Study 2: Behavioral Intentions and Expectations Data Analysis

Author

Jamie C. Lee

Published

March 10, 2026

Summary

This document presents descriptive analyses for Study 2, which examined how experimental manipulation — via easy vs. hard quiz conditions using either an estate planning quiz (EP) or Raven’s Progressive Matrices (RM) — affects behavioral intentions (BI) and behavioral expectations (BE) related to estate planning as well as information-seeking behavior.

Sample. 800 participants were recruited on Prolific for a 2x2 between-subjects design. Participants were screened by age (30-65 years old), number of previous submissions on Prolific (100+), and approval rate (100%). The sample is balanced by gender (50-50 split). Our sample has an average age of 43.5 (SD=9.3), with most participants falling between the ages 30 and 44. Most participants have not completed either a will or a trust, and 41.1% of the sample have also not named beneficiaries for any of their financial accounts. 20.5% of participants reported having a will, and 10.2% of participants reported having a trust.

Manipulation checks. The easy/hard difficulty manipulation appears to have worked reasonably well, as indicated by participants’ perceived difficulty ratings and their objective performance on the quizzes, particularly in the Raven’s Matrix conditions. However, the confidence variable did not vary across conditions as expected: hard conditions produced slightly higher rather than lower measured confidence. This pattern is likely driven by the way the confidence variable is constructed, which depends on the difference between expected and actual performance. A more detailed discussion of this issue is provided in Section 2 of the document.

Behavioral intentions (BI) and behavioral expectations (BE). BI and BE were moderately-to-strongly correlated across estate planning actions, supporting the view that they are related but distinct constructs. Accordingly, the main analyses focus on the BE variables. Overall, average BE variables were low across experimental conditions, falling below the midpoint of the scale. In other words, participants reported a low likelihood of taking estate planning-related actions in the near term (e.g., drafting a will, creating a trust, talking to others about estate planning, seeking out information, etc.).

Condition assignment does not appear to meaningfully affect BI or BE. In contrast, prior estate planning engagement is more strongly associated with BE values: participants who have previously taken estate planning-related actions generally report higher likelihoods of taking additional actions in the future. This pattern suggests that prior engagement may signal greater readiness to continue progressing through the estate-planning process, and that informational interventions may be more effective when targeted toward individuals who have already begun engaging. However, even among participants with prior engagement, average BE values remain relatively low and still fall below the midpoint of the scale.

Information seeking behavior. Hover time across the eight estate planning content areas did not appear to differ meaningfully across experimental conditions. Hover behavior did not appear to be strongly related to prior estate planning engagement; however, participants with higher confidence in estate planning processes tended to spend less time hovering over informational content.

1 Demographic and Behavioral Variables

1.1 Age

As Carolyn anticipated, even though I screened for 30-65 year olds, without stratification, our sample ended up including more participants in the lower end of this range. This pattern is evident in Figure 1, which shows the continuous age distribution, and in Figure 2, which groups participants into five-year intervals.

1.1.1 Continuous Distribution

Our sample has a mean age of 43.5 (SD = 9.3).

1.1.2 Grouped Distribution

Participants were grouped into 5-year age bands. The largest group was 30-34 (n = 164, 20.5%), and the smallest was 60-65 (n = 49, 6.1%).

1.2 Prior Estate Planning Engagement

1.2.1 Estate Planning Completion Stage

Participants indicated their current stage of estate planning on a 6-point scale, ranging from having taken no steps (1) to having completed one or more documents (5); 6 = “I prefer not to answer.”

Table 1. Distribution of Estate Planning Completion Stage
Response Label N %
1 I have not taken any steps. 243 30.4
2 I have thought about it but haven't done anything yet. 232 29.0
3 I have started learning or talking about it. 128 16.0
4 I have begun some planning. 96 12.0
5 I have completed one or more estate planning documents. 100 12.5
6 I prefer not to answer. 1 0.1

The most commonly reported stages were Response 1 (“I have not taken any steps.”; n = 243, 30.4%) and Response 2 (“I have thought about it but haven’t done anything yet.”; n = 232, 29%). Combined, these two stages account for more than 50% of the sample, suggesting that most participants have yet to complete one of the primary estate planning documents (will or trust) and are still early in the estate planning process.

Table 1 frequencies are visualized in Figure 3 below.

1.2.2 Specific Estate Planning Actions

Participants indicated which specific estate planning actions (more broadly defined here) they had previously completed. A value of 1 indicates the action was taken; blank values indicate the action has not been taken.

Table 2. Prior Estate Planning Actions
Action N %
Has a Will 164 20.5
Has a Trust 82 10.2
Named Beneficiaries 421 52.6
Has Power of Attorney 125 15.6
No Prior Planning 323 40.4
Preferred Not to Respond 7 0.9

Of the various actions inquired about, the most common action was naming beneficiaries (52.6% of participants). Consistent with the completion stage frequencies reported above, a decent share of participants (40.4%) reported no prior estate planning involvement.

Among the four core estate planning actions (will, trust, beneficiary designations, and power of attorney), participants completed an average of 1 action (SD = 1.1).

Table 3. Number of Prior Estate Planning Actions Completed (out of 4)
Actions Completed N %
0 329 41.1
1 284 35.5
2 96 12.0
3 48 6.0
4 43 5.4

Table 3 frequencies are visualized in Figure 4 below.

2 Manipulation Checks: Difficulty, Actual Performance, and Confidence

The experimental conditions are coded as follows:

  • Condition 1 = Easy Estate Planning Quiz

  • Condition 2 = Hard Estate Planning Quiz

  • Condition 3 = Easy Raven’s Matrix

  • Condition 4 = Hard Raven’s Matrix.

If the easy/hard manipulation was successful, participants assigned to the hard conditions (2 and 4) should report higher perceived difficulty, score lower on the assessment, and show lower confidence than those assigned to the easy conditions (1 and 3). Generally, this is what we find. However, consistent with pre-testing results, the easy vs. hard estate planning quiz seems to be a milder manipulation compared to the Raven’s Matrix tests. Questions should be further refined to improve the strength of the manipulation.

2.1 Difficulty

Following completion of the quiz, participants were asked: To what extent did you find the questions easy or difficult to answer correctly? Difficulty was measured on a 7-point scale from 1: Very easy to 7: Very difficult.

Overall, hard conditions (2 and 4) show higher difficulty ratings than easy conditions (1 and 3), consistent with a successful manipulation. This difference is more pronounced in the Raven’s Matrix conditions than in the Estate Planning Quiz conditions, suggesting that the estate planning manipulation can be further strengthened.

Table 4. Difficulty by Condition
Condition N Mean SD Min Max
Cond 1 (Easy EP) 202 4.97 1.35 1 7
Cond 2 (Hard EP) 202 5.36 1.57 1 7
Cond 3 (Easy RM) 198 4.52 1.47 1 7
Cond 4 (Hard RM) 198 5.57 1.40 1 7

The distributions of difficulty ratings for each condition are visualized in Figure 5 below.

2.2 Actual Performance

In addition to differences in perceived difficulty ratings, the easy/hard manipulation should lead to differences in participants’ actual performance, measured by the number of quiz questions answered correctly (out of 8 questions total).

Overall, easy conditions (1 and 3) resulted in higher scores than hard conditions (2 and 4), consistent with the perceived difficulty differences observed above. Again, the score difference is more pronounced in the Raven’s Matrix conditions than in the Estate Planning Quiz conditions.

Table 5. Actual Correct by Condition
Condition N Mean SD Min Max
Cond 1 (Easy EP) 202 4.72 1.65 1 8
Cond 2 (Hard EP) 202 3.38 2.06 0 8
Cond 3 (Easy RM) 198 5.86 1.89 1 8
Cond 4 (Hard RM) 198 3.34 1.78 0 8

The distributions of correct responses for each condition are visualized in Figure 6 below.

2.3 Confidence

2.3.1 Initial Confidence Measure

Confidence was initially calculated by subtracting each participant’s actual number of correct responses from their speculated number of correct responses. Thus, positive values (> 0) indicate overconfidence (participants believed they answered more questions correctly than they actually did), while negative values (< 0) indicate underconfidence.

In theory, hard conditions (2 and 4) should produce lower confidence than easy conditions (1 and 3). However, participants in the hard conditions also achieved lower actual scores, and because this confidence measure is defined relative to these scores, the expected pattern does not clearly emerge in the observed data. In addition, a substantial share of participants are perfectly calibrated (confidence score = 0; 25.6% of the sample), which likely attenuates differences across conditions.

Table 6. Frequency Distribution of Confidence (Full Sample)
Score N %
-6 1 0.1
-5 2 0.2
-4 6 0.8
-3 36 4.5
-2 95 11.9
-1 173 21.6
0 205 25.6
1 147 18.4
2 78 9.8
3 34 4.2
4 14 1.8
5 4 0.5
6 3 0.4
7 1 0.1
8 1 0.1

Somewhat counterintuitively, the hard conditions appear to produce slightly higher measured confidence than the easy conditions (see Table 7). Although this pattern appears inconsistent with prior findings (e.g., Hadar et al., 2013), it can be explained by the construction of the confidence metric: because it is defined as the difference between expected and actual performance, lower actual scores in the hard conditions mechanically push the measure upward.

At the same time, participants’ expected scores (how many questions they think they got correct) should also be lower in the hard conditions. If both expected and actual scores shift downward together, the difference between them - our confidence measure - may remain relatively unchanged across conditions. This is consistent with what we observe in the data (see Table 7 and Figure 7). Table 8 shows that participants’ speculated number of correct responses declines in the hard conditions, which closely resembles the pattern in actual performance shown in Table 5.

Together, these observations suggest that this initial confidence metric may not exclusively capture the intended construct of interest. Therefore, we use a different measure of confidence described in the next section.

Table 7. Confidence by Condition
Condition N Mean SD Min Max
Cond 1 (Easy EP) 202 -0.33 1.32 -6 4
Cond 2 (Hard EP) 202 0.41 1.73 -4 5
Cond 3 (Easy RM) 198 -0.41 1.72 -4 7
Cond 4 (Hard RM) 198 0.35 1.91 -5 8

Table 8. Speculated Number Correct (guess_correct) by Condition
Condition N Mean SD Min Max
Cond 1 (Easy EP) 202 4.39 1.71 0 8
Cond 2 (Hard EP) 202 3.79 1.71 0 8
Cond 3 (Easy RM) 198 5.44 1.78 0 8
Cond 4 (Hard RM) 198 3.69 1.70 0 8

2.3.2 Alternative Confidence Measure

This section uses exploratory factor analysis (EFA) to identify latent structure across eight self-reported confidence items — three reflecting global confidence in estate planning generally, and five reflecting specific confidence about particular topics or decisions. Factor scores are then used as predictors of all the BI and BE outcomes in the subsequent sections.

2.3.2.1 Confidence Item Descriptions

All items were collected on 7-point scales.

Table 9. Confidence Item Descriptions
Group Item Name Item Wording Scale
Global Global: Overall I feel like I have a good overall understanding of estate planning. 1: Strongly disagree – 7: Strongly agree
Global Global: Comfort I would feel comfortable explaining the basics of estate planning to someone else. 1: Strongly disagree – 7: Strongly agree
Global Global: Confident I feel confident in my understanding of how estate planning works. 1: Strongly disagree – 7: Strongly agree
Specific Specific: No Will How well do you feel you understand how assets are distributed if someone dies without a will? 1: Not at all – 7: Very well
Specific Specific: Beneficiaries How well do you feel you understand how beneficiary designations determine who receives assets? 1: Not at all – 7: Very well
Specific Specific: Right of Survivorship How well do you feel you understand how joint ownership with right of survivorship affects who inherits property? 1: Not at all – 7: Very well
Specific Specific: POA How well do you feel you understand the role of powers of attorney when someone becomes incapacitated? 1: Not at all – 7: Very well
Specific Specific: Will vs. Trust How well do you feel you understand how wills and trusts differ in how assets are distributed? 1: Not at all – 7: Very well

2.3.2.2 EFA: Item Descriptives and Correlations

People, on average, do not seem very confident in their understanding of estate planning. However, across the 8 confidence items, there appears to be moderate-to-strong inter-item correlations, which is consistent with the presence of latent structure and supports factor analysis.

Table 10. Descriptives for Confidence Items
Item N Mean SD Min Max
Global: Overall 800 3.29 1.74 1 7
Global: Comfort 800 2.94 1.81 1 7
Global: Confident 800 3.22 1.72 1 7
Specific: No Will 800 3.35 1.82 1 7
Specific: Beneficiaries 800 3.93 1.88 1 7
Specific: Right of Survivorship 800 3.21 1.84 1 7
Specific: POA 800 4.16 1.81 1 7
Specific: Will vs. Trust 800 3.18 1.85 1 7

Items are moderately-to-strongly correlated overall, with global items tending to correlate more strongly with each other than with specific items, suggesting the presence of at least two latent factors.

Table 11. Inter-Item Correlation Matrix
Item Global: Overall Global: Comfort Global: Confident Specific: No Will Specific: Beneficiaries Specific: Right of Survivorship Specific: POA Specific: Will vs. Trust
Global: Overall 1.000 0.866 0.941 0.671 0.636 0.665 0.585 0.731
Global: Comfort 0.866 1.000 0.862 0.645 0.582 0.644 0.508 0.716
Global: Confident 0.941 0.862 1.000 0.672 0.640 0.678 0.591 0.737
Specific: No Will 0.671 0.645 0.672 1.000 0.655 0.699 0.618 0.717
Specific: Beneficiaries 0.636 0.582 0.640 0.655 1.000 0.673 0.711 0.673
Specific: Right of Survivorship 0.665 0.644 0.678 0.699 0.673 1.000 0.582 0.713
Specific: POA 0.585 0.508 0.591 0.618 0.711 0.582 1.000 0.624
Specific: Will vs. Trust 0.731 0.716 0.737 0.717 0.673 0.713 0.624 1.000

2.3.2.3 EFA: Parallel Analysis and Factor Solution

To determine the number of factors to retain, a parallel analysis was conducted. This method compares the eigenvalues from the observed data with eigenvalues generated from 500 randomly simulated datasets with the same sample size and number of variables. Factors are retained when the eigenvalue from the observed data exceeds the corresponding simulated eigenvalue. The scree plot below (Figure 9) displays both the observed and simulated eigenvalues and suggests retaining two factors.

Parallel analysis suggests retaining 2 factors.

From the factor loadings reported in Table 12 and Figure 10, Factor 1 appears to capture confidence in specific EP processes, whereas Factor 2 reflects more general or global confidence in one’s understanding of EP.

Table 12. EFA Factor Loadings (oblimin rotation, ML estimation)
Item Factor 1 Factor 2 Communality
Global: Overall -0.016 0.985 0.945
Global: Comfort 0.043 0.858 0.797
Global: Confident 0.017 0.954 0.937
Specific: No Will 0.756 0.080 0.676
Specific: Beneficiaries 0.892 -0.071 0.698
Specific: Right of Survivorship 0.742 0.091 0.669
Specific: POA 0.843 -0.086 0.600
Specific: Will vs. Trust 0.652 0.230 0.723
Loadings < |0.30| suppressed for clarity. Communality = h².
Table 13. Variance Explained by Factor Solution
Factor SS Loadings Prop. Variance Cum. Variance
ML2 Factor 1 3.202 0.400 0.400
ML1 Factor 2 2.843 0.355 0.756

2.3.2.4 Factor Scores by Condition

The experimental manipulation does not appear to have meaningfully affected participants’ self-reported confidence in their EP understanding.

Table 14. Factor Score Means (SD) by Condition
Easy EP
Hard EP
Easy RM
Hard RM
Factor Cond 1 Cond 2 Cond 3 Cond 4
Factor 1 -0.065 (0.918) -0.109 (0.957) 0.094 (1.011) 0.083 (0.935)
Factor 2 0.023 (0.967) -0.162 (0.954) 0.086 (1.038) 0.056 (0.976)

Pairwise t-tests confirm that the conditions did not meaningfully affect participants’ self-reported confidence.

Table 15. Pairwise t-Tests: Factor Scores by Condition (Holm-Corrected)
Factor Comparison t df p (raw) p (Holm) sig
Factor 1 Cond 1 vs. 2 (Easy vs. Hard EP) 0.469 401.3 0.639 1.000
Cond 3 vs. 4 (Easy vs. Hard RM) 0.105 391.6 0.916 1.000
Cond 1+2 vs. 3+4 (EP vs. RM) -2.593 795.4 0.010 0.058 †
Factor 2 Cond 1 vs. 2 (Easy vs. Hard EP) 1.940 401.9 0.053 0.221
Cond 3 vs. 4 (Easy vs. Hard RM) 0.302 392.5 0.763 1.000
Cond 1+2 vs. 3+4 (EP vs. RM) -2.015 794.8 0.044 0.221
Welch's t-tests. sig column based on Holm-corrected p. † p < .10, * p < .05, ** p < .01, *** p < .001

3 Behavioral Intentions and Behavioral Expectations

Following the manipulation, participants were asked a series of questions about actions they might consider taking related to estate planning in the future. Specifically, they reported both their behavioral intentions and behavioral expectations for 7 estate planning-related actions.

Behavioral intentions (BI, bi_ prefix) capture what participants say they intend to do. Participants indicated whether they presently intend to perform each behavior within the next week using a 9-point scale anchored at 1: No, definitely do not intend and 9: Yes, definitely do intend.

Behavioral expectations (BE, be_ prefix) capture what participants expect they will actually do. Participants reported how likely it is that they will perform each behavior within the next week using a 9-point scale anchored at 1: Extremely unlikely and 9: Extremely likely. Prior literature suggests these constructs are related but conceptually distinct (Warshaw & Davis, 1985; Tanner & Carlson, 2009).

3.1 BI and BE Descriptives by Condition

3.1.1 Aggregated across Full Sample

Overall, both behavioral intentions (BI) and behavioral expectations (BE) are low across all conditions, with average values falling below the midpoint of the scale. This suggests that participants generally report a low likelihood of taking estate planning-related actions in the near term. However, BI and BE values are slightly higher in the EP quiz conditions than in the RM conditions, indicating that participants who completed the estate-planning quiz report somewhat greater intentions and expectations to take related actions.

Table 16. Behavioral Intentions by Condition — Mean (SD)
Easy EP
Hard EP
Easy RM
Hard RM
Variable Cond 1 Cond 2 Cond 3 Cond 4
Will 3.16 (2.83) 3.07 (2.71) 2.81 (2.75) 2.61 (2.47)
Beneficiaries 3.71 (3.06) 3.65 (2.9) 2.98 (2.89) 2.64 (2.54)
Trust 2.77 (2.54) 2.68 (2.46) 2.48 (2.49) 2.11 (1.97)
POA 3.21 (2.89) 2.91 (2.62) 2.71 (2.59) 2.4 (2.41)
Talk 3.42 (2.78) 3.6 (2.76) 3.38 (2.87) 3.19 (2.64)
Lawyer 2.7 (2.44) 2.55 (2.34) 2.53 (2.55) 2.23 (2.17)
Info 4.23 (3.08) 4.33 (3.03) 3.73 (2.91) 3.6 (2.72)
Table 17. Behavioral Expectations by Condition — Mean (SD)
Easy EP
Hard EP
Easy RM
Hard RM
Variable Cond 1 Cond 2 Cond 3 Cond 4
Will 2.46 (2.43) 2.43 (2.35) 2.2 (2.21) 1.97 (1.99)
Beneficiaries 3.2 (2.93) 2.94 (2.63) 2.47 (2.48) 2.33 (2.3)
Trust 2.15 (2.14) 2.1 (1.98) 1.94 (1.92) 1.66 (1.53)
POA 2.52 (2.42) 2.22 (2.25) 2.07 (1.92) 1.93 (1.98)
Talk 3.03 (2.61) 3.23 (2.67) 2.92 (2.52) 2.79 (2.47)
Lawyer 2.23 (2.12) 2.09 (2) 2.02 (2.07) 1.75 (1.69)
Info 3.75 (3.04) 3.88 (2.97) 3.21 (2.7) 3.13 (2.55)

The box plots in Figure 12 below suggest that although the mean and median BI and BE values do not differ substantially across conditions, the distribution of scores appears somewhat higher in the EP quiz conditions. This pattern suggests that while the manipulation may not shift the overall distribution of responses, it may increase intentions or expectations for a subset of participants. This possibility motivates additional analyses of specific subgroups - for example, splitting participants based on their prior estate planning engagement, as explored in the following subsection and subsequent sections.

Because behavioral expectations (BE) have been shown to be more predictive of actual future behavior, all subsequent analyses focus on BE variables. BI variables are therefore omitted from the main analyses. Additional robustness checks examining the relationship between BI and BE variables are reported in Appendix C.

3.1.2 Split by Prior EP Completion Stage

Altogether, the tables below suggest that participants who report being in the stage of “I have started learning or talking about estate planning” tend to have higher BEs across the 7 estate planning-related behaviors. This pattern appears to hold regardless of condition assignment or manipulation. However, these average BEs still rarely exceed the scale midpoint (5).

3.1.2.1 BE: Will

BE: Will — Mean (SD) by EP Completion Stage and Condition
By Condition
EP Completion Stage Overall Easy EP Hard EP Easy RM Hard RM
No steps taken 1.41 (1.2) 1.41 (1.28) 1.64 (1.54) 1.4 (1.12) 1.21 (0.68)
Thought about it 2.28 (2.22) 2.62 (2.56) 2.51 (2.23) 2.06 (2.1) 1.95 (1.92)
Started learning/talking 3.34 (2.71) 3.9 (2.82) 3.46 (2.98) 3.22 (2.64) 2.72 (2.27)
Begun some planning 2.9 (2.5) 3.12 (2.76) 2.42 (2.18) 3.45 (2.86) 2.27 (1.53)
Completed EP documents 2.35 (2.64) 2.18 (2.38) 2.75 (2.85) 1.65 (1.87) 2.79 (3.19)

3.1.2.2 BE: Beneficiaries

BE: Beneficiaries — Mean (SD) by EP Completion Stage and Condition
By Condition
EP Completion Stage Overall Easy EP Hard EP Easy RM Hard RM
No steps taken 1.81 (1.85) 1.9 (2) 2.3 (2.28) 1.62 (1.62) 1.44 (1.32)
Thought about it 2.88 (2.61) 3.44 (3.05) 3.17 (2.62) 2.23 (2.15) 2.6 (2.39)
Started learning/talking 3.73 (2.91) 4.5 (3.05) 3.92 (2.99) 3.69 (2.82) 2.76 (2.6)
Begun some planning 3.3 (2.71) 4.08 (3.06) 2.27 (1.64) 3.76 (3.21) 2.87 (2.03)
Completed EP documents 2.86 (3.02) 3.27 (3.27) 3.33 (3.27) 1.96 (2.37) 2.96 (3.1)

3.1.2.3 BE: Trust

BE: Trust — Mean (SD) by EP Completion Stage and Condition
By Condition
EP Completion Stage Overall Easy EP Hard EP Easy RM Hard RM
No steps taken 1.34 (1.12) 1.38 (1.34) 1.52 (1.41) 1.31 (0.96) 1.14 (0.56)
Thought about it 1.96 (1.75) 2.32 (2.12) 2.32 (1.98) 1.64 (1.48) 1.57 (1.16)
Started learning/talking 2.88 (2.42) 3.33 (2.88) 2.59 (2.22) 3.41 (2.51) 2.17 (1.89)
Begun some planning 2.2 (2.07) 2 (1.5) 2.23 (2.2) 2.41 (2.47) 2.07 (1.98)
Completed EP documents 2.12 (2.35) 2.41 (2.67) 2.25 (2.4) 1.62 (1.88) 2.25 (2.49)

3.1.2.4 BE: POA

BE: POA — Mean (SD) by EP Completion Stage and Condition
By Condition
EP Completion Stage Overall Easy EP Hard EP Easy RM Hard RM
No steps taken 1.43 (1.28) 1.48 (1.32) 1.46 (1.27) 1.52 (1.38) 1.27 (1.14)
Thought about it 2.21 (2.13) 2.6 (2.43) 2.45 (2.31) 1.68 (1.45) 2.06 (2.08)
Started learning/talking 2.86 (2.36) 3.47 (2.7) 2.92 (2.75) 2.91 (1.8) 2.1 (1.86)
Begun some planning 2.85 (2.57) 3.35 (2.88) 2.08 (1.9) 3.28 (2.75) 2.53 (2.5)
Completed EP documents 2.49 (2.65) 2.91 (2.81) 2.75 (3.05) 1.73 (2.03) 2.64 (2.67)

3.1.2.5 BE: Talk

BE: Talk — Mean (SD) by EP Completion Stage and Condition
By Condition
EP Completion Stage Overall Easy EP Hard EP Easy RM Hard RM
No steps taken 1.8 (1.63) 1.74 (1.49) 2.11 (1.92) 1.59 (1.21) 1.76 (1.77)
Thought about it 3.08 (2.48) 3.25 (2.69) 3.4 (2.57) 2.96 (2.4) 2.75 (2.25)
Started learning/talking 4.38 (2.83) 4.3 (2.79) 4.46 (3.05) 4.28 (2.69) 4.48 (2.86)
Begun some planning 4.08 (2.76) 3.81 (2.91) 4.04 (2.72) 4.48 (2.92) 3.87 (2.39)
Completed EP documents 2.9 (2.77) 3.32 (2.95) 3.04 (2.85) 2.42 (2.56) 2.89 (2.82)

3.1.2.6 BE: Lawyer

BE: Lawyer — Mean (SD) by EP Completion Stage and Condition
By Condition
EP Completion Stage Overall Easy EP Hard EP Easy RM Hard RM
No steps taken 1.35 (1.11) 1.44 (1.43) 1.41 (1.26) 1.29 (0.75) 1.24 (0.86)
Thought about it 2 (1.93) 2.32 (2.18) 2.45 (2.19) 1.81 (1.97) 1.48 (1.09)
Started learning/talking 2.86 (2.32) 3.23 (2.39) 2.68 (2.32) 3.22 (2.51) 2.31 (1.97)
Begun some planning 2.5 (2.29) 2.62 (2.12) 1.88 (1.61) 2.9 (2.82) 2.6 (2.44)
Completed EP documents 2.19 (2.4) 2.32 (2.51) 2.38 (2.53) 1.62 (1.83) 2.46 (2.67)

3.1.2.7 BE: Info

BE: Info — Mean (SD) by EP Completion Stage and Condition
By Condition
EP Completion Stage Overall Easy EP Hard EP Easy RM Hard RM
No steps taken 2.21 (2.05) 2.3 (2.16) 2.84 (2.43) 2 (1.81) 1.73 (1.61)
Thought about it 3.87 (2.82) 4.29 (3.17) 4.42 (3.07) 3.28 (2.37) 3.48 (2.47)
Started learning/talking 5.14 (2.8) 5.2 (3) 5.08 (2.96) 4.94 (2.86) 5.38 (2.41)
Begun some planning 4.1 (2.87) 3.96 (2.88) 3.73 (2.86) 4.66 (3.22) 3.93 (2.19)
Completed EP documents 3.09 (3.11) 4 (3.57) 3.79 (3.41) 2.04 (2.36) 2.75 (2.88)

3.1.3 Split by Prior Specific EP Actions

There may be a small effect of prior engagement in specific EP behaviors on behavioral expectations (BE) for future engagement.

In each table below, participants are divided into two groups: those who have not previously taken the specified EP action (e.g., creating a will) and those who have. Within each of these groups, participants are further split by experimental condition. Each row reports the average BE value, participants’ reported likelihood of taking the action within the next week, measured on a 9-point scale (1: Extremely unlikely to 9: Extremely likely).

Overall, the results suggest that the EP quiz may slightly increase BE values relative to the Raven’s Matrix conditions, with the easy EP quiz appearing somewhat more effective. However, there may be a stronger pattern with respect to prior behavior: participants who have previously engaged in EP-related actions generally report higher BE values for future actions. This pattern suggests that prior engagement may be an important predictor of readiness to take additional steps, and that informational interventions may be more effective when targeted to individuals who have already begun engaging with EP.

That said, average BE values remain relatively low - below the midpoint of the scale - even among participants with prior EP engagement.

3.1.3.1 Has a Will

BE Outcomes by Has a Will and Condition — Mean (SD)
Has not done this
Has done this
Variable Easy EP Hard EP Easy RM Hard RM Easy EP Hard EP Easy RM Hard RM
Will 2.22 (2.23) 2.29 (2.26) 2.04 (2.01) 1.73 (1.6) 3.35 (2.93) 3 (2.61) 2.79 (2.76) 3 (2.96)
Beneficiaries 3.08 (2.89) 2.9 (2.61) 2.39 (2.34) 2.21 (2.15) 3.65 (3.05) 3.1 (2.73) 2.79 (2.93) 2.82 (2.82)
Trust 2.03 (2.07) 2.06 (1.92) 1.84 (1.69) 1.54 (1.29) 2.63 (2.32) 2.3 (2.21) 2.3 (2.58) 2.13 (2.26)
POA 2.29 (2.25) 2.11 (2.12) 1.9 (1.61) 1.77 (1.76) 3.37 (2.8) 2.65 (2.68) 2.7 (2.7) 2.61 (2.65)
Talk 2.77 (2.51) 3.1 (2.61) 2.8 (2.41) 2.72 (2.4) 4 (2.79) 3.75 (2.84) 3.37 (2.88) 3.11 (2.74)
Lawyer 2.03 (1.97) 2.02 (1.88) 1.9 (1.85) 1.56 (1.39) 2.98 (2.48) 2.35 (2.41) 2.47 (2.7) 2.53 (2.46)
Info 3.64 (3.05) 3.86 (2.91) 3.14 (2.56) 3.12 (2.5) 4.14 (3) 3.95 (3.27) 3.47 (3.19) 3.16 (2.8)

3.1.3.2 Has a Trust

BE Outcomes by Has a Trust and Condition — Mean (SD)
Has not done this
Has done this
Variable Easy EP Hard EP Easy RM Hard RM Easy EP Hard EP Easy RM Hard RM
Will 2.45 (2.43) 2.36 (2.3) 2.21 (2.2) 1.97 (1.97) 2.52 (2.54) 3.17 (2.75) 2.16 (2.3) 2 (2.28)
Beneficiaries 3.19 (2.93) 2.95 (2.65) 2.46 (2.44) 2.34 (2.3) 3.29 (2.99) 2.83 (2.43) 2.6 (2.83) 2.22 (2.39)
Trust 2.12 (2.13) 2.03 (1.89) 1.89 (1.84) 1.62 (1.45) 2.48 (2.18) 2.89 (2.65) 2.28 (2.41) 2 (2.2)
POA 2.47 (2.37) 2.21 (2.25) 2.02 (1.84) 1.92 (1.98) 2.95 (2.78) 2.28 (2.35) 2.44 (2.42) 2.06 (2.1)
Talk 2.94 (2.57) 3.24 (2.68) 2.9 (2.5) 2.84 (2.48) 3.76 (2.91) 3.17 (2.57) 3.08 (2.74) 2.28 (2.35)
Lawyer 2.17 (2.04) 2.04 (1.97) 2.01 (2.02) 1.73 (1.64) 2.71 (2.7) 2.61 (2.25) 2.12 (2.39) 1.94 (2.1)
Info 3.67 (2.98) 3.84 (2.94) 3.28 (2.7) 3.25 (2.59) 4.38 (3.53) 4.33 (3.4) 2.76 (2.76) 1.94 (1.83)

3.1.3.3 Named Beneficiaries

BE Outcomes by Named Beneficiaries and Condition — Mean (SD)
Has not done this
Has done this
Variable Easy EP Hard EP Easy RM Hard RM Easy EP Hard EP Easy RM Hard RM
Will 2.16 (2.19) 2.29 (2.17) 2.17 (2.21) 1.74 (1.67) 2.73 (2.6) 2.56 (2.5) 2.23 (2.21) 2.2 (2.25)
Beneficiaries 2.53 (2.58) 2.82 (2.56) 2.42 (2.44) 1.9 (1.76) 3.79 (3.1) 3.04 (2.69) 2.52 (2.52) 2.76 (2.67)
Trust 2.01 (2.11) 1.91 (1.56) 1.7 (1.56) 1.44 (1.14) 2.28 (2.16) 2.28 (2.28) 2.14 (2.16) 1.87 (1.82)
POA 2.12 (2.09) 2.05 (1.95) 2.03 (1.94) 1.73 (1.8) 2.88 (2.63) 2.37 (2.49) 2.1 (1.91) 2.14 (2.14)
Talk 2.33 (2.16) 2.93 (2.48) 2.89 (2.56) 2.43 (2.26) 3.65 (2.83) 3.51 (2.81) 2.95 (2.5) 3.15 (2.62)
Lawyer 2 (1.91) 1.92 (1.67) 1.93 (2.05) 1.39 (1.07) 2.43 (2.27) 2.25 (2.25) 2.09 (2.08) 2.1 (2.08)
Info 3.14 (2.79) 3.9 (2.84) 3.16 (2.62) 2.85 (2.5) 4.29 (3.15) 3.87 (3.1) 3.26 (2.78) 3.41 (2.59)

3.1.3.4 Has Power of Attorney

BE Outcomes by Has Power of Attorney and Condition — Mean (SD)
Has not done this
Has done this
Variable Easy EP Hard EP Easy RM Hard RM Easy EP Hard EP Easy RM Hard RM
Will 2.43 (2.37) 2.4 (2.3) 2.27 (2.22) 1.97 (1.96) 2.64 (2.76) 2.61 (2.63) 1.85 (2.12) 1.96 (2.2)
Beneficiaries 3.17 (2.91) 2.94 (2.59) 2.58 (2.55) 2.32 (2.28) 3.33 (3.08) 2.94 (2.85) 1.97 (2.1) 2.39 (2.48)
Trust 2.14 (2.09) 2.12 (1.9) 1.99 (1.96) 1.69 (1.58) 2.24 (2.37) 2.03 (2.4) 1.67 (1.73) 1.43 (1.23)
POA 2.45 (2.29) 2.19 (2.15) 2.07 (1.93) 1.86 (1.85) 2.88 (3.01) 2.35 (2.78) 2.09 (1.93) 2.39 (2.66)
Talk 2.98 (2.55) 3.13 (2.59) 2.91 (2.54) 2.73 (2.39) 3.3 (2.96) 3.77 (3.03) 3 (2.44) 3.18 (2.92)
Lawyer 2.15 (1.99) 2.05 (1.87) 2.15 (2.17) 1.73 (1.62) 2.64 (2.68) 2.29 (2.61) 1.39 (1.27) 1.86 (2.07)
Info 3.67 (2.98) 3.83 (2.88) 3.3 (2.71) 3.18 (2.52) 4.15 (3.32) 4.16 (3.47) 2.79 (2.64) 2.86 (2.77)

3.2 Relationship between Confidence Level and BE (using the Alternative Confidence Measure)

3.2.1 Specific EP Confidence

Higher confidence across the five specific aspects of estate planning is associated with greater expected likelihood of engaging in each of the 7 estate planning-related behaviors. However, this relationship appears to be driven largely by the fact that both factors (specific and global) of confidence are correlated with prior EP engagement (see Table 18 below). This suggests that the experimental manipulation may not have meaningfully affected confidence and/or that this particular measure of confidence may instead reflect a more stable characteristic shaped by prior behavior.

Table 18. Correlations: Confidence Factors × Prior EP Engagement
Factor Variable N r t df p sig
Factor 1 EP Completion Stage (1-5) 799 0.482 15.545 797 0 ***
Prior Number of Actions Completed (0-4) 799 0.430 13.455 797 0 ***
Factor 2 EP Completion Stage (1-5) 799 0.535 17.884 797 0 ***
Prior Number of Actions Completed (0-4) 799 0.482 15.546 797 0 ***
ep_completion values of 6 excluded. † p < .10, * p < .05, ** p < .01, *** p < .001

The relationship between confidence and BE appears to be fairly noisy with poor model fit (see scatterplots in Figures 13 and 15 below). Thus, I would not treat it too seriously for now.

Table 19. Regressions: Factor 1 → BE Outcomes
Outcome N B SE R2 p sig
BE: Will 800 0.690 0.080 0.086 0 ***
BE: Beneficiaries 800 0.669 0.094 0.060 0 ***
BE: Trust 800 0.580 0.068 0.084 0 ***
BE: POA 800 0.670 0.076 0.088 0 ***
BE: Talk 800 0.745 0.091 0.077 0 ***
BE: Lawyer 800 0.635 0.070 0.095 0 ***
BE: Info 800 0.719 0.102 0.059 0 ***
† p < .10, * p < .05, ** p < .01, *** p < .001

3.2.2 Global EP Confidence

Like the specific EP results in the previous section, higher general confidence around estate planning is associated with greater expected likelihood of engaging in each of the 7 estate planning-related behaviors. However, again, this relationship appears to be driven largely by the fact that confidence is correlated with prior EP engagement.

Table 21. Regressions: Factor 2 → BE Outcomes
Outcome N B SE R2 p sig
BE: Will 800 0.603 0.078 0.070 0 ***
BE: Beneficiaries 800 0.544 0.092 0.042 0 ***
BE: Trust 800 0.533 0.066 0.076 0 ***
BE: POA 800 0.614 0.074 0.079 0 ***
BE: Talk 800 0.651 0.089 0.063 0 ***
BE: Lawyer 800 0.564 0.068 0.079 0 ***
BE: Info 800 0.572 0.100 0.040 0 ***
† p < .10, * p < .05, ** p < .01, *** p < .001

3.3 Information Seeking Behavior

Hover time across 8 informational content areas (will, trust, probate, beneficiary, POA, starting estate planning, timing, and strategy) was converted from milliseconds to seconds (÷ 1,000) and log-transformed using log(x+1) to address the right-skewness of the data. Overall, there does not appear to be any meaningful differences across conditions.

3.3.1 Distribution of Logged Hover Time (Aggregated across Conditions) by Topic

Unlike the results from Study 1, a substantial share of participants in this study (approximately 20–30%) did not hover to view the information for each topic. There is no clear evidence that participants who did not hover had more prior EP engagement than those who did. Moreover, across the full distribution of hover times, there does not appear to be a relationship between prior EP engagement and logged hover time (see Table 22 below).

Table 22. Correlations: Hover Time × EP Engagement (Holm-Corrected)
Topic Predictor N r t df p (raw) p (Holm) sig
Will EP Completion Stage 799 -0.081 -2.300 797 0.022 0.326
Prior Actions Completed 799 -0.080 -2.276 797 0.023 0.326
Trust EP Completion Stage 799 -0.021 -0.592 797 0.554 1.000
Prior Actions Completed 799 -0.043 -1.227 797 0.220 1.000
Probate EP Completion Stage 799 -0.012 -0.332 797 0.740 1.000
Prior Actions Completed 799 -0.024 -0.688 797 0.492 1.000
Beneficiary EP Completion Stage 799 -0.012 -0.330 797 0.742 1.000
Prior Actions Completed 799 -0.004 -0.103 797 0.918 1.000
POA EP Completion Stage 799 -0.038 -1.075 797 0.283 1.000
Prior Actions Completed 799 -0.032 -0.898 797 0.370 1.000
Starting EP EP Completion Stage 799 -0.031 -0.887 797 0.376 1.000
Prior Actions Completed 799 -0.079 -2.241 797 0.025 0.329
Timing EP Completion Stage 799 -0.060 -1.688 797 0.092 1.000
Prior Actions Completed 799 -0.101 -2.866 797 0.004 0.068 †
Strategy EP Completion Stage 799 0.030 0.856 797 0.392 1.000
Prior Actions Completed 799 0.001 0.025 797 0.980 1.000
Holm correction applied across all tests. ep_completion values of 6 excluded. † p < .10, * p < .05, ** p < .01, *** p < .001

3.3.2 Logged Hover Time Descriptives by Condition

Logged hover time across topics does not appear to differ meaningfully by condition. Pairwise t-tests indicate that these differences are not statistically significant. For brevity, these analyses are not included in the document.

Table 23. Logged Hover Time by Topic and Condition — Mean (SD)
Easy EP
Hard EP
Easy RM
Hard RM
Topic Easy EP Hard EP Easy RM Hard RM
Will 2.32 (1.22) 2.31 (1.21) 2.22 (1.17) 2 (1.25)
Trust 1.45 (1.05) 1.62 (1.02) 1.26 (1.04) 1.3 (1.02)
Probate 1.5 (1.18) 1.63 (1.17) 1.44 (1.17) 1.44 (1.19)
Beneficiary 1.56 (1.25) 1.56 (1.26) 1.26 (1.2) 1.29 (1.23)
POA 1.23 (1.06) 1.2 (1.03) 0.96 (1) 0.94 (0.99)
Starting EP 1.25 (1.05) 1.25 (1.02) 1.13 (1.02) 1.11 (1.05)
Timing 1.01 (0.9) 1.11 (0.93) 1.05 (0.91) 0.93 (0.93)
Strategy 0.83 (0.78) 0.83 (0.74) 0.72 (0.74) 0.73 (0.72)

3.3.3 Relationship between Confidence and Hover Time

Overall, across most EP topics, participants with higher confidence tended to spend less time hovering over the informational content.

Table 24. Correlations: Hover Time × Confidence Factors (Holm-Corrected)
Topic Predictor N r t df p (raw) p (Holm) sig
Will Confidence Factor 1 800 -0.148 -4.233 798 0.000 0.000 ***
Confidence Factor 2 800 -0.165 -4.712 798 0.000 0.000 ***
Trust Confidence Factor 1 800 -0.101 -2.878 798 0.004 0.029 *
Confidence Factor 2 800 -0.143 -4.084 798 0.000 0.001 **
Probate Confidence Factor 1 800 -0.046 -1.298 798 0.195 0.389
Confidence Factor 2 800 -0.098 -2.796 798 0.005 0.032 *
Beneficiary Confidence Factor 1 800 -0.083 -2.362 798 0.018 0.074 †
Confidence Factor 2 800 -0.116 -3.309 798 0.001 0.009 **
POA Confidence Factor 1 800 -0.095 -2.705 798 0.007 0.035 *
Confidence Factor 2 800 -0.113 -3.213 798 0.001 0.011 *
Starting EP Confidence Factor 1 800 -0.126 -3.601 798 0.000 0.003 **
Confidence Factor 2 800 -0.158 -4.522 798 0.000 0.000 ***
Timing Confidence Factor 1 800 -0.134 -3.830 798 0.000 0.002 **
Confidence Factor 2 800 -0.145 -4.126 798 0.000 0.001 **
Strategy Confidence Factor 1 800 -0.043 -1.216 798 0.224 0.389
Confidence Factor 2 800 -0.066 -1.859 798 0.063 0.190
Holm correction applied across all tests. † p < .10, * p < .05, ** p < .01, *** p < .001

4 Appendix

4.1 Appendix A: Randomization Check

Age and prior estate planning involvement are compared across the four experimental conditions to assess whether randomization was successful.

4.1.1 Age by Condition

Table A1. Age by Condition
Condition N Mean SD Min Max
Condition 1 202 44.3 9.7 30 65
Condition 2 202 43.8 9.3 30 64
Condition 3 198 42.8 9.2 30 63
Condition 4 198 43.0 9.0 30 65

Age appears broadly comparable across conditions.

4.1.2 Prior Estate Planning by Condition

Table A2. Prior Estate Planning Actions by Condition (N, %)
Action Condition 1 Condition 2 Condition 3 Condition 4
Has a Will 43 (21.3%) 40 (19.8%) 43 (21.7%) 38 (19.2%)
Has a Trust 21 (10.4%) 18 (8.9%) 25 (12.6%) 18 (9.1%)
Named Beneficiaries 107 (53%) 106 (52.5%) 109 (55.1%) 99 (50%)
Has Power of Attorney 33 (16.3%) 31 (15.3%) 33 (16.7%) 28 (14.1%)
Table A3. Mean Number of Prior Estate Planning Actions by Condition
Condition N Mean SD
Condition 1 202 1.01 1.12
Condition 2 202 0.97 1.05
Condition 3 198 1.06 1.19
Condition 4 198 0.92 1.12

The distribution of prior estate planning actions appears broadly similar across conditions, suggesting successful randomization with respect to prior engagement.

4.2 Appendix B: Additional Manipulation Checks

Motivation, enjoyment, guessing, and total time on task are examined across the four conditions. These variables should be relatively comparable across conditions, to confirm that the manipulation did not inadvertently affect other constructs.

4.2.1 Motivation

Table B1. Motivation by Condition
Condition N Mean SD Min Max
Cond 1 (Easy EP) 202 6.13 1.24 1 7
Cond 2 (Hard EP) 202 6.00 1.22 1 7
Cond 3 (Easy RM) 198 6.03 1.31 1 7
Cond 4 (Hard RM) 198 5.71 1.46 1 7

4.2.2 Enjoyment

Table B2. Enjoyment by Condition
Condition N Mean SD Min Max
Cond 1 (Easy EP) 202 4.28 1.77 1 7
Cond 2 (Hard EP) 202 4.29 1.74 1 7
Cond 3 (Easy RM) 198 4.46 1.91 1 7
Cond 4 (Hard RM) 198 3.63 1.85 1 7

4.2.3 Total Time on Task

Total time on task (condition_time_sec) is defined as the sum of time in seconds spent across all 8 quiz pages.

Table B3. Total Time on Task (seconds) by Condition
Condition N Mean SD Min Max
Cond 1 (Easy EP) 202 213.44 139.20 71.3 1184.1
Cond 2 (Hard EP) 202 255.24 163.45 68.8 1514.1
Cond 3 (Easy RM) 198 244.08 123.66 76.2 777.2
Cond 4 (Hard RM) 198 333.13 238.89 17.9 1831.3

Participants did not appear to spend substantially more time in the hard conditions compared to the easy ones, suggesting that any increased difficulty was not primarily driven by additional time on task.

4.3 Appendix C: Relationship between BI and BE

Based on prior literature, BI and BE are expected to be positively correlated, with BE values typically slightly lower than BI values. This section of the appendix examines whether the two measures show this expected pattern.

4.3.1 Correlations Between BI–BE Counterparts

Moderate-to-high but sub-1.0 correlations support treating BI and BE as related but distinct constructs, consistent with prior literature.

Table C1. Pearson Correlations Between BI and BE Counterparts
Action N r (BI–BE)
Will 800 0.731
Beneficiaries 800 0.786
Trust 800 0.716
Talk to Family 800 0.798
Consult Lawyer 800 0.760
Seek Information 800 0.823
POA 800 0.710

4.3.2 Signed Difference Scores (BI − BE)

Positive values indicate that participants report stronger intentions than expectations (BI > BE), while negative values indicate the opposite; a value of zero reflects perfect alignment between the two. As expected, BI values are, on average, slightly higher than their BE counterparts, which is consistent with patterns documented in prior literature.

Table C2. Signed Difference Scores (BI − BE): Descriptives
Action N Mean SD Min Max
Will 800 0.65 1.86 -8 8
Beneficiaries 800 0.51 1.82 -6 8
Trust 800 0.55 1.68 -7 8
POA 800 0.62 1.88 -8 8
Talk 800 0.40 1.70 -7 8
Lawyer 800 0.48 1.55 -6 8
Info 800 0.48 1.73 -6 8
Table C3. Signed Difference Scores (BI − BE) by EP Completion Stage — Mean (SD)
Action No steps Thought about it Started learning Begun planning Completed docs
Will 0.47 (1.45) 0.79 (1.89) 0.67 (2.1) 0.62 (1.92) 0.75 (2.28)
Beneficiaries 0.35 (1.37) 0.62 (1.98) 0.42 (1.99) 0.64 (1.75) 0.67 (2.21)
Trust 0.38 (1.22) 0.6 (1.88) 0.54 (1.78) 0.69 (1.59) 0.73 (2.05)
POA 0.45 (1.39) 0.74 (1.92) 0.74 (2.15) 0.72 (1.93) 0.54 (2.36)
Talk 0.26 (1.32) 0.58 (1.69) 0.36 (1.95) 0.34 (1.97) 0.46 (1.97)
Lawyer 0.3 (1.24) 0.52 (1.58) 0.59 (1.63) 0.62 (1.78) 0.55 (1.82)
Info 0.32 (1.34) 0.56 (1.72) 0.44 (2.01) 0.77 (2.11) 0.45 (1.78)
Positive values = BI > BE; negative values = BE > BI; zero = perfect alignment.