You are studying the relationship between productivity and length of employment. Both variables are coded into three categories. You compute a chi-square test of independence and find the observed significance level to be 0.35. A personnel consultant does a similar study using the same criteria categories. He calculates a chi-square test of independence and finds an observed significance level of 0.002. You examine his results and notice that the percentage of cases in each cell of the table is almost identical to the percentages that you observed. You conclude that he doesn't know how to calculate a chi-square value. Give another explanation why his chi-square could differ from yours.