RESEARCH METHOD

Data collection
Samples
Measures
Analyses

Data collection

The data were collected using the Alcohol and Public Health Research Unit's CATI (computer-assisted telephone interviewing) system. Respondents were aged 14-65 years and were randomly selected from the Auckland telephone free-calling area. The report is based on surveys which were undertaken annually from 1990 through to 1999. Interviewing took place between mid-November and mid-December each year.

Further details on the research method and the CATI system are included in Appendix C.

Samples

The sample size and response rate averaged across the surveys was 1204 and 68%. Data for each year are presented in Appendix D.

The demographic composition of the samples generally showed a high degree of consistency from measure to measure. The data are included as Appendix B. There was some variation in the proportion of men and women. Also since 1994 the survey contained higher proportions of university educated and higher income people in keeping with changes in the population of Auckland. The statistical analysis controlling for these variations is discussed later in this section (under the heading ‘Analyses’) and in Appendix E.

Measures

Frequency and quantity of alcohol consumption:

It was first ascertained whether respondents had consumed any alcoholic beverage in the previous 12 months. Those who had were then asked to report on drinking in 15 standard locations plus any additional locations they used. For each place that had been one of their drinking locations in the previous 12 months, they were asked what they would drink on a typical occasion at this location. This information was then used to estimate the frequency of drinking, the quantity consumed on a typical occasion and the annual volume of alcohol consumed (see Appendix F for further details).

Location of Drinking

Data collected for different drinking locations allowed for trends in popularity and drinking patterns in different locations to be analysed.

Frequency of drinking larger quantities and drinking enough to feel drunk

Another question ascertained how often people drank relatively large quantities of alcohol. Large amounts were defined as six or more drinks for men and four or more for women. A ‘drink’ was defined as the equivalent of one can or stubbie bottle of beer, one can of ready to drink, one largish (140ml) glass of wine, or a double hotel nip of spirits, all of which approximately equate to 15 mls of absolute alcohol. The quantity consumed on a typical occasion is reported in the text as number of drinks.

A separate question asked people how often they drank enough to feel drunk.

Concern over own drinking

The drinkers were also asked whether, compared with a year ago, they were drinking more, less, or the same. They were also asked how happy they were with their current level of drinking compared with the level they felt was right for them.

Problems from own drinking

The drinkers were also asked how many times in the previous 12 months they had experienced each of 15 consequences from their drinking which, for convenience, will be referred to as ‘problems’. These ranged from more minor problems such as hangovers, influences on performance at work and arguments, to more serious consequences such as getting into fights and motor vehicle accidents. These data were used to produce two measures for the analyses. The ‘types of problems’ measure – how many of the 15 problems were mentioned. The ‘number of problems’ was the summed frequency of reported problems across all 15 items.

Problems from others’ drinking

All persons, including abstainers, were asked how much harmful effect other people's drinking was having on five areas of their life: home life, friendships and social life, financial position, health and work. Respondents were also asked about four specific consequences of others’ drinking: being involved in a motor vehicle accident that involved someone else’s drinking; being involved in some other type of accident causing injury or major damage that involved someone else’s drinking; being physically assaulted by someone who had been drinking; and sexually harassed by someone who had been drinking.

Concern about the drinking of others

All respondents were asked, whether in the last 12 months, they had been seriously concerned about the drinking of friends, relatives or acquaintances.

Possible reasons for increased and decreased consumption

Those who said they were drinking more than the year before were read a list of possible reasons for drinking more and asked to indicate the reasons that applied to them. There was a similar procedure for those drinking less.

Attitudes to alcohol

A measure was included to try and get some indication of a normalisation process in which alcohol is increasingly viewed as a normal everyday food and drink product and less as a drug with associated problems. Statements read ‘Beer/wine/spirits is/are a suitable drink for most times of the day’, and it was scored on a five point agree-disagree scale for each of the three beverages.

Changes in consumption of different beverages

For each of beer, wine, spirits, and low alcohol beer, drinkers were asked if they were drinking more, less or the same compared with a year ago.

Home production of alcohol

Respondents were asked about the home production of beer and spirits.

Demographic sub-groups:

As well as examining the data for all drinkers, analyses have also been undertaken to examine trends for different demographic and consumption sub-groups. The number in each sub-group varied a little because of the variations in total sample size for each survey. The average number in each survey and the average number of drinkers is shown in brackets after each category. This included four age within gender groups: males 14-29 years (217 in total sample and 188 drinkers each year), males 30-65 years (343; 304), females 14-29 years (224; 180) and females 30-65 years (414; 342). Youth drinkers (161; 122) were analysed as a separate group with 14-19 year old young men and women combined (sample sizes were considered insufficient to separate the men and women).

Analyses were also undertaken based on three equal sized income groups: higher (373; 283), medium (373; 320) and lower (373; 350). Income was personal income from all sources. The income thresholds and average incomes were computed by first aggregating the incomes data. This was necessary as respondents had only been asked to state a range in which their income falls (for example $35,000 to $39,999) rather than their exact income. The income cut-off points between the three income groups were computed by assuming respondents were equally distributed across intervals. By this method, the higher income group was found to be those with an income of more than $31,283 p.a. in 1990 (the threshold had risen to $40,518 p.a. by 1999). The middle income group was defined as those with an income above $13,642 p.a. ($18,846 by 1999) and up to $31,283 p.a., and those with an income of $13,642 p.a. or less were the lower income group.

Respondents were also grouped by employment status. They were divided into the employed and the non-employed. The employed were subdivided into two groups – higher and lower status occupations. The higher status occupations (382; 544) comprised the top three occupational categories from Statistics New Zealand’s Standard Classification of Occupations. These were: 1) legislators, administrators; 2) managers, professionals; and 3) technicians and associated professionals. The lower status occupations (460; 357) comprised the people in occupations coded 4 to 9: 4) clerks; 5) service and sales workers; 6) agriculture and fisheries workers; 7) trades workers; 8) plant and machine operators and assemblers; and 9) elementary occupations. The non-employed (360; 308) comprised the retired, the unemployed, parents at home, and others.

Analyses

The data were subjected to logarithmic transformation and trends were subjected to analysis of covariance or logistic regression analyses. For ease of interpretation, data in graphs are shown in the form of geometric means. In the text, averages (arithmetic means) are given for key variables. The arithmetic mean takes into account the influence of the people at the highest and lowest consumption level (see Appendix E for further explanation).

The analyses in this report have controlled for age and gender, to remove any possibility that variations in the age/gender composition of the samples in the different surveys would affect the comparison of results between surveys (age and gender are the two demographic variables having the greatest influence on drinking patterns). It was also found that there was variation in consumption data depending on the time the data were collected. Because the proportion of the sample interviewed each month varied slightly between surveys, the month of collection (November or December) has also been controlled for. Given the increase in the percentage of the population with university degrees and higher incomes, the effect of this has also been analysed.

Where trends are referred to in this report, a statistically significant effect at <0.05 has been found.

Top | Back | Next | Home