• If we forget about gender and lump people together, then still 50% like SUVS and 50% prefer Sports Cars.
• Look at the Male Column:
• Only 20% like SUVs. Most like Sports Cars!
• It is reversed for the women!
• But - if we combine them it, is still a 50-50 split.
• Each column sure does differ from 50-50, doesn't it?
• They look significantly different from 50-50.
• That's a visual Chi-square.
• Car preference is Dependent on Gender.
• Gender flunked the Test of Independence
• But car dealers in this scenario better know about gender. {sentence is a bit awkward, maybe drop the "but" and say the dealers better know about this "gender difference"?}

You can see that the percents {percentages?} of men in women in each cell

(little box) are quite different from 50-50. Thus the O (observed) {data}

differs quite a bit from the E (expected).

Chi-square would be significant.

Nuances of Calculating the Chi-square!

This page isn't about such. It is about the the {extra the} concept. But for those of you who are interested.

You need to calculate the expected for each cell, separately.

Here's the formula: