Using JASP To Conduct a Chi-Square Screencast Transcript

Okay. In this video, we’re going to use JASP to conduct a chi-square statistic. And in this example data, we asked participants two questions. One to rate their physical activity as low, moderate or vigorous. And then we asked them to rate their fruit consumption. How much fruit do you consume in a day and a week? I’m not sure exactly what it was, but they rate their fruit consumption as low, medium or high. And we want to see if those two variables are independent of one another or if there’s some type of relationship there. And since both variables are categorical, we’re going to run a chi-square. And in JASP, you find this under frequencies. We construct a frequency table and it’s one of the options under frequency tables. So I’m going to put fruit consumption as my rows. It’s going to tabulate. I’ll have three rows, low, medium, and high. And then I’m going to put physical activity as my column. So I’m going to have three columns, low, moderate, and vigorous.

Then I’m going to ask for statistics. I’m going to ask for the chi-square. So you see you, I have significant results, but I need a little more data here. So under cells, I can ask for the expected counts. So there is my table. I have a significant chi-square value. And in my opinion, JASP makes you work a little bit hard to figure to interpret this because what I would like as a residual, the difference between the actual count and the expected count, because we usually are expected to interpret the highest residuals, okay? So you can do this by hand and just calculate the difference between 69 and 51. And the difference between 206 and 212, just go down the rows. To save us a little time, I just put it in Excel. You don’t need to do this.

But this is our data in Excel and I added a residual column where I just subtract the expected count from the actual count. And those are my differences. And we can’t do a post hoc test and say, “Where are the significant differences?” We know there’s a difference within these two variables, but we can just report the highest residuals. And where do you draw the line? Do you report all the residuals? Probably not.

In this case we have nine, that’s an awful lot. I would think at least the top three highest residuals would be fair. And in this case, I’ve highlighted those in red. So people who consumed a low amount of fruit and had low physical activity, the actual count was 69. We would expected far fewer. The difference or the residual was 17. And then people who ate a lot of fruit and had low activity, we had an actual count of 14 in that category. We would expected far more and the residual of −12.82. And then the next highest residual was over here in vigorous activity in high fruit consumption. Based on both being equal or both being equivalent, we would’ve expected that count to be 169. In that cell, we actually got 157, so fewer than expected, 11.8, fewer than expected. So just report, in this case, the significant chi-square. And then in this case, it’s up to you. There may be a logical one, maybe all the double digit residuals, but I just reported the top three. And that was residual of 17, 12 and 11.8. So that is using JASP to conduct a chi-square.