negative correlation between the number of self-check unlimi

[baronbaik]

"There is negative correlation between the number of self-check unlimited lanes and the average customer age"

Screenshot:
http://www3.picturepush.com/photo/a/132 ... -IR-10.png


My question:
Store G has self-check unltd lanes 0 and avg customer age 38.8.
Store E has self-check unltd lanes 4 and avg customer age 42.5.

Why isn't that enough to say: False. This isn't negative correlation?

[RonPurewal]

"correlation" doesn't mean "absolutely perfect linear relationship" ... it just means "correlation".

e.g., most taller men have bigger feet, and most shorter men have smaller feet. so, there's a correlation between height and shoe size.
now, i'm only 5'10" and wear a size 15 shoe (yes, i'm a freak), and my brother is 6'4" and wears a size 12 shoe. while weird, these observations certainly don't destroy the entire correlation between shoe size and height.

i think part of the problem here is that you didn't sort the table according to either of the two things that are actually being compared here.
try sorting the table according to either "self-check unlimited lanes" or "average customer age", and the general correlation -- even if there are one or two exceptions -- should be quite clear.

[baronbaik]

Thanks for getting back to me, I appreciate it very much.

try sorting the table according to either "self-check unlimited lanes" or "average customer age"

I believe I sorted by "self-check unlimited" (third column from left), isn't that correct?

"absolutely perfect linear relationship" ... it just means "correlation".

(...)

-- even if there are one or two exceptions -- should be quite clear.

Fair enough, one or two exceptions are okay. To help me with my takeaway, where do I draw the line?

Should I assume that if most of the items follow the linear relationship, then the relationship is correlated?

[tim]

Yes, it looks like you correctly sorted by self-check unlimited; I'm not sure what your question is beyond that.

As for a correlation, you will almost certainly fail this type of question if you are looking for a hard cutoff point regarding how much the data have to fit a particular pattern. The whole point of correlation (at least on the GMAT) is to see if there is a GENERAL trend in the data - one variable GENERALLY increasing or decreasing as the other one increases or decreases. If you're honest with yourself, you'll probably realize you have no trouble identifying whether a general trend exists.

[ernieedwardo856]

That's good i appreciate your information.

[RonPurewal]

.

[keshavlalitchaudhary]

If you're honest with yourself, you'll probably realize you have no trouble identifying whether a general trend exists.

Is there a definite rule/approach we can follow? What do you mean being 'honest with yourself'? I am trying to answer this question not find the meaning of life or get in touch with God.

[cgentry]

I believe Tim's comment was merely meant to indicate that the GMAT IR questions will use data sets that either (a) show fairly blatant relationships, if you sort by the relevant category, or (b) show fairly blatant lack of relationships.

In other words, "if you're honest with yourself" probably meant "let go of the need for an absolute, unvarying line of data" and allow for the fact that real life data is messy. This means IR data sets will have some messiness. But the messiness of the data set won't obscure any existing trend.