Link between Internet use and Loneliness

JuliusCaesar

from the BPS bulletin:

---

5. Large, longitudinal study finds tentative links between internet use and loneliness

---

Internet use is growing at a phenomenal rate and much ink has been spilled by commentators forecasting the psychological consequences of all this extra web-time. A lot of that comment is mere conjecture whilst many of the studies in the area are cross-sectional, with small samples, producing conflicting results. The latest research contribution comes from Irena Stepanikova and her colleagues and involves a massive sample, some of whom were followed over time. The results suggest that more time on the internet is associated with increased loneliness and reduced life satisfaction. However, it's a complicated picture because the researchers' different outcome measures produced mixed results.

Over thirteen thousand people answered questions about their internet use, loneliness and life satisfaction in 2004 and in 2005. They'd been chosen at random from a list of US land-line numbers. The majority of the people quizzed in 2004 were different from those quizzed in 2005, but 754 people participated in both phases, thus providing some crucial longitudinal data.

An important detail is that the researchers used two measures of internet use. The first 'time-diary' method required participants to consider six specific hours spread out over the previous day and to estimate how they'd spent their time during those hours. The other 'global recall' measure was more open-ended and required participants to consider the whole previous twenty-four hours and detail as best they could how they'd used that time.

The cross-sectional data showed that participants who reported spending more time browsing the web also tended to report being lonelier and being less satisfied with life. This association was larger for the time-diary measure. The strength of the association was modest, but to put it in perspective, it was five times greater than the (inverse) link between loneliness and amount of time spent with friends and family. Turning to web-communication, the global recall measures showed that time spent instant messaging, in chat rooms and news groups (but not email) was associated with higher loneliness scores. For the time-diary measure, it was increased email use that was linked with more loneliness.

The longitudinal data showed that as a person's web browsing increased from 2004 to 2005, their loneliness also tended to increase (based on the global recall measure only). Both measures showed that increased non-email forms of web communication, including chat rooms, also went hand in hand with increased loneliness. Finally, more web browsing over time was linked with reduced life satisfaction by the time-diary measure, whilst more non-email web communication over time was linked with reduced life satisfaction by the global recall measure.

Perhaps the most important message to come out of this research is that the results varied with the measure of internet use that was used - future researchers should note this. The other message is that more time browsing and communicating online appears to be linked with more loneliness, the two even increase together over time. However, it is important to appreciate that we don't know the direction of causation. Increased loneliness may well encourage people to spend more time online, rather than web time causing loneliness. Or some other factor could be causing both to rise in tandem. It's worth adding too that the web/loneliness link held even after controlling for time spent with friends and family. So if more web use were causing loneliness, it wasn't doing it by reducing time spent socialising face-to-face.

'We are hopeful that our study will stimulate future research ... ,' the researchers said, 'but at this point any claims suggesting that as Internet use continues to grow in the future, more people will experience loneliness and low life-satisfaction would be premature.' _________________________________

Stepanikova, I., Nie, N., & He, X. (2010). Time on the Internet at home, loneliness, and life satisfaction: Evidence from panel time-diary data Computers in Human Behavior, 26 (3), 329-338

http://dx.doi.org/10.1016/j.chb.2009.11.002

A sad day for me......

JuliusCaesar

On the other hand.....

Xluna

Socially inept, depressed and highly introverted people spend alot of time on the internet. No **** Sherlock.:pac:

cupofjoe

First year undergrad psychology student could pick this study apart.

Valmont

Could you elaborate further, cupofjoe? What do you think is wrong with this study? I find some first year undergrads to be impetuously critical, but then again, perhaps you know some outstanding first years. The study seems quite unassuming to me. They present some interesting correlations they have found which could definitely be used as a base for further experimental research. I don't really see anything a first year psychology student could "pick apart" with good reason? Then again, I don't have access to the full article from science direct.

cupofjoe

This whole study is based on correlation-causation - you're taught that about 4 statistics lectures into psychology undergrad. They even state it themselves in their discussion - they cannot say anything about whether people use the internet more because they are already lonely or have low life satisfaction versus whether their use of the internet creates their loneliness/low life satisfaction; or whether the measures that co-varied with loneliness are causal or not. So basically all they can say is that some of the things they decided to measure co-varied with loneliness and some didn't...and that’s exactly what the research they bring in during their intro says too – some of it finds internet use increases loneliness/depression while some find it decreases it, but this avenue tells you nothing about causation.

They only asked about "loneliness" and "life satisfaction", are they the only measures of psychological well-being? They do show high loadings onto “psychological well-being” for the scales they used but if they only used loneliness on its own they would have gotten these high loadings too and that doesn’t mean they have actually tapped into psychological well being at all.

They based their study on 6 periods of time through one single day - so their entire correlation of loneliness and life satisfaction is based on how these people spent one single day in their lives…or two if they were included in the longitudinal part.

In selecting the questions to use in their scale of loneliness '...we selected only four items with high item-scale correlations according to Russell (1996) and high face validity as judged by the authors and several colleagues experienced in survey research' (p. 332), so basically lets just choose ourselves the ones that will give us high loneliness values and get rid of the ones that might not...and of course then their alpha coefficients were reasonably high (.82, .80, .80, .80). They also seem to only provide coefficients for the entire sample, their data included ages from 18-70, it might have been interesting to see if they checked the coefficients for age groups considering a lot of scales are developed specifically for different age groups – of course I could be corrected on this one because I’m not going to look up the data on the development of that scale.

I’ve only quickly read the study once and it’s not crap or anything but I just find that studies showing correlational data tend to spread around the internet and on radio stations as if it's found and proven something miraculous. If I have mistated or got their study completely wrong though I would be more than happy to be corrected!

Valmont

cupofjoe wrote: »

I’ve only quickly read the study once and it’s not crap or anything but I just find that studies showing correlational data tend to spread around the internet and on radio stations as if it's found and proven something miraculous.

I agree, many newspapers and laymen will misinterpret a correlational study as implying some form of causation. The point I was making is that the authors were very careful to point out the correlational nature of their study and I have found that interesting correlational studies tend to end with a statement of this nature:

'We are hopeful that our study will stimulate future research ...'

I think that is an important factor to consider when looking at a correlational study; whether the evidence provided will allow other researchers to construct an experimental project around the area to see if any causal factors are actually involved.

JuliusCaesar

Exactly. Post hoc non ergo propter hoc. They do acknowledge this, as any good study would. Shows the importance of reading a study!

cupofjoe

I get your point about them clarifying that nothing should be concluded - but you also asked how a student could find flaw with the study.

That said, I don't buy that just because a researcher puts a statement that they hope their research stimulates others means it's a good piece of research...that's bog standard article etiquette. I've seen studies use a 10 minute sample of video footage to judge whether children engaged in stereotyped behaviour to compare them to chronological and mental age matched controls; the whole comparison was based on how those kids spent 10 minutes of their lives...their whole study was based around that...them sticking "this research should progress others" doesn't mean it was good research. Read a bunch of undergrad psychology projects and they'll all say the same - at least the ones listening up in their research methods modules...

I have no problem with the study in terms of replication, as that is important - but my whole point was based on the fact that it has big methodological flaws.

Valmont

cupofjoe wrote: »

- but my whole point was based on the fact that it has big methodological flaws.

Big methodological flaws that you have yet to point out. I'm not letting this one slide. Impetuous and over zealous criticism of research articles really gets to me.

cupofjoe wrote: »

This whole study is based on correlation-causation

Wrong. It is a correlational study. I'm presuming you didn't read the entire article considering the authors were careful to mention that:

"'but at this point any claims suggesting that as Internet use continues to grow in the future, more people will experience loneliness and low life-satisfaction would be premature."

cupofjoe wrote: »

they cannot say anything about whether people use the internet more because they are already lonely or have low life satisfaction versus whether their use of the internet creates their loneliness/low life satisfaction; or whether the measures that co-varied with loneliness are causal or not.

Exactly, this is a correlational study. If you think correlational studies are pointless in general, then please create a new thread so we can tackle that point separately.

cupofjoe wrote: »

and that’s exactly what the research they bring in during their intro says too – some of it finds internet use increases loneliness/depression while some find it decreases it,

cupofjoe wrote: »

but this avenue tells you nothing about causation.

Again, it is not supposed to- It is a correlational study.

You missed the point that this study used a very large sample, over 13,000 individuals. The early studies did not have as large a sample. It would have been impractical for the initial studies in this area to use 13,000 individuals when they were not even sure what they would find. The early results allowed the present researchers to commit to a larger sample. The next step after extensive correlational evidence is to construct an experimental project, within which the question can be asked again and the putative presence of any causal factors can be deduced. Sometimes, without correlational studies, many experimental research questions may not have testable implications of the general hypothesis. In terms of this time-line of research development, the current study was more than adequate.

Induction is an important part of the scientific method and as such the aggregate from many different experiments often forms the basis for a new theory. Rarely will you find one single piece of research opening up and polishing off a research area in one shot. So if you some studies seem pointless or pedantic to you, look closer.

By all means, keep analysing studies for methodological flaws but don't patronise the authors of this article for supposedly making basic mistakes when, ironically, it is you who are conflating correlation and causation in your criticism by implying that the former is useless without the latter. You have already proved your initial statement incorrect considering a misunderstanding of the purposes of a correlational study (stats 101 right?) seem to underpin your own "critique".

cupofjoe

Valmont wrote: »

Big methodological flaws that you have yet to point out. I'm not letting this one slide. Impetuous and over zealous criticism of research articles really gets to me.

I've already stated flaws that you seem to have omitted from your critique of my critique; or sorry, my "critique".

1. Selecting to choose which items to use in a scale will inflate a correlation - personally I take it as a red flag when I see researchers choosing which items to use, I will forever be wary when this is done and it's a big flaw in my book.

2. Deciding to base an entire study on how people spent one single day in their lives, this is also a big limitation in my book.

Can you please clarify why these are not issues? I'm actually genuinely hoping you might provide evidence or a reference why they aren't - I love being proven wrong (I'm being 100% genuine, I am not being sarcastic at all).

Valmont wrote: »

You missed the point that this study used a very large sample, over 13,000 individuals. The early studies did not have as large a sample. It would have been impractical for the initial studies in this area to use 13,000 individuals when they were not even sure what they would find. The early results allowed the present researchers to commit to a larger sample. The next step after extensive correlational evidence is to construct an experimental project, within which the question can be asked again and the putative presence of any causal factors can be deduced. Sometimes, without correlational studies, many experimental research questions may not have testable implications of the general hypothesis. In terms of this time-line of research development, the current study was more than adequate.

I already said that what they did is fine in terms of replication - this is not an issue I am contending.

Valmont wrote: »

Induction is an important part of the scientific method and as such the aggregate from many different experiments often forms the basis for a new theory. Rarely will you find one single piece of research opening up and polishing off a research area in one shot. So if you some studies seem pointless or pedantic to you, look closer.

My point is not that this research should never have been done - it's the methodology that I have issues with, as stated above. So this is also not something I am contending.

I don't understand your last sentence though, can you expand on "look closer" please.

Valmont wrote: »

By all means, keep analysing studies for methodological flaws but don't patronise the authors of this article for supposedly making basic mistakes when, ironically, it is you who are conflating correlation and causation in your criticism by implying that the former is useless without the latter. You have already proved your initial statement incorrect considering a misunderstanding of the purposes of a correlational study (stats 101 right?) seem to underpin your own "critique".

I would say the only person here being patronising is you.

I'll "criticise" and will continue to do so regardless of whether you take it to be "patronising" to the authors, but that's your decision to take it that way. Is part of research/science not to attempt to identify limitations that might have lead to research outcomes or do you just take what the results section says without checking the methodology? A lot of people don't bother with the method section when that's the heart of the study and will tell you the most about the research in my opinion.

Conflating the correlation-causation is about the only thing I'd partially concede, but only in terms of my statement "This whole study is based on correlation-causation"; that is how I chose to take the research and was not the goal of the researchers. It's been made clear in your posts that they didn't conclude anything and it was also in the original post too so believe me that wasn't missed, my last post was a reference to this - but again just because research might stimulate more research or progresses an area of research doesn't mean the original study had no flaws.

Could you show my where I stated this was “pointless” or “pedantic” please? I believe your personal desire to defend this might be clouding your judgement and reading of my argument – if you read my last post I clearly stated no issue with it in terms of replication.

I take a pessimistic attitude to research and I think it's actually a very prudent attitude to have - my first goal for everything I read is to find the flaws and I pray that my "impetuous" and "over zealous" criticism will stay with me forever

Chriskavo

Socially inept, depressed and highly introverted people spend alot of time on the internet. No **** Sherlock.

Not being smart but judging by the amount of posts you have made you could just as easily be referring to yourself. I know that I spend far too much time online but I do enjoy a social life and am happily married. I think sweeping remarks like that are stereotypical and by and large not true. I'm sure their is quite a lot more people out there who spend more time on the Internet but wont admit it.