The primary hypothesis is to test when the Verbal Overshadowing effect disappears. To test this, we analyze the data based on each of the 5-10-15- and 20- minute breakdowns.
We oversampled participants, such that our final sample consisted of N = 6,196).
On average, across all timings, there was a significant verbal overshadowing effect (*F*(1, 5664) = 12.13, *p* = .0005).
anova correct verbalize if timing != 0
Number of obs = 5666 R-squared = 0.0021
Root MSE = .445567 Adj R-squared = 0.0020
Source | Partial SS df MS F Prob > F
-----------+----------------------------------------------------
Model | 2.40889266 1 2.40889266 12.13 0.0005
|
verbalize | 2.40889266 1 2.40889266 12.13 0.0005
|
Residual | 1124.47568 5664 .198530311
-----------+----------------------------------------------------
Total | 1126.88457 5665 .19892049
When it comes to the timing, manipulation, we found the predicted results in the sense that there was no verbal overshadowing effect at the earliest timings (5min: *b* (1234) = .003, *p* = .899) and we successfully replicated the verbal overshadowing effect at the 20 minute timing (*b*(1067) = -.058, *p* = .032, 95%CI = -.111 to -.005).
When it comes to the specific timing, there was no evidence for a Verbal Overshadowing effect at 10 minutes (*b*(1140) = -.006, *p* = .833). A significant VO effect emerged at 15 minutes (*b*(1134) = -.075, *p* = .004, 95%CI = -.126 to -.024). This suggests that the interference effect that occurs due to verbalizing does not operate at 0 minutes (Alogna et al., 2014)
Furthermore, participants were randomly assigned to either wait 20 minutes before verbalizing and perform the verbalzing task for 5 minutes or for one minute. The magniutde of the VO effect was similar if not nomminally larger for those who verbalized for 1 minute instead of 5 minutes (*b*(1081) = -.079, *p* = .003, 95%CI = -.131 to -.027). This represents a nice methodological move forward for conducting VO studies online, that people can be assigned to verbalize for 1 minute instead of 5 minutes.
by timing: regress correct i.verbalize if timing > 0, vce(robust)
-------------------------------------------------------------------------------------------------------------------------------------------------------------
-> timing = 0
no observations
-------------------------------------------------------------------------------------------------------------------------------------------------------------
-> timing = 1
Linear regression Number of obs = 1083
F( 1, 1081) = 8.73
Prob > F = 0.0032
R-squared = 0.0079
Root MSE = .44145
------------------------------------------------------------------------------
| Robust
correct | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
1.verbalize | -.0789448 .026723 -2.95 0.003 -.1313797 -.02651
_cons | .304878 .0192326 15.85 0.000 .2671405 .3426156
------------------------------------------------------------------------------
-------------------------------------------------------------------------------------------------------------------------------------------------------------
-> timing = 5
Linear regression Number of obs = 1236
F( 1, 1234) = 0.02
Prob > F = 0.8989
R-squared = 0.0000
Root MSE = .45434
------------------------------------------------------------------------------
| Robust
correct | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
1.verbalize | .0032846 .0258495 0.13 0.899 -.0474292 .0539984
_cons | .2887789 .0184247 15.67 0.000 .2526317 .3249261
------------------------------------------------------------------------------
-------------------------------------------------------------------------------------------------------------------------------------------------------------
-> timing = 10
Linear regression Number of obs = 1142
F( 1, 1140) = 0.04
Prob > F = 0.8328
R-squared = 0.0000
Root MSE = .44906
------------------------------------------------------------------------------
| Robust
correct | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
1.verbalize | -.0056212 .0266192 -0.21 0.833 -.0578493 .0466069
_cons | .2822878 .0193509 14.59 0.000 .2443204 .3202553
------------------------------------------------------------------------------
-------------------------------------------------------------------------------------------------------------------------------------------------------------
-> timing = 15
Linear regression Number of obs = 1136
F( 1, 1134) = 8.27
Prob > F = 0.0041
R-squared = 0.0073
Root MSE = .43963
------------------------------------------------------------------------------
| Robust
correct | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
1.verbalize | -.0750777 .0261093 -2.88 0.004 -.1263058 -.0238497
_cons | .3019538 .019366 15.59 0.000 .2639565 .3399511
------------------------------------------------------------------------------
-------------------------------------------------------------------------------------------------------------------------------------------------------------
-> timing = 20
Linear regression Number of obs = 1069
F( 1, 1067) = 4.60
Prob > F = 0.0321
R-squared = 0.0043
Root MSE = .44115
------------------------------------------------------------------------------
| Robust
correct | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
1.verbalize | -.0581536 .0270997 -2.15 0.032 -.1113282 -.0049789
_cons | .2960784 .0202342 14.63 0.000 .256375 .3357819
------------------------------------------------------------------------------
Next, we explore the effect of verbalizing on saying the perpretator is 'not present'.
Here we find that the VO effect on saying the perp is not present was not present after 5 minutes of a distractor task (*b*(1234) = .028, *p* = .247) but was present at 10 minutes and beyond (10min: *b*(1140) = .06, *p* = .016, 95%CI = .109 to .011; 15min: *b*(1134) = .062, *p* = .017, 95%CI = .113 to .011; 20min: *b*(1067) = .155, *p* < .001, 95%CI = .21 to .1). Furthermore, those in the 20min condition who only verbalizied for 1 minute instead of 5 were also more likely to say that the perp was not present (*b*(1081) = .093, *p* = .001, 95%CI = .147 to .039).
by timing: regress notpresent i.verbalize if timing > 0, vce(robust)
-------------------------------------------------------------------------------------------------------------------------------------------------------------
-> timing = 0
no observations
-------------------------------------------------------------------------------------------------------------------------------------------------------------
-> timing = 1
Linear regression Number of obs = 1083
F( 1, 1081) = 11.54
Prob > F = 0.0007
R-squared = 0.0107
Root MSE = .44644
------------------------------------------------------------------------------
| Robust
notpresent | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
1.verbalize | .0929027 .027347 3.40 0.001 .0392435 .1465618
_cons | .2351916 .0177187 13.27 0.000 .2004246 .2699586
------------------------------------------------------------------------------
-------------------------------------------------------------------------------------------------------------------------------------------------------------
-> timing = 5
Linear regression Number of obs = 1236
F( 1, 1234) = 1.34
Prob > F = 0.2468
R-squared = 0.0011
Root MSE = .42539
------------------------------------------------------------------------------
| Robust
notpresent | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
1.verbalize | .0280214 .024185 1.16 0.247 -.0194268 .0754695
_cons | .2227723 .0169169 13.17 0.000 .1895833 .2559613
------------------------------------------------------------------------------
-------------------------------------------------------------------------------------------------------------------------------------------------------------
-> timing = 10
Linear regression Number of obs = 1142
F( 1, 1140) = 5.85
Prob > F = 0.0157
R-squared = 0.0051
Root MSE = .42309
------------------------------------------------------------------------------
| Robust
notpresent | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
1.verbalize | .0603813 .0249572 2.42 0.016 .0114142 .1093485
_cons | .202952 .017291 11.74 0.000 .1690263 .2368778
------------------------------------------------------------------------------
-------------------------------------------------------------------------------------------------------------------------------------------------------------
-> timing = 15
Linear regression Number of obs = 1136
F( 1, 1134) = 5.75
Prob > F = 0.0166
R-squared = 0.0050
Root MSE = .43822
------------------------------------------------------------------------------
| Robust
notpresent | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
1.verbalize | .0623189 .0259868 2.40 0.017 .0113313 .1133064
_cons | .2291297 .017728 12.92 0.000 .1943463 .263913
------------------------------------------------------------------------------
-------------------------------------------------------------------------------------------------------------------------------------------------------------
-> timing = 20
Linear regression Number of obs = 1069
F( 1, 1067) = 31.04
Prob > F = 0.0000
R-squared = 0.0279
Root MSE = .4574
------------------------------------------------------------------------------
| Robust
notpresent | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
1.verbalize | .1550317 .0278262 5.57 0.000 .1004315 .209632
_cons | .2313725 .0186911 12.38 0.000 .194697 .2680481
------------------------------------------------------------------------------
In an exploratory test to understand the nature of the effects, we ran an analysis testing what is the probability of being correct conditional on choosig anyone, by whether someone verbalized. Meaning, we eliminated all people who chose 'not present' form the analysis and re-ran the VO analysis. The only significant effect to occur was when participants were delayed 15 minutes (*b*(838) = -.072, *p* = .03, 95%CI = -.136 to -.007). All other timings were statistically nonsignificant (all *p*s > .07).
Finally, on average 99% of our participants indicated that they took the study seriously.
by timing: regress correct i.verbalize if timing > 0 & notpresent == 0, vce(robust)
-------------------------------------------------------------------------------------------------------------------------------------------------------------
-> timing = 0
no observations
-------------------------------------------------------------------------------------------------------------------------------------------------------------
-> timing = 1
Linear regression Number of obs = 781
F( 1, 779) = 3.24
Prob > F = 0.0724
R-squared = 0.0041
Root MSE = .48278
------------------------------------------------------------------------------
| Robust
correct | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
1.verbalize | -.0623759 .0346662 -1.80 0.072 -.1304262 .0056743
_cons | .3986333 .0233981 17.04 0.000 .3527024 .4445641
------------------------------------------------------------------------------
-------------------------------------------------------------------------------------------------------------------------------------------------------------
-> timing = 5
Linear regression Number of obs = 943
F( 1, 941) = 0.33
Prob > F = 0.5637
R-squared = 0.0004
Root MSE = .48599
------------------------------------------------------------------------------
| Robust
correct | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
1.verbalize | .0182806 .0316516 0.58 0.564 -.0438353 .0803966
_cons | .3715499 .0222892 16.67 0.000 .3278076 .4152922
------------------------------------------------------------------------------
-------------------------------------------------------------------------------------------------------------------------------------------------------------
-> timing = 10
Linear regression Number of obs = 874
F( 1, 872) = 0.43
Prob > F = 0.5117
R-squared = 0.0005
Root MSE = .48186
------------------------------------------------------------------------------
| Robust
correct | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
1.verbalize | .0213989 .0325958 0.66 0.512 -.0425764 .0853743
_cons | .3541667 .0230367 15.37 0.000 .3089529 .3993805
------------------------------------------------------------------------------
-------------------------------------------------------------------------------------------------------------------------------------------------------------
-> timing = 15
Linear regression Number of obs = 840
F( 1, 838) = 4.70
Prob > F = 0.0304
R-squared = 0.0056
Root MSE = .47839
------------------------------------------------------------------------------
| Robust
correct | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
1.verbalize | -.071508 .0329809 -2.17 0.030 -.1362429 -.0067732
_cons | .3917051 .023459 16.70 0.000 .3456598 .4377504
------------------------------------------------------------------------------
-------------------------------------------------------------------------------------------------------------------------------------------------------------
-> timing = 20
Linear regression Number of obs = 735
F( 1, 733) = 0.01
Prob > F = 0.9436
R-squared = 0.0000
Root MSE = .48758
------------------------------------------------------------------------------
| Robust
correct | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
1.verbalize | .002551 .0360528 0.07 0.944 -.0682281 .0733301
_cons | .3852041 .0246127 15.65 0.000 .3368843 .4335239
------------------------------------------------------------------------------
**References**
Alogna, V. K., Attaya, M. K., Aucoin, P., Bahník, Š., Birch, S., Bornstein, B., ... & Carlson, C. (2014). Registered replication report: Schooler & Engstler-Schooler (1990). Perspectives on Psychological Science, 9(5), 556-578.