The Sixth Amendment guarantees all Americans the right to an impartial jury. Yet typical methods for selecting jurors fall short of ensuring that constitutional right: the impartial jury remains an ideal rather than a reality.

While strikes for cause eliminate prospective jurors who express overt bias, the biggest problem—the real problem—is hidden bias. By design, human beings make rapid judgments about other people upon first sight; among them are trustworthiness and likeability. These judgments quickly transform into “gut feelings”, which lead people to unconsciously filter new information in a way that confirms their original, and often erroneous, impression. These processes come into play regardless of good intentions to be fair and open-minded. In voir dire, asking prospective jurors if they can be impartial, if they can ignore pretrial publicity and put aside their opinions about a case and their feelings about a defendant, is at best futile. At worst, it drives bias underground. Our goal is to find better ways to identify hidden bias before someone takes a seat in the jury box.

To this end, we have performed several studies to reveal underlying attitudes in prospective jurors by examining the effect of question wording in change of venue surveys and voir dire. (If you are not familiar with our previous research and would like to learn about other questioning techniques we have examined, short summaries of a few of our studies appear in Appendix 1. In addition, you can read our TJE articles on prehabilitation and question wording.

Our most recent experiment involves data from nine change of venue surveys performed by the first author over the last decade. Although these were all high publicity criminal cases, due to the nature of the wording and question variables, the lessons learned will surely apply equally well to civil cases.

The Study

In criminal cases, the single most important bias issue, the one around which all others orbit, is whether a prospective juror or survey respondent believes a defendant is guilty. In a civil case, the central bias question is whether a person favors one side over the other. Many voir dire and survey questions indirectly test the focal point of bias: “Do you think big corporations are out of control in this country?” “Are there too many trivial lawsuits these days?” “What evidence have you heard about this case?”

But of course there are direct ways to ask fundamental bias questions as well. In a change of venue survey for a criminal case, for example, the wording of the direct question has traditionally resembled the wording suggested in our ASTC Practice guidelines: “Based on what you have read or heard, do you think [name] is definitely not guilty, probably not guilty, probably guilty, or definitely guilty?”

After performing several change of venue surveys in criminal cases, the first author suspected that the traditionally worded guilt/innocence question led many people to automatically give the culturally expected answer—that a person is innocent until proven guilty beyond a reasonable doubt—rather than taking some time to reflect on their true feelings about the defendant’s guilt or innocence.

What led to this suspicion? First, of those who did not say they thought the defendant was probably or definitely guilty, virtually no one chose innocent as their answer. Instead, many volunteered a third option, the legally appropriate but perhaps disingenuous, “innocent until proven guilty.” Indeed, a large number of those respondents disclosed guilty bias in their responses to later survey items. For example, they affirmed statements such as he confessed, the community thinks he’s guilty, he will be convicted, the police found the murder weapon in his apartment, and so on.

Those observations led to my supplementing the traditional guilt/innocence question with a “water cooler” version, one that though it’s more casual, the less official tone might reduce the likelihood of knee-jerk answers. My variation of the water cooler question was: “If you had to say you lean one way or the other right now about the [ROTATE] guilt or innocence of _______, which way would you lean?”

In subsequent surveys I saw that whatever percentage had answered Guilty in response to the Traditional guilt/innocence question, another ten or fifteen percent answered guilty when the Lean question came up a little later.

The current experiment was a formal test of the question, “Across nine change of venue surveys, individually and collectively, does the water cooler/lean question tap into additional Guilty bias above and beyond that uncovered by the traditional guilty bias question?”

We tested an additional possibility with the most recent survey of the nine (performed in April and May of 2016). Callers asked roughly half the respondents the Lean question only, while the other half were asked both versions as usual—that is, they answered the Traditional question, then those who did not commit to either Innocent or Guilty were asked the Lean question. We wondered whether skipping the Traditional question and going straight to the more casual Lean question would result in just as many admissions of guilty bias as would asking the two questions sequentially. If that turned out to be true, simply asking the Lean question in surveys and voir dire could uncover a great deal of guilty bias rather efficiently. (Although it might still be necessary to retain both questions in COV surveys to adhere to traditional guidelines).

Method

From 2006 through 2016, in cooperation with polling companies, the first author conducted nine landline and cell phone change of venue surveys concerning high pretrial publicity murder cases. Each survey had a target of 400 respondents, for margins of error under 5%. In two counties with small populations we had to settle for substantially smaller numbers.

After a series of screening and familiarity-with-the-crime questions came the traditional Guilty Bias question. For those who answered either Guilty or Innocent, an open-ended “why” question followed. The Lean question was asked of respondents who had not committed to innocent or guilty (except, as mentioned above, for a subset of respondents in the most recent survey).

Results

For each case separately as well as for the nine cases together, the Lean question resulted in a statistically significant increase in Guilty opinions. The increases ranged from about 9% to 19%, for an average increase of about 14%. Case by case and overall results appear in Figure 1. (See Table 1 in Appendix for data and statistical tests.)

Admissionspfguiltybias_645

Figure 1. Percent increase in admissions of Guilty opinion from traditional Guilty question to Lean Guilty question.

Note: All increases are statistically significant.

The additional variable we tested in Case 9—to determine whether the Lean question alone would yield a similarly high rate of Guilty Bias compared with the Traditional question along with the Lean question—resulted in an affirmative finding. Of the 139 respondents who answered only the Lean question, 67.6% (94) said they believed the defendant was guilty. As can be seen in Figure 1, for those answering both questions, the cumulative rate was nearly identical to the rate for the Lean question alone (66.4%; 79/119).

Discussion

What does an increase in the percentage of people admitting guilty bias mean in practical terms? Translating numbers into words for a few of the cases makes the significance clear:

  • 29% to 45% – from over a quarter to almost half (Case 5)
  • 49% to 68% – from about half to over two thirds (Case 6)
  • 42% to 57% – from less than half to more than half (Case 7)

Increases of these magnitudes in COV surveys could make the difference between a change of venue for your client being denied or granted. Furthermore, using the Lean question in jury questionnaires and voir dire — though not directly tested here — is also likely to reveal prospective jurors with a guilty bias, leading to dismissals for cause.

Asking biased jurors whether they can be impartial despite their opinions and gut feelings is not merely pointless, it puts your client in jeopardy. Revealing hidden bias in prospective jurors before seating them on a jury will help ensure your client’s right to a fair trial and strengthen the integrity of our justice system.


Mykol C. Hamilton, Centre College

Kate Zephyrhawke, Hillsborough Community College

References

Dexter, H. R., Cutler, B. L., & Moran, G. (1992). A test of voir dire as a remedy for the prejudicial effects of pretrial publicity. Journal of Applied Social Psychology,22, 819-832.

Hamilton, M., Augustus, A. N., & Melloan, J. (2011, March). Voir dire, change of venue, and “legal desirability”: Hidden jury bias in high PTP cases. Poster presented at the American Psychology-Law Society International Conference, Miami, FL.

Hamilton, M.C., Linden, E., Pitt, M., & Robbins, E. (2014, August). The ubiquity of premature rehabilitation: How “prehabilitation” leads prospective jurors to hide their biases, The Jury Expert, 26(3), 48-65.

Hamilton, M. C., & Henize, H. (February, 2013). Prehabilitation, Individual Voir Dire, and Detecting Prospective Juror Bias: The Trayvon Martin Case. Presented at the American Psychology-Law Society Conference, Portland, OR.

Hamilton, M.C., & Zephyrhawke, K. (2015, November). Revealing juror bias without biasing your juror: Experimental evidence for best practice survey and voir dire questions. The Jury Expert, 27(4). http://www.thejuryexpert.com/2015/12/revealing-juror-bias-without-biasing-your-juror-experimental-evidence-for-best-practice-survey-and-voir-dire-questions/

Moran, G., & Cutler, B. L. (1991). The prejudicial impact of pretrial publicity. Journal of Applied Social Psychology, 21, 345-367.

Apppendix 1

Background concerning our previous research:

The theme that unites findings from many of our studies is a voir dire technique we call “prehabilitation,” or the attempt to rehabilitate prospective jurors because they may potentially be biased. Research shows that plain old rehabilitation does not work well (e.g., Dexter, Cutler, & Moran, 1992; Moran & Cutler, 1991), and prehabilitation is worse. Not only does prehabilitation fail to remedy bias, but it drives bias underground. Judges prehabilitate when they introduce voir dire by drilling into prospective jurors their duty to be fair and objective, as well as challenging their abilities: “You must listen only to evidence presented in court.” “We want to know whether you’re capable ofpresuming innocence.” “It will be your job as a juror to set aside any preconceived notions.” “We’re here to see ifyou can follow the law.” The attorneys and/or judge then continue with a series of leading, prehabilitative questions in the same vein—can you, are you capable of, will you be able to … fulfill your duties, do what the law requires, meet your responsibilities?

In response, prospective jurors minimize or deny their bias. It’s no surprise that they avoid responses like no I can’t be fair, I refuse to follow the law, I’m not open-minded… Prehabilitation thus defeats the central purpose of voir dire, which is to seat a fair jury by striking biased prospective jurors.

Descriptions of four previous studies:

1. Students read about the death of Trayvon Martin, which had happened just the previous month (Hamilton & Henize, 2013). Half imagined talking to a group of friends about the case (a variation of the “water-cooler approach”); half imagined they were prospective jurors and read a judge’s prehabilitative voir dire introduction.

Those in the friends/water cooler condition, as compared with those in the prehabilitative judge introduction condition, leaned more strongly toward George Zimmerman’s having committed murder, were more certain that the defendant would not receive a fair trial, and were more certain that it would be difficult to presume Zimmerman’s innocence.

2. We found that prehabilitation in voir dire introductions by judges is ubiquitous. In ten cases across the U.S. and one in Canada, every judge used prehabilitative techniques for every one of the 604 prospective jurors (Hamilton et al., 2014 TJE paper on ubiquity of à prehabilitation)

3. Hamilton & Zephyrhawke (2015 TJE paper, question wording) found that willingness to admit difficulty with the presumption of innocence or with putting aside pretrial information is strongly affected by question wording in COV surveys.

For example, more bias was uncovered with “If you … had to decide whether _____ is guilty, you might have some trouble putting aside opinions…” than with If you … had to decide whether _____ is guilty,… you could put aside opinions…” Also, asking people how difficult it might be to “assume he is not guilty” rather than using the legalistic phrase “assume he is innocent until proven guilty” increased admissions bias.

4. People are less likely to admit guilty bias in voir dire than in COV surveys, when heavy prehabilitation occurs in a judge’s introduction and in questioning (Hamilton, Augustus, and Melloan, 2011). In one of the murder cases reported in the body of the current paper, the judge declined to change the venue. Therefore, a comparison could be made of bias admissions by survey respondents versus prospective jurors. An equally high 91% of both groups were familiar with the case, yet six times more survey respondents (42%) than prospective jurors (7%) admitted guilty bias.

Appendix 2

Table 1. Increases in admissions of Guilty opinion from traditional Guilty question to Lean Guilty question, with significance information.

Case #

Traditional Guilty question

% Guilty Answers to

(# of Guilty answers/N*)

Lean Guilty question

% increase in Guilty answers

(# of Lean Guilty answers/N)

Total

Total % Guilty answers

(total # Guilty answers/N)

Significance

Chi square statistic (1df); probability

X2 p

1

70.1%

(281/401)

9.7%

(39/401)

79.8%

320/401

9.52

= .002

2

29.8%

(90/302)

9.3%

28

39.1%

(118/302)

5.35

= .02

3

54.6%

(216/396)

9.3%

37

63.9%

(253/396)

6.78

= .009

4

24.8%

(67/270)

15.1%

41

40,0%

(108/270)

13.53

= .0002

5

28.7%

(114/397)

16.1%

64

44.8%

(178/397)

21.5

< .00001

6

48.7%

(194/398)

19.0%

76

67.8%

(270/398)

29.07

< .00001

7

23.2%

(92/395)

15.4%

61

38.7%

(153/395)

21.3

< .00001

8

42.0%

(168/400)

14.8%

59

56.8%

(227/400)

16.82

< .00001

9

48.7%

(58/119)

17.6%

21

66.4%

(79/119)

6.88

= .009

All

41.6%

(1280/3078)

13.8%

426

55.4%

(1706/3078)

117.47

< .00001

*N = total number of respondents minus number who declined to answer the Guilty question.