Socially unbiased parenting support on the Internet - Semantic Scholar

Blackwell Science, LtdOxford, UKCCHDChild: Care, Health and Development0305-1862Blackwell Publishing Ltd, 200420043114352Original ArticleParenting support on the InternetA. Sarkadi and S. Bremberg

Original Article

Socially unbiased parenting support on the Internet: a cross-sectional study of users of a large Swedish parenting website A. Sarkadi*† and S. Bremberg†‡ *Department of Public Health and Caring Sciences, Uppsala University, Uppsala, †National Institute of Public Health, Olof Palmes gata, Stockholm, and ‡Department of Public Health Sciences, Karolinska Institute, Stockholm, Sweden Accepted for publication 29 July 2004

Abstract

Keywords parenting, social support, Internet, social class, gender bias Correspondence: Anna Sarkadi, MD, PhD, Department of Public Health and Caring Sciences, Uppsala University, Uppsala Science Park SE-751 85, Uppsala, Sweden E-mail: Anna.Sarkadi@pubcare. uu.se

Background The Internet is becoming increasingly recognized as a source of social support. Parents of children with cancer and autism have been shown to find social support online, and many parents of healthy infants and children seek information about parenting online. However, access and use of the Internet is greater among socio-economically advantaged groups, a phenomenon known as the ‘digital divide’. Our aim was to investigate whether users of a Swedish general parenting website perceived support in the parenting role and, if so, whether this support was socially biased because of the digital divide phenomenon. Methods Users of the largest Swedish parenting website were asked to participate in the study. A total of 2221 users completed the anonymous survey posted on the website during a one-week period. Results Most respondents (95%) were female (mean age 30.6 years). Respondents’ educational level was slightly, but not significantly, higher than that in the general population, whereas 68% had income levels at or under the national average, contradicting the intuitive hypothesis that users would be socio-economically privileged. Perceived social support, measured by the Interpersonal Support Evaluation List (ISEL) appraisal subscale, indicated high perceived support. Living without a partner and having lower levels of income and education increased perceived support. The perception that other parents’ opinions are more valuable than the advice of experts was influential in the regression equation for ISEL scores, indicating that peer help is important in online social support. Conclusions Internet use for general parenting issues in Sweden, mainly by women, does not seem to follow the digital divide phenomenon. Therefore, the internet provides an exciting opportunity for future infant and child public health work. The lack of fathers, however, was a surprising finding and introduces a gender bias into this seemingly socially unbiased medium.

Original Article © 2005 Blackwell Publishing Ltd

43

44

A. Sarkadi and S. Bremberg

Introduction Social support online The potential of the Internet to be a source of social support is becoming increasingly evident (Larkin 2000; LaCoursiere 2001). Women with breast cancer who used the Internet for seeking information about their disease experienced more social support and less loneliness than women who did not use the Internet at all or used it for other purposes (Fogel et al. 2002). In a randomized controlled study for online support of self-management in type 2 diabetes, patients who had been assigned to online support groups experienced higher diabetes-specific social support than did patients who received diabetes information only or a personal self-management coach online (Barrera et al. 2002). Persons with HIV (Reeves 2000) and carers of persons with Alzheimer’s disease (White & Dorman 2000) have also been shown to find online social support useful. Additionally, it has been found that people with mental disorders report being able to discuss issues on the Internet that they would choose not to disclose in face-to-face contacts (Kummervold et al. 2002).

Internet-based parenting support – is it socially biased? Evidence indicates that the use of the Internet as a communicative and adaptive means of sharing and disseminating information is unequally distributed in different socio-economic groups (Rice 2002). This phenomenon, known as the ‘digital divide’, has been extensively studied and analysed. Studies from Australia (Hellwig & Lloyd 2000), the USA (Papadakis 2001) and Britain (Burrows et al. 2000) indicate that Internet use is greater in socioeconomically advantaged groups. Parents of children with cancer (Han & Belcher 2001) and parents of children with stigmatizing disabilities have found support online (Mickelson 1997), but these parents were more likely to be highly educated and have high incomes. In a study using the clientele of a general paediatric emergency unit, Internet access and use was linearly correlated with income (Mandl et al. 2000).

Shedding light on additional social bias, an overview of paediatric patient education materials on the Internet revealed that most information is written on a higher level than would be appropriate for an average adult reader in terms of education and intellectual abilities (D’Alessandro et al. 2001). There is no doubt that the Internet is an important source of information for many parents (Simpson 1997). The question is whether the Internet can provide general support for parents in their parenting role and, if so, would this support be socially biased because of the digital divide phenomenon.

Aims Based on the above overview, the following two questions arose and thus comprised the principal aims of our study:

1 Can the Internet provide general social support in the parenting role in a Swedish setting?

2 Is social interaction and support online a privilege of the socially advantaged or is it accessible and/or useful for parents from a broader range of social groups?

Methods The website We chose to examine users of the largest Swedish parenting website, FöräldraNätet (English translation, the Parents Network), which can be found at http://wwwalltforforaldrar.com. The website is privately owned, financed principally by advertisements and has existed since 1997. There are 250 discussion groups on the site, and the target group of the website is parents of children of all ages, including adolescents. The discussion groups cover a wide variety of topics and issues, including sexual and mental health, twin and adoption groups, pain management at delivery, child abuse, bullying, home school, divorce, sports, and various adult subjects. Anyone is allowed to access the message boards of the discussion groups; however, registration is required if one wishes to write his or her own message. There is a

© 2005 Blackwell Publishing Ltd, Child: Care, Health & Development, 31, 1, 43–52

Parenting support on the Internet

community manager whose major objective is to ensure that the discussion groups function according to the conduct rules of the site (e.g. use of appropriate language and a respectful attitude towards other users).

The questionnaire A web-based questionnaire was posted on the main page of the website and was available for completion during a week’s time span in early December 2002. A message was posted on the message board of each discussion group to alert users to the ongoing survey. The questionnaire was designed according to the principles described by Dillman and Bowker (1999), i.e. it was simple and selfinstructive with no mandatory answers and easy to browse and click for completion. The questionnaire, which was pilot-tested before its launch on the website, included 62 items and took 10–12 min to complete (Table 1). Social support was measured by a specified and revised version of the appraisal support subscale of the Interpersonal Support Evaluation List (ISEL) (Cohen et al. 1985). The ISEL was chosen as a measure of social support because it was short; it examined the functional aspects of social support; it had been adequately tested (including for social bias); and it had shown good internal consistency (Cronbach’s alpha: 0.88–0.90). The ISEL also had good validity and predictive capacity in terms of, for example, psychological and physical symptomatology, as compared to other instruments used to measure social support (Orth-Gomér & Undén 1987). The appraisal subscale was employed (as opposed to the tangible support and belonging support subscales) because it had most relevance for the specific setting of this study. After personal communication with the principal author, Dr Cohen, all 10 items of the ISEL appraisal subscale were rephrased in order to specifically refer to the website as the source of support (Table 1). Self-esteem was measured by the 10-item selfesteem subscale of the ISEL. Questions referred to the person’s relationship to his/her friends in real life (Table 1). The self-esteem subscale was included after the recommendation of Dr Cohen as this subscale had consistently shown correlation

45

with the measured social support (Cohen et al. 1985). Sum scores for both the appraisal and selfesteem subscales ranged from 0 to 30. A sum larger than 20 was interpreted as indicative of high perceived support/self-esteem (2 or more points for each item), after personal communication with Dr Cohen. The authors received the data from the respondents as an Excel file from the programmer at the website. Users were not pre-selected for answering the questionnaire, but could freely choose to click on the link leading them to the first page of the form. Thus, the sampling procedure was not as in traditional surveys (Groves 1989), but on a purely voluntary basis, implying a certain sampling bias among those who completed the questionnaire.

Data analysis Descriptive statistics were prepared and, where relevant, compared with data from national statistics (Statistics Sweden 2002) or with data from the membership database of the website (e.g. age and gender of the users). Because 95% of respondents were women, comparisons with the Swedish statistics databases are primarily based on women, if not otherwise stated. Differences were generally tested using the chi-square test. In determining effects of multiple variables on perceived appraisal support, a multiple linear regression analysis was performed with the ISEL appraisal score as the dependent variable. A probability level of P < 0.05 was considered to be significant.

Results Respondents The questionnaire was completed partly or in total by 2499 persons. Of these, 2221 (89%) had completed more than two pages of the five-page questionnaire, thus providing meaningful data for analyses. Although excluded from the more elaborate analyses, the 278 individuals (or 11%) who had failed to complete the questionnaire were used to investigate the bias affecting self-selection to fill in the questionnaire, as measured by comparisons of gender, age, geographical location, parental sta-


46


Table 1. Items included in the questionnaire, n = 62 n

Item

Background 14 Socio-economic status/Civic status Pattern of website use 6 Member/Frequency of visits ‘Netiquette’ 5 Anonymity – the use of pseudonyms ‘Flaming’ – exposition, risk, behaviour in situation ‘Lurkers’ – public place with private messages Own constructions of peer support 6 1. Do you use the Parents Network as a primary source of advice? 2. If you have received any specific advice, were satisfied/not satisfied? 3. The opinions of other parents are more important than the advice of experts 4. It is good to give other people support by answering their posts 5. It is good to read what others have written on message boards 6. It is a problem for me to get advice from people I haven’t met before Social support* Appraisal subscale (Only parents to children under the age of 18 years) 10 1. There are several people on the Parents Network that I trust to help solve my problems concerning my children 2. There is no one on the Parents Network whom I feel comfortable talking to about intimate personal problems 3. There really is no one on the Parents Network who can give me an objective view of how I’m handling my problems 4. I feel that there is no one on the Parents Network I can share my most private worries and fears with 5. There is someone on the Parents Network I can turn to for advice about handling problems with my family 6. When I need suggestions on how to deal with a personal problem, I know someone on the Parents Network I can turn to 7. There is someone on the Parents Network I could turn to for advice about making career plans or changing my job 8. There really is no one on the Parents Network I can trust to give me good financial advice 9. If a family crisis arose, it would be difficult to find someone on the Parents Network who could give me good advice about how to handle it 10. There is at least one person I know on the Parents Network whose advice I really trust Self-esteem subscale 10 1. Most of my friends are more interesting than I am 2. There is someone who takes pride in my accomplishments 3. Most people I know think highly of me 4. I think that my friends feel that I’m not very good at helping them solve their problems 5. I am as good at doing things as most other people are 6. In general, people do not have much confidence in me 7. I am more satisfied with my life than most people are with theirs 8. Most of my friends are more successful at making changes in their lives than I am 9. I am closer to my friends than most other people are to theirs 10. I have a hard time keeping pace with my friends Internet use pattern 8 Frequency Errands Socializing Research ethics 3 Informed consent Publishing statements from discussion groups *Definitely true, probably true, probably false, definitely false. Items scored from 0 to 3. Values of 2–3 indicative of high perceived support/self-esteem. Sum scores ranged 0–30. A sum larger than 20 was interpreted as indicative of high perceived support/ self-esteem (2 or more points for each item).

tus and marital status. Men were more likely than women to discontinue filling in the survey (15% vs. 10%; P < 0.0001) as were individuals not currently parenting children under the age of 18 years as compared with those who were currently

parenting children (13% vs. 10%; P < 0.05). Those who were married were more likely to complete the questionnaire than those who were cohabitants without being married or persons living alone (92%, 88%, and 86%, respectively; P < 0.01). Age



and geographical location did not appear to affect completion of the questionnaire. Additionally, using the time slot for filling in the questionnaire, respondents were dichotomized into those who submitted the survey during the first 24 h of its launch on the site and those who did so later during the week. The first group of respondents was significantly more likely to provide complete (usable) questionnaires than those who submitted the questionnaire later in the week (93% vs. 86%; P < 0.0001), a finding suggesting that frequent users are overrepresented among the respondents to this questionnaire. The following data are thus based on 2221 respondents, hereafter also referred to as users. When only the female or male respondents are referred to, this will be clearly indicated.

Descriptive statistics Age and gender Users were mainly young, with 12% under 24 years of age and 66.5% between the ages of 25 and 34 years. Those 35–44 years old represented 20% of the users, whereas only 1.5% were 45 years or older. The mean age of all the respondents in the study was 30.6 years (range 16–56 years). Almost all respondents were women (94.9%) and only 3.3% were men (1.8% of the users chose not to indicate their gender). Noteworthy is that, according to the member registry of the website, 13.5% of members are male and 86.5% female. Thus, there are more male registered members at the site than indicated by the gender distribution of respondents. The difference is statistically significant (P < 0.001) and clearly indicative of the fact that women are overrepresented in the questionnaire data. Mean age of the members registered on the website, on the other hand, was 32 years, so the questionnaire, where mean age of respondents was 30.6 years, should be regarded as representative of users from this aspect.

Sociodemographic profile of users Most respondents were married or had a partner (Table 2), while only 5.3% were lone parents (5.1%

47

for the mothers in the sample). This percentage (5.1%) is considerably lower than would be expected based on the general Swedish population, where the proportion of mothers rearing their children alone is 16.9%. The respondents had somewhat higher levels of education than women aged 24–44 years in the general Swedish population; however, the difference was not statistically significant, P = 0.10. A stratified comparison for different age groups revealed that for the youngest respondents (16– 24 years) and those aged 35–44 years, the difference in educational level between users of the website and women in the general Swedish population was statistically significant, P < 0.01. For the largest group of 25–34 years olds (66.5%), however, the difference was not statistically significant, even though more users had some post-high school training than corresponding women in the general Swedish population (58% vs. 40%; P = 0.076). Another important measure of socio-economic status, namely income, showed that a considerable proportion of users had lower than average incomes. This finding could be because most users were young and were women; almost a third were on maternity leave (which means these women are receiving 80% of their normal income), and 10% studied full-time, implying living on a student allowance. Nonetheless, the finding that 68% of respondents earned on or below average salaries indicates that this is not a very privileged or elite subgroup of society. Only 7% of respondents had incomes in the highest available category. The region of the capital city, Stockholm, was somewhat overrepresented, with 36% of respondents residing in the area compared with the expected 23% that is based on population statistics (Statistics Sweden 2002).

The subsample of men Proportionately, few men (3.3%; n = 86) answered the questionnaire, and it is therefore of special interest to describe this subsample. The men respondents were somewhat older than the women respondents (mean age 34.9 years, range 17– 56 years, compared to 30.3 years for the women, P < 0.05), and a higher number of the men had


48


Table 2. Characteristics of the sample population, n = 2221* Variable Civic status Married Cohabiting without marriage Partnered, not living together Widow Living alone Education Elementary school, no high school degree (9 years) High school degree (12 years) College or other postgraduate education < 3 years (13–15 years) College or university education ≥ 3 years (15–18 years) Income level 30 000 SEK (ª3330€) Number of children (mean =1.78) 1 2 3 4 Employment status Full-time worker Full-time parental leave Part-time worker Full-time studies Sick-leave/disability pension Unemployed Retired

Number

Percent

1010 996 40 2 115

46.7 46.0 1.8 0.1 5.3

123 810 498 662

6 39 24 31

476 348 583 353 165 142

23 16.8 28.2 17.1 8 6.9

825 665 227 106

45.3 36.5 12.5 5.8

551 609 291 164 197 80 11

27 29 20 10 9.5 4 0.5

*Not all respondents answered all questions. †This is the range of average income.

salaries in the highest category as compared with the women (23% vs. 6.3%; P < 0.001), despite that only 54% of men were full-time workers and that educational levels were similar in both genders. There were also more lone parents among the men, not only proportionally (17% lone parents among the fathers and 5.1% among the mothers, P < 0.01) but also compared with numbers in the general Swedish population (17% vs. the expected 7% of lone parent fathers, P < 0.01).

population (Statistics Sweden 2003). Most respondents used the Internet daily (92%) and 37% reported spending 3 h or more per day online. The majority used their computers at home for logging on to Parents Network, but 17% reported accessing the website both from work and home. Logons were distributed rather evenly during the day, but 13% indicated that they used the website mostly in the evenings.

Social support online Internet use Respondents were frequent and confident users of the Internet, with more using the web for banking, official communications, shopping, bookings, and socializing than would be expected based on the information technology survey of the Swedish

Only respondents with children under the age of 18 years (n = 2030) were asked to complete the ISEL appraisal subscale (see Table 1 for details). Of these 2030 respondents, 1674 (83%) gave enough answers to allow ISEL scoring. The average score of perceived support was 21.5 points (SD = 4.9),



which represents a high general perception of support from Parents Network. Similarly, respondents reported generally high levels of self-esteem (M = 21.4; SD = 3.8). We attempted to further investigate factors affecting perceived support in a linear regression equation. Separate equations were performed for men, women and the group as a whole. A manual entry procedure was used (age was kept in the model for adjustment) to determine which variables had a significant effect on the dependent variable. As expected, no differences were observed in the model for the women only and in the model for all respondents (because 95% of respondents were female), which is why we present a main model based on all respondents with valid ISEL scores. The final main model is based on a sample of n = 1655, where n = 19 (1%) outliers (ISEL scores < 9) have been excluded to allow for a normal distribution of residuals and homogeneity of variance. The model explains 26% of the variance in the dependent variable (Table 3). Mean Cook’s distance, indicating the presence of influential observations and is required to be < 1, was 0.001 (range 0.000–0.013).

49

After adjusting for age, the results of the statistical analysis revealed that the social and economic variables of living without a partner, having lower levels of education and income, and having several children related significantly to the ISEL appraisal support scores (Table 3). As expected, ISEL selfesteem scores were also related to perceived support: higher ratings of self-esteem were related to higher perceived support. Website-related factors that significantly affected perceived support were being a registered member, and having regular contact with other users. In addition, two of our own constructions related to peer support proved significant for perceived support: using the Parents Network as a primary source of information or advice and the perception that the opinions of other parents are more valuable than expert advice. The latter was the most influential explanatory variable in the regression equation in terms of B coefficient. A similar, age-adjusted regression model was done for men by using the same procedure as above (Table 4). After exclusion of a single outlier, the model was based on n = 54 subjects and explained 48% of the variance in the ISEL scores. The two lowest income levels (out of five) were the only

Table 3. Linear regression equation for Interpersonal Support Evaluation List (ISEL) appraisal score as dependent variable. The model is based on the sample as a whole, n = 1490*, R2 = 26% Variable

Coefficient B

Significance

Constant Demograpic variables Age of respondent in years (kept in model for adjustment) Living without a partner† Cohabiting without marriage Number of children No high school diploma‡ High school graduate