¶1 This case asks us to determine whether, by orally instructing a jury that it must fill in a blank verdict form, a trial court improperly coerced the jury to reach a verdict. The Court of Appeals concluded that the trial court’s action coerced the jury, violating the defendant’s right to a fair trial and thus meriting a new trial. In this case, because the jury had announced its unanimity prior to the judicial conduct at issue, we find no judicial coercion possible, and accordingly we reverse the Court of Appeals’ decision and reinstate the defendant’s conviction.
FACTS AND PROCEDURAL HISTORY
¶2 The State charged Tyrone Ford with two counts of child rape based on two occasions when he allegedly had sex with a minor, L.A.K., in August and September 2006. Clerk’s Papers (CP) at 19. After both sides rested at trial, the jury retired to deliberate at 7:47 p.m. and returned with its verdict at 2:01 p.m. the next day. CP at 79; VI Verbatim Report of Proceedings (VRP) at 433. Then, the following occurred:
THE COURT: Will the presiding juror please rise. Has the jury reached a unanimous verdict?
*187THE PRESIDING JUROR: Yes.
THE COURT: Would you pass the verdict forms to my bailiff, and then you may be seated, sir.
Gentlemen, I’ll dispense with the reading of the caption heading.
“We, the jury, find the defendant, Tyrone Ford, guilty of the crime of Rape of a Child in the Third Degree as charged in Count Two.”
(Pause; reviewing documents. 2:05 p.m.) Gentlemen, sidebar.
(Bench conference; not recorded.) THE COURT: I’m sending the jury back to the jury room. Verdict form No. 1 is completely blank. It must be filled in. Please go with Dorothy.
(Jurors reenter courtroom. 2:09 p.m.) THE COURT: Presiding juror, have you reached a verdict as to Count One?
THE PRESIDING JUROR: Yes.
THE COURT: Was it a unanimous verdict?
THE PRESIDING JUROR: Yes.
THE COURT: Why don’t you go ahead and give it to Dorothy, then.
(Pause; reviewing document.) Okay, as to Count One:
“We, the jury, find the defendant, Tyrone Ford, guilty of the crime of Rape in the Second Degree as charged in Count One.”
IV VRP at 389-92; VI VRP at 434. The jurors were then individually polled and each confirmed the verdict. Nothing in the record indicates that any juror expressed any disagreement with either verdict during polling.
¶3 This was relatively short. The sidebar occurred at 2:05 p.m., and the jury returned by 2:09 p.m. VI VRP at 434. After sending the jury back, the trial court considered the possible explanations for the blank form: the jury may have forgotten to fill out the form, they were deadlocked on count *188one, or they reached a “not guilty” verdict on count one. IV VRP at 391. The court decided to advise the jurors that if they had any questions, they should submit them to the court. But before that advice could be relayed, the jury returned with a unanimous guilty verdict on count one. Ford did not object to any of these events or instructions.
¶4 Ford appealed his conviction, arguing that, among other things, the trial court erred by directing the jury to fill out the verdict form for count one. The Court of Appeals, in a split decision, reversed Ford’s conviction for count one, concluding that the trial court’s instruction to the jury regarding the verdict form was a manifest constitutional error that was substantially likely to have affected the outcome of Ford’s trial. State v. Ford, 151 Wn. App. 530, 540-41, 213 P.3d 54 (2009).
¶5 We granted the State’s petition for review. State v. Ford, 168 Wn.2d 1005, 226 P.3d 781 (2010).
ISSUE
¶6 Did the trial court’s instruction that the jury must fill out the blank verdict form constitute coercion and thus violate the defendant’s right to a fair trial?
ANALYSIS
¶7 Appellate courts typically will not consider an issue raised for the first time on appeal. RAP 2.5(a); State v. Kirkman, 159 Wn.2d 918, 926, 155 P.3d 125 (2007). However, an error may be raised for the first time on appeal if it is a manifest error affecting a constitutional right. RAP 2.5(a)(3). To demonstrate such an error, the defendant must show that the error actually prejudiced his rights at trial. Kirkman, 159 Wn.2d at 926. A claim of judicial coercion affecting a jury verdict is such an error that we will review, and our analysis is guided by precedent.
¶8 To prevail on a claim of improper judicial interference with the verdict, a defendant “must establish a *189reasonably substantial possibility that the verdict was improperly influenced by the trial court’s intervention.” State v. Watkins, 99 Wn.2d 166, 178, 660 P.2d 1117 (1983). This requires an affirmative showing and may not be based on mere speculation. We consider the totality of circumstances regarding the trial court’s intervention into the jury’s deliberations. Watkins, 99 Wn.2d at 177-78; State v. Boogaard, 90 Wn.2d 733, 739-40, 585 P.2d 789 (1978). Before we can do so, the defendant must first establish that the jury was still within its deliberative process.
¶9 Ford has not made this threshold showing, and so he cannot show the type of influence or coercion necessary to establish that the court improperly influenced the jury’s verdict. Nothing in the record before us suggests that the jury was deadlocked or experiencing any difficulty in reaching a decision. What we have is the opposite. The jurors twice indicated their unanimity.1 The jurors were polled. Each juror affirmed agreement with the verdict. There is no room for judicial coercion or influence because, as the record shows, the jurors had reached their verdict. And as far as the end result of completing the verdict form, they could just as conceivably have returned a “not guilty” verdict.
¶10 The Court of Appeals’ majority analogized the facts of Ford’s case to those in Boogaard, a case that did, in fact, involve a deadlocked jury being interrupted in its deliberations and coerced by the court to reach a verdict. The jury in that case began its deliberations in midafternoon, and when no verdict had been rendered by 9:30 p.m., the judge sent the bailiff to inquire about the jury’s status. The indication was that its vote at that time was 10-2. The judge then *190summoned the jury to the courtroom and proceeded to ask each juror, one by one, whether the jury could reach a verdict in half an hour. All but one juror responded affirmatively, and after returning to deliberations, the jury, not surprisingly, reached a verdict within the half hour time frame. We reasoned that by questioning each juror individually, the judge made clear his preference that they reach a verdict within a half-hour, thus improperly interfering in the deliberative process and coercing them into reaching a verdict. Boogaard, 90 Wn.2d at 733-39. Here, the Court of Appeals’ majority opined that a similar coercion occurred; by directing the jury to fill out the blank verdict form, the trial court coerced the jury into reaching a verdict.
¶11 The Court of Appeals’ analogy is misplaced. Here, unlike in Boogaard, there was no judicial interruption of deliberations. Deliberations had ended, which we know because the jury indicated its unanimity before the judge sent the jury back to fill out the verdict form. The court asked if the jury had reached a “unanimous verdict,” and the presiding juror responded that they had. Judge Wulle did not specify to which count he referred, so if the jury was hung on one count, it is very likely they would have said so. When the jury returned, the court again asked if the jury had reached a unanimous verdict. Following the reading of the verdict on count one, each juror was individually polled and confirmed agreement. Thus, unlike in Boogaard, here we are not presented with the obvious effects of judicial influence in the jury’s deliberations.
¶12 Next, Ford argues, and the Court of Appeals agreed, that the trial court in this case violated CrR 6.15(f)(2), which prohibits a trial court from suggesting to a jury that it must reach an agreement. That rule provides:
After jury deliberations have begun, the court shall not instruct the jury in such a way as to suggest the need for agreement, the consequences of no agreement, or the length of time a jury will be required to deliberate.
CrR 6.15(f)(2). Ford contends that the trial court violated this rule by instructing the jury that the verdict form for *191count one “must be filled in.” He argues that because the verdict form specified that the blank should be filled in with either “guilty” or “not guilty” the court’s instruction to fill in the form effectively commanded the jury to agree unanimously one way or the other, when the blank verdict form they had returned was an appropriate and acceptable verdict. This argument begs the question. Here, the indication was that the jury had reached a unanimous verdict and this shows deliberations were complete. Because the jury had finished its deliberations, CrR 6.15(f), titled “Questions from Jury During Deliberations,” has no application.
¶13 Ford urges us to consider the time the jury spent filling out the verdict form as evidence of further jury deliberation. Ford insists that because the jury was away for five minutes, when it should have taken only seconds to correct the error if it was indeed inadvertent, any hold-out jurors must have been pressured to vote with the majority. There are three obvious problems with Ford’s argument. First, this short time frame suggests the opposite: there were no further deliberations, which is consistent with the jury’s announcing a unanimous verdict. Second, this would also suggest the presiding juror was misleading the court when asked if the jury had reached a unanimous verdict and the presiding juror twice responded affirmatively. Third, the jurors were individually polled following the verdict delivery and each affirmed the verdict as his or her own. Moreover, pointing to the short time frame is not the affirmative showing that Ford conceded at oral argument he would need to make but instead requires speculation regarding the jury’s deliberations.2 While we can imagine some instances where a greater period of jury absence may indicate further deliberations, we expect it would be coupled with a directive from the court to reach agreement or a suggestion of a time limit for deliberations, such as occurred in Boogaard. It remains Ford’s burden to show the *192jury was undecided when sent back to the jury room, and he has not made this showing.
¶14 The Court of Appeals’ majority concluded that the trial court’s oral instruction to the jury conflicted with the written jury instructions by suggesting the need for agreement. However, review of the written instructions demonstrates that the trial court’s instructions merely restated instructions the jury had already been given in written form. As both Ford and the Court of Appeals’ majority observed, the written jury instructions stated that the jury “need not unanimously agree____” CP at 34. But the written instructions also informed the jury that it “must fill in the blank provided in each verdict form the words ‘not guilty’ or the word ‘guilty’ . . . . ” CP at 35. This written instruction appears to impart the same meaning as the trial court’s oral instruction that verdict form one “must be filled in.” Ford cannot argue that the written instructions improperly coerced the jury into agreeing, so it is hard to see how the court’s oral instruction — which was substantively identical to the written instruction — could have done so.
¶15 Finally, Ford argues that juries are presumed to follow the instructions provided. E.g., State v. Ervin, 158 Wn.2d 746, 756,147 P.3d 567 (2006). Ford contends that the return of a blank verdict form shows the jury was following instruction 12, which included, as mentioned above, that the jury need not unanimously agree that all acts were proved. But Ford’s argument conflicts with the jury’s representation of reaching a unanimous verdict. Furthermore, another instruction, which we also presume the jury followed, stated that “you should not change your honest belief as to the weight or effect of the evidence solely because of the opinions of your fellow jurors, or for the mere purpose of returning a verdict.” CP at 24. Rather than pick and choose which instruction we feel the jury followed, we presume the jurors followed all instructions and were being honest both when it announced a unanimous verdict and when each was individually polled as to that verdict.
*193CONCLUSION
¶16 Judicial coercion must include an instance of actual conduct by the trial judge during jury deliberations that could influence the jury’s decision. To make such a claim, a defendant must first make a threshold showing that the jury was still within its deliberative process. Second, though related, the defendant must affirmatively show that the jury was at that point still undecided. Third, the defendant must show judicial action designed to force or compel a decision, and fourth, the impropriety of that conduct. Finally, if raised for the first time on appeal, a defendant must show that such interference rises to the level of manifest error, such that it actually prejudiced the constitutional right to a fair trial. No such showing has been established in this case. Accordingly, we reverse the Court of Appeals’ decision and reinstate Ford’s conviction.
Alexander, Owens, and J.M. Johnson, JJ., concur.This is not to say, as the dissent suggests, that an initial indication of unanimity from the jury is conclusive proof that deliberations have ended, foreclosing all claims of judicial coercion. “ ‘Until a verdict is received and filed for record, the trial court may send the jury back to consider and clarify or correct mistakes appearing on the face of the verdict.’ ” State v. Badda, 68 Wn.2d 50, 61, 411 P.2d 411 (1966) (quoting Beglinger v. Shield, 164 Wash. 147, 152, 2 P.2d 681 (1931)). Contrary to the structural definition of “deliberations” the dissent advances, a jury should be able to fix mistakes without judicial coercion being claimed in every instance.
This court should not delve into the jury deliberation process. E.g., State v. Depaz, 165 Wn.2d 842, 854, 204 P.3d 217 (2009).