Skip to main content
NCBI home page
Innovation in Aging logoLink to Innovation in Aging
. 2023 Mar 7;7(3):igad022. doi: 10.1093/geroni/igad022

Social Media Use May Not Be As Bad As Some Suggest: Implication for Older Adults

Shelia R Cotten 1,2,✉, Reza Ghaiumy Anaraky 3, Amy M Schuster 4
Editor: J Tina Savla
PMCID: PMC10168587  PMID: 37179658

Abstract

Background and Objectives

Social media use (SMU) has increased over the past decade among older adults. Cross-sectional studies report SMU is related to negative mental health outcomes (i.e., depression). Given depression is the most common mental health issue for older adults and increases the risk of morbidity and mortality, determining longitudinally whether SMU is related to increased depression is critical. This study examined the longitudinal relationship between SMU and depression.

Research Design and Methods

Data from 6 waves (2015–20) of the National Health and Aging Trends Study (NHATS) were analyzed. Participants included a nationally representative sample of U.S. older adults, aged 65 years and older (N = 7,057). We used a Random Intercept Cross-Lagged Panel Modeling (RI-CLPM) framework to examine the relationship between primary outcomes of SMU and depression symptoms.

Results

No pattern was found of SMU predicting depression symptoms, or depression symptoms predicting SMU. The major driver of SMU in each wave was SMU in the previous wave. On average, our model accounted for 3.03% of the variance in SMU. The major predictor of depression in each wave was the pre-existing depression. Our model accounted for an average of 22.81% of the variance in depressive symptoms.

Discussion and Implications

The results suggest that SMU and depressive symptoms are driven by the previous patterns of SMU and depression, respectively. We did not find any patterns of SMU and depression influencing each other. NHATS measures SMU with a binary instrument. Future longitudinal research should employ measures that account for SMU duration, type, and purpose. These findings suggest that SMU may not be related to negative health outcomes, in this case, depression, for older adults.

Keywords: Depression, Internet use, Social media effects, Technology

Translational Significance: Research on older adults’ social media use (SMU) and depressive symptoms has predominantly used cross-sectional data. Results are conflicting regarding whether a causal relationship exists and, if so, whether the association is positive or negative. Using a binary measure for SMU, we found that SMU does not increase or decrease depressive symptoms over time for older adults. SMU may not be as bad for older adults as some researchers and the media would suggest. Future longitudinal studies should use more comprehensive SMU measures that include duration, type, and purpose of use.

As technology use has proliferated across age groups, especially among older adults, predictions about the impacts of this use on society have also increased. In particular, social media use (SMU) has generated substantial debate as to its impacts on well-being, including depression. Social media is a type of digital platform (e.g., social networking sites, virtual communities, social gaming, video sharing) where users can distribute content, connect, and interact with other users online (Aichner et al., 2021). Social media has been described by some as a technology that is destroying society and the mental health of those who use it (Orlowski, 2020), whereas other scholars note that SMU has the potential to support social connections, regardless of geographic distance, and offset health constraints (Antonucci, Ajrouch, & Manalel, 2017).

Forty-five percent of community-dwelling older adults reported using social media in 2021, an increase from 37% in 2018 (Pew Research Center, 2022). Given the increasing rate of SMU among older adults (Auxier & Anderson, 2021), understanding the impacts of this use on mental health, especially depression, is critical. Depression is the most prevalent mental health issue among older adults (Helvik et al., 2016; Kok & Reynolds, 2017; Orlowski, 2020) and increases older adults’ risk for morbidity (Taylor, 2014) and mortality (Vanoh et al., 2016). Yet, symptoms of depression are often undiagnosed in older adults (Kok & Reynolds, 2017) as they are thought to be a normal part of aging (e.g., due to chronic illness and loss; NAMI, 2009). Given the potential seriousness of depression and the increasing SMU among older adults, in this study, we examine the association between SMU and older adults’ risk of depression.

Social Media Use by Older Adults

Approximately 73% of community-dwelling individuals in the United States aged 50–64 years and almost half (45%) of those 65 years and older report using social media (Auxier & Anderson, 2021). Older adults who use social media do so several times a week, typically; however, many use it daily (Aarts et al., 2015; Auxier & Anderson, 2021). For smartphone-using older adults, SMU is reported to be the main activity on their devices (Busch et al., 2021). However, the frequency of SMU has been found to decline with age. For example, 50- to 64-year-old survey respondents reported higher SMU for different platforms such as YouTube (83%), Facebook (73%), and Pinterest (38%) compared to those 65 years and older who reported less SMU for YouTube (49%), Facebook (50%), and Pinterest (18%; Auxier & Anderson, 2021).

Older adults’ SMU is mainly driven by their desire to meet their social and emotional needs (Hülür & Macdonald, 2020). For example, SMU provides older adults with another way to connect with family and friends (Jung et al., 2017; Sinclair & Grieve, 2017). Older adults who use social media have been found to have increased social support and greater life satisfaction (Bell et al., 2013; Khoo & Yang, 2020; Kim & Shen, 2020; Leist, 2013; Sala et al., 2021). In addition, older adults who use social media feel more socially connected (Sinclair & Grieve, 2017). Moreover, increased daily SMU is associated with a more positive mood for older adults with smaller social networks (Kim & Fingerman, 2022).

Social Media Use and Depressive Symptoms

Two challenges exist in prior research regarding the association between SMU and depressive symptoms in older adults: determining (1) whether a causal relationship exists between SMU and depression and (2) whether the association between SMU and depression is positive or negative (Ang & Chen, 2019; Cotten, Schuster, & Seifert, 2021; Fu & Xie, 2021; Hofer & Hargittai, 2021; Kouvonen et al., 2021; Meshi, Cotten, & Bender, 2020). First, the association between SMU and depression is unclear. Some studies found no association between SMU and depressive symptoms for older adults (Guzman & Mentes, 2017; Schwaba & Bleindorn, 2021; Sharifian et al., 2021). Other studies, however, reported an association between SMU and depressive symptoms, suggesting that SMU led to depressive symptoms (Cenko, Kaufman, & Manini, 2021; Fu & Xie, 2021; Lewin et al., 2023; Smith et al., 2021). In contrast, Kouvonene and colleagues (2021) reported an association with depressive symptoms leading to SMU. Regardless of the associations reported between SMU and depressive symptoms, most of the studies used a cross-sectional study design, which prevents the establishment of whether a causal relationship actually exists between SMU and depressive symptoms (Cenko, Kaufman, & Manini, 2021; Fu & Xie, 2021; Guzman & Mentes, 2017; Hofer & Hargittai, 2021; Kouvonen et al., 2021; Lewin et al., 2023; Sharifian et al., 2021; Smith et al., 2021). Ang and Chen (2019) employed a longitudinal design, though they used only two waves (2013–14) of data. They reported that SMU was protective against the negative effects of pain on depression. However, scholars argue that having only two waves of data is insufficient in longitudinal designs and may lack credibility (Ployhart & MacKenzie, 2014). Our study is unique as we seek to address this first conflict, the causal relationship between SMU and depression among older adults, by analyzing six waves of the National Health and Aging Trends Study (NHATS) data.

The second challenge is that prior research does not reach a consensus as to whether the association between SMU and depressive symptoms is positive or negative. Studies have reported a negative association between SMU and depression (e.g., Fu & Xie, 2021; Kouvonen et al., 2021; Yang et al., 2021). Fu and Xie (2021) analyzed data from a 2017 survey of Chinese older adults and found that higher SMU was associated with having lower levels of depression. Kouvonen et al. (2021) studied data from the Care, Health and Ageing of Russian-speaking Minority (CHARM) study and reported that a lower likelihood of SMU was associated with having higher depressive symptoms. Yang and colleagues (2021) surveyed Hong Kong residents during the height of cornonavirus 2019 (May to June 2020) and reported that for older adults, more SMU was linked to lower levels of depression. Conversely, other research reports a positive association between SMU and depression (e.g., Perlis et al., 2021). Perlis et al. (2021) reported a positive association between two specific types of SMU (i.e., TikTok and Snapchat) and depressive symptoms. However, their sample was not solely limited to an older adult population (mean age = 55.8).

The ambiguity as to whether the association between SMU and depressive symptoms is positive or negative could also be due to the SMU measures used in the studies, which analyze overall use and not the type or purpose of use. Studies analyzing type of use reported that active SMU (e.g., posting) was positively associated with depressive symptoms (Chhatwani, Kumar Mishra, & Rai, 2023; Lewin et al., 2023), whereas passive SMU (e.g., browsing) was negatively associated with depressive symptoms (Lewin et al., 2022). Hofer and Hargittai (2021) analyzed purpose of use and found that older adults using social media were more prone to check in on someone who was suddenly missing from an online group and older adults who were more likely to answer questions in a social media forum reported higher depressive symptoms.

Overall, the literature lacks evidence on whether a causal relationship exists between SMU and depression among older adults and, if so, the direction of the causal relationship (Cotten et al., 2021). Given the unclear relationship between SMU and depression and the need for longitudinal designs in this area of research, we seek to determine if there is a causal relationship between SMU and depressive symptoms among older adults (Challenge #1) using six waves (2015–20) of data from the NHATS. In addition, we sought to determine whether there was a positive or negative association between SMU and depressive symptoms (Challenge #2). Addressing these questions can help us hypothesize how SMU and depressive symptoms influence one another and enable us to provide better insights on whether and how we can use technology to support older adults’ mental well-being.

Method

Data and Study Sample

This longitudinal study included data from the NHATS which annually surveys a nationally representative sample of individuals 65 years and older to understand trends in late-life functioning (Freedman & Kasper, 2019). We used data from Waves 5–10 (2015–20). These six waves were selected as (1) the variable assessing the use of online social networking sites became first available in Wave 3 (2013), and (2) the sample was replenished in Wave 5 (2015) with 4,182 participants to account for individuals who were no longer participating because the original sample of 8,245 in Wave 1 (2011). In this study, we included participants who were present in Wave 5. Those who were living in nursing homes or responses that were recorded by a proxy were also removed (1,277 individuals). Our sample includes 7,057 older adult respondents, although this number decreased in the consequent waves as some participants dropped out or passed away.

Measures

SMU was assessed with one-item measure: In the last month have you gone on the Internet or online to visit social network sites? The survey defined social media as websites individuals use to keep in touch with their friends, like Facebook. Response options included 1 (yes) and 0 (no). The Patient Health Questionnaire-2 (PHQ-2) was used in the NHATS to assess depressive symptoms (Kroenke et al., 2003) with a two-item measure: Over the last month, how often have you felt down, depressed, or hopeless? Over the last month, how often have you had little interest or pleasure in doing things? Response options on a 4-point Likert scale ranged from 0 (not at all) to 3 (nearly every day). We used a sum score of these two items ranging from 0 to 6 as the PHQ-2 variable. Furthermore, we used several control variables. NHATS measures the limitations in activities of daily living (ADLs) by asking the extent of difficulty respondents experienced in using the toilet, eating, getting dressed, and showering or washing up by themselves and without help during the last month. The responses were recorded with a four-item Likert scale: none, a little, some, or a lot. We used the sum score of these four measures to control for limitations in ADLs. In addition, NHATS records a self-reported health variable by asking about participants’ general health with a Likert scale of four items: excellent, very good, good, fair, or poor. We used this variable as a continuous variable with higher scores equating to better health. Lastly, participants reported sociodemographic information including age, gender, race, marital status, and education. Age (65–107 years) was measured as a continuous variable. Gender (1 [male], 0 [female]), race (White [reference category], Black, Other), and marital status (0 [never married, widowed, or divorced], 1 [living with partner or married]) were dummy coded. Education was measured from no school to a graduate degree, for which we used a continuous variable with a higher value denoting higher education.

Analytic Strategy

To examine the relationship between SMU and depression over time, we used a Random Intercept Cross-Lagged Panel Model (RI-CLPM) framework (Hamaker et al., 2015; Lucas, 2023). The RI-CLPM includes autoregressive effects and cross-lagged effects. Autoregressive effects allow the state of a unit to be a function of its past state. Cross-lagged effects assess whether one variable at time t can predict another variable at time t + 1. In addition, the RI-CLPM can show whether deviations of a variable at a within-person level will result in changes in another variable or not (e.g., if individuals who start using social media will experience a change in their risk of depression). Many scholars have adopted this approach when studying SMU (e.g., Orben et al., 2019; van der Schuur et al., 2019). For example, van der Schuur et al. (2019) used this methodology and found that SMU frequency does not significantly predict sleepiness, whereas the stress resulting from SMU predicts subsequent sleepiness. In our case, we studied whether SMU and depressive symptoms at time t + 1 are a function of SMU and depressive symptoms at time t (see Figure 1). All the analyses were conducted in R v4.2 and Lavaan package.

Figure 1.

Figure 1.

Open in a new tab

Random Intercept Cross-Lagged Panel Modeling (RI-CLPM) results for relationship between social media use and depressive symptoms across six waves*. PHQ-2 = Patient Health Questionnaire-2; SMU = social media use. *To keep Figure 1 readable, we did not include the control variables in the figure. See Tables 2 and 3 for the full results with the control variables.

Results

Participants in the 2015 wave were, on average, 77.76 years old (standard deviation [SD] 7.62). The majority of the sample was White (81.4%), female (53.3%), had at least a high school education (68.97%), and was single (50.60%). Furthermore, the average scores for limitations in ADLs and the self-reported overall health were 0.33 (SD 0.76) and 3.23 (SD 1.05), respectively. Of the dependent variables, the average score for PHQ-2 was 0.89 (SD 1.30), and there were 1,530 social media users in the sample. Table 1 reports the descriptive statistics across all the study waves.

Table 1:

Demographic Characteristics of the Sample

Variable 2015 2016 2017 2018 2019 2020
(N = 7,057) (N = 5,897) (N = 5,161) (N = 4,589) (N = 4,111) (N = 3,524)
Gender, N (%)
 Female 4,060 (57.50) 3,406 (55.76) 2,969 (57.53) 2,650 (57.75) 2,384 (57.99) 2,042 (57.95)
 Male 2,997 (42.50) 2,491 (42.24) 2,192 (42.47) 1,939 (42.25) 1,727 (42.01) 1,482 (42.05)
Race, N (%)
  White 4,867 (68.97) 4,150 (70.37) 3,633 (70.39) 3,255 (70.93) 2,926 (71.17) 2,553 (72.44)
  Black 1,431 (20.28) 1,177 (19.95) 1,045 (20.24) 913(19.89) 817 (19.87) 689 (19.55)
 Other 759 (10.76) 570 (9.66) 483 (9.35) 421 (9.17) 368 (8.95) 282 (8.80)
Living with partner, N (%) 3,489 (49.40) 2,877 (48.78) 2,475 (47.95) 2,171 (47.30) 1,915 (46.58) 1,631 (46.28)
SMU, N (%)
 User 1,530 (44.77) 1,419 (49.88) 1,292 (51.49) 1,190 (52.10) 1,115 (53.12) 1,103 (56.56)
 Nonuser 1,741 (53.23) 1,426 (50.12) 1,217 (48.51) 1,094 (47.90) 984 (46.88) 847 (43.44)
Age, M (SD) 77.76 (7.62) 78.46 (7.44) 79.10 (7.20) 79.74 (6.99) 80.34 (6.79) 80.75 (6.48)
Education, M (SD) 4.06 (1.73) 4.08 (1.73) 4.10 (1.73) 1.73 (4.12) 4.13 (1.73) 4.18 (1.74)
Overall health, M (SD) 3.23 (1.05) 2.72 (1.53) 2.36 (1.68) 2.10 (1.75) 1.86 (1.75) 1.63 (1.77)
Limitations in ADLs, M (SD) 0.33 (0.76) 0.34 (0.78) 0.36 (0.80) 0.37 (0.81) 0.41 (0.86) 0.40 (0.86)
PHQ-2, M (SD) 0.89 (1.30) 0.88 (1.28) 0.89 (1.30) 0.85 (1.24) 0.85 (1.25) 0.95 (1.32)
Open in a new tab

Notes: ADLs = activities of daily living; PHQ-2 = Patient Health Questionnaire-2; SMU = social media use. The PHQ-2 measured depression symptoms. N’s vary depending on the level of missing data across the variables.

There were minimal differences found between the participants who used social media compared to those who did not. Both were, on average, the same age (74.0 years, SD = 6.0 SMU; 75.7 years, SD = 6.6 non-SMU), mainly White (82.3% SMU; 80.3% non-SMU), and had similar levels of education (41.9% SMU; 42.3% non-SMU—at least a bachelor’s degree). Although the participants who used social media were predominantly female (60.9%), the nonsocial media using participants were mostly male (53.0%).

We fit an RI-CLPM model controlling for age, gender, marital status, race, limitations in ADLs, and the self-reported measure of overall health. The model fit indices, shown in Table 2, show an overall acceptable fit (χ2(217) = 615.147 p <.001, comparative fit index (CFI) = 0.970, Tucker–Lewis index (TLI) = 0.953, root mean square error of approximation (RMSEA) = 0.016, 90% confidence interval [CI: 0.015, 0.017], standardized root mean squared residual (SRMR):0.035; Ghaiumy Anaraky et al., 2021; Hu & Bentler, 1999). Furthermore, the RI-CLPM does not show any cross-lagged causal relationship patterns between SMU and depression. This means that the effects of depressive symptoms at time t on SMU at time t + 1 are not consistently significant (ps > .05). In addition, our model does not suggest any consistent effects of SMU at time t, on depressive symptoms at time t + 1 (ps > .05). However, we found several significant autoregressive effects where depressive symptoms and SMU at time t can predict depressive symptoms and SMU at time t + 1, respectively (ps < .001).

Table 2:

The Fit Indices for Constrained and Unconstrained Models

Model χ2 CFI/TLI RMSEA 90% CI [LL, UL] SRMR
Unconstrained model χ2(217) = 615.147, p < .001 0.970/0.953 0.016 [0.015, 0.017] 0.035
Constrained Model χ2(265) = 681.250, p < .001 0.969/0.960 0.015 [0.014, 0.016] 0.035
Open in a new tab

Note: CI = confidence interval, LL = lower limit, UL = upper limit.

We also conducted a second model, constraining the lagged and autoregressive effects (Hamaker et al., 2015). Similar to the unconstrained model, this model shows an overall acceptable fit (χ2(265) = 681.250, p < .001, CFI = 0.969, TLI = 0.960, RMSEA = 0.015, 90% CI: [0.014, 0.016], SRMR = 0.035; Ghaiumy Anaraky et al., 2021; Hu & Bentler, 1999). The constrained and unconstrained models are not significantly different, as ∆CFI < 0.010, ∆RMSEA < 0.015, and ∆SRMR < 0.030 (Chen, 2007). Therefore, we accept the constrained model and only report this model. However, the overall results of the constrained model are not different than the unconstrained model, suggesting no significant cross-lagged effects and significant autoregressive effects (ps < .001). In addition, PHQ and SMU random intercepts are not significantly correlated (B = 0.008, p = .292). We present these results in Figure 1, Table 3 (for PHQ-2 DV), and Table 4 (for SMU DV).

Table 3:

The Full Random Intercept Cross-Lagged Panel Modeling (RI-CLPM) Results for PHQ-2.

Variable 2015 2016 2017 2018 2019 2020
β SE p Value β SE p Value β SE p Value β SE p Value β SE p Value β SE p Value
Between level
 Age −0.006 0.002 0.012 −0.007 0.002 0.003 −0.004 0.003 0.182 −0.002 0.003 0.554 −0.002 0.003 0.512 0.001 0.004 0.698
 Gender (vs female) −0.085 0.034 0.013 −0.099 0.037 0.008 −0.107 0.038 0.004 −0.101 0.038 0.008 −0.183 0.042 <.001 −0.119 0.046 0.010
 Race: Black (vs White) −0.007 0.045 0.873 0.010 0.051 0.849 0.070 0.051 0.174 0.109 0.053 0.040 0.007 0.055 0.902 0.040 0.061 0.511
 Race: Other (vs White) 0.105 0.058 0.072 0.105 0.062 0.089 0.116 0.067 0.085 0.001 0.067 0.991 0.079 0.072 0.271 0.081 0.094 0.389
 Education −0.033 0.010 0.001 −0.043 0.010 <.001 −0.049 0.011 <.001 −0.031 0.010 0.003 −0.047 0.011 <.001 −0.039 0.013 0.003
 Living with partner (vs living alone) −0.253 0.038 <.001 −0.220 0.039 <.001 −0.168 0.040 <.001 −0.161 0.040 <.001 −0.126 0.044 0.004 −0.148 0.049 0.003
Within level
 Limitations in ADLs 0.284 0.015 <.001 0.284 0.015 <.001 0.284 0.015 <.001 0.284 0.015 <.001 0.284 0.015 <.001 0.284 0.015 <.001
 Overall health −0.284 0.011 <.001 −0.284 0.011 <.001 −0.284 0.011 <.001 −0.284 0.011 <.001 −0.284 0.011 <.001 −0.284 0.011 <.001
 PHQ-2 (prior wave) — — — 0.126 0.014 <.001 0.126 0.014 <.001 0.126 0.014 <.001 0.126 0.014 <.001 0.126 0.014 <.001
 SMU (prior wave) — — — −0.016 0.047 0.739 −0.016 0.047 0.739 −0.016 0.047 0.739 −0.016 0.047 0.739 −0.016 0.047 0.739
Open in a new tab

Notes: ADLs = activities of daily living; PHQ = Patient Health Questionnaire; SE = standard error; SMU = social media use.

Table 4:

The Full Random Intercept Cross-Lagged Panel Modeling (RI-CLPM) Results for Social Media Use

2015 2016 2017 2018 2019 2020
β SE p Value β SE p Value β SE p Value β SE p Value β SE p Value β SE p Value
Between level
 Age −0.010 0.002 <.001 −0.010 0.002 <.001 −0.009 0.002 <.001 −0.012 0.002 <.001 −0.010 0.002 <.001 −0.009 0.002 <.001
 Gender (vs female) −0.142 0.021 <.001 −0.135 0.021 <.001 −0.098 0.022 <.001 −0.103 0.023 <.001 −0.115 0.024 <.001 −0.135 0.025 <.001
 Race: Black (vs White) −0.108 0.031 <.001 −0.114 0.032 <.001 −0.068 0.034 0.048 −0.082 0.036 0.022 −0.147 0.036 <.001 −0.131 0.037 <.001
 Race: Other (vs White) −0.048 0.037 0.196 −0.071 0.040 0.079 −0.092 0.043 0.032 −0.068 0.044 0.126 −0.050 0.045 0.264 −0.062 0.047 0.187
 Education 0.005 0.006 0.426 −0.005 0.006 0.370 −0.006 0.006 0.345 −0.017 0.006 0.007 −0.002 0.006 0.723 −0.002 0.007 0.791
 Living with partner (vs living alone) 0.027 0.020 0.183 0.036 0.021 0.076 0.019 0.022 0.377 −0.022 0.021 0.310 0.010 0.023 0.649 0.020 0.023 0.391
Within level
 Limitations in ADLs −0.005 0.008 0.543 −0.005 0.008 0.543 −0.005 0.008 0.543 −0.005 0.008 0.543 −0.005 0.008 0.543 −0.005 0.008 0.543
 Overall health 0.006 0.006 0.293 0.006 0.006 0.293 0.006 0.006 0.293 0.006 0.006 0.293 0.006 0.006 0.293 0.006 0.006 0.293
 PHQ-2 (prior wave) — — — −0.006 0.006 0.364 −0.006 0.006 0.364 −0.006 0.006 0.364 −0.006 0.006 0.364 −0.006 0.006 0.364
 SMU (prior wave) — — — 0.185 0.024 <.001 0.185 0.024 <.001 0.185 0.024 <.001 0.185 0.024 <.001 0.185 0.024 <.001
Open in a new tab

Notes: ADLs = activities of daily living; PHQ = Patient Health Questionnaire; SE = standard error; SMU = social media use.

Discussion

Media hype and some researchers would have us believe that social media is destroying society, particularly in terms of negative mental health consequences. We investigated this issue among older adults as depression is the most common mental health issue for older adults and increases the risk of morbidity and mortality. In this study, we attempted to address two challenges in the literature: (1) whether a causal association exists between older adults’ SMU and their depressive symptoms, and (2) the association between older adults’ SMU and depressive symptoms (i.e., if there is a positive or negative correlation between the two measures). We utilized six waves (2015–20) of NHATS data, with a nationally representative sample of U.S. older adults aged 65 years and older, to examine whether a causal relationship exists between SMU and depressive symptoms. Due to the mixed results in the literature as to the relationship between SMU and depression for older adults (see Cotten et al., 2021, review article), we did not pose specific hypotheses at the onset of this study. However, knowing the causal direction in the interplay between SMU and depressive symptoms suggests different implications to stave off depression and the negative effects of depression. For example, if SMU influences depressive symptoms, it is crucial to identify use patterns that lead to depressive symptoms and encourage practices that can mitigate depressive symptoms. On the other hand, if depression symptoms or risk affects SMU, we may be able to use SMU data as a means of depressive symptoms diagnosis.

Our results did not show evidence of a causal relationship between SMU and depressive symptoms among older adults. SMU does not increase or decrease older adults’ depressive symptoms and resulting depressive symptoms over time, nor do depressive symptoms affect SMU over time. These findings were consistent with previous cross-sectional studies (Guzman & Mentes, 2017; Sharifian et al., 2021). However, these findings do not support the limited prior research that suggests causal relationships between SMU and depression (Fu & Xie, 2021; Kouvonen et al., 2021; Lewin et al., 2022; Vuorre et al., 2021).

Although our results do not suggest any causal relationship between SMU and depression among the NHATS sample of older adults included in this study, we cannot refute the possibility of any causal relationship between SMU and depressive symptoms in general terms. For example, if we assume that SMU influences depressive symptoms for the general population, our results imply that older adults are not using social media in a manner that either puts them at risk for depression or that protects them from depression. For example, previous research shows that SMU can promote well-being when older adults use social media to connect with social ties, exchange social support, and stay connected with the lives of their friends and family members (Casanova et al., 2021; Sinclair & Grieve, 2017). Not finding a negative causal relationship between SMU and depression perhaps suggests that older adults do not predominantly use social media for these activities or that they do not use them frequently enough to yield an impact on depression. Thus, they do not receive depression-mitigating gratifications. Unfortunately, our data do not allow us to examine either of these postulates.

In addition to studying the causal relationship between SMU and depressive symptoms across consecutive waves, we examined the correlation between SMU and depressive symptoms in the same waves. The introduction section outlined this point as the second challenge in the literature, where some studies report a positive association and others report negative associations between SMU and depressive symptoms. Our results did not suggest any significant and consistent correlations between SMU and depressive symptoms. Figure 1 shows the only significant association between SMU and depressive symptoms occurred in the 2017 wave of data, suggesting that higher SMU in that particular wave was associated with having lower depressive symptoms. However, this effect was not significant in other waves.

We suspect the lack of significant findings may be due to the cursory way that SMU was assessed in NHATS. SMU was simply operationalized through a binary variable of use versus no use. The problem with this operationalization is that survey respondents are placed into the category of social media user, whether they use social media almost constantly or use it only once a month. In an ideal situation, however, researchers need measures assessing the frequency, type, amount, and purpose of SMU. This can help the model account for more refined use patterns, making us more confident in interpreting these results. Though we suggest that more refined measures are needed, some prior studies, though not focused on older adults, that used a better instrument for measuring SMU also did not find any association between SMU and depressive symptoms. For example, Morin-Major and colleagues (2016) did not find any significant association between adolescents’ Facebook-use frequency and depressive symptoms in a cross-sectional study. Frison and colleagues (2016) did not find any association between adolescents’ negative Facebook-use experiences and depressive symptoms in their two-wave longitudinal study. Although these studies do not focus on the older adult population, their results are in line with the present study. Future studies should include more detailed SMU measures to allow for more robust analysis and interpretation. Even future waves of the NHATS can replace this coarse social media measure with a question measuring the frequency of SMU on a Likert scale (with 0 indicating no use) or estimates in hours of daily or weekly SMU.

Although we did not find consistent causal associations between SMU and depressive symptoms, we found depressive symptoms and SMU reinforced themselves over time (Slater, 2007). This result aligns with the literature depicting SMU as a habit (Hu et al., 2018; Soror et al., 2021) because SMU in a particular wave is the best predictor of subsequent SMU. We found similar habit-wise patterns relating to depressive symptoms. In a particular wave, the best predictor of depressive symptoms is the prior depressive symptoms measure.

Limitations

Although this is the most rigorous study to date that has examined the longitudinal nature of the relationship between SMU and depression in the older adult population, there are limitations to note. A significant limitation is the way that SMU is measured in the NHATS study. The SMU measure only assessed use within the last month, and not the frequency, duration, type, or purpose of use, or which social media sites are used, all of which Cotten and colleagues have noted are important when examining SMU among older adults (Cotten et al., 2021). In addition, depression symptoms are operationalized using a very brief and short screener (PHQ-2). Future research should aim for a more holistic understanding of SMU as well as examine a range of well-being outcomes and how they may change over time. Furthermore, the NHATS survey is administered on a yearly basis. Closing the gaps between the waves (e.g., measuring the variables on a monthly or more frequent basis), though this would be very costly, can help to identify when, how, and under what circumstances SMU and depression may be affecting each other over time.

In addition, the NHATS data set is comprised of self-reported responses. Some research suggests that technology use measures may be prone to recall bias. The validity of self-reported data can be threatened by recall bias (Deng et al., 2019; Ellis et al., 2019). Future studies can use applications that track social media usage on participants’ personal computers and smartphones to avoid recall biases in measuring SMU, though using these applications is more invasive and costlier in terms of study administration.

Conclusion

This longitudinal study examines the causal relationships between SMU and one aspect of mental health—depression. Although it may be the case that SMU can help older adults stay connected with their social ties, SMU does not connote either positive or negative impacts on depressive symptoms over time in this NHATS sample of older adults. These findings suggest that popular press reports and cross-sectional studies of the harmful effects of SMU on mental health may be exaggerated, for older adults at least. However, our lack of findings may be due to the coarse measurement instruments included in the NHATS surveys for SMU and depression. Future research should study the relationship between SMU and depression with more granular instruments that capture more nuanced data from respondents.

Contributor Information

Shelia R Cotten, Department of Sociology, Anthropology, and Criminal Justice, Clemson University, Clemson, South Carolina, USA; Department of Communication, Clemson University, Clemson, South Carolina, USA.

Reza Ghaiumy Anaraky, Department of Technology Management and Innovation, Tandon School of Engineering, New York University, New York City, New York, USA.

Amy M Schuster, Department of Sociology, Anthropology, and Criminal Justice, Clemson University, Clemson, South Carolina, USA.

Funding

None declared.

Conflict of Interest

None declared.

References

  1. Aarts, S., Peek, S. T. M., & Wouters, E. J. M. (2015). The relation between social network site usage and loneliness and mental health in community-dwelling older adults. International Journal of Geriatric Psychiatry, 30(9), 942–949. doi: 10.1002/gps.4241 [DOI] [PubMed] [Google Scholar]
  2. Aichner, T., Grünfelder, M., Maurer, O., & Jegeni, D. (2021). Twenty-five years of social media: A review of social media applications and definitions from 1994 to 2019. Cyberpsychology, Behavior, and Social Networking, 24(4), 215–222. doi: 10.1089/cyber.2020.0134 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Ang, S., & Chen, T. Y. (2019). Going online to stay connected: Online social participation buffers the relationship between pain and depression. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 74(6), 10S20–1031. doi: 10.1093/geronb/gby109 [DOI] [PubMed] [Google Scholar]
  4. Antonucci, T. C., Ajrouch, K. J., & Manalel, J. A. (2017). Social relations and technology: Continuity, context, and change. Innovation in Aging, 1(3), igx029. doi: 10.1093/geroni/igx029 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Auxier, B, & Anderson, M. (2021). Social media use in 2021. https://www.pewresearch.org/internet/2021/04/07/social-media-use-in-2021/
  6. Bell, C., Fausset, C., Farmer, S., Nguyen, J., Harley, L., & Fain, W. B. (2013, May). Examining social media use among older adults. In Proceedings of the 24th ACM conference on hypertext and social media (pp. 158–163). doi: 10.1145/2481492.2481509 [DOI]
  7. Busch, P. A., Hausvik, G. I., Ropstad, O. K., & Pettersen, D. (2021). Smartphone usage among older adults. Computers in Human Behavior, 121, 106783. doi: 10.1016/j.chb.2021.106783 [DOI] [Google Scholar]
  8. Casanova, G., Zaccaria, D., Rolandi, E., & Guaita, A. (2021). The effect of information and communication technology and social networking site use on older people’s well-being in relation to loneliness: review of experimental studies. Journal of Medical Internet Research, 23(3), e23588. doi: 10.2196/23588 [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Cenko, E., Kaufmann, C., & Manini, T. (2021). COVID-19 related media consumption and mental health in older adults. Innovation in Aging, 5(Suppl 1), 1006. doi: 10.1093/geroni/igab046.3581 [DOI] [Google Scholar]
  10. Chen, F. F. (2007). Sensitivity of goodness of fit indexes to lack of measurement invariance. Structural Equation Modeling: A Multidisciplinary Journal, 14(3), 464–504. doi: 10.1080/10705510701301834 [DOI] [Google Scholar]
  11. Chhatwani, M., Kumar Mishra, S., & Rai, H. (2023). Active and passive social media usage and depression among the elderly during COVID-19: does race matter? Behaviour & Information Technology, 42(2), 215–226. doi: 10.1080/0144929X.2022.2045359 [DOI] [Google Scholar]
  12. Cotten, S. R., Schuster, A. M., & Seifert, A. (2021). Social media use and well-being among older adults. Current Opinion in Psychology, 45, 101293. doi: 10.1016/j.copsyc.2021.12.005 [DOI] [PubMed] [Google Scholar]
  13. Deng, T., Kanthawala, S., Meng, J., Peng, W., Kononova, A., Hao, Q., Zhang, Q., & David, P. (2019). Measuring smartphone usage and task switching with log tracking and self-reports. Mobile Media & Communication, 7(1), 3–23. doi: 10.1177/2050157918761491 [DOI] [Google Scholar]
  14. Ellis, D. A., Davidson, B. I., Shaw, H., & Geyer, K. (2019). Do smartphone usage scales predict behavior? International Journal of Human–Computer Studies, 130, 86–92. doi: 10.1016/j.ijhcs.2019.05.004 [DOI] [Google Scholar]
  15. Freedman, V. A., & Kasper, J. D. (2019). Cohort profile: The National Health and Aging Trends Study (NHATS). International Journal of Epidemiology, 48(4), 1044–1045. doi: 10.1093/ije/dyz109 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Frison, E., Subrahmanyam, K., & Eggermont, S. (2016). The short-term longitudinal and reciprocal relations between peer victimization on Facebook and adolescents’ well-being. Journal of Youth and Adolescence, 45(9), 1755–1771. doi: 10.1007/s10964-016-0436-z [DOI] [PubMed] [Google Scholar]
  17. Fu, L., & Xie, Y. (2021). The effects of social media use on the health of older adults: An empirical analysis based on 2017 Chinese general social survey. Healthcare, 9, 1143. doi: 10.3390/healthcare9091143 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Ghaiumy Anaraky, R., Li, Y., & Knijnenburg, B. (2021). Difficulties of measuring culture in privacy studies. Proceedings of the ACM on Human–Computer Interaction, 5(CSCW2), 1–26. doi: 10.1145/347952236644216 [DOI] [Google Scholar]
  19. Guzman, A. A., & Mentes, J. C. (2017). Depression and social media use among middle aged and older us adults. Innovation in Aging, 1(Suppl 1), 1179. doi: 10.1093/geroni/igx004.4296 [DOI] [Google Scholar]
  20. Hamaker, E. L., Kuiper, R. M., & Grasman, R. P. (2015). A critique of the cross-lagged panel model. Psychological Methods, 20(1), 102–116. doi: 10.1037/a0038889 [DOI] [PubMed] [Google Scholar]
  21. Helvik, A. S., Bjørkløf, G. H., Corazzini, K., Selbæk, G., Laks, J., Østbye, T., & Engedal, K. (2016). Are coping strategies and locus of control orientation associated with health-related quality of life in older adults with and without depression? Archives of Gerontology and Geriatrics, 64, 130–137. doi: 10.1016/j.archger.2016.01.014 [DOI] [PubMed] [Google Scholar]
  22. Hofer, M., & Hargittai, E. (2021). Online social engagement, depression, and anxiety among older adults. New Media and Society, 1–18. doi: 10.1177/14614448211054377 [DOI] [Google Scholar]
  23. Hofer, M., & Hargittai, E. (2021). Online social engagement, depression, and anxiety among older adults. New Media and Society, 1–18. doi: 10.1080/10705519909540118 [DOI] [Google Scholar]
  24. Hu, T., Stafford, T. F., Kettinger, W. J., Zhang, X. P., & Dai, H. (2018). Formation and effect of social media usage habit. Journal of Computer Information Systems, 58(4), 334–343. doi: 10.1080/08874417.2016.1261378 [DOI] [Google Scholar]
  25. Hülür, G., & Macdonald, B. (2020). Rethinking social relationships in old age: Digitalization and the social lives of older adults. American Psychologist, 75(4), 554–566. doi: 10.1037/amp0000604 [DOI] [PubMed] [Google Scholar]
  26. Jung, E. H., Walden, J., Johnson, A. C., & Sundar, S. S. (2017). Social networking in the aging context: Why older adults use or avoid Facebook. Telematics and Informatics, 34(7), 1071–1080. doi: 10.1016/j.tele.2017.04.015 [DOI] [Google Scholar]
  27. Khoo, S. S., & Yang, H. (2020). Social media use improves executive functions in middle-aged and older adults: A structural equation modeling analysis. Computers in Human Behavior, 111, 106388. doi: 10.1016/j.chb.2020.106388 [DOI] [Google Scholar]
  28. Kim, C., & Shen, C. (2020). Connecting activities on Social Network Sites and life satisfaction: A comparison of older and younger users. Computers in Human Behavior, 105, 106222. doi: 10.1016/j.chb.2019.106222 [DOI] [Google Scholar]
  29. Kim, Y. K., & Fingerman, K. L. (2022). Daily social media use, social ties, and emotional well-being in later life. Journal of Social and Personal Relationships, 0, 1–20. doi: 10.1177/02654075211067254 [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Kok, R. M., & Reynolds, C. F. (2017). Management of depression in older adults: A review. Journal of the American Medical Association, 317(20), 2114–2122. doi: 10.1001/jama.2017.5706 [DOI] [PubMed] [Google Scholar]
  31. Kouvonen, A., Kemppainen, L., Ketonen, E. L., Kemppainen, T., Olakivi, A., & Wrede, S. (2021). Digital information technology use, self-rated health, and depression: Population-based analysis of a survey study on older migrants. Journal of Medical Internet Research, 23(6), e20988. doi: 10.2196/20988 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Kroenke, K., Spitzer, R. L., & Williams, J. B. (2003). The Patient Health Questionnaire-2: Validity of a two-item depression screener. Medical Care, 1284–1292. [DOI] [PubMed] [Google Scholar]
  33. Leist, A. K. (2013). Social media use of older adults: A mini-review. Gerontology, 59(4), 378–384. doi: 10.1159/000346818 [DOI] [PubMed] [Google Scholar]
  34. Lewin, K. M., Meshi, D., Schuster, A. M. & Cotten, S. R. (2023) Active and passive social media use are differentially related to depressive symptoms in older adults. Aging & Mental Health, 27(1), 176–183. doi: 10.1080/13607863.2022.2068133 [DOI] [PubMed] [Google Scholar]
  35. Lucas, R. E. (2023). Why the cross-lagged panel model is almost never the right choice. Advances in Methods and Practices in Psychological Science, 6(1), 1–22. doi: 10.1177/25152459231158378 [DOI] [Google Scholar]
  36. Meshi, D., Cotten, S. R., & Bender, A. R. (2020). Problematic social media use and perceived social isolation in older adults: A cross-sectional study. Gerontology, 66(2), 160–168. doi: 10.1159/000502577 [DOI] [PubMed] [Google Scholar]
  37. Morin-Major, J. K., Marin, M. F., Durand, N., Wan, N., Juster, R. P., & Lupien, S. J. (2016). Facebook behaviors associated with diurnal cortisol in adolescents: Is befriending stressful? Psychoneuroendocrinology, 63, 238–246. doi: 10.1016/j.psyneuen.2015.10.005 [DOI] [PubMed] [Google Scholar]
  38. National Alliance on Mental Illness [NAMI]. (2009). Depression in older persons: Fact sheet. NAMI. [Google Scholar]
  39. Orben, A., Dienlin, T., & Przybylski, A. K. (2019). Social media’s enduring effect on adolescent life satisfaction. Proceedings of the National Academy of Sciences, 116(21), 10226–10228. doi: 10.1073/pnas.1902058116 [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Orlowski, J. (2020, September 27). We need to rethink social media before it’s too late. We’ve accepted a Faustian Bargain. The Guardian. https://www.theguardian.com/commentisfree/2020/sep/27/social-dilemma-media-facebook-twitter-society [Google Scholar]
  41. Perlis, R. H., Green, J., Simonson, M., Ognyanova, K., Santillana, M., Lin, J., Quintana, A., Chwe, H., Druckman, J., Lazer, D., Baum, M. A., & Della Volpe, J. (2021). Association between social media use and self-reported symptoms of depression in US adults. JAMA Network Open, 4(11), e2136113. doi: 10.1001/jamanetworkopen.2021.36113 [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Pew Research Center (2022). Smartphone ownership and social media use among older adults continue to grow. https://www.pewresearch.org/fact-tank/2022/01/13/share-of-those-65-and-older-who-are-tech-users-has-grown-in-the-past-decade/ft_2022-01-13_techbyage_01/
  43. Ployhart, R., & MacKenzie, W. (2014). Two waves of measurement do not a longitudinal study make. In Lance C., Vandenberg R. (Eds.), More statistical and methodological myths and urban legends (pp. 85–99). Routledge. [Google Scholar]
  44. Sala, E., Cerati, G., & Gaia, A. (2021). Are social media users more satisfied with their life than non-users? A study on older Italians. Ageing & Society, 43(1), 76–88. doi: 10.1017/S0144686X21000416 [DOI] [Google Scholar]
  45. Schwaba, T., & Bleidorn, W. (2021). Log on and prosper? Little evidence for codevelopment between psychological adjustment and technology use in older adulthood. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 76(1), 67–77. doi: 10.1093/geronb/gbaa073 [DOI] [PubMed] [Google Scholar]
  46. Sharifian, N., Kraal, A. Z., Zaheed, A. B., Sol, K., Morris, E. P., & Zahodne, L. B. (2021). Measurement invariance of social media use in younger and older adults and links to socioemotional health. Innovation in Aging, 5(2), 1–11. doi: 10.1093/geroni/igab009 [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Sinclair, T. J., & Grieve, R. (2017). Facebook as a source of social connectedness in older adults. Computers in Human Behavior, 66, 363–369. doi: 10.1016/j.chb.2016.10.003 [DOI] [Google Scholar]
  48. Slater, M. D. (2007). Reinforcing spirals: the mutual influence of media selectivity and media effects and their impact on individual behavior and social identity. Communication Theory, 17(3), 281–303. doi: 10.1111/j.1468-2885.2007.00296.x [DOI] [Google Scholar]
  49. Smith, K., Chappell, A., Spalding, R., Wilson, J., Haliwa, I., & Strough, J. (2021). Older age and other protective factors of mental health during the COVID-19 pandemic in the US. Innovation in Aging, 5(S1), 720725–720. doi: 10.1093/geroni/igab046.2688 [DOI] [Google Scholar]
  50. Soror, A., Steelman, Z. R., & Turel, O. (2021). Exhaustion and dependency: A habituation-sensitization perspective on the duality of habit in social media use. Information Technology & People, 35(1), 67–95. doi: 10.1108/itp-11-2019-0603 [DOI] [Google Scholar]
  51. Taylor, W. D. (2014). Depression in the elderly. New England Journal of Medicine, 371(13), 1228–1236. doi: 10.1056/nejmcp1402180 [DOI] [PubMed] [Google Scholar]
  52. Van der Schuur, W. A., Baumgartner, S. E., & Sumter, S. R. (2019). Social media use, social media stress, and sleep: Examining cross-sectional and longitudinal relationships in adolescents. Health Communication, 34(5), 552–559. doi: 10.1080/10410236.2017.1422101 [DOI] [PubMed] [Google Scholar]
  53. Vanoh, D., Shahar, S., Yahya, H. M., & Hamid, T. A. (2016). Prevalence and determinants of depressive disorders among community-dwelling older adults: Findings from the towards useful aging study. International Journal of Gerontology, 10(2), 81–85. doi: 10.1016/j.ijge.2016.02.001 [DOI] [Google Scholar]
  54. Vuorre, M., Orben, A., & Przybylski, A. K. (2021). There is no evidence that associations between adolescents’ digital technology engagement and mental health problems have increased. Clinical Psychological Science, 9(5), 823–835. doi: 10.1177/2167702621994549 [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Yang, X., Yip, B. H., Mak, A. D., Zhang, D., Lee, E. K., & Wong, S. Y. (2021). The differential effects of social media on depressive symptoms and suicidal ideation among the younger and older adult population in Hong Kong during the COVID-19 pandemic: Population-based cross-sectional survey study. JMIR Public Health and Surveillance, 7(5), e24623. doi: 10.2196/24623 [DOI] [PMC free article] [PubMed] [Google Scholar]

Articles from Innovation in Aging are provided here courtesy of Oxford University Press

RESOURCES