• No se han encontrado resultados

Participants for the current study were nine of the same group who took part in Experiment 4 (5M, 4F; mean age = 21.1, sd = 2.3). Ten participants completed data collection for Experiment 4, but one of these was lost to the study before data collection was completed due to illness.

Like for Experiment 4, the participants for the current study also represented a subset of those who had participated in the larger Experiment 2. Unlike

Experiment 4 no data was drawn from the earlier studies for use in the current study.

Participants were paid AUD $50.00 for each night of their participation in the current study (total AUD $100.00).

Materials

Equipment for polysomnographic recording and the delivery of sounds was the same as for the earlier experiments. Alterations were made to the sound delivery system to allow for the presentation of a flickering light stimulus. This included adjustments that allowed the intensity of the light to be delivered in increasing steps according to the method of limits. The program was also altered to allow the simultaneous presentation of signals.

sound are not available. The central question of the current study concerned the naturalistic quality and value of this signal, rather than the range of frequencies it generates.

Flickering Light

The flickering light was a 20 watt halogen bulb which was mounted upon custom built frame. The frame consisted of a horizontal arm projecting at right angles from a metal stand with telescopic adjustment available to control the height. This allowed it to be mounted directly above the pillow as shown in Figure 6.1.

Figure 6.1. Configuration of flickering light over pillow.

The stand was supported by weights placed at the footings to ensure that it would not move or cause injury to participants by overbalancing on top of them. In keeping with the modified method of limits, the light started at a very low level and was increased in strength incrementally every 30 seconds. The intensity

Weights Telescopic adjustment

Light

of the light emitted at each step is shown in Table 6.1. The measurements were taken at a distance of 1m from the pillow against the background of a clean white sheet of A4 paper. All measurements were taken with a standard light meter. The corresponding sound levels are also included and indicate the levels when light or sound were administered either singly, or simultaneously. Note that a range of intensities is included due to the flickering nature of the light.

Table 6.1. Signal strength at each step of the modified method of limits

Level Time (seconds) Sound Intensity

(dBA) Light Intensity (lux) 1 0-30 35 0-1 2 30-60 40 2-3 3 60-90 45 5-6 4 90-120 50 13-15 5 120-150 55 21-26 6 150-180 60 29-36 7 180-210 65 42-53 8 210-240 70 57-72 9 240-270 75 74-93 10 270-300 80 91-114 11 300-330 85 110-138 12 330-360 90 136-170 13 360-570 95 153-202

From that point the light steadily increased until it reached peak intensity. The US NFPA 72 (2002) standard outlines requirements that must be met by a strobe light that is being used to alert the hearing impaired to the possibility of a fire, but these relate to the intensity at the light (110 candela if the strobe is mounted at more that 24 inches from the ceiling) and not to the received intensity at the pillow. For the current study a flickering, rather than a strobe light is being used and measurements are made at the pillow, therefore it is not known how this relates to the standard.

Unlike for the sound, it was not necessary to make a specific file for each

increment of light intensity. The different levels of light were manipulated by a control box varying voltage input that was connected between the light and the signal delivery system on the laptop. This control box was designed and made by staff of CESARE.

Procedure

Data for the current study was collected simultaneously with data collection for Experiment 4. This was explained to participants when they were recruited and information on both studies was sent out and consent was obtained at the same time. Payment for participation in each study was made separately, meaning that individuals actually received a total of AUD $100.00 for each of the two nights of participation comprised of AUD $50.00 for each study (grand total AUD $200.00 for both nights).

Data for the current study was always collected directly after the completion of data collection for Experiment 4 on both nights of the study. This meant that a total of five signals were administered to each participant on each night, and the two signals for Experiment 4 were always collected first and second. The three signals for the current study included the naturalistic house fire sound from

house fire sound together with the flickering light. These signals were presented in counterbalanced order (always as the third, fourth, or fifth signal presented) across two alcohol conditions (sober and approximately .08 BAC) over two nights.

Since data was collected in tandem between the current study and Experiment 4 details of procedures for polysomnographic recording, sound stimulus

calibration and presentation, and the administration of alcohol can be found there.

An important point of departure in the procedure for data collection that distinguishes the current study from the all preceding experiments was that

awakenings were carried out in stage 2 sleep rather than stage 4. This was done

because it was uncertain whether five awakenings would be possible in stage 4 on a single night. It was most important when alcohol was administered because it is known that alcohol continues to be absorbed rapidly for a period of 30 to 45 minutes after the last drink is taken before BAC reaches its peak level, and then begins to decline in a linear fashion at the rate of about .015 per hour (Sadler, 1999). It was possible that attaining five periods of stage 4 sleep may have taken a considerable amount of time for some participants which would have the

consequence that data was being compared across different alcohol conditions within the one night. For example signal one may have been delivered within the first half hour of the sleep period, but signal five may not have been delivered until four or five hours later. In this circumstance the BAC when the first signal was delivered would be expected to fall between .08 and .09, but for the fifth

time taken to collect five stage 4 awakenings. Both of these difficulties were minimised in previous studies by collecting less data points and by appropriate counterbalancing. Since it was necessary to collect data for five signals,

scheduling awakenings for the current study in stage 2 allowed this to be

completed in a considerably shorter period of time which was hoped to minimise methodological concerns.

It must be noted that although it was known that BAC was at .08 when

participants went to bed, the BAC would have varied between individuals before data collection was commenced for the current study depending upon how long it took to collect the two data points for Experiment 4 beforehand. The first two stage 4 awakenings usually happened quite quickly, so it was estimated that BAC would have remained at a moderate level, and was likely to fall within the range of around .08 and .04. This was considered within tolerance limits because no difference had been found in behavioural response times between the .08 and .05 BAC for Experiment 2.

The additional feature of this study was the inclusion of a light stimulus. The intensity of the flickering light was calibrated at the pillow in each room

measured against the background of a clean white sheet of A4 paper. The use of the white paper allowed a standardised background across bedrooms in order to minimise differences in light intensity caused by reflecting off varying colours and patterns between pillow cases. The equipment was set up according to the configuration shown in Figure 6.1, and the paper and light meter were placed in the centre of the pillow, directly under the light. The light was commenced at increment six (29 – 36 lux), which corresponded to the calibration level of the sound signals (60dBA). The light meter was then checked and if the reading was too high (greater than 36 lux) the telescopic adjustment was used to raise the height of the light, thereby decreasing its intensity. If the reading was too low

(less than 29 lux), then the height of the light was lowered in order to increase its intensity. These adjustments continued until the desired intensity was reached. It was endeavoured to calibrate the light at as close to the higher level of 36 lux as possible.

Data Analysis

A 3x2 repeated measures ANOVA was used to calculate differences in

awakening times. The within subjects factors were ‘stimulus’ with three levels corresponding to the naturalistic house fire sound, the flickering light, and the combination of the two, and ‘alcohol’ with two levels, sober and .08 BAC. The between subjects factor of sex was not included because complete data was collected for only four female participants which makes the comparison

questionable. All statistics were calculated using SPSS version 14. The output can be viewed as Appendix 8. Light was entered into the ANOVA as the last variable so that simple contrasts could be made between this variable and the other two. This was recalculated with simultaneous presentation of the two sounds entered last. The small overall sample size of the current study should also be noted when interpreting results.

RESULTS

It was not appropriate to compare AATs to international standards for the current study as had been the procedure for the previous studies because it was not known how the light stimulus related to standard requirements. Instead the number of participants who slept through all levels of the signals are presented. For the naturalistic house fire sound all participants responded when sober (at or below 75dBA), as well as when alcohol was given (one at a level above 80dBA). For the flickering light four (44.4%) participants slept through all levels when sober, and eight (88.9%) slept through when alcohol was administered. When the signals were combined once again all participants responded when sober (at or below 75dBA for the sound), and none slept through when under the influence of alcohol (one at a level above 80dBA for the sound). What is most apparent from this data is that the flickering light on its own performed poorly in all conditions. It is also apparent that the naturalistic house fire sound performed considerably better than expected if the results of the previous experiments are taken into account.

Means and standard deviations for behavioural response times (measured in seconds) were calculated and are shown in Table 6.2. As for the previous studies behavioural response time was recorded as 600 seconds if the person failed to respond.

Table 6.2. Descriptive statistics for behavioural response time in seconds according to stimulus type and alcohol level (N = 9).

ALCOHOL