Adolescent computer use: Approach, avoidance, and parental control

Abstract

The nature and extent of computer use in an Australian regional sample of adolescents was examined in relation to individual characteristics and parental control. High engagement with computers, problematic use of computers, and the use of computers for different purposes were related to general behavioural tendencies of approach and avoidance represented by the behavioural inhibition system (BIS)/behavioural activation system (BAS) measures and to measures of parental control provided by the Parental Bonding Instrument. Results indicated that age, gender, and the BIS/BAS measures were useful constructs in explaining variation in computer use generally, and in explaining the diversity of reasons for computer use. Parental control was only weakly related to outcome behaviours. It was concluded that problematic computer use and hours spent at the computer show some similarities with substance use except for the strength of the relationships and the role of the BIS.

The rapid absorption of new communication technologies into our everyday lives indicates the benefits and rewards that ensue for many people from the use of these technologies. But that same rapid absorption arouses the fear that some people will be drawn to these rewards to the extent that their social relationships, or health, or education, or work-related achievements will suffer (Widyanto & Griffiths, 2006; Young, 1999). The extremely rapid uptake of videos, computer games, internet, and mobile phones by adolescents suggests that if detrimental effects from the use of these technologies can occur, emerging signs should be evident within this group. The present study investigated computer use in adolescents with a focus on differentiating problematic use from high engagement (Charlton, 2002).

Recently, internet addiction has been argued by some to be a new clinical phenomenon (Beard, 2005; Kraut et al., 1998; Young, 2004) with identifiable problematic effects on health, marriage, education, and work, while others remain sceptical (e.g., Blaszczynski, 2006; Shaffer, Hall, & Vander Bilt, 2000; Southwell & Doyle, 2002; Widyanto & Griffiths, 2006). Widyanto and Griffiths (2006) point out that even if internet addiction exists, it is not the internet itself that is the object of the addiction but rather the stimuli that are available via the internet. Hence, even if problematic computer use can be identified, it should be viewed as an indicator of more difficult-to-measure emotional attachment to stimuli that are made available by using a computer. Southwell and Doyle (2002) emphasise that an understanding of internet use or computer use more generally must recognise that both positive and negative effects can be identified and that explanations will depend on characteristics of the individual, the type of computer use, and “societal time or space” (p. 391). Individual characteristics and circumstances of the computer user are likely to differ depending on the type of stimulation being accessed.

The presence of detrimental or problematic effects from repeated engagement in a behaviour has been described as indicative of the person being “addicted” to the behaviour (Charlton, 2002; Charlton & Danforth, 2007; Orford, 2001; Widyanto & Griffiths, 2006). In particular, addictive use of a technology can be distinguished from high engagement with a technology by the presence of signs of conflict, withdrawal, relapse, and behavioural salience. These criteria are also known as core criteria. Peripheral, or high engagement, criteria, include tolerance, euphoria, and cognitive salience, and are common to both addictive use of a technology and high engagement with a technology (Charlton, 2002; Charlton & Danforth, 2007).

Charlton (2002) used items tapping the Brown (1997) criteria for behavioural addiction such as cognitive and behavioural salience, euphoria, tolerance, withdrawal, financial problems, and conflict, with items from the Charlton and Birkett (1995) assessment of computer engagement in an exploratory factor analysis to distinguish between these two concepts. From a sample of 193 men and 198 women with an average age of 26 years, a three-factor solution identified as Engagement (28% of variance), Addiction (11%), and Comfort (4%) was described. Engagement items referred to being good at computing, liking the challenge of computing, spending increasing amounts of time computing, and positive thoughts about computing. The Addiction items referred to problematic effects from computing such as interfering with other engagements, or social activities, or work, or failing to get enough sleep, and arguments over how much time was being spent with computing. The Comfort factor was related to how comfortable or relaxed a person was with using computers. A possible developmental model was proposed in which high engagement with computing activities precedes an addiction stage. Milder aspects of addiction such as tolerance, euphoria, and cognitive salience may be present during high engagement but progress to withdrawal symptoms, relapse, conflict, and behavioural salience in the addiction stage.

Carver and White (1994) and Carver, Sutton, and Scheier (2000) proposed a model of behaviour and affect based on approach and avoidance that has its origins in the work of Gray (1972, 1981, 1987). The behavioural inhibition system (BIS) responds to signals of punishment and other aversive states with heightened anxiety, which tends to inhibit ongoing behaviour. The behavioural activation system (BAS) responds to signals of reward and other desirable states by strengthening or directing ongoing behaviour and may be experienced as impulsivity. The two systems are believed to have different neural substrates and have distinct influences on behaviour. Carver et al. propose that these two motivational systems constitute “basic building blocks that underlie the complexity of human behaviour” (p. 741). With this in mind, the nature and extent of individual adolescent computer use could well be linked to the predominance of one or the other of these two behavioural tendencies.

Behavioural activation and behavioural inhibition have been related to depression and anxiety (Diego, Field, & Hernandez-Reif, 2001; Johnson, Turner, & Iwata, 2003; McFarland, Shankman, Tenke, Bruder, & Klein, 2006), psychopathy (Muris, Meesters, de Kanter, & Timmerman, 2005; Newman, MacCoon, Vaughan, & Sadeh, 2005), attention-deficit – hyperactivity disorder (Quay, 1997), internalising and externalising behaviour (Colder & O'Connor, 2004), eating disorders (Kane, Loxton, Staiger, & Dawe, 2004) and various forms of substance use and abuse (Franken, Muris, & Georgieva, 2006; Johnson et al., 2003; Knyazev, 2004). Greater involvement with computers, and particularly computer games, was expected to be predicted by the BAS measure because the positive, engaging stimulation afforded in these situations is likely to elicit approach behaviour. But the contribution of behavioural inhibition to different types of stimulation being accessed through the computer remains unclear.

The relationship between approach and avoidance tendencies and computer use may be moderated by the child's family environment. After all, high engagement or even addiction to computer use would be difficult in a family situation where the parents closely monitor behaviour and intervene when such behaviour appears excessive. Conversely, a less controlling family situation might allow greater indulgence in desired end states. The Parental Bonding Instrument (PBI) was developed by Parker, Tupling, and Brown (1979) to assess the parent – child relationship in terms of two dimensions. A primary parental care dimension (caring vs. indifference/rejection) and a secondary overprotection dimension (control/overprotection vs. allowance of autonomy and independence) were identified in a mixed sample of 150 mothers and 148 fathers ranging in age from 17 to 40 years using retrospective self-report. Since then the PBI has been used extensively in survey-based research designs.

The present investigation used the overprotection subscale only. The argument was made that excessive time spent and engagement with computers is likely to attract the attention of parents who are likely to intervene to limit the accessibility of the behaviour. Conflict is then likely to follow, which might be reflected in significant positive associations between this measure and problematic computer use.

Research into the relationship between the overprotection dimension of the PBI and behavioural measures has produced several findings. Maternal overprotection was a significant factor in predicting bulimic eating attitudes in girls (Ahmad, Waller, & Verduyn, 1994), in predicting scores on agoraphobia and social phobia scales (Parker, 1979), in differentiating alcoholics, narcotic addicts, and controls (Bernadi, Jones, & Tennant, 1989), in predicting introversion, body image, and neuroticism (Cubis, Lewin, & Dawes, 1989), and in predicting gambling status (Grant & Won Kim, 2002). Findings for paternal overprotection have not been as strong (Enns, Cox, & Clara, 2002; Kendler, Sham, & Maclean, 1997; Parker, 1983). A limitation of much previous work in this area is the use of adults to retrospectively assess their parenting, which often occurred many years previously. Parental control may have a more influential effect in the immediate environment of adolescents.

Finally, research has demonstrated gender differences in computer game playing and in the selection of television shows, toys, and outdoor activities more generally (Cherney & London, 2006; Chou & Tsai, 2007; Hartmann & Klimmt, 2006). These findings indicate that boys are more motivated to play computer games especially if the games have more violent or competitive content and less socially meaningful content. For example, Chou and Tsai (2007) found that adolescent boys in Taiwan had significantly higher motivation to play computer games and reported greater entertainment value and enjoyment from playing than girls. In addition, boys reported more negative impacts on their studies and their relationships with their parents and teachers than girls.

In summary, the present study examined the nature and extent of adolescent computer use in relation to temperament characteristics represented by the BIS/BAS and measures of parental control represented by the PBI overprotection scale. The BIS/BAS measures and parental control have been linked to similar problematic behaviours but very few studies have examined both constructs in the one study. Specific hypotheses included the expectation that BAS would predict greater computer engagement, more problematic use, and greater use of computer games. Greater parental control would predict greater problematic computer use. Gender differences were also expected to arise, with one expectation being that boys would be more involved in the use of computer games than girls.

Method

Participants

One hundred and seventy-eight secondary school students ranging in age from 12 to 19 years (M = 14.7 years, SD = 1.60) were recruited from two schools on the eastern seaboard of New South Wales (NSW). The catchment area for the schools includes a large regional town populated predominantly by Caucasian Australians with minorities of Aboriginal and Asian – Australian students. There were 95 boys and 81 girls in the sample (with the gender response missing for two students).

Materials

The questionnaire package consisted of the BIS/BAS scales, the PBI overprotection scales for mother and father, the computer activity questions (based on Charlton, 2002), computer use expectations questions, type and extent of computer use questions, and several demographic questions relating to age, gender, use of cigarettes and alcohol. These latter two questions were included to compare predictors of these substances with predictors of computer use.

BIS/BAS scales

Carver and White (1994) developed 20 items to measure behavioural inhibition and behavioural activation. All items were rated on a 4-point scale, from 1 (strongly agree) to 4 (strongly disagree), and presented in a randomised format. The seven BIS items form a single scale and example items include “Criticism or scolding hurts me quite a bit” and “I worry about making mistakes”. Carver and White report internal reliability at α = .74 for US college students. The present study found α = .69 for Australian adolescents.

The 13 BAS items form three subscales: Reward Responsiveness (five items), Drive (four items), and Fun Seeking (four items). Carver and White (1994) report Cronbach αs = .73, .76, and .66 respectively, the present study found .69, .70, and .63. The BIS was also reported to correlate .28 with Reward, −.12 with Drive, and −.08 with Fun. The present study found correlations of .49, −.05, and −.04, respectively. Minor wording alterations were made to two items to adjust for culture and age appropriateness. Only the total BAS score is used in this report (α = .77, BIS – BAS correlation, r = .20).

Parental Bonding Instrument

The PBI (Parker et al., 1979) was developed as a measure of the relationship between a child and his or her parents. Only the overprotection subscale of the PBI was used as described earlier and the wording of items was adapted to reflect the present tense of the adolescent situation. For this subscale, Parker et al. reported a split-half correlation of .74 (present study, α = .83 for mothers and .87 for fathers).

Computer activity questions

The 14-item scale used to measure adolescent computer use was derived from the analysis provided by Charlton (2002). Charlton identified three factors of Engagement, Addiction, and Comfort in college student's reports of their computing activities. The seven items with the highest loadings on the Engagement factor and the seven items with the highest loadings on the Addiction factor were selected (Table I). The Engagement items represent questions about the predominance and salience of computing activities in a person's life without reference to negative effects. The Addiction items, in contrast, all refer to problems that have occurred from using computers. Charlton reported the correlation between the two factors as r = .38 after an oblique rotation (present study correlation between engagement and problematic use, r = .49). A 5-point response scale, strongly agree; agree; neither agree or disagree; disagree; and strongly disagree was used. Internal reliability for both the Engagement scale and the Problematic scale was α = .72.

Table I. Items from Charlton (2002) forming the measures of computer engagement and problematic computer use

Engagement	Problematic
1. I tend to want to spend more and more time using computers	1. I often don't get enough sleep because of my computing activities
2. I often feel a buzz of excitement while computing	2. Arguments have sometimes happened at home because of the time I spend on computing activities
3. I hardly ever think about computing when I am not using a computer^†	3. I have tried but can't cut back on the time I spend computing
4. It is important for me to be good at computing	4. I never miss meals because of my computing activities^†
5. The less I have to do with computers the better^†	5. When I am not using a computer I often feel agitated
6. I like the challenge that learning to use a computer presents	6. Computing activities have sometimes interfered with my school work
7. It would not matter to me if I never used a computer again^†	7. My social life has sometimes suffered because of my computer activities
Note: ^†Reverse scored item.

Computer use expectancies

As another measure of engagement with computers, eight items asked about expected emotional effects from the expectation of being able to use a computer later in the day. The items were grouped into four positive reinforcement expectation items (“Relax and grow calm”, “Enjoy the challenges offered”, “Become excited”, “Have totally new experiences”) and four negative reinforcement expectation items (“Avoid feeling depressed”, “Escape from boredom”, “Stop feeling angry or irritable”, “Avoid stress or anxiety”). Items were rated on a 5-point scale from very unlikely, fairly unlikely, neither unlikely nor likely, fairly likely, to very likely.

An estimate of the total amount of time spent using a computer over the last week was obtained by asking students to complete a table of the daily use that had occurred over the previous seven day. These figures were then summed. Further questions asked for the student's age, gender, age when computer use was initiated, whether the student thought their recent computer use was increasing or decreasing, and frequency of cigarette and alcohol use (0 = not at all; 1 = tried once or twice; 2 = monthly; 3 = weekly; 4 = daily).

Finally, students were asked to rate their use of computers for each of the following nine purposes: computer games (e.g., Xbox, GameCube, Nintendo, Sega, GameBoy, handheld games including mobile phones); writing; music; Internet games; Internet email or chat room; Internet news or other information; art; Internet gambling; or spreadsheets and calculation. A 4-point scale (0 = never; 1 = sometimes; 2 = moderately often; 3 = very often) was used. The use of computers to access internet pornography was considered but dropped because of school disapproval.

Procedure

During development of the questionnaire, pilot checks were conducted with children and adults to assess the age and culture appropriateness of the items and format and the completion time required. Some minor modifications were implemented. The final package required approximately 15 min to complete. Ethics approval for the research was obtained from the Human Research Ethics Committee at the University of New England.

Teachers were provided with detailed protocols to observe that included instructions for students, example questions, and procedures for collecting completed surveys. Teachers then chose a convenient time to conduct a brief introductory session and explain the research to students. Information sheets and consent forms were provided for caregivers, and students were asked to return the consent forms for the data collection session if they chose to participate. Alternative activities were designed for those not participating. During the questionnaire completion session, anonymity was emphasised, unmarked opaque envelopes were provided for sealing each survey, and sealed boxes were provided for deposition of the completed survey at the end of the session. Students were asked to keep the questionnaire content confidential and to respect the confidentiality of fellow students. Collection boxes were received from faculty heads within a day or two of completion.

Results

All data were initially screened for entry errors, missing values, and severely skewed distributions using SPSS. Entry errors were remedied, missing values on scale items were estimated as the mean of the three cross-sample most highly correlated items (if more than two item responses were missing then the case was deemed missing entirely), and two variables were severely skewed. One of these was Hours spent computing, which was converted to ranks, and the other asked about the use of computers for Internet gambling. This question was dropped from further analysis because of the very low incidence of any use of computers for this purpose (never = 92.7%, sometimes = 3.4%, moderately often = 1.7%, very often = 1.1%).

Preliminary analysis of the cognitive expectations items indicated little differentiation in terms of the positive and negative reinforcement components. The two components were correlated, r = .66, and when all items were combined into the one scale of positive expectations about later computer use, α = .82. The analysis reported therefore uses the combined total scale for this measure.

Gender comparisons found that boys reported using computers during the previous week for approximately 8.2 hr (Mdn = 7.00) whereas girls' reported computer use was 5.8 hr (Mdn = 3.79). The percentage of boys who reported using a computer >2 hr/day was 18.7% while 6.2% reported their average use as >4 hr/day. For girls, the corresponding figures were 10.3% and 1.3%. The number of boys using a computer for ≥2 hr/day (n = 18) compared to girls (n = 8) was not significant, χ²(1) = 2.44, p = .118. Other major gender differences included the finding that girls scored significantly higher than boys on the BIS, t(175) = −4.48, p < .001, on father control, t(168) = −2.34, p = .02, and greater use of computers for music and chat, t(173) = −2.80, p = .006. Boys were noted for more engagement with computers, t(171) = 2.35, p = .02, heavier use of computers for games, t(173) = 4.09, p < .001, and also higher use of alcohol, t(175) = 2.61, p = .010.

Pearson correlations between predictor variables and computer use variables are presented in Table II. Greater problematic computer use was related to being older and to greater mother control, and, to some degree, higher scores on the BAS. Weaker relationships were found for computer engagement including higher scores on the BAS, and greater mother control and father control. Stronger relationships were found for cigarette and alcohol use, with the BAS and BIS figuring prominently. Higher scores on the BAS were associated with more hours spent computing. Engagement scores and problematic use scores were correlated with Hours spent computing, r = .45 and r = .31, respectively (Spearman correlation). Cognitive expectancies about the later use of a computer were unrelated to the predictor variables included here. Whether the student reported a recent increasing or decreasing trend in their use of computers also had little association with the predictors used here. These two variables were dropped from later analyses.

Table II. Zero-order correlations between major variables

Explanatory variables	Computer use variables					Substance use
	Engage	Problem	Hours	Trend	Expect	Cigarette use	Alcohol use
Age	.04	.20**	.26***	.05	.05	.28***	.45***
Gender (M = 0, F = 1)	−.18*	−.13	−.14	.10	−.09	−.13	−.19**
BAS	.18*	.18*	.21**	.13	−.09	.26***	.26***
BIS	.01	.11	−.02	.14	.07	−.20**	−.26***
Mother control	.17*	.24**	.07	.09	−.03	.05	.05
Father control	.17*	.15	.09	.18*	−.06	.09	.07
Notes: BAS = behavioural activation system; BIS = behavioural inhibition system; “Expect” = expected positive emotional effects from the possible use of a computer later that day; “Trend” = recent increase or decrease in computer use as perceived by the student.
Hours spent computing is ranked.
*p < .05; **p < .01; ***p < .001.

The first major analysis compared the prediction of engagement with computers with the prediction of problematic use of computers using multiple regression. The use of alcohol and cigarettes and hours spent computing (ranked) are included for comparison (Table III). In general the BAS and BIS were strongly predictive (in opposite directions) of cigarette and alcohol use but were only weakly predictive of computer use. Parental variables were not uniquely predictive of computer engagement, alcohol use, or cigarette use. However, maternal control was predictive of problematic computer use. The gender difference (being male predicted reports of higher alcohol use) disappeared when age was included in the regression equation.

Table III. Multiple regression of predictors and measured variables

	Computer Engagement	Computer Problematic	Hours spent computing	Cigarette use	Alcohol use
Predictors
Age	−.03	.16*	.24**	.29***	.44***
Gender (M = 0, F = 1)	−.22*	−.13	−.04	.02	.04
BAS	.15	.16*	.20*	.28***	.31***
BIS	.09	.16*	−.03	−.23**	−.30***
Mother control	.13	.25**	.04	−.02	−.01
Father control	.13	.04	.08	.14	.12
ΔR^2	4.6%*	6.7%**	1.0%	1.7%	1.3%
Total R^2	11.3%**	17.1%***	11.8%**	20.8%***	35.3%***
Notes: BAS = behavioural activation system; BIS = behavioural inhibition system.
Table entries are standardised coefficients; ΔR^2, change in R^2 for parental control variables as a block.
*p < .05; **p < .01; ***p < .001.

Several interactive regression models representing the moderating influence of mother control and father control on the relationship between BIS and BAS and computer engagement and problematic computer use were tested (O'Connor, 1998). Very few of the interactions reached significance, with the most revealing finding being that high mother control in the presence of high activation or inhibition predicted greater problematic computer use (2.9% and 3.9% explained variance respectively).

The last eight items in the survey asked about the frequency of use of different types of computer applications. Using principal components analysis with varimax rotation, four pairs of items were clearly identifiable as the use of computers for games (computer games, internet games, 25.4%), use of computers for news (news or other information, spreadsheets and calculation, 19.2%), use of computers for art (art, writing, 15.7%), and use of computers for music (music, internet email or chat room, 12.6%) for a total of 72.9% explained variance. Four new variables were computed from the average of the two variables that comprised each pair.

Finally, multiple regression analysis was used again to predict each of the different use types (Table IV). This analysis clearly indicates the different nature of the four types of computer use. Being female predicted greater use of music and chat and to some extent also art and writing, whereas being male predicted greater use of computer games (computer and internet-based). The BAS was a uniquely significant predictor of music and chat use, while the BIS was a significant (negative) predictor of games use. Engagement scores were positively associated with the use of computers for games, for news and spreadsheets, and for art and writing. Problematic computer use was negatively associated with the use of computers for news and spread sheets and art and writing. Variables measuring parental control were unrelated to the various computer use types.

Table IV. Multiple regression results for different computer use types

	Games	News/Spreadsheet	Art/Writing	Music/Chat
Predictors
Age	−.32***	.19*	.05	.31***
Gender (M = 0, F = 1)	−.26**	−.02	.21*	.33***
BAS	.03	.17*	.02	.26**
BIS	−.20**	−.10	.03	.02
Mother control	.02	−.04	.02	.01
Father control	−.08	−.06	−.02	.02
Engagement	.34***	.33***	.24**	.07
Problematic	.13	−.25**	−.25**	.09
Total R^2	34.8***	16.1%**	9.6%*	22.6%***
Notes: BAS = behavioural activation system; BIS = behavioural inhibition system.
Table entries are standardised coefficients.
*p < .05; **p < .01; ***p < .001.

Discussion

The nature and extent of computer use in adolescents was examined in relation to individual characteristics of gender, age, temperament, and perceived parental control. There was only a tendency for boys to report using computers more than girls but there was a marked difference in the relationship between gender and type of computer use. Girls reported accessing music and chat services more frequently than boys, while boys reported accessing game-playing activities more frequently than girls. Younger adolescents were associated with greater game playing while older adolescents were associated with greater use of music and chat services. These results indicate that the nature of the stimuli being sought through computer use varies greatly with gender and maturation.

Fundamental temperament characteristics of approach behaviour and stimulus seeking as assessed by the BAS were uniquely predictive of computer use for music and chat purposes while the BIS was negatively predictive of game playing. Although the present data are correlational, perhaps those individuals with greater sensitivity to punishment cues shy away from game playing while those high in approach temperament seek rewards from social contact and music. Surprisingly, the BAS was not uniquely predictive of games use. The BIS and BAS were far more predictive of adolescent use of cigarettes and alcohol, with higher BAS and lower BIS each uniquely predictive of greater use. These results indicate that the conceptual basis of the BIS and BAS has meaning in the understanding of the attraction of different types of computer use for different people. Older age and higher BAS scores predicted greater problematic computer use, which is similar to the prediction of substance use, although much weaker. The main difference lies in the role of the BIS, which is highly negatively predictive of cigarette and alcohol use but is weakly positive for predicting problematic computer use. Again, with the caution that these data are correlational, perhaps the illegality and social prohibitions associated with cigarette and alcohol use by adolescents entice some and repel others in a way that is not found with computer use.

Perceptions of parental control as assessed by the PBI overprotection scale were not related to the types of computer use or to substance use. But greater perceptions of maternal control were associated with greater problematic effects from computer use. Perhaps mothers monitor computer use behaviour more carefully than fathers, or are more available to monitor computer use behaviour than fathers, and intervene more readily when the behaviour appears to interfere with other responsibilities, thereby arousing some conflict. Of course, the correlational nature of the data also allows for the possibility that greater controlling behaviour by the mother results in greater use of computers as a way of retreating or escaping from the control. Some other generally weak associations were found between the PBI scales and outcome measures that disappeared when other variables were included in the analysis. There was little evidence of moderating effects of parental control on the relationship between approach and avoidance behaviours on the one hand and computer use on the other, except for a suggestion that high mother control in the presence of high BIS or high BAS was associated with higher problematic computer use.

The conceptual distinction between high engagement with computers and problematic use of computers was useful and although in this sample only weakly predictive relationships were found, the findings do suggest that male subjects with higher behavioural activation tendencies may incur more problematic effects as they get older. If Charlton (2002) and Charlton and Danforth (2007) are correct in suggesting that high engagement with no major negative consequences (but with mild tolerance, euphoria, and cognitive salience) precedes addictive use (where conflict and problems begin), then addictive use of computers may not be evident at the ages covered in the present sample. Further research needs to determine if highly engaged adolescent users develop into addicted adult users.

An interesting pattern of findings was of high engagement associated with higher use of computers for news and spreadsheets and art and writing, while greater problematic effects were associated with less use of each of these (Table IV). Less problematic effects were associated with greater use of computers for purposes such as calculation, art, and writing, perhaps because of less conflict with parents when their child is using a computer for these purposes. Some support for this comes from the positive (but not quite significant) direction of the findings for problematic effects associated with games use and music and chat use, which we positively associated with greater computer use for these purposes. More research into the differentiation of computer use type including different types of games, chat and text services would be beneficial. In addition, the accessing of stimuli related to gambling, pornography, and violence were not included in the present study but are probably of even greater interest. Two boys and one girl (1.7% of total) reported using the internet moderately often for gambling while a further two boys reported gambling on the internet very often (1.1%). These figures are consistent with reports of problem gambling rates in adults (Orford, 2001).

Limitations of the present study include the use of self-report methodology, which restricts the inference of causality with any of the reported associations, and the lack of detail with computer use. In particular, whether adolescents even had a computer at home was not specifically asked, although examination of the use diary for the previous week and the rates of non-school use of computers indicated that only 3 – 4% reported no weekend use of computers and only low to moderate non-school uses. Other detail includes whether broadband or dial-up access is available at home, whether computer games were predominantly online or offline, whether movies or photographic purposes for computer use should be identified, and whether multiple computers are available at home. Another limitation relates to what parents see as excessive use of computer technology and the nature of the restrictions that parents actually impose on computer use at home. In addition, the PBI asks about general perceptions of parental control, not control specifically related to computer use.

Despite these limitations the present study identifies some important and some less important variables for understanding the nature of adolescent involvement with new technologies. This is an important and growing area of research that may benefit from the starting point provided here.

Further research needs to also consider the effect of culture on computer use. The present study was based on a regional population and the same pattern of results is unlikely to be found with urban students. Similarly, nationality differences are likely to exist in how computers are used and for what purposes (Southwell & Doyle, 2002). Perhaps peer relationships are more important than parental. Perhaps other more direct parental variables such as parents' use of computers may have more relevance in predicting problematic effects.

In conclusion, the present study found considerable diversity in adolescent computer use. Age, gender, and approach and avoidance tendencies predominantly explained this diversity. Measures of parental control were less related to the variables assessed here but some complex relationships were found that are worthy of more investigation.