Effect of Cognitive Load on Delayed Recall and Recognition Memory

Observed values of the Delayed Recall and Delayed Recognition memory were considered as the dependent variables of MANCOVA, while Cognitive Load was considered as the fixed factor. As before, both Need for Cognition and Working Memory Capacity were entered as covariates of the model.

Table 5-24 – Results of the Multivariate Analysis of Cognitive Load Effects on Delayed Memory

Variable Wilks’

𝚲

F Sig. 𝜼_𝒑^𝟐

Cognitive Load .95 3.60 .01 .03

Need for Cognition .99 0.33 .72 .00

Working Memory Capacity .97 4.41 .01 .03

154

Table 5-25 - Results of Between Subject Effects on Delayed Recall and Recognition Memory

Delayed Recall Memory Delayed Recognition

Memory df F Sig. 𝜼_𝒑^𝟐 df F Sig. 𝜼_𝒑^𝟐 Factor

Cognitive Load 2 5.49 .00 .04 2 2.94 .05 .02

Covariates

Need for Cognition 1 0.48 .49 .00 1 0.07 .78 .00 Working Memory Capacity 1 5.91 .01 .02 1 4.55 .03 .02

According to the results of the multivariate analysis presented in Table 5-24, the existence of a significant mean difference between Cognitive Load groups (Λ = .95, 𝐹(2, 279) = 3.60, 𝑝 < .05) was revealed, considering both delayed recall and recognition memory together. Similar to the results of the previous analysis, only Working Memory Capacity appeared to have a significant impact as a covariate (Λ = .97, 𝐹(1,279) = 4.41, 𝑝 < .05).

The between-subject analysis (see Table 5-25) revealed an existence of a significant effect of Cognitive Load on Delayed Recall Memory (𝐹(2, 279) = 5.49, 𝑝 < .01).

However, the effect size appeared to be small (𝜂_𝑝² = .04). Surprisingly, the mean differences of Delayed Recognition memory between the groups of Cognitive Load were only marginally significant (𝐹(2, 279) = 2.94, 𝑝 < .10) and the effect size also appeared to be small (𝜂_𝑝² = .02).

As in the case of Hypothesis 8, Hypothesis 9 also predicted a non-linear relationship between Cognitive Load and the delayed memory constructs. Therefore, the means of the dependent variables were plotted against the respective levels of the independent variable as shown in Figure 5-7 and Figure 5-8. Accordingly, the relationship appeared to be negatively linear. This was further confirmed by the contrast analysis on Recall Memory (𝑝_{𝐿𝑖𝑛𝑒𝑎𝑟} < .01; 𝑝_{𝑄𝑢𝑎𝑑𝑟𝑎𝑡𝑖𝑐}> .05) and Recognition Memory(𝑝_{𝐿𝑖𝑛𝑒𝑎𝑟} <

.05; 𝑝_{𝑄𝑢𝑎𝑑𝑟𝑎𝑡𝑖𝑐}> .05).

155

Table 5-26 – Pair-wise Comparison Matrix of Cognitive Load Groups on Delayed Memory

Cognitive Load Groups Moderate High 𝚫𝑴 Sig. 𝚫𝑴 Sig.

Delayed Recall

Low -0.22 .36 -0.48 .00

Moderate -0.26 .20

Delayed Recognition

Low -0.17 .55 -0.35 .04

Moderate -0.18 .55

The mean differences between the Cognitive Load groups were consistent with those of the Immediate Recall memory. In other words, a significant mean difference in Delayed Recall Memory was found only between low and high levels of Cognitive Load (Δ𝑀_{𝐿𝑜𝑤→𝐻𝑖𝑔ℎ} = −0.48, 𝑝 < .01), but the mean differences between the moderate level and respective low and high levels of Cognitive Load were not significant (Δ𝑀_{𝐿𝑜𝑤→𝑀𝑜𝑑} = −0.22, 𝑛𝑠 ; Δ𝑀_{𝑀𝑜𝑑→𝐻𝑖𝑔ℎ} = −0.26, 𝑛𝑠). A similar effect was found with Delayed Recognition Memory as well. Despite finding a marginally significant value in between-subject analysis (see Table 5-25), the pair-wise comparison indicated a significant mean difference for Delayed Recognition Memory between low and high levels of Cognitive Load (Δ𝑀_{𝐿𝑜𝑤→𝐻𝑖𝑔ℎ} = −0.35, 𝑝 < .05).

However, the remaining group differences were not significant (see Table 5-26).

The result of this analysis did not provide support for Hypotheses 9a and 9b. Instead of moderate Cognitive Load being associated with the highest mean value, the low and high levels of cognitive load conditions were associated with the highest and lowest delayed memory respectively.

156

Figure 5-7 - Mean Profile of Delayed Recall Memory for Cognitive Load

Figure 5-8 - Mean Profile of Delayed Recognition Memory for Cognitive Load

157 5.6.6.2 Effect of Cognitive Load on Repeated Measures

The final set of hypotheses tested in this study pertained to the repeated measure, or the difference between the immediate memory and the delayed memory. Hypotheses 9c and 9d predicted that the difference between the recall and recognition memory would be significant at low and high levels of Cognitive Load, but not at the moderate level.

These hypotheses were tested using repeated measures MANCOVA. Accordingly, a within-subject variable was created (Time) with two levels and two measurement variables, Recall and Recognition memory. Immediate Recall, Delayed Recall, Immediate Recognition and Delayed Recognition were assigned to each level of the within-subject variables (i.e., immediate and delayed). As in the other analyses, the Cognitive Load grouping variable was used as the fixed factor and Need for Cognition and Working Memory Capacity were considered as the covariates.

Table 5-27 – Results of the Multivariate Analysis of Cognitive Load Effects on Repeated Memory Measures

Within Subject Effect Wilks’

𝚲

F Sig. 𝜼_𝒑^𝟐 Within-Factor

Time .98 1.98 .14 .01

Between-Subject Factor - Interaction

Time * Cognitive Load .96 3.78 .00 .03

Table 5-28 – Results of Within-Subjects Contrasts: Immediate vs. Delayed

df F Sig. 𝜼_𝑷^𝟐

Time

Recall 1 3.95 .04 .01

Recognition 1 .04 .85 .00

Time * Cognitive Load

Recall 2 6.61 .00 .05

Recognition 2 1.24 .29 .01

158

Figure 5-9 – Means Profile of Repeated Recall Memory for Cognitive Load

Figure 5-10 - Means Profile of Repeated Recognition Memory for Cognitive Load

159

According to the results presented in Table 5-27, time did not have any direct effect on the repeated measure. In other words, irrespective of Cognitive Load, memory did not seem to differ across time(Λ = .98, 𝐹(2, 278) = 1.98, 𝑝 = .14). However, the data indicated a significant interaction effect between Cognitive Load and time (Λ = .96, 𝐹(4, 558) = 3.74, 𝑝 < .01, 𝜂_𝑝² = .03). This suggested the existence of a significant mean difference between Cognitive Load groups in either [both] recall or [and]

recognition memory (see Table 5-28). Thus, a respective Univariate test examined this interaction. Accordingly, it was found that differences across Cognitive Load existed only for Recall(𝐹(2, 279) = 6.61, 𝑝 < .01, 𝜂_𝑝² = .05) but not for Recognition Load (see Table 5-29). The mean difference indicated a negative difference, the reasons for which will be further discussed in Chapter 6.

Table 5-29 – Pair-wise Comparisons for Within-subject Recall and Recognition Memory

a – Sidak adjustment for multiple comparisons was applied

The within-subject analysis indicated a significant difference between the immediate and delayed memory of low and high groups of Cognitive Load while there was no significant difference in moderate Cognitive Load. However, a similar pattern did not appear in Recognition memory. Hence, Hypothesis 9c was partially supported while Hypothesis 9d was not.

160 5.7 SUMMARY OF HYPOTHESIS TESTING

A summary of the hypotheses testing is given in Table 5-30.

Table 5-30 - Summary of Hypotheses Tests Results

Hypothesis Supported

H1 Incongruent music in an advertisement will generate incongruent cognitions, thereby increasing the felt Psychological Discomfort.



H2 A complex message results in a higher level of perceived Cognitive Load than does a simple message in an externally-paced media.



H3 Congruent music will have Gestalt effects on message processing and hence will attenuate the felt Cognitive Load.



H4 The higher the Psychological Discomfort, the greater the demands for cognitive resources, which will thereby increase the felt Cognitive Load.



H5 Increased levels of dissonance experienced as psychological discomfort lead to negative valance of attitude towards advertisement.



H6 When the perceived cognitive load increases, it will lead to negative valance of attitude towards the advertisement.



H7 The effect on Cognitive Load is such that;

a: SC has a lower effect on Cognitive Load than that of SI.  b: SI and CC have similar effects on Cognitive Load.  c: CI has a higher effect on Cognitive Load than that of CC.  H8 The relationship between Cognitive Load and …

a: Immediate Recall memory will mimic an inverted “U” shape.  b: Immediate Recognition memory will mimic an inverted “U”

shape.



H9 The relationship between Cognitive Load and …

a: Delayed Recall Memory will mimic an inverted “U” shape.  b: Delayed Recognition Memory will mimic an inverted “U”

shape.

 c: The difference between immediate and delayed recall memory

will be significant at low and high levels of cognitive load, but not at the moderate level.

P

d: The difference between immediate and delayed recognition memory will be significant at low and high levels of cognitive load, but not at the moderate level.



161 5.8 ANALYSES OF RECALL COMPONENTS

As mentioned before, Immediate and Delayed Recall Memory consisted of Category, Brand, and Message. Nonetheless, such a measure did not explain the individual behaviour of the respective recall components. Though the behaviour of these individual components was not hypothesised in the conceptual model, the next section presents an analysis of these separate components.