LSAT trends show increase in test-takers and project modest 2018 JD enrollment increase

In my last post, I looked at the law school enrollment figures for 2017. What might happen in 2018?

While LSAT test-takers are up, it's worth emphasizing that an increasing percentage of test-takers are repeaters, not first-time test-takers. On the flip side, the number of schools accepting the GRE as an alternative to the LSAT may understate the number of law school applicants next year.

More importantly than LSAT test-takers increasing, however, is their quality. I emphasized this years ago: the quality of the applicant pool matters in much the way that the quantity does. Professor Jerry Organ has helpfully examined the increase in quality.

(It's worth noting that LSAC changed its data for law school applicants in 2016; it explains, "Archived data for 2015 and prior years include applicants for the fall term only and also include deferrals; therefore, archived data are not comparable to current data." They are, however, close enough for our present comparative purposes; and 2016-2017 are comparable, albeit I only have an estimate for 2017 right now.)

Let's also provide some comparisons in recent LSAT & enrollment data. We saw 1L JD enrollment largely flat for the fourth straight year, and the overall law school enrollment figure may well have bottomed out.

But LSAT test-takers have increased each year since 2015: from 101,600, to 105,900, to 109,400, with a projected 125,000 test-takers this cycle. LSAT test-takers are not proportionately translating into applicants; indeed, despite a 3.3% increased in LSATs administered last year, applicants actually declined slightly, and matriculants increased only 0.8%. Part of this, as I've identified, is attributable to increased numbers of repeaters taking the LSAT. But there are other reasons why LSATs administered are not translating into applicants--reasons I could only speculate about at this time. In part, low quality test-takers may have contributed to inflated LSAT statistics, but we may be seeing a reversal.

That said, surely such a significant increase in the percentage of LSAT test-takers would yield at least some increase in applicants and matriculants--particularly given the quality of those test-takers. Only time will tell. For now, stagnant JD enrollment is the status quo, and law schools can look forward to a glimmer of hope for some improvement in 2018.

2017 law school enrollment: JD enrollment flat, nearly 1 in 7 are not in the JD program

The 2017 law school enrollment figures have been released, and they reveal flat JD enrollment and a sharp uptick in non-JD enrollment.

In contrast, total JD enrollment is at its lowest point since 1974, when 105,708 students were enrolled in just 157 ABA-accredited law schools. Enrollment dropped slightly from last year, down to 110,156.

1L enrollment is actually slightly up, from 37,107 last fall to 37,398 this year. It's the fourth straight year of enrollment in the 37,000-range.

Earlier I predicted that non-JD legal enrollment would decline this year due to uncertainty in immigration and travel rules from the new presidential administration. That is emphatically not the case. Instead, there's a whopping 20% increase in non-JD enrollment, from 13,677 in 2016 to 16,428 this fall. Perhaps some of this arises from the jump in non-JD online degrees, particularly "masters of legal studies"-type degrees.

The growth has been explosive in recent years. When coupled with the decline and flattening of JD enrollment, the relative figures are, in my view, staggering. 13% of all students enrolled in law schools are not a part of a JD program--nearly 1 in 7 students. That's up from 11% last year, 10.3% in 2015, and 9.1% in 2017.

I've earlier wondered about a coming reckoning for non-JD legal education, a market largely unregulated by the American Bar Association and with essentially no disclosure of student inputs or outcomes. And I wonder how long this trajectory might continue.

In light of this enrollment data, I'll shortly project some things about the Class of 2018.

Is the ABA any good at evaluating judicial nominees?

The American Bar Association ("ABA") has long been evaluating federal judicial nominees, and it has received some scrutiny for how it goes about doing so. There have been empirical studies to show that Republican-nominated judicial candidates tend to receive lower scores than Democratic-nominated candidates, which offer their own limitations.

...as an aside, I've also found it interesting to dig through the ratings of those who appeared on President Donald Trump's "list" of prospective Supreme Court nominees:

Brett Kavanaugh: revised rating Q (sm), WQ (min) (backstory on downgraded rating here)

Thomas Hardiman: Q (sm), NQ (min) [on nomination to Third Circuit, WQ (1 abstention)]

Raymond Kethledge: WQ (sm), Q (min)

Amul Thapar: WQ (1 abstention)

Diane Sykes: WQ (sm), Q (min)

Steven Colloton: Q (sm), WQ (min), NQ (min)

Raymond Gruender: Q

Neil Gorsuch: WQ

Timothy Tymkovich: Q (sm), NQ (min)

Bill Pryor: Q (sm), NQ (min)

Federico Moreno: Q

To be fair, there are different traits that might make one a good district court judge, appellate judge, and Supreme Court justice. But it's worth noting, I think, that the very candidates a Republican presidential administration considers as most worthy of a Supreme Court nomination received, on the whole, fairly middling grades from the ABA.

Back to the topic at hand. I want to set those aside for a moment these debates, and look instead at something else. Is the ABA any good at doing what it purports to do?

As the Standing Committee on the Federal Judiciary reports, "the Committee focuses strictly on professional qualifications: integrity, professional competence and judicial temperament." The goal" to "ensure that the most qualified persons serve on the federal judiciary."

Is the ABA any good at that?

In part, that's because the ABA is typically looking backward at a candidate's record, then trying to project it forward to how the ABA believes that person will behave as a judge. It might be the case that past performance is an indicator of future success, but it also might be the case that the ABA is relying on weak measures of "qualifications."

One problem is the "rating" system itself, which lacks any nuance and instead offers the kind of thumbs-up/thumbs-down (and thumbs-sideways) of a movie review. Yes, there are probably several minutes of thoughtful film commentary that could precede that final rating, but, here, the ABA actually leaves all that commentary on the cutting room floor. All we have are opaque inputs and a single output.

One of the criteria that the ABA uses is "experience," and it includes some hard-and-fast proxies for experience: "The Committee believes that a prospective nominee to the federal bench ordinarily should have at least twelve years’ experience in the practice of law." This isn't a terribly thoughtful criterion, even if it has the advantage of being a fairly clear rule. That said, one would be hard-pressed to think a rule like this does very much to fill out the term "qualified" or "not qualified." After all, Roger Ebert might well have said, "If a movie comes in under an hour twenty, I give a thumbs down. But if we have no rush to get younger judges, then perhaps it's a fairly harmless criterion.

Additionally, the committee makes other kinds of ex ante determinations about what makes a good judge, like "substantial courtroom and trial experience as a lawyer or trial judge is important." These tend to skew the judiciary toward those with more practical experience, true; it also skews toward litigators and trial lawyers. For appellate judicial nominees, the ABA places "somewhat less emphasis on the importance of trial experience as a qualification for the appellate courts." It prizes certain types of experience: "While the Committee recognizes that civic activities and public service are valuable experiences for a prospective nominee, they are not a substitute for significant experience in the practice of law in either the private or public sector."

For those presidents who pre-screened their lists of applicants with the ABA, the results can be frustrating. President Barack Obama saw the ABA reject 14 of his prospective judicial nominees as "not qualified" his first three years in office. As Obama administration officials complained, "In particular, they have questioned whether the panelists — many of whom are litigators — place too much value on courtroom experience at the expense of lawyers who pursued career paths less likely to involve trials, like government lawyers and law professors."

Now, perhaps these ABA litigators are right, and perhaps their criteria are superior. Could that be measured? That would be a new and valuable place for future ABA studies. But that is also difficult to quantify. Allow me to offer a few thoughts.

First, we have a handful of notoriously bad-behaving judges we can examine.

Thomas Porteous was rating unanimously "qualified" (not "well qualified"), but he was impeached and removed for committing perjury by signing false financial disclosure forms and abusing his judicial office.

Samuel Kent was unanimously rated "well qualified," but he was impeached and later resigned from office for lying about sexual misconduct involving female employees.

Mark Fuller, in contrast, received a "qualified" rating with a minority "not qualified," before resigning after an investigation involving allegations about spousal abuse.

Second, I looked to a couple of the examples cited recently of more controversial nominees, and then I examined what litigants had to say about those judges in the Federal Judicial Almanac.

Roger Benitez received a substantial majority "not qualified," with a minority "qualified." Here's highlights from the Federal Judicial Almanac on him:

Alison Nathan received a majority "qualified," with a minority "not qualified." From the Almanac:

As a law professor who has to read (often inconsistent) student evaluations of my own performance each semester, I'm well aware of the limitations evaluations like these might present. But, in my view, they reflect, on the whole, that both judges are, with some possible weaknesses, capable and competent (even "qualified") judges. And, of course, perhaps someone will counter that using the Almanac has an entirely different set of flaws to evaluate judges.

Now, I have no idea how many Type I & Type II errors come out of the ABA judicial nominating process, at least to the regard that it's trying to anticipate who is "qualified" for the bench." I just cherry-picked a few examples, and I made no effort to dig deeper.

Additionally, these judges are hardly a random sample. They are selected, at times, to comport with the ABA criteria, and, with a couple of recent Republican administration exceptions, exclude candidates unanimously rated "not qualified." Other political reasons sink some nominees. Some items arise so far in the future that perhaps the ABA could not reasonably have been thought to be capable of evaluating that candidate's qualifications.

That said, I do think there would be tremendous value in examining the Federal Judicial Almanac entries of recent nominees and trying to compare them with ABA ratings. I assume, but perhaps I am wrong in the assumption, that the correlation between "qualified" ratings and the feedback from litigants is uneven. Additionally, I wonder if, over the course of a judge's career, the judge's capabilities (particularly instincts regarding exercises of discretion) improve to a degree that lessen the significance of any shortcomings. (Admittedly, learning on the job may be cold comfort to some early litigants!)

All this is to say, I wonder, setting aside the political critiques of the ABA at the moment, whether its evaluation process is even doing what it's designed to do in the first place.

A small data point on an emergency designation in a California election

Emergencies and elections don't get along well. The threat of emergencies, like acts of terrorism or massive weather events, remain a concern in elections, but we seem to lack many structures in place to handle such events.

Fires in Sonoma County, California prompted Governor Jerry Brown on October 19, 2017 to sign an executive order declaring an all mail ballot election for the November 7 election.

I was a bit skeptical of what I thought to be a fairly late emergency declaration and wondered how it might play out. Granted, it's an admittedly low-turnout election, but on the surface it appears that canceling in-person voting had a negligible impact, if any, on turnout.

The 2013 election had 6364 absentee and 1248 precinct ballots for 35.2% turnout. The 2015 election had 7003 absentee and 1235 precinct ballots for 33.0% turnout. Surely, a high percentage of voters already casting absentee ballots helps minimize any damage from canceling precinct voting. The final results this election were 6590 absentee for 31.2% turnout--a decline, but no bigger than the previous off-year-over-year decline.

I'm certainly not equipped to address matters like the competitiveness of the candidates or contentiousness of the issues or any of the many other confounding variables that could affect turnout. But, it's a small data point to consider in the larger scheme of thinking about how to handle emergencies and elections.

Sexual misconduct, Senate precedent, and the Expulsion Clause

I recently wrote a piece at Law & Liberty on the Senate's power to expel Roy Moore in the event he wins the Alabama Senate special election. He has been accused of sexual misconduct, and calls for him to resign have not been heeded. Some have wondered about whether the Senate might refuse to seat him; I explain in the piece why that is inconsistent with the constitutional understanding set forth in Powell v. McCormack.

But what about the power to expel a member once seated? The Constitution says this:

Each House may determine the Rules of its Proceedings, punish its Members for disorderly Behavior, and, with the Concurrence of two-thirds, expel a Member.

I explain in the piece that there are few contours to the power to expel, except that it must occur by a two-thirds vote. That's mostly a procedural check, and the substance has largely been left to the Senate to flesh out. (Professor Brian Kalt shares similar thoughts in a Wall Street Journal opinion.)

I identify a couple of precedents that would limit the Senate's power to expel Mr. Moore. First, the Senate has typically wondered whether it has the power to expel members for conduct that arose prior to the candidate taking office. (Some earlier debates actually focused on whether it had to arise during that existing term in office and not from a preceding term, but recent Senate investigations have moved away from that view.) Second, the Senate has generally refused to expel a member for conduct known to the voters at the time of the election, the notion being that it's not for the Senate to expel a member with such baggange sent by the voters.

The Senate has not expelled a member since the Civil War, but it has investigated members who resigned before facing an expulsion vote, like Senator Bob Packwood in 1995.

New allegations have arise in recent weeks. Representative John Conyers and Senator Al Franken have been accused of sexual misconduct, and campaigns arose to urge their resignations.

Mr. Conyers heeded such calls this week and resigned, rather than face an ethics investigation and, perhaps, subsequent expulsion vote. His conduct arose during his terms in office (as he has been in the House for several decades).

Mr. Franken is facing new calls from a number of his fellow Senators to resign. Some allegations of his sexual misconduct arose from events prior to taking office, but certainly were not widely known to voters. He is also facing an ethics investigation.

It is worth noting how the rather underdefined precedents shaping Congress's practices under the Expulsion Clause may be experiencing a rather significant change right before our eyes. Ethics investigations and calls for resignation for (at least some) conduct arising prior to taking office suggest that a good number of Senators would apply the Clause in a way that the Senate has been reluctant to do in the past.

This is not to say that this is a wrong view, just to say that the Senate is, as is within its authority, deciding the proper scope of its authority under the Clause. It may be using informal means, like social media campaigns, political calls for resignations, and political party threats (perhaps offering primary challengers, withholding campaign funding, or revoking committee assignments), in ways that do not impact the Expulsion Clause.

But in the event Mr. Moore is seated, and the Senate begins to look at these recent precedents involving allegations of sexual misconduct prior to a candidate taking office, it may be moving toward a new understanding of its own power under the Clause.

That may not be the case, of course. Expulsion is an understandably extraordinary remedy, requiring significant consensus, and we have seen the Senate bluster recently and fail to carry through. These other tools at the disposal of the parties and the Senate may make the Senate reluctant to carry through with expulsion of Mr. Franken (if he does not resign in the next several hours) or Mr. Moore (if he is elected). But it is worth considering how recent weeks may be shaping, before our very eyes, a renewed attention to the Senate reconsidering its past practices and reinvigorating the Expulsion Clause.

A change in calculating pass rates for the California bar exam

Good news from the California bar: the overall bar pass rate rose year-over-year from 43% to 49.6%. Or... did it?

The State Bar of California made a small change to how it calculates the passing rate of bar exam test-takers. In April 2017, it adopted the following change:

It was moved, seconded and duly carried that beginning with the February 2017 administration of the California Bar Examination applicants who did not complete all portions of the examination not be included in the pass/fail statistics published at the time results from the examination are published; and that for an examination to be considered complete, applicant must have achieved a grade of at least 40 on their answers to each question on the examination.

The change is a sensible one: if a test-taker walks out in the middle of the exam, it doesn't seem terribly sensible to include that test-taker as a failure. That's not usually what we'd think about in terms of failure rates; instead, those who sat through the whole exam, answered all the questions, and tried to pass the bar would be the ones whose success rates we'd like to evaluate. A quotation from Karen Goodman on the Committee of Bar Examiners in the Daily Journal was consistent with this: "It seemed like if people did not finish the test, they should not count against the pass rate." (Of course, I suppose, the person did fail!)

At the same time, instituting this new change could make it appear that bar pass rates were higher than they actually were, because the new pass rates are going to be higher than old pass rates due to the change in methodology.

The February 2017 overall pass rate was reported at 34.5%, when under the old methodology it would have been 33.9% (a 0.6-point difference). 78 did not complete the exam

For July 2017, 66 did not complete the exam. That lifted the overall percentage who passed from 49.19% to 49.57%. A California bar representative also informed me that the July 2016 exam had 89 who did not complete the exam, a pass rate of 43.57% v. 43.07%.

(It's worth emphasizing this difference is probably even smaller today because the bar has been shorted from three days to two as of July 2017, making it more likely that more individuals will finish the exam.)

This is a very modest advantage to all schools in reporting their overall pass rates--odds are that one dropout in 200 can bump a school's overall score by a point (when rounded). And it offers a very modest (if slightly deceptive) improvement to the current state of affairs when considering bar passage rates in California. It makes comparisons across years slightly disparate.

But, in an era nearly obsessed with almost any numerical change in bar exam statistics, this one is worth highlighting for future consideration. The true year-over-year comparison is 43.6% to 49.6% (+6 points), or 43.1% [sic; that's the percentage shared with me!] to 49.2% (+6.2 points), not 43.0% to 49.6% (+6.6 points). In future years, the comparison will be easier to make.

A critically important legal question looms over Alabama Senate race: what happens if Roy Moore withdraws?

Election law informs political strategy. The choices that parties, candidates, and voters make may change depending on the legal consequences of those actions.

A critically important legal questions looms over the Alabama Senate race. If embattled candidate Roy Moore withdraws, it may have one of two effects. There is uncertainty about which effect will take place. And it could entirely change the political strategy of Republicans.

Imagine for a moment, and it requires little imagination, that Mr. Moore remains in the race. Republican leaders have a choice: stick with Mr. Moore, or back a single write-in candidate. (A third might be to simply endorse Democratic candidate Doug Jones, but I'll assume Republicans want to keep the seat in the hands of someone who'll caucus with them.) There isn't much to think about in this scenario. It's purely a question of political strategy.

Imagine, instead, Mr. Moore withdraws from the race, or the party withdraws him as its nominee. (I should pause on the last point to note that I hardly know how the party has a mechanism to do so, and whether it can yank the rug out of its own nominee, apart from the fact that state law apparently authorizes it; I won't address how it might do so, or whether Mr. Moore might challenge it.)

What is the strategic decision from the Republican Party then? In my view, it all turns on what Alabama Code 17-6-21(c) means. And while the Secretary of State John Merrill has said one thing about what the law says, I'm not sure that's accurate. And if it isn't accurate, the strategy changes completely.

If Mr. Merrill's interpretation is correct: if an ineligible candidate (dead, disqualified, or withdrawn) receives the most votes, the election is declared null and void, and a new special election would need to be called. That is consistent with a long line of precedent in Alabama, as recently articulated in law as 2001.

In this scenario, the best Republican strategy is to "punch Moore." Consider the 2006 Florida race involving disgraced representative Mark Foley, where his name appeared on the ballot but actually stood for stand-in replacement Joe Negron, and where the slogan "punch Foley for Joe" instructed voters that voting for the disgraced candidate would be the best choice--because it wasn't for the disgraced candidate at all. The same held true in Missouri in 2000, where voters chose the deceased Mel Carnahan and knew that the governor would appoint a replacement (Mr. Carnahan's wife, he promised) before the special election to replace him.

A write-in candidacy is not as good a choice. That is, because some early votes have already been cast for Mr. Moore; finding a capable candidate is dicey; and getting write-ins generally is a challenge... why not reset the clock?

If my interpretation is right: HB 62 was enacted in 2014. It changed the rules for recounts for late-withdrawing candidates only. Its amended text, as I've emphasized, provides: "In the event that a candidate submits a notification of withdrawal after the applicable deadline, the name of the candidate shall remain on the ballot and the appropriate canvassing board may not certify any votes for the candidate." (Similar language is in (b) when a party withdraws a nominee.) That means, the second-best vote-getter actually wins--because there are no votes for the withdrawn candidate, because no votes for that candidate have been counted.

In this scenario, the best Republican strategy is to back a write-in candidate. Early votes for Mr. Moore have already been cast, which is a problem. But, if Mr. Jones would be the winner, their sole priority would be to back a write-in candidate who could win--not votes for Mr. Moore that would be thrown out.

And a strategy to vote for Moore would lead to a Jones win. It's exactly the reverse incentives.

The resolution of this question is of critical importance. Litigation assuredly will ensue: if Mr. Merrill's interpretation is heeded, Mr. Jones would surely sue to claim that he would win, and perhaps seek a declaration before the election. Voters would be confused, weakening Republican chances in the event they wait. Mr. Jones's supporters might be surprised if Mr. Merrill's interpretation carries the day and he is fighting an invisible candidate through December 12.

Regardless of which rule is the right one under Alabama law, it is critically important that this question be addressed sooner rather than later.

Sorting out the Alabama Senate election: could the governor reschedule the special election?

Whew. After musing about possibilities in the Alabama Senate special election, and digging deeper into vote-counting for a withdrawn candidate, a new scenario has popped up: the governor postponing the election. As reported in the New York Times:

State law gives the governor broad authority to set the date of special elections, and Ms. Ivey, who is a Republican, already rescheduled the Senate election once, after inheriting the governor’s office in April when her predecessor, Robert Bentley, resigned in a sex and corruption scandal. Ms. Ivey’s advisers have not ruled out exercising that power again, according to Republicans in touch with her camp, but she has signaled that she would like reassurances of support from the White House before taking such an aggressive step.
. . .
But there is no apparent precedent for rescheduling an election so close to the planned vote, Republicans acknowledged. In addition to state election laws, such an extreme step could also run afoul of federal voting rights law.

The thought goes, I suppose, that a delay would give extra time for Mr. Moore to withdraw (voluntarily or involuntarily), and such a withdrawal, which might occur outside of the 76-day withdrawal window, would permit the party to replace Mr. Moore's name on the printed ballot.

Could this actually be the case? I'm not so sure. (I'll only address here a proposal to postpone the election, not to cancel it, and it would not require any action on the part of the legislature. I won't address whether independent Alabama constitutional rights would prevent this action from the governor in a voting rights context. I'll also emphasize, a I did in my last post, that this appears quite speculative, as Governor Kay Ivey has indicated no intention of changing the election date. Finally, I'll only discuss Alabama state law. I won't discuss the possibility that this act might run afoul of federal law or the United States Constitution.)

First, Ms. Ivey "rescheduled" the special election after Governor Robert Bentley resigned. Mr. Bentley had previously scheduled the special election to coincide with the 2018 regularly-scheduled primary and general elections. Each did so ostensibly pursuant to Alabama Code 36-9-8: "Whenever a vacancy occurs in the office of senator of and from the State of Alabama in the Senate of the United States more than four months before a general election, the Governor of Alabama shall forthwith order an election to be held by the qualified electors of the state to elect a senator of and from the State of Alabama to the United States Senate for the unexpired term. If the vacancy occurs within four months of but more than 60 days before a general election, the vacancy shall be filled at that election. If the vacancy occurs within 60 days before a general election, the Governor shall order a special election to be held on the first Tuesday after the lapse of 60 days from and after the day on which the vacancy is known to the Governor, and the senator elected at such special election shall hold office for the unexpired term."

But Mr. Bentley faced a legal challenge to his scheduling of the special election. He claimed that he had authority to push the election back to the general election. Challengers noted that he must "forthwith order an election" if the vacancy occurred "more than four months before a general election," and that only in instances where the vacancy occurred between four months and 60 days of the next scheduled general election could the governor hold the special election concurrent with the general election. A memorandum from the Legislative Reference Service concurred.

These are interesting matters of statutory interpretation under state law. But if Mr. Bentley lacked the power in the first place (that is, his original writ of election was inconsistent with Alabama law), then Ms. Ivey's pronouncement for an earlier special election was not a "change" or a "reschedule" of the special election; it was the first valid writ of election issued. While the Times describes this as a "rescheduled" election, the legal effect matters. If Mr. Bentley's original election lacked the force of law, Ms. Ivey practically (but not legally) rescheduled the election.

So, is there anything that constrains the governor? At least some constraints appear in the statute and its interpretation (apparently flouted by Mr. Bentley). But consider other relevant portions of the messy Election Code.

Consider constraints on the governor's power under Alabama Code 36-9-9: "The Governor . . . must give notice of a special election to elect a senator for an unexpired term in the same manner and for the same time as is prescribed for special elections to fill a vacancy in the office of members of the House of Representatives in Congress."

That refers to Alabama Code 17-15-3: "All special elections provided for by this chapter are to be ordered by the Governor, who must issue writs of election, directed to the judge of probate of the counties in which such election is required to be held and must specify therein the district or county in which, and the day on which, such election is to be held; the cause and object of the same; the name of the person in whose office the vacancy has occurred and, in all cases in which a special election is directed in a district composed of more than one county, such election must be directed to be held on the same day in each county." And Alabama Code 17-15-4, "The Governor must give notice of any special election for representatives in Congress, or state officers, by proclamation." Finally, Alabama Code 17-15-7, "Special elections are to be held and conducted, the returns thereof made and certificates given, and, unless otherwise expressly provided, regulated in all respects by the provisions in relation to general elections."

It's very hard for me to read these statutes together to suggest that the Alabama legislature has empowered the governor to schedule and reschedule, unilaterally, without constraint, special elections. First, 36-9-8 already places some timing restraints on the governor. Second, portions of Title 17 limit the discretion given to the governor, as special election are largely designed to resemble general elections.

The Seventeenth Amendment includes this relationship between the governor and the legislature: "When vacancies happen in the representation of any state in the Senate, the executive authority of such state shall issue writs of election to fill such vacancies: Provided, that the legislature of any state may empower the executive thereof to make temporary appointments until the people fill the vacancies by election as the legislature may direct.

I think the best way of harmonizing these statutes, and the way that events have transpired so far, is that Ms. Ivey's scheduling of the special election was the first valid proclamation from the governor (and that Mr. Bentley's proclamation was void under state law); and that she lacks the power to change the date. But, others may differ... and I'm always happy to consider competing theories or a more complete understanding of state statutes.