AMP_ex5
Milka Waniak
2025-10-27
Demographics
In this experiment,we probed whether (a) the awareness effect emerges on the third-party AMP – First-person AMP: Pdiff(pleasant) = 0.350 (median = 0.353; SD = 0.358). Third-party AMP: Pdiff(pleasant) = 0.348 (median = 0.389; SD = 0.380) – how much more likely participants are to say “pleasant” after a pleasant prime vs. an unpleasant prime – Pdiff = 0.35 means pleasant primes increased “pleasant” responses by 35 percentage points compared to unpleasant primes –First-person AMP: P(aware) = 0.477 (median = 0.471; SD = 0.342). Third-party AMP: P(aware) = 0.566 (median = 0.583; SD = 0.250) —This is the proportion of trials where participants said they were “influenced” (vs “not_influenced”) (b) awareness effects on the first-person and third-party AMPs are related to each other.
##
## === FIRST-PERSON AMP ===## AMP Score (Pdiff pleasant):## Mean: 0.6## Median: 1## SD: 0.5477226##
## Proportion with positive AMP effect:##
## FALSE TRUE
## 0.4 0.6##
## Awareness (P(aware)):## Mean: 0.5529412## Median: 0.5## SD: 0.2666595##
## === THIRD-PERSON AMP ===## AMP Score (Pdiff pleasant):## Mean: 0.6## Median: 1## SD: 0.5477226##
## Proportion with positive AMP effect:##
## FALSE TRUE
## 0.4 0.6##
## Awareness (P(aware)):## Mean: 0.5166667## Median: 0.4444444## SD: 0.2442234
##
## === TRIAL-LEVEL ANALYSIS ===##
## Model comparison: Response effect## Data: exp_long
## Models:
## m1: aware_response_fact ~ 1 + (1 | participant_id)
## m2: aware_response_fact ~ trial_response_fact + (1 | participant_id)
## npar AIC BIC logLik -2*log(L) Chisq Df Pr(>Chisq)
## m1 2 422.66 430.37 -209.33 418.66
## m2 3 424.66 436.23 -209.33 418.66 0.0023 1 0.9615##
## Model comparison: Prime type effect## Data: exp_long
## Models:
## m2: aware_response_fact ~ trial_response_fact + (1 | participant_id)
## m3: aware_response_fact ~ trial_response_fact + block_name + (1 | participant_id)
## npar AIC BIC logLik -2*log(L) Chisq Df Pr(>Chisq)
## m2 3 424.66 436.23 -209.33 418.66
## m3 4 424.80 440.23 -208.40 416.80 1.8544 1 0.1733##
## Model comparison: Response × Prime Type interaction## Data: exp_long
## Models:
## m3: aware_response_fact ~ trial_response_fact + block_name + (1 | participant_id)
## m4: aware_response_fact ~ trial_response_fact * block_name + (1 | participant_id)
## npar AIC BIC logLik -2*log(L) Chisq Df Pr(>Chisq)
## m3 4 424.80 440.23 -208.40 416.80
## m4 5 416.63 435.92 -203.32 406.63 10.168 1 0.001429 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1##
## Model comparison: AMP target effect## Data: exp_long
## Models:
## m4: aware_response_fact ~ trial_response_fact * block_name + (1 | participant_id)
## m5: aware_response_fact ~ trial_response_fact * block_name + amp_target + (1 | participant_id)
## npar AIC BIC logLik -2*log(L) Chisq Df Pr(>Chisq)
## m4 5 416.63 435.92 -203.32 406.63
## m5 6 418.29 441.44 -203.15 406.29 0.3432 1 0.558##
## Model comparison: Three-way interaction## Data: exp_long
## Models:
## m5: aware_response_fact ~ trial_response_fact * block_name + amp_target + (1 | participant_id)
## m6: aware_response_fact ~ trial_response_fact * block_name * amp_target + (1 | participant_id)
## npar AIC BIC logLik -2*log(L) Chisq Df Pr(>Chisq)
## m5 6 418.29 441.44 -203.15 406.29
## m6 9 417.72 452.44 -199.86 399.72 6.5675 3 0.08704 .
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1##
## Pairwise comparisons:## block_name = Positive, amp_target = First person:
## contrast estimate SE df z.ratio p.value
## pleasant - unpleasant -1.050 0.542 Inf -1.936 0.0528
##
## block_name = Negative, amp_target = First person:
## contrast estimate SE df z.ratio p.value
## pleasant - unpleasant -0.200 0.519 Inf -0.386 0.6996
##
## block_name = Positive, amp_target = Third party:
## contrast estimate SE df z.ratio p.value
## pleasant - unpleasant -0.805 0.516 Inf -1.559 0.1191
##
## block_name = Negative, amp_target = Third party:
## contrast estimate SE df z.ratio p.value
## pleasant - unpleasant 1.674 0.557 Inf 3.007 0.0026
##
## Results are given on the log odds ratio (not the response) scale.##
## === CORRELATIONS ===##
## Correlation between first-person and third-party awareness:##
## Pearson's product-moment correlation
##
## data: participant_scores$shareAwareSelf and participant_scores$shareAwareThird
## t = 3.6876, df = 3, p-value = 0.03457
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
## 0.1135008 0.9937898
## sample estimates:
## cor
## 0.9051294##
## Correlation between first-person and third-party AMP scores:##
## Pearson's product-moment correlation
##
## data: participant_scores$ampDiffSelf and participant_scores$ampDiffThird
## t = 0.29277, df = 3, p-value = 0.7888
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
## -0.8389648 0.9144660
## sample estimates:
## cor
## 0.1666667
1. Model
Comparisons - What Predicts Awareness? Response effect (pleasant vs
unpleasant choice): NOT significant (p = 0.96)
Whether participants chose “pleasant” or “unpleasant” doesn’t affect awareness reports Prime type effect (positive vs negative prime): NOT significant (p = 0.17)
The valence of the prime alone doesn’t affect awareness Response × Prime Type interaction: SIGNIFICANT (p = 0.001) ⭐
This is the key awareness effect! Awareness depends on the combination of prime and response People notice influence when their response matches the prime AMP target effect (first-person vs third-person): NOT significant (p = 0.56) ⚠️
This is surprising! Unlike the original study, you didn’t find higher awareness in third-person Original study found P(aware) was higher in third-party (0.566) vs first-person (0.477) Three-way interaction: Marginally significant (p = 0.087)
Suggests the awareness pattern might differ slightly between first and third-person, but weakly 2. Pairwise Comparisons - When Do People Report Awareness? Looking at pleasant vs unpleasant responses:
First person, Positive prime: p = 0.053 (marginal)
After positive primes, people slightly more aware when choosing unpleasant (goes against the prime) First person, Negative prime: p = 0.70 (not significant)
No difference Third party, Positive prime: p = 0.119 (not significant)
No difference Third party, Negative prime: p = 0.003 ⭐⭐
Strong effect! After negative primes, people are MORE aware when choosing pleasant (going against the prime) Estimate = 1.674 on log-odds scale (large effect) Pattern: People seem most aware when their response contradicts the prime (especially for negative primes in third-person)
- Correlations Correlation between first-person and third-person awareness: r = 0.91 (p = 0.035)
Very high correlation! People who report high awareness in first-person also report high awareness in third-person This is within-subject consistency
## Warning: Using `size` aesthetic for lines was deprecated in ggplot2 3.4.0.
## ℹ Please use `linewidth` instead.
## This warning is displayed once every 8 hours.
## Call `lifecycle::last_lifecycle_warnings()` to see where this warning was
## generated.## Warning: The `size` argument of `element_rect()` is deprecated as of ggplot2 3.4.0.
## ℹ Please use the `linewidth` argument instead.
## This warning is displayed once every 8 hours.
## Call `lifecycle::last_lifecycle_warnings()` to see where this warning was
## generated.## Warning: Removed 9 rows containing missing values or values outside the scale range
## (`geom_point()`).
## # A tibble: 39 × 7
## participant_id perspective_label prime_type response_type prop_aware n_trials
## <chr> <chr> <chr> <chr> <dbl> <int>
## 1 2s0b60be First-person AMP Negative … Pleasant 1 4
## 2 2s0b60be First-person AMP Negative … Unpleasant 0.154 13
## 3 2s0b60be First-person AMP Positive … Pleasant 0.333 15
## 4 2s0b60be First-person AMP Positive … Unpleasant 0 2
## 5 2s0b60be Third-party AMP Negative … Pleasant 0.444 9
## 6 2s0b60be Third-party AMP Negative … Unpleasant 0.556 9
## 7 2s0b60be Third-party AMP Positive … Pleasant 0.467 15
## 8 2s0b60be Third-party AMP Positive … Unpleasant 0.333 3
## 9 30f31afm First-person AMP Negative … Pleasant 1 14
## 10 30f31afm First-person AMP Negative … Unpleasant 1 3
## # ℹ 29 more rows
## # ℹ 1 more variable: condition <fct>## # A tibble: 8 × 6
## perspective_label prime_type response_type mean_prop_aware sd_prop_aware
## <chr> <chr> <chr> <dbl> <dbl>
## 1 First-person AMP Negative prime Pleasant 0.68 0.460
## 2 First-person AMP Negative prime Unpleasant 0.608 0.388
## 3 First-person AMP Positive prime Pleasant 0.622 0.407
## 4 First-person AMP Positive prime Unpleasant 0.625 0.415
## 5 Third-party AMP Negative prime Pleasant 0.644 0.310
## 6 Third-party AMP Negative prime Unpleasant 0.376 0.211
## 7 Third-party AMP Positive prime Pleasant 0.440 0.372
## 8 Third-party AMP Positive prime Unpleasant 0.572 0.380
## # ℹ 1 more variable: n_participants <int>