1 BIOSTATISTICS 5.6 TEST OF HYPOTHESIS 2 BIOSTATISTICS • TERMINAL OBJECTIVE: • 5.6 Perform a test of significance on a hypothesis using Chi-square test. 3 BIOSTATISTICS • STATE THE PURPOSE OF A: 5.6.1 2X2 contingency table. 5.6.2 2x2 expected table. 4 Purpose - Contingency • General – Public health professionals use contingency tables to display data used in calculating measures of association and tests of statistical significance. 5 Purpose - Contingency – Used to study the association between exposure and disease with the observed frequencies. In basic terms, the observed table shows a relationship between exposure and outcome (ill or well). 6 Purpose - Expected • General – Computes the frequencies we would if there is no relationship between exposure and outcome. – Determines which test statistic is used on the hypothesis. • Chi-square • Fisher's exact test 7 BIOSTATISTICS • 5.6.3 Complete a 2x2 contingency table from observed data. 8 Completing A 2x2 Contingency Table • Data is derived from frequency distribution table, such as a Food Specific Attack Rate Table, or other two variable table. 9 Completing A 2x2 Contingency Table • Basic Format – Composed of four outlined square cells. – Disease status is designated at the top of table. – Exposure status is designated along side of table. 10 Completing A 2x2 Contingency Table Format Outcome Exposure Yes No Total Yes a b H1 No c d H2 Total V1 V2 N 11 Completing A 2x2 Contingency Table • Presenting a 2x2 contingency table – Title • Appropriate for identification. Addresses what, where, and when. • Follows rules of table construction. 12 Completing A 2x2 Contingency Table – Headings • Rows and columns labeled for exposure and outcome, respectively. 13 Completing A 2x2 Contingency Table – Printing • Double line above header, single line below. • No internal lines are needed. • Single line below the row for totals. 14 Completing A 2x2 Contingency Table Example OUTBREAK ASSOCIATED WITH EATING TURKEY, USS ERASMUS B DRAGON, 25 NOV 04 Gastroenteritis Ill Well Total Ate turkey 97 36 133 Did not eat turkey 2 23 25 Total 99 59 158 15 Computing A 2x2 Expected Table • COMPUTE: 5.6.4 Data for a 2x2 expected table. 16 Computing A 2x2 Expected Table • Obtain data from observed table. • Format 17 Computing A 2x2 Expected Table Format Disease Exposure Yes No Total Yes a' b' a' + b' No c' d' c' + d' Total a' + c' b' + d' N 18 Computing A 2x2 Expected Table • Formula – – – – a´ = (H1)(V1)/N b´ = (H1)(V2)/N c´ = (H2)(V1)/N d´ = (H2)(V2)/N – Note: Row and column totals equal the observed totals. 19 Computing A 2x2 Expected Table • Evaluation – If any one of the cells (a´ through d´) is less than 5, the Fisher's exact test is used. – When all cells are 5 or greater, the Chi-Square test is used. 20 Computing A 2x2 Expected Table • Example of expected table Gastroenteritis Ill Well Total Ate turkey 83.34 49.66 133 Did not eat turkey 15.66 9.34 25 Total 99 59 158 21 Computing A 2x2 Expected Table – a' = (133)(99)/158 = 83.34 – b' = (133)(59)/158 = 49.66 – c' = (25)(99)/158 = 15.66 – d' = (25)(59)/158 = 9.34 22 Computing A 2x2 Expected Table • 5.6.5 The value of Chi-square from a 2x2 contingency table. 23 Calculating Chi-square • Once the 2X2 contingency table is completed, Chi-Square is computed by substituting the values in the table into the Chi-Square equation. 24 Calculating Chi-square Equation 2= N[|(a d)-(b c)|-N/2]2 (a+b)(c+d)(a+c)(b+d) 25 Calculating Chi-square • Steps: – – – – Substitute the values into the equation. Perform the functions in the parentheses first. Subtract one-half of N from this total. Square the value within the brackets. 26 Calculating Chi-square – Multiply that number by "N". – Simplify the denominator by multiplying the totals. – Carry out the remaining division. – Round off to the nearest hundredth. 27 Calculating Chi-square • Example using TABLE 5.6A χ2= 158[((97*23)-(2*36))-158/2]2 – 158[2159-158/2]2 – 158[2080]2 – 158[4326400] 28 Calculating Chi-square – 683571200 – 19421325 – 683571200/19421325 = 35.1969 = 35.20 29 Calculating Chi-square • 5.6.6 Define the null (HØ) and alternative (HA) hypotheses. 30 Defining Hypothesis • Definition (statistical) – Statement about the relationship between probability distributions. • Educated guess or an idea as to what may be going on in a particular situation. 31 Defining Hypothesis • Two types – Null (HØ) – Alternate (HA) 32 Defining Hypothesis • Null hypothesis: – There is no association between two factors under consideration. It may be due to chance. 33 Defining Hypothesis • Alternate hypothesis: – There is an association between the factors under consideration. It is not due to chance. 34 Hypothesis • 5.6.7 Interpret the test of significance on the null hypothesis. 35 Hypothesis • Chi-Square test: – Either proves or disproves the null hypothesis. – When the null hypothesis is disproved, then the alternative hypothesis is selected. 36 Interpreting The Test Of Significance • Test of significance – Either proves or disproves the null hypothesis. – When the null hypothesis is disproved, then the alternate hypothesis is selected. 37 Interpreting The Test Of Significance • P value – The P value is the probability that our result will occur due to chance. – Chi-square calculates a value which represents a known P value. 38 Interpreting The Test Of Significance • Interpretation – If the Chi-Square value is greater than 3.84 (P 0.05), then the null hypothesis is rejected and the alternate hypothesis is accepted. • There is a statistically significant association between the two factors. 39 Interpreting The Test Of Significance – If the Chi-Square value is less than or equal to () 3.84, the alternative hypothesis is rejected in favor of the null hypothesis. • The association between the two factors is NOT statistically significant. 40 Interpreting The Test Of Significance – A Chi-Square value > 6.63 (P 0.01) is considered highly significant. 41 You have just finished the last presentation in Biostatistics! Tomorrow: Practice 42 43 You have just finished the last presentation in Biostatistics! Tomorrow: Practice 44