CAPÍTULO II: FAMILIA MULTIPROBLEMÁTICA CONCEPTO, TERMINOLOGÍA E
FAMILIA MULTIPROBLEMÁTICA CONCEPTO, TERMINOLOGÍA E INTERVENCIÓN DESDE EL TRABAJO SOCIAL
2.6 Cuestiones relacionales y sus implicaciones para la intervención psicosocial
Statistical analyses were conducted on reaction times (i.e., the time duration between the presentation of a target to the key-press response from subjects.) obtained from the lexical decision task. Across all forty participants, the overall mean error rate was 11.74% (SD = 5.35) (902 trials/7680 trials). Only the critical trials were included for the reaction time analyses, and incorrect responses (3.68%) (53 trials/1440 trials) and responses over two standard deviations (5%) (72 trials/1440 trials) were excluded. The overall mean of self-reported TSM percent is 39.25%, and the mean value for old subjects is 48.06% and that for young subjects is 32.05%.
Figure 4.2: Mean reaction time of all subjects. Result of the citation target case is on the left, and that of the sandhi target case on the right.
Figure 4.2 shows mean reaction time of all the subjects for each prime type in two different target cases. The error bars represent one standard error above and below the mean. In the citation target case, u51 primes had the longest mean reaction time suggesting a potential inhibitory priming effect, whereas s55 primes yielded the short- est mean reaction time indicating a possible facilitatory priming effect. The mean reaction time with control primes as a reference lies in between the two. Reaction times with s55 primes and control primes are significantly different from each other (β=-128.10, SE=55.01, t=-2.329, p=0.0202). In the sandhi target case, u51 primes induced the shortest mean reaction time, followed by s55 primes, while control primes had the longest mean reaction time. However, the differences among the three are not significant. Reaction times with s55 primes and u51 primes are only marginally significantly different from each other (β=163.67, SE=87.66, t=1.867, p=0.0623).
Figure 4.3: Mean reaction time of different generation groups. In each panel, result of the citation target case is on the left, and that of the sandhi target case on the right.
The result of the old age group is shown on the left while that of the young subject group is on the right. In the case of citation targets, i.e. on the left of each panel, the results of old subjects and young subjects pattern together. However, for the old age group reaction times with the three types of primes do not differ significantly; for young subjects, s55 primes (the green bar) elicited significantly shorter reaction times than control primes, the red bar (β=-165.06, SE=75.71, t=-2.18, p=0.0299), suggesting a facilitatory priming effect. In the case of sandhi targets, for old subjects, u51 primes (the blue bar) yielded the shortest reaction times, which were significantly shorter than those with s55 primes (β=334.67, SE=147.86,t=2.263, p=0.024287) and with control primes (β=311.61, SE=147.86, t=2.107, p=0.035863) respectively. The difference between reaction times with s55 primes and control primes is not significant. For young subjects, there are only marginally significant differences between reaction times with control primes and u51 primes (β=-178.4, SE=102.8,t=-1.734, p=0.0837)
and between reaction times with control primes and s55 primes (β=-199.1, SE=102.8, t=-1.936, p=0.0536). Note that for old subjects, the difference between reaction times with u51 primes in two target cases is also significant (β=353.57, SE=169.78,t=2.083, p=0.0377).
A series of linear mixed-effects models were applied to participants’ log-transformed reaction times using the lme4 package in R. The dependent variable was the log- transformed reaction time (logRT), and there were prime type, target type, and gen- eration as categorical independent variables, and TSM percent and rating as continu- ous independent variables. Participant, item, and gender were considered as random variables. In total sixteen models were created and compared using the Likelihood ratio tests. The full model and the resulting final model are formulated in (1). The results showed that the interaction among prime type, TSM percent, and generation was significant (χ2=6.2308, df=2, p=0.04436 *), and so were the two random vari- ables, participant (χ2=132.59, df=1, p=< 2.2e-16 ***) and item (χ2=23.855, df=1, p=1.039e-06 ***).
(1) a. Full: logRT∼targetType*primeType*TSM percent*generation*rating + (1|participant) + (1|item) + (1|gender)
b. Final: logRT∼primeType*TSM percent*generation + (1|participant) + (1|item)
Table 4.2: Results of a linear mixed models analysis for old subjects. Estimate Std. Error t-value (Intercept) 5.57683 0.28894 19.301 primeTypes55 -0.45561 0.23253 -1.959 primeTypeu51 0.01194 0.23652 0.050 TSM percent 0.17650 0.54384 0.325 primeTypes55:TSM percent 0.86649 0.44386 1.952 primeTypeu51:TSM percent -0.35215 0.45195 -0.779
of linear mixed-effects models and Likelihood ratio tests were applied. The final model for old subjects was logRT ∼ primeType*TSM percent + (1|participant) + (1|item), while the final model for young subjects was simply the intercept with two random variables: logRT ∼ 1 + (1|participant) + (1|item). Table 4.2 shows the results of the final model for old subjects. Generally speaking, as compared with control primes, reaction times with s55 primes were shorter (i.e., a facilitatory effect) while reaction times with u51 primes were longer (i.e., an inhibitory effect), reflecting what is observed in the citation target case in Figure 4.3. Meanwhile, reaction times increased when the TSM usage percentage increased, and the effect of prime types was modulated through the TSM usage percentage. With s55 primes, when TSM percent increased a unit, reaction times increased by 0.87 log unit, but with u51 primes, the effect was reversed; that is, when TSM percent increased a unit, reaction times decreased by 0.35 log unit. All these effects were only seen in the data of old subjects and not in the data of young subjects.