# Statisticians: would this look reasonably smart, stupid, or simply nonsensical?

**URL:** <https://boards.straightdope.com/t/statisticians-would-this-look-reasonably-smart-stupid-or-simply-nonsensical/608129>\
**Category:** In My Humble Opinion\
**Created:** [January 2, 2012, 9:58am UTC](https://boards.straightdope.com/t/statisticians-would-this-look-reasonably-smart-stupid-or-simply-nonsensical/608129 "2012-01-02T09:58:24Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![Koxinga](https://avatars.discourse-cdn.com/v4/letter/k/4af34b/32.png) [@Koxinga](https://boards.straightdope.com/u/Koxinga)\
**Post date:** [January 2, 2012, 9:58am UTC](https://boards.straightdope.com/t/statisticians-would-this-look-reasonably-smart-stupid-or-simply-nonsensical/608129/1 "2012-01-02T09:58:24Z")

</div>

It’s been years since I studied statistics, but I’d like to put a simple statistical illustration in a short report I’m writing. I’m not basing my whole analysis on this, but I’d like to spice it up with this kind of chart. And yet, I’ve forgotten enough statistics that I must turn to you to ask, would I be demonstrating proficiency or ignorance here?

To wit: I’d like to create two linear regressions for similar sets of data, to the effect that Set A is demonstrated to have a closer relationship with “variable X” than Set B. I’m not actually interested in “variable X” but I want to show that its influence is decreased for Set B, such that I can support the idea that _another_ variable, call it “variable Y,” is at work in B but not in A.

And I’ll do so just by presenting the two simple scatter plots and saying “Set A shows an r-squared of 0.792 [real value] and Set B has an r-square of 0.654 [ditto], so _something else_ must be at work for Set B.” And then go on to talk about that something else.

Smart, stupid, or does it not make sense at all?

---

<div class="post-metadata">

**Author:** ![Iridescent\_Orb](https://avatars.discourse-cdn.com/v4/letter/i/9de053/32.png) [@Iridescent\_Orb](https://boards.straightdope.com/u/Iridescent_Orb)\
**Post date:** [January 2, 2012, 1:10pm UTC](https://boards.straightdope.com/t/statisticians-would-this-look-reasonably-smart-stupid-or-simply-nonsensical/608129/2 "2012-01-02T13:10:16Z")

</div>

Perhaps you could add variable Y to both regressions. Ideally you will see the coefficient for X (and its significance) is greater in Set A than in Set B, and the coefficient for Y (and its significance) is greater in Set B than in Set A. This would provide good support for your hypothesis.

---

<div class="post-metadata">

**Author:** ![Baracus](https://avatars.discourse-cdn.com/v4/letter/b/2bfe46/32.png) [@Baracus](https://boards.straightdope.com/u/Baracus)\
**Post date:** [January 2, 2012, 1:49pm UTC](https://boards.straightdope.com/t/statisticians-would-this-look-reasonably-smart-stupid-or-simply-nonsensical/608129/3 "2012-01-02T13:49:26Z")

</div>

I am a little bit rusty on stats, but I don’t think what you have is sufficient. You have shown that there is a difference between set A and set B but I think you need more analysis to show that it is a statistically significant difference. Hopefully someone who knows more stats can tell you how to do that.