Thank goodness for laptops and wireless internet. For this post I had to dress my young, whippersnapping self as a black stew-pot and climb onto a very high horse. Balancing a desktop would have been nigh impossible, especially with all the kettles watching. This pot is stewing a rant (with a soupçon of rave) on some basic number sense.
In the tapestry of words, numbers are “hard” and “empty” pieces of glass capable of shredding the entire cloth. They can be woven into the fabric successfully, but only with careful attention to their particular natures. Take “five,” for example. It’s “hard” because its specificity—one more than four, one less than six and not two, not three, not seventy-eight, not anything else but five—accompanies it everywhere it goes. It’s “empty” because it doesn’t carry anything else: it’s not inherently big or small or good or bad or whatever. Attempting to exploit numbers’ connotation of hardness without accommodating their specificity or emptiness undermines our conversations.
Which brings me to the title. I have encountered, in both cyber- and meat-spaces, sentences something like this: “The fact that [number] women spoke at general conference shows that the church [verb] women.” There were 561 talks in the 20 conferences from April 1997 to October 2006, of which women gave 41, or 7%.  To begin to deploy this number meaningfully we must explain what it represents and establish a comparison value.
There’s more than one way to count talks and sex, so we should announce which way we picked, usually in long, dry footnotes. In some cases, determining sex is difficult—witness many family history efforts (Holy gender confusion, Batman!)—but it’s not a problem for conference.  Whether to count audit and statistical reports as talks is not so obvious. If I’m analyzing what portion of formal instruction comes from women, it could make sense to exclude the reports as not representing authoritative exposition; if I’m studying role modeling I might include them since there’s no doctrinal reason a woman couldn’t chair the audit committee or—for the love of all that is virtuous, lovely, and gender-stereotyped—be the secretary. Whatever the choice, the reader has to know it, just as they must be painfully clear about any formulas. As small as 7% is, it’s even smaller if you think we’re including the non-general meetings in the total.  Then again, I might measure something else entirely. For some situations, time at the pulpit could be a more useful metric than trips thereto. 
Numbers, being “empty,” also (almost) always need context; for that, they need another number, the expected value. Three examples illustrate how the expected value can change a number’s meaning. The first two expectations are, respectively, that the overall percentage of talks given by women tracks the sex representation in (1) the general membership, which is (let’s say) 50% female, and (2) the conference speaker pool, which includes, on average, 117 individuals, of whom 9, or 8%, are female.  The third expectation attempts to control for rank by counting only the horizontal part of the org-chart, where 15 general officers, of whom 9, or 60%, are female, fulfill (approximately) parallel assignments. 
Compared to the 50%-female general population, the 7% female speaker frequency indicates that women are under-represented by a factor of seven. Compared to the 8%-female conference speaker pool, the 7% female speaker frequency shows women and men to be near parity. For the third example, the target group gave 47 talks, of which 41, or 87%, were by women.  Thus, compared to the 60% expected value, women speak too often in conference by a factor of about 1.5. Three different expectations lead to three very different conclusions.
Take-home message: It is not meet that numbers should be alone. If you find a lone number in the gardens of your writing, give it companionship meet for its needs: (1) a clear explanation of how and why, of all the numbers, in all the worlds, you chose this one and (2) an expected or comparison value.
Intelligible conversation requires a common language. Since numbers are “empty,” we must redefine what they represent for each conversation. If we don’t, we’ll talk past each other with the equivalent of mutually unintelligible grunts; if we do, we free ourselves to focus on our differing assumptions—where most conversations break down—and maybe to come to a greater understanding of each other and the cosmos.
And there you have it. I could probably rant some more (sig figs, anyone? polygamy rates? missionary numbers?), but I’ll slither off the high horse for now and go plan how to avoid using numbers in public for a while so as not to embarrass myself. I’m afraid to look at my last post, which includes several numbers.
 Gender representation in conference is, I think, an important question with serious consequences. I am using it here, however, merely as an example. I counted only talks given in the four general sessions of each conference and excluded audit reports, statistical reports, sustainings, prayers, and videos. I used the tables of contents for the conference editions of the Ensign available at lds.org.
 Determining speaker sex from the conference editions of the Ensign is easy due to gendered titles, dimorphic dress (the report has pictures), and pre-exclusion of the sexually ambiguous.
 Including P, RS, and YW meetings while otherwise counting as before, the total is 764 talks, with 104 women-delivered, or 13.6%. So, if you assume I’m counting the non-general meetings as part of conference, you’ll understand the female speaker frequency to be about half what it actually is.
 Suppose, hypothetically, that thirty women each gave a two minute talk and fourteen men occupied the remaining seven hours. Even though 68% of the speakers were women, the portion of the total time they occupied (13%) tells us more about the volume of doctrine coming through women. The relative time per capita tells us something different yet.
 The conference speaker pool includes general authorities and general officers. The number of authorities changes every conference but did not seem to shift significantly over the decade in question. Using the “General Authorities” chart from the 2001 through 2006 conference editions and the fact that the general presidencies were continuously staffed in this period, I calculated the average number of potential conference speakers to be 117, taking care not to double-count Seventies who were simultaneously serving as general officers. The 9 female speakers come from the Relief Society (3), Young Women (3), and Primary (3) general presidencies. The male speakers come from the 1st Presidency (3), Quorum of the Twelve (12), 1st and 2nd Quorums of the Seventy with Presidents (average, 87), Presiding Bishopric (3), and Young Men (3) and Sunday School (3) general presidencies. This method of counting underestimates the size of the speaker pool since the published chart reflects the changes made at the conference while the speaker pool draws from those on the chart plus any who were released at that conference. I predict (without actually counting) that the undercount is negligible (in terms of overall sex representation percentages).
 The officers are the RS, YW, YM, SS, and Primary general presidencies. Recall that before 2004 the YM and SS officers were also Seventies.
 As before, I only counted talks given in the four general sessions. The number of talks given per organization was: YM, 1; SS, 5; RS, 14; YW, 14; P, 13. The premise of the example is compromised by the fact that of the six talks given by males, three were given by officers who were simultaneously one of the Presidents of the Seventy.