2-Crosstabulation AND Chi-square Test.pptVIP

Cross- Tabulation Chi-square test Cross Tables Cross tables are used to investigate the (possible) relation between two nominal or ordinal variables. Is income class related to store preference? Are retail students more successful in getting a job than hotel students? Cross-table with observed frequencies Example: is income class related to store preference? Low Middle high Huatang 30 55 15 100 Carfour 45 75 30 150 Kelong 15 30 40 85 90 160 85 335 Dependent and independent variables A cross table contains a dependent and an independent variable. The table can be read in two “directions”: 1) is customer income profile different between stores? (stores= dependent) 2) Does store preference differ between income groups?( income groups= independent) Cross-table with percentages low middle high Huatang 33% 34% 18% carfour 50% 47% 35% Kelong 17% 18% 47% 90= 100% 160= 100% 85= 100% 335 Column percentages Row percentages Low Middle High Huatang 30% 55% 15% 100= 100% carfour 30% 50% 20% 150= 100% Kelong 18% 35% 47% 85= 100% 335 Row or Column Percentages? The rule is : Always use percentages in the direction of the independent variable; It is common to put the independent variable in the columns. Chi-square test(1)-variables Two categorical variables, with two or more categories in each. For example: gender (male/female); and smoker (Y/N). The lowest expected frequency in any cell should be 5 or more. Less stringent criteria: at least 80% of cells have expected frequencies of 5 or more. 1X2 or 2X2 table needs at least 10; 2X2 violates the assumption should use Fisher’s exact probability test. Continuity Correction. Chi-square test(1) The Chi-square test( -test) is used in a cross-table to calculate how likely it is that the two variables are not related. First the hypothesis is formulated: H0: The variables are not related H1: The variables are related Chi-square test(2) Next the so-called value is calculated: This -value