|Sometimes there are things which simply cannot be left alone.
I feel that as a professional I have to take a position.
One of the most recent issues I have taken exception with is the report
This report claims to demonstrate conclusively that the Internet, in opposition to the claims made for it, actually increases depression and loneliness.
I say it is a flawed study and the conclusions are unwarranted. A variation of this article appeared in the op-ed section of the Pittsburgh Post-Gazette on Sunday, 27 September, 1998.
First off, you don't have to take my word for it. Read the original report yourself at
and decide for yourself if I'm wrong in my position.
One of the questions that has been posed to me is regarding my qualifications to comment on such a report. I was specifically asked, did I participate in the research, was I part of the project, and do I know anything at all about psychology? I was even asked by one correspondent if I had an enmity towards any of the researchers. I was not a researcher on the project, I was not part of the project, and I never met any of them except Bill Scherlis, and we were friends when he and I were on the faculty at CMU. Since then, I've simply lost touch, since he spent many years in Washington DC and did not return to the University until after I had left.
As far as qualification, I lived with a psychology major for three years. My undergraduate roommate was a psych major, and convinced me to take a couple courses in the subject as my non-major qualifying courses. In addition, I had many conversations with him on the design of experiments, read books on the subject, and was often a subject of various experiments. At CMU, I became interested in Cognitive Psychology during my graduate studies, and at one point even was working on a PhD dissertation on attempting to study the mental models programmers have about their programs, with my goal being to incorporate such knowledge into a more user-oriented debugger. After several years of study, I abandoned this topic since I felt I needed another three years of intense study to be able to do a dissertation in the area, and my student research funding would run out long before this could come to pass. Nonetheless, I took courses on Cognitive Psychology from the likes of Allen Newell and Herbert A. Simon (Nobel laureate). Even after receiving my PhD, I took a couple graduate seminars in the psychology department, and participated in and commented on numerous experiments. In addition, I have a background in "hard science"; my undergraduate degree was in mathematics (including two semesters of statistics), but I had enough credits to take a minor in physics (the year I graduated the school discontinued the awarding of "minor degrees"). The heart of such science is measurement, and statistics play an important role. Therefore, I can additionally comment on the significance of the results with some confidence that I know what I am talking about.
Below is my editorial on this research.
I did not feel it was appropriate to comment on the "Internet Depression" paper until I had read it. I have now read what is given as a "preliminary draft" version of the paper, and I find the results deeply suspect. Not the science, which is extremely well done, but the conclusion itself.
The draft of the paper that I read is found athttp://homenet.hcii.cs.cmu.edu/progress/HN.impact.10.htm, for anyone else who wants to read it. [Note: since then, the APA paper cited in the previous URL has become available.]
The results are: after one to two years of using the Internet, individuals exhibited reduced social interactions and increased depression. Consequently, use of the Internet is responsible for reduced social interaction and increased depression.
The most significant defect about this study, and one I find overwhelming, is that there is no control group. There is no way to say if the results are due to the Internet, due to the economy, due to the specific demographics of the group chosen, or could be induced by any activity with one or more characteristics of the Internet. In fact, for one of the results, there is no evidence to suggest that the variance is not due to simple statistical variance in the result.
The study showed teens used the Internet more than adults. No surprises here. Greater use of the Internet showed subsequent declines in family communication. No surprises here. In fact, Im surprised that this is touted as a significant result. What is appalling about this study is a fundamental theme of this critique: there is no control group. What this study really appears to show is that the introduction of any new, fascinating, individual activity decreases family communication. The Internet is merely one instance of such a new, fascinating, individual activity, and why it is singled out as being a Bad Influence when compared with, say, chess playing, ham radio, computers without Internet access, amateur astronomy, building model airplanes, or rebuilding a car makes me wonder. The conclusion is
Using the Internet decreases family communication.
Without a control group, it is impossible to say if the real conclusion should have been
Any new, self-absorbing activity decreases family communication.
Another non-surprise: people who are extroverted use the Internet less. They probably also watch TV less, and compared to other groups, such as model airplane builders, probably spend less time building model airplanes.
Now, about this "loneliness" measure. There are those of us who do not have wide social circles. The smallness of our social circle is in no way dependent upon Internet access or lack thereof. Why, exactly, having a large social circle is a Good Thing and a small social circle is a Bad Thing is somewhat obscure. [Note: the following week I attended a lecture by the principle author; see my commentary below]. However, the measure of loneliness is quite bizarre. I don't know if the researchers noted an odd fact: there are only 24 hours in a day, and we sleep, on the average, eight of them away. We work or go to school another eight. Allowing for breakfast, dinner, and early bedtimes, this leaves most of us with the interval of 6pm to 11pm to engage in whatever else in life interests us. If we devote a significant part of that five hours to engaging in some new, fascinating, personal activity, guess what: we will not interact with our social circle, family or otherwise, as much. The conclusion is stated as
Using the Internet increases loneliness.
Without a control group, it is impossible to say if the real conclusion should have been
Any new, self-absorbing activity decreases social interaction.
There appears to be an unstated assumption here: decreasing social interaction is identical to increasing loneliness.
Well, perhaps not. Based on the suggested questionnaire based on questions such as "who can I turn to", perhaps loneliness has a different interpretation. But the data reported suggests that people who use the Internet more showed increased loneliness. It had already reported that teens use the Internet more than adults. Over the course of a year or two, many teens find that they have increased loneliness as gaps form between themselves and their peers or their parents (this is given passing consideration in the conclusions, and dismissed, for reasons I cannot comprehend). Without a control group of people who had no new, fascinating technology to play with, but represented similar demographics, it is impossible to say if the conclusion is not really
Two years in the life of a teen can increase feelings of loneliness.
Any teen involved in a new, self-absorbing activity will have feelings of increased loneliness.
Any teen involved in a new, self-absorbing activity will have decreased social interaction.
In the section titled "Evaluating the causal claim", the authors state that "increases in Internet usage were associated with larger increases in loneliness and larger declines in social support for teenagers than for adults". There is nothing to show that increases in time spent playing video games has the same effect, or for that matter, increases in times spent reading books, playing the piano, or drawing cartoons.
Finally, there is the consideration of depression. The study postulates the conclusion
Using the Internet increases depression
Again, I emphasize, there is no control group. The report states "those with initial higher stress reported greater increases in depression". How is this correlated with Internet usage as the cause? Could it not be true that
Involvement in any new, self-absorbing activity which has opportunity for failure can increase depression.
For that matter, without a control group of similar demographics, how do we know what caused the depression? This is an age of downsizing. A moderately well-off adult can certainly feel depression if their salary remained flat, there was no promotion, or their job is in danger. And without a breakdown by age, and for that matter, how many teens were concerned with SATs, Getting Into The College Of Their Choice (or had recently failed to do so), or were feeling depressed about leaving home to go to college. For that matter, one respondent told about a college-age daughter who felt homesick. It is clear that the Internet did not cause this. I see no way we distinguish Internet usage from:
The depression of teens can increase over a one or two year period
In a sample of similar demographics, depression has generally increased over the last two years.
Given that about half the study group was female, the reality is (political correctness notwithstanding) that some women are prone to monthly mood swings. For that matter, the study did not reveal if any of the females involved became pregnant or birthed a child during the study, or had a two-year-old child at some point during the study (our friends who have gone through the "terrible twos" assure us it is very depressing at times), so there is no way to validate or refute the conclusions:
There was an increase in depression caused by hormonal factors at the time the second set of tests were administered
There were family experiences that increased feelings of depression by the time the second set of tests were administered.
The truly appalling conclusion is "Greater use of the Internet was associated with statistically significant declines in social involvement and with increases in loneliness". While this is true, and supported by the data, singling out the Internet is an unwarranted conclusion. Back in the 1960s, when I was a teen, nobody considered doing a study on the effects of model railroading, model airplanes, chess, electronics, ham radio, rebuilding cars, stamp collecting, fly tying, reading science fiction, playing the piano, photography (including darkroom work) or amateur astronomy on teens. Many of my peers were involved in exactly one of these activities, to the virtual exclusion of almost everything else. The Internet is merely the latest of a long list of fascinating, self-absorbing, singular activities that people engage in. And while it is a bit more pervasive, it has not been contrasted with similarly pervasive activities such as non-Internet computer usage or video games, nor with any of the aforementioned activities, or with groups that do not get involved with activities with similar characteristics. Therefore, I claim that the singling out of the Internet as the cause of such problems is misleading and inappropriate. The only valid conclusion I see is that any activity that is performed in isolation can lead to declines in social involvement and with increases in loneliness. Given the aforementioned five hours a day in which we can engage in activities other than working/school, sleeping, and eating, an detailed study is not required to come to these conclusions.
One phenomenon reported was that the Internet is used to form "weak social ties" rather than "strong social ties". The question as to whether weak ties are "social ties" I think is open to question. Many of the "social" correspondences I had over the Internet were more for our mutual entertainment than for forming relationships of any sort; the online equivalent of cocktail-party chatter. Of course, both of us realized that we where simply exchanging pleasantries, or holding debates, but I never saw it as being anything other than that. I would no more have expected the Internet to be a method of forming friendships with my correspondents than I would expect that my reading a book was forming a friendship with the author. And nobody in the 1950s or 1960s debated policy about ham radio operators, and whether or not using ham radio led to increases in depression or loneliness. For that matter, the use of Morse code was probably more personally distancing than the typed correspondences in chat rooms.
I strongly suspect that the problems are not that people are substituting weak ties for strong ones, but that they are substituting one activity for others that would form strong ties. The social phenomenon of the Internet may be irrelevant here. Consider a stamp collector of the 1950s with an extensive correspondence (physical mail) with other stamp collectors. If he (stamp collectors were mostly male, in that era, or so it seemed) spent all his evenings studying his collection and writing letters, instead of going bowling with his buddies, would we have felt a need to redesign the technology of stamp collecting or establish educational policies about stamp collecting, or have public policy debates about stamp collecting? Or for that matter, about cruise ships, bed-and-breakfasts, neighborhood bars, and cocktail parties, most of which seem to be for the purpose of forming weak social relationships. (I should point out that some people tend to avoid cruise ships, bed-and-breakfasts, neighborhood bars, and cocktail parties because of a desire to avoid superficial social interactions).
The authors pose that having technology for "finding people based on their attributes" is far less well supported than the existing technologies for finding products and information. The authors seem to think this is a deficiency in the Internet technology. I think this is a good thing. I get far too much junk email already, and having anyone be able to locate me "based on my attributes" is quite undesirable. Mass marketers would love to target me (and lots of people like me) "based on my attributes". Id much rather have search engines based on finding products and information; these are something their creators want to make available, publicly. The thought of a search engine that could find me "based on my attributes" horrifies me; it would make me vulnerable to all sorts of people whom I would rather not encounter in any way.
My fundamental conclusion about this paper is that it is a very-well-done study that is totally meaningless until it is validated by similar studies in other areas, including having a control group for which there were no new, self-absorbing activities present. The conclusions pinpointing the Internet as the source of loneliness and depression have no meaning unless it can be shown that no other new, self-absorbing activities can lead to the same effect.
Joseph M. Newcomer
Joseph M. Newcomer has a PhD in Computer Science from CMU, has been a CMU faculty member, and was one of the founding scientists of the Software Engineering Institute. For more than a decade he has been a self-employed consultant and software developer and is the co-author of two books on programming Microsoft Windows.
There are several other observations. For example, the significant result is that social interactions decreased from 24 people in the past 30 days to 23 people in the past 30 days. This is said to be statistically significant. But where is it written that statistical significance is social significance? I can think of lots of people that I'd rather not talk to, and the impact on me, personally, of not talking to them, is nonexistent to positive. If using the Internet is an option to going out and talking to someone I'm somewhere between indifferent to and trying to avoid, yes, I'll use the Internet.
For that matter, can you actually enumerate all the people you interacted with socially in the last month? We have a subjective evaluation of social interaction which is based on a theory that people can remember their last 30 days accurately enough to know exactly how many people they interacted with. Yet the differences are so minute (one person!) that simple statistical variance could produce figures as meaningful. And, for that matter, what about the previous month? Perhaps they interacted with two fewer people, or two more. My own variance of interaction over the course of a year can vary by a factor of 2 to 10 comparing any two individual months, yet a variance of 4% is touted as socially significant by this study.
A week after my piece was published, I went to a public talk presented by the principle author, Kraut. When I confronted him with the accusation that the premise that a large social circle was important to happiness was unsubstantiated, he responded by telling me that "30 years' worth of research demonstrate that on the average people with larger social circles are less lonely and less depressed than people with small social circles" Q.E.D. I think this is specious. If it is true, then 50% of the population is either indifferent to the size of their social circle or non-lonely, non-depressed in spite of a small social circle (I think I'm in this latter set). The conclusion also does not suggest that genuinely lonely people have small social circles because the factors that make them lonely also mitigate against their forming larger social circles. Therefore, I think the fundamental premise that assumes that large social circles are important for happiness is shaky (it is worth noting that "happiness" and "contentment" are not measured factors anywhere; even the tests for loneliness show a cultural bias: "I feel there is no one I can turn to" implicitly suggests that people who feel they have no one to turn to think of this as a serious handicap, rather than an opportunity for self-reliance). [Click here to return to the paper if you got here via the hyperlink]
And while the percentages are statistically significant, they are all very small. All statistical significance says is that "we are absolutely certain that these numbers are correct", and in no way can place any judgment that the numbers have any meaning. As a practicing engineer, I care very much about factors of 2, or orders of magnitude. I do not care very much for factors of 4%.
While the authors admit that they need to do a comparative study, they have chosen television as the alternative to study. Not video games. Not ham radio. Not chess. Not music. They plan to compare a completely passive entertainment form to any highly interactive activity (I see little difference between the Internet and learning to play the piano) and draw some conclusion from this. They also point out that the loss of social contact caused by the Internet may lead to social isolation. Apparently the impact of the automobile, shopping malls, home air conditioning, the telephone, etc., are not felt to be significant, but the terrible, terrible Internet is a demon because of some statistically significant but socially irrelevant numbers came from the study. All this can lead to is a bunch of totally unqualified people (particularly school boards and parent groups), including neo-Luddites, getting involved in a decision process they are remarkably ill-qualified to pursue, and the potential damage that such people can do by believing that this report says something important (particularly when it seems to be attractive to the anti-technology nut fringe) is far more dangerous to our society than the possible social impact of the Internet.
|For another view of the impact of technology on society, see The Victorian Internet : The Remarkable Story of the Telegraph and the Nineteenth Century's On-Line Pioneers, by Tom Standage. While it is true that technology changes society (for example, James Burke's Connections), we sometimes cannot determine the impact for decades, and trying a priori to contain the changes because they violate our preconceived notions of what is "right" is pointless. History has already demonstrated this conclusively. Imagine if society had to decide to control the impact of cell phones and personal computers at the time the point-contact transistor was being invented (see Riordan & Hoddenson's history, Crystal Fire for a marvellous history of the invention of the transistor).|