Fourth Court of Appeals
San Antonio, Texas
CONCURRING OPINION
No. 04-23-00099-CV
Daisy Campos RODRIGUEZ,
Appellant
v.
Ricardo “Richie” RANGEL, Jr.,
Appellee
From the 49th Judicial District Court, Webb County, Texas
Trial Court No. 2022-CVK-001669-D1
Honorable Susan D. Reed, Judge Presiding
Opinion by: Patricia O. Alvarez, Justice
Concurring Opinion by: Rebeca C. Martinez, Chief Justice
Sitting: Rebeca C. Martinez, Chief Justice
Patricia O. Alvarez, Justice
Sandee Marion, Chief Justice (Ret.) 1
Delivered and Filed: November 13, 2023
I respectfully concur in the judgment. While I agree with the majority that the trial court
did not err in determining the “true outcome” of the election, I would subtract illegal votes for
Daisy 2 from the manual recount, rather than from the election night vote count or the CVR, as the
majority does. The manual recount was the “final canvass,” and the record lacks legally sufficient
evidence of a “mistake” to disregard it. See TEX. ELEC. CODE ANN. § 221.003(a)(2)(C).
1
The Honorable Sandee Marion, Chief Justice (Retired) of the Fourth Court of Appeals, sitting by assignment of the
Chief Justice of the Texas Supreme Court. See TEX. GOV’T CODE ANN. §§ 74.003, 75.002, 75.003.
2
I refer to Daisy by her first name for ease of reference.
Concurring Opinion 04-23-00099-CV
Consequently, I would sustain Daisy’s second issue. Nevertheless, I concur in the judgment
because the vote difference based on the manual recount was eleven, and the trial court did not
abuse its discretion by determining that at least twelve of the illegal votes cast were for Daisy.
I. BACKGROUND
The trial court found:
3. . . . The Court ordered the election administrator to provide the Cast Vote
Record (“CVR”) to all parties. At trial, Elections Administrator Jose Castillo
testified that he had the most faith in the CVR as accurately reflecting the vote. The
Court found him to be knowledgeable and credible.
4. The Court finds the Election Night Count and the CVR as the accurate count
of the votes and should be the vote used in the determination of this cause.
In Daisy’s second issue, she complains that these factual findings constitute an abuse of
discretion because they improperly supplant the legislatively-preferred manual recount for the
general election night results and they are not supported by legally sufficient evidence. I agree on
both points.
II. LEGISLATIVELY-DETERMINED STARTING POINT
Section 213.033(a) of the Texas Election Code establishes that the election night results
constitute the original canvass. However, under section 213.033(a), a recount “that changes the
number of votes received for a particular candidate” “void[s]” the “original canvass,” and “the
new canvass is the official canvass for the election.” See id. § 213.033(a). Here, the manual
recount changed the number of votes received for the candidates. Thus, it became the “new
canvass,” from which the election was determined, and it was the “final canvas,” as that term is
defined in the Election Code. See id. § 1.005(5) (“‘Final canvass’ means the canvass from which
the official result of an election is determined.”). This final canvass could be disregarded only for
a legislatively-determined reason, such as if the election officer “made a mistake.” See id.
§ 221.003(a)(2)(C).
-2-
Concurring Opinion 04-23-00099-CV
Rangel references Slusher v. Streater, 896 S.W.2d 239, 242 (Tex. App.—Houston [1st
Dist.] 1995, no writ.), for the proposition that a “trial court has the discretion to disregard a final
canvass if there was a mistake in the counting of ballots or if some votes were cast illegally.” True
enough, but neither the Election Code nor Slusher allow us to begin our analysis from a point other
than from the final canvass. See TEX. ELEC. CODE ANN. § 221.003(a) (“The tribunal hearing an
election contest shall attempt to ascertain whether the outcome of the contested election, as shown
by the final canvass, is not the true outcome . . . .”). 3
III. LEGALLY INSUFFICIENT EVIDENCE OF A MISTAKE
Instead of beginning with the manual recount and setting it aside upon a showing that an
election officer “made a mistake,” the trial court set the manual recount aside on the basis that
Castillo “had the most faith in the CVR as accurately reflecting the vote.” Castillo’s ultimate
opinion that he “ha[s] more confidence in the Election Night results and the CVR” than the manual
recount rests on three parts: (1) that the election night results and the CVR are “independent;” (2)
the CVR “verifies” the “accuracy” of the election night results; and (3) the “independent” and
“verified” election night results establish that they are “more accurate” than the manual recount.
All three of these discrete opinions, alone or combined, are legally insufficient to establish that the
manual recount is not the “true outcome” because an election officer “made a mistake.” See id.
1. Castillo’s “assumption” that the election night results and the CVR are
“independent” lacks a reliable foundation and is merely his ipse dixit.
On voir dire and then on cross examination by Daisy, Castillo testified:
DAISY: Now, the data in this CVR is the same — it’s the exact-same data
that you had on Election Night?
3
In Slusher, the trial court, in accordance with section 221.008 of the Election Code and at a hearing that was not
transcribed, inspected ballot boxes in open court and determined that the official canvass was off by one vote. Slusher,
896 S.W.2d at 241; see TEX. ELEC. CODE ANN. § 221.008 (allowing trial court to determine correct vote count from
ballot boxes). Thus, the trial court in Slusher, unlike here, witnessed the recount.
-3-
Concurring Opinion 04-23-00099-CV
CASTILLO: It’s the same number. I guess . . . [ellipses in original]
...
DAISY: And the data from the CVRs, that is the same data that was used to
make the reports for Election Night? It’s just a different report,
right?
CASTILLO: I’m not sure. I didn’t — I have not asked that question from our
vendor, but I’m assuming — from every other question that I have
asked them — that they’re independent.
The trial court sought clarification:
COURT: May I ask one little question —
RANGEL: Yes, Your Honor.
COURT: — because I’m a little confused about something? When it goes
into that scanner, is the count coming off of that ballot that is in the
scanner and going to the result, and then the image is simply a stored
image?
CASTILLO: Correct, Judge.
COURT: And then, eventually, that CVR that is created is done off of that
stored image.
CASTILLO: Of the stored image. Yes, Judge.
(Emphasis added). On cross examination by Daisy, Castillo acknowledged that he did not know
how the voting system put together the numbers.
We have long held — under the lower preponderance of the evidence standard — that
“[e]xpert opinion cannot be based upon mere guess or speculation, but must have a proper factual
basis. Ochs v. Martinez, 789 S.W.2d 949, 958 (Tex. App.—San Antonio 1990, writ denied)
(emphasis added). Castillo’s “guess,” response of “not sure,” and “assum[ption]” fail to meet that
long-settled — and less exacting — rule requiring that an expert provide more than a guess or
speculation. Rangel, bearing the burden of proof, failed to lay any foundation for Castillo’s
“assumption” that the election night results and the CVR are “independent.” Cf. Flores v. Cuellar,
-4-
Concurring Opinion 04-23-00099-CV
269 S.W.3d 657, 660 (Tex. App.—San Antonio 2008, no pet.) (“To overturn an election, the
contestant must prove by clear and convincing evidence that voting irregularities materially
affected the election results.”); Merrell Dow Pharms., Inc. v. Havner, 953 S.W.2d 706, 714 (Tex.
1997) (“If the foundational data underlying opinion testimony are unreliable, an expert will not be
permitted to base an opinion on that data because any opinion drawn from that data is likewise
unreliable.”). Accordingly, there is legally insufficient clear and convincing evidence supporting
Castillo’s discrete opinion that the election night results and the CVR are “independent.”
2. Castillo undermines his legally insufficient “assumption” by admitting that the
CVR does not “verify” the accuracy of the election night results. Therefore,
inaccuracies from scanning votes reported in the election night results will be
repeated on the CVR.
The majority posits that the CVR “verified” the election night results. Rodriguez v. Rangel,
No. 04-23-00099-CV, slip op. at 23 (Tex. App.—San Antonio Nov. 13, 2023, no pet. h.).
However, by Castillo’s admission the CVR does not double check the election night results by
independently reviewing each ballot. On voir dire and on cross-examination by Daisy, Castillo
testified:
DAISY: And so, if that scanner reads it in properly [sic], there is no way of
double-checking it without looking at the actual ballot visually, like
a manual recount?
RANGEL: Your Honor, I need to object to that question insofar as there are no
pleadings in this case challenging the Election Department’s
equipment; [Daisy] filed a General Denial only; and in our
pleadings, Your Honor, we mentioned the residency issue and
another issue with regards to the precinct that we’ll get into it later.
So, I object insofar that this [is] irrelevant because there are no
supportive pleadings regarding any problems with the election
equipment, Your Honor.
DAISY: Judge, it is relevant because we are trying to support the first count
as the official count; and, there is a recount, which is the actual
official count; and then there was a judicial recount. So, this first
vote is extremely relevant because the witness has said — over our
objection — that he thinks, in his opinion, that it’s accurate.
-5-
Concurring Opinion 04-23-00099-CV
COURT: You are overruled.
RANGEL: Thank you, Your Honor.
DAISY: The question was: If the scanner reads — does it read the ballot for
the particular race — or reads it in properly [sic], for example,
giving the race to the wrong candidate — then that’s the data that is
going to be stored in the equipment, correct?
CASTILLO: If it reads it wrong, yes; but, there is no indication that it read it
wrong.
...
DAISY: And that’s why the only reason to — the only way of double-
checking this is by looking at the ballots manually?
CASTILLO: I would say looking at them manually, or visually go through each
cast vote record of the PDF picture.
DAISY: Which hasn’t been done?
CASTILLO: I haven’t done it.
The admission Castillo made under examination by Daisy establishes that any notion that
the CVR “verifies” the election night results is legally insufficient under the lower preponderance
of the evidence standard in two related respects. First, no reasonable factfinder could disregard
the admission. See City of Keller v. Wilson, 168 S.W.3d 802, 827 (Tex. 2005) (providing that a
“legal-sufficiency review in the proper light must credit favorable evidence if reasonable jurors
could, and disregard contrary evidence unless reasonable jurors could not”); see also Jackson v.
Morrison, No. 03-14-00419-CV, 2015 WL 4195578, at *2 (Tex. App.—Austin Jul. 8, 2015, no
pet.) (mem. op.) (recognizing that a deemed admission may conclusively negate a claim). Second,
it is part of an internally inconsistent expert opinion, which also constitutes no evidence. See Gen.
Motors Corp. v. Iracheta, 161 S.W.3d 462, 470–72 (Tex. 2005). In General Motors, the Texas
Supreme Court wrote:
-6-
Concurring Opinion 04-23-00099-CV
Iracheta attempts to borrow from each of her experts pieces of opinion that seem to
match, tie them together in an ill-fitting theory, discard the unwanted opinions,
disregard the fact that the experts fundamentally contradicted themselves and each
other, and then argue that this is some evidence to support the verdict. Inconsistent
theories cannot be manipulated in this way to form a hybrid for which no expert
can offer support.
Id. at 472 (citing Volkswagen of Am., Inc. v. Ramirez, 159 S.W.3d 897, 911 (Tex. 2004)) (emphasis
added). If a litigant cannot piece together inconsistent opinions from multiple experts, he cannot
piece together inconsistent opinions from a single expert.
3. A discrepancy between the victory margin is not, in and of itself, clear and
convincing evidence that a “mistake” was made during the manual recount
and conversely that the election night results are more “accurate.”
To be clear: every ballot cast in the race for District 2 was counted in the manual recount.
Nevertheless, the majority finds a “ballot counting discrepancy,” “overvotes discrepancy,” and the
“accuracy of [the] manual recount return sheet” by stacking a series of unreasonable inferences on
top of each other. This impermissible inference stacking violates the rules governing a review of
the evidence under the clear and convincing standard, the legislative prerogative articulated by
section 221.003(a)(2)(C) of the Election Code, and the equal inference rule.
The manual recount revealed that Daisy’s margin of victory grew over the election night
results by five votes, to a total victory margin of eleven, by Daisy picking up an additional vote
and Rangel dropping six votes. Visually, the difference appears as:
Votes Election Manual
Night Recount
Ricardo “Richie” Rangel Jr. 1,951 1,945
Daisy Campos Rodriguez 1,957 1,956
Subtotal 3,908 3,901
At first blush, the manual recount appears to have not counted seven ballots because there are
seven fewer votes attributed to any candidate. However, the “missing” seven ballots are
-7-
Concurring Opinion 04-23-00099-CV
represented in the undervote total, which grew by exactly seven votes — from 243 to 250. These
undervotes represent ballots in which neither of the two candidates were selected.
The majority’s “overvotes discrepancy” analysis and its assertion of the “accuracy of [the]
manual recount return sheet” is similarly unavailing. The manual recount sheet provides:
It has no row to record overvotes. The majority holds that the trial court “could have observed
that the manual recount return sheet was prepared specifically for this recount, it could have
included a line showing the number of overvotes, but it did not report any.” Rodriguez, slip op. at
21 (emphasis added). There is no evidence supporting the majority’s two “could haves.” The
record is silent as to why the pre-printed manual recount sheet does not include an overvote row.
A reasonable inference from this fact is that the talliers did not manually write down the four
overvotes. The majority references sections 213.011 4 and 213.012 5 of the Election Code as
“evidence” that the trial court “could have accepted the manual recount return sheet report as an
accurate statement of the manual recount results . . . which showed the talliers reported counting
4
Section 213.011 of the Election Code provides: “The recount committee shall count the votes separately by precinct.”
TEX. ELEC. CODE ANN. § 213.011.
5
Section 213.012 of the Election Code provides:
(a) After the recount is completed, the recount committee chair shall prepare a report of the
committee’s vote count and sign the report. Votes shall be reported separately by precinct.
(b) The chair shall deliver one copy of the report to the recount supervisor and one copy to the
general custodian of election records.
(c) The copies of the report shall be preserved for the period for preserving the precinct
election records.
TEX. ELEC. CODE ANN. § 213.012.
-8-
Concurring Opinion 04-23-00099-CV
only 4,151 ballots.” Id. at 22 (emphasis added). However, sections 213.011 and 213.012 are
statutes; they are not evidence.
Castillo’s testimony, in fact, supports a conclusion that every ballot was counted during
the manual recount. When asked on redirect examination by Rangel to explain these two
variances, Castillo testified:
Hum, my honest opinion is human error. The manual recount which — there were
4,155 ballots that were counted for that race; and, we had — I guess our data
showed, let’s say from box precinct, let’s say 359, there should have been 25 ballots
for District [2].
So, we opened up the ballots — me and my staff — and we pulled them out
manually, all 25 ballots. So, for example, if we had 25 ballots, we didn’t turn them
over to the talliers until we had all the 25. Now, if we were missing one, then we
would go through the ballots, again, to make sure we found it.
So, we turned all 359 ballots to the talliers. Now, the talliers, I guess, they
got tired — their eyes or they counted wrong — that’s something to think of
because the complete number of ballots were handed over to the talliers; and I
would just imagine that it was just human error, they were tired, and they marked
it wrong for that count.
The majority highlights Castillo’s testimony as part of the evidence that the manual recount
— coupled with the CVR “verifying” the accuracy of election night results — is the source of the
“ballot discrepancy” or “mistake.” However, Castillo admitted that the CVR does not verify the
election night results. Therefore, the only other piece of evidence is the discrepancy alone. This
is where the majority’s analysis violates the equal inference rule, which provides that:
Testimony that gives rise to any number of inferences, none more probable than
another, is legally insufficient to support the inference of a fact. Likewise, when
the circumstances are equally consistent with either of two facts, neither fact may
be inferred. [The factfinder] may not simply speculate that a particular inference
arises from the evidence.
United Rentals N. Am., Inc. v. Evans, 668 S.W.3d 627, 642 (Tex. 2023) (internal citations and
quotation marks omitted). Under the equal inference rule, a change in Daisy’s victory margin does
not alone establish where any potential mistake was made. Instead, a reasonable inference explains
-9-
Concurring Opinion 04-23-00099-CV
that seven fewer votes were attributed to a particular candidate in the manual recount because the
manual recount found seven additional undervotes. This is a reasonable inference because it
reconciles all of the credible evidence before the trial court. Likewise, as to the “overvotes
discrepancy,” if Castillo is correct in that his staff fastidiously ensured that the manual recount
talliers counted each ballot precinct-by-precinct, then the talliers would have been handed every
ballot — including any that contained an overvote. Thus, reconciling all of the evidence before
the trial court, a reasonable inference is that the manual recount talliers did not report overvotes
on the manual recount sheet because there was no line to record this data.
4. “I guess, they got tired” is not legally sufficient evidence of a “mistake” made
during the manual recount.
Section 221.003(a)(2)(C) of the Election Code is a legislative prerogative that may be
overcome only by clear and convincing evidence of, among other things, a “mistake” — not faith, 6
not unverified accuracy, 7 and not unfounded confidence. 8 The majority’s legal sufficiency review
errs by accepting Castillo’s conclusory, speculative, and internally inconsistent opinions as a
foundation upon which to stack unreasonable inferences. Once all the legally insufficient evidence
and unreasonable inferences are stripped away, all that remains is Castillo’s “guess” that the
manual recount talliers “got tired.” However, this too is legally insufficient. Therefore, I would
sustain Daisy’s second issue. 9
Rebeca C. Martinez, Chief Justice
6
The third factual finding provides that “Elections Administrator Jose Castillo testified that he had the most faith in
the CVR as accurately reflecting the vote.” (Emphasis added).
7
The fourth factual finding provides that “[t]he Court finds the Election Night Count and the CVR as the accurate
count of the votes and should be the vote used in the determination of this cause.” (Emphasis added).
8
Castillo testified that “I have more confidence in the Election Night results and the CVR.” (Emphasis added).
9
I would not address Daisy’s first issue — that the trial court abused its by discretion admitting “expert testimony”
from Castillo — because of my disposition of her second issue. See TEX. R. APP. P. 47.1 (“The court of appeals must
hand down a written opinion that is as brief as practicable but that addresses every issue raised and necessary to final
disposition of the appeal.”).
- 10 -