearson product-moment correlation coefficient is a measure of the co-relation between two variables x and y.  Pearson's r reflects the intensity of linear relationship between two variables. It ranges from +1 to -1.  r near 1 : Positive Correlation  r near -1 : Negative Correlation  r near 0 : No or poor correlation Pearson’s Linear Correlation Coefficient r

40. Cov(x,y) : Covariance between x and y  x et  y : Standard deviation of x and Standard deviation of y n : Number of observations Pearson’s Linear Correlation Coefficient r

41.   corr lpat lassets lrd lrdi lpat_assets

42.  Is  significantly different from 0 ?  H 0 : r x,y = 0  H 1 : r x,y  0  t* : if t* > t  with (n – 2) degree of freedom and critical probability α (5%), we reject H 0 and conclude that r significantly different from 0. Pearson’s Linear Correlation Coefficient r

43.   pwcorr lpat lassets lrd lrdi lpat_assets, sig

44.  Assumptions of Pearson’s r  There is a linear relationships between x and y  Both x and y are continuous random variables  Both variables are normally distributed  Equal differences between measurements represent equivalent intervals. We may want to relax (one of) these assumptions Pearson’s Linear Correlation Coefficient r

45. Spearman’s Rank Correlation Coefficient ρ  Spearman's rank correlation is a non parametric measure of the intensity of a correlation between two variables, without making any assumptions about the distribution of the variables, i.e. about the linearity, normality or scale of the relationship.   near 1 : Positive Correlation   near -1 : Negative Correlation   near 0 : No or poor correlation

46. d² : the difference between ranks of paired values of x and y n : Number of observations  ρ is simply a special case of the Pearson product-moment coefficient in which the data are converted to ranks before calculating the coefficient. Spearman’s Rank Correlation Coefficient ρ

47.  spearman lpat lassets lrd lrdi lpat_assets

48. Spearman’s Rank Correlation Coefficient ρ  spearman lpat lassets lrd lrdi lpat_assets, stats(rho p)

49. Pearson’s r or Spearman’s ρ ?  Relationship between tastes and levels of consumption on a large sample? (ρ)  Relationship between income and Consumption on a large sample? (r)  Relationship between income and Consumption on a small sample? Both (ρ) and (r)

50.  Analyse  Corrélation  Bivariée  Click on Pearson Pearson’s Linear Correlation Coefficient r

51.  Analyse  Corrélation  Bivariée  Click on “Spearman” Spearman’s Rank Correlation Coefficient ρ