We spend a lot of time debating the characteristics of generations—are baby boomers really selfish and entitled, are millennials really narcissists, and the latest, has the next generation (whatever it is going to be called) already been ruined by cellphones? Many academics—and many consultants—argue that generations are distinct and that organizations, educators, and even parents need to accommodate them. These classifications are often met with resistance from those they supposedly represent, as most people dislike being represented by overgeneralizations, and these disputes only fuel the debate around this contentious topic.
But the science on “generations” does not back up these distinctions. In fact, solid evidence supporting generations, their characteristics, or even their existence, is lacking. In short, the science shows that generations are not a thing.
It is important to be clear what not a thing means. It does not mean that people today are the same as people 80 years ago or that anything else is static. Times change and so do people. However, the idea that distinct generations capture and represent these changes is unsupported.
What is a generation? Those who promote the concept define it as a group of people who are roughly the same age and who were influenced by a set of significant events. These experiences supposedly create commonalities, making those in the group more similar to each other and more different from other groups now and from groups of the same age in the past.
In line with the definition, there is a commonly held perception that people growing up around the same time and in the same place must have some sort of universally shared set of experiences and characteristics. It helps that the idea of generations intuitively makes sense. But the science does not support it. In fact, most of the research findings showing distinct generations are explained by other causes, have serious scientific flaws, or both.
Before continuing, I’d like to acknowledge that it is somewhat ironic to use generational labels and research to argue that generations do not exist. But because proving the negative is impossible, I am forced to a certain extent to rely on the research that has been done on generations (and some of it is sound) to show that generations are not a thing.
That said, let’s start with one of the biggest problems in generations research—inappropriate attribution of cause. Most studies that purport to show generational differences likely found something else. For example, millennials score lower on job satisfaction than Gen Xers, but are millennials really a less satisfied generation? Early in their careers, Xers were also less satisfied than baby boomers. As people get older, they are more likely to leave jobs they do not like and migrate toward ones they do. Ask a 22-year-old in her first job and a 42-year-old in her fourth about job satisfaction. Would you be surprised that the older one, who has had the chance to move around, explore, and advance in her career, likes her job more? This is an age effect, not a generational effect.
Want another one? Perhaps the most commonly cited generational effect is the so-called epidemic of narcissism among young people today. (Of course, narcissism cannot be an epidemic as it is not an infectious disease, but I digress.) Numerous books, articles, and pundits have claimed that millennials are much more narcissistic than young people in the past.
Guess what? The science does not back this up either. Our research shows that while narcissism among young people did increase slightly through the mid-2000s (about 1.8 points on a 40-point scale), it is now back to where it was in the 1980s. That’s right, on average, millennials are no more narcissistic now than Xers or boomers were when they were in their 20s, and one study has even found they might be less so than generations past. While millennials today may be more narcissistic than Xers or boomers are today, that is because young people are pretty narcissistic regardless of when they are young. This too is an age effect.
Final example. Research shows that millennials joining the Army now show more pride in their service than boomers or Xers did when they joined 20-plus years ago. Is this a generational effect? Nope. Everyone in the military now shows more pride on average than 20 years ago because of 9/11. The terrorist attack increased military pride across the board. This is known as a period effect and it doesn’t have anything to do with generations.
Another problem—identifying true generational effects is methodologically very hard. The only way to do it would be to collect data from multiple longitudinal panels. Individuals in the first panel would be measured at the start of the study and then in subsequent years with new panels added every year thereafter, allowing assessment of whether people were changing because they were getting older (age effects), because of what was happening around them (period effects), or because of their generation (cohort effects). Unfortunately, such data sets pretty much do not exist. Thus, we’re never really able to determine why a change occurred.
Generations researchers have collected data using other approaches besides multiple longitudinal panels (e.g., cross-sectional studies and time-lagged panels). Among their many issues, a serious concern with any group-based data such as these is that they can suffer from what can be called a Within and Between Analysis problem. The problem arises comparing groups (i.e., generations) to each other when individuals within the groups vary a lot. An example, national culture, helps explain this problem.
According to one national-culture model, people from the United States are, on average, relatively individualistic, indulgent, and uncomfortable with hierarchical order. Conversely, people from China are generally group-oriented, restrained, and comfortable with hierarchy. However, these countries are so large and diverse that they each have millions of individuals who are more similar to the “averages” of the other country than to their own. There is more variation within countries than between them, and thus figuring out exactly what characteristics truly represent a nation’s culture is almost impossible.
Generations data has the same problem. Are some millennials narcissistic? Are some boomers selfish? Sure, but there are many who are not and whose profiles mirror other generations. Hence, saying someone is a millennial doesn’t say anything about them as an individual because the variation within millennials (or Xers or boomers) is often more extensive than the variation between generations.
Given these design and data issues, it is not surprising that researchers have tried a variety of different statistical techniques to massage (aka torture) the data in an attempt to find generational differences. Studies showing generational differences have used statistical techniques like analysis of variance (ANOVA) and cross-temporal meta-analysis (CTMA), neither of which is capable of actually attributing the differences to generations.
The statistical challenge derives from the problem we have already raised—generations (i.e., cohorts) are defined by age and period. As such, mathematically separating age, period, and cohort effects is very difficult because they are inherently confounded with one another. Their linear dependency creates what is known as an identification problem, and unless one has access to multiple longitudinal panels like I described above, it is impossible to statistically isolate the unique effect of any one factor. Thus, studies that have used these statistical techniques on cross-sectional data or time-lagged panels and that have reported finding some generational difference may be capturing age (e.g., job satisfaction) or period (e.g., pride in the Army), but there is no way they provide statistical evidence for cohort (e.g., narcissism) effects.
One recent effort investigated the statistics problem, applying several different techniques to the same data sets, and found that different statistics produced different results. For example, depending on whether ANOVA, CTMA, or another technique called cross-classified hierarchical linear modeling (CCHLM) was used and when the data was collected, a researcher might variously conclude that employees with the lowest work stress were the GI generation, the silent generation, or that everyone, including boomers and Xers, were all equally stressed. Or looking at job satisfaction again through a different lens, ANOVAs generally showed that millennials were lowest, CTMAs sometimes showed millennials were lowest and sometimes Xers were lowest, and CCHLM found either a small generational effect or nothing at all. It is hard to argue that generations are a thing if evidence for them depends so heavily on which statistic is used.
OK, so what? Even if generations are not a thing, does it matter? What’s the big deal with this kind of introspection? Isn’t it harmless? There are several problems with this thinking.
First, relying on flawed generational science leads to poor advice and bad decisions. An analogy: Women live longer than men, on average. Why? They engage in fewer risky behaviors, take better care of themselves, and have two X chromosomes, giving them backups in case of mutations. But if you are a man and you go to the doctor and ask how to live longer, she doesn’t tell you, “Be a woman.” She says eat better, exercise, and don’t do stupid stuff. Knowing the why guides the recommendation.
Now imagine you are a manager trying to retain your supposedly job-hopping, commitment-averse millennial employees and you know that Xers and boomers are less likely to leave their jobs. If you are that manager, you wouldn’t tell your millennial employees to “be a boomer” or “grow older” (nor would you decide to hire boomers or Xers rather than millennials—remember that individuals vary within populations). Instead, you should focus on addressing benefits, work conditions, and other factors that are reasons for leaving.
Second, this focus on generational distinctions wastes resources. Take the millennials-as-commitment-averse-job-hoppers stereotype. Based on this belief, consultants sell businesses on how to recruit and retain this mercurial generation. But are all (or even most) millennials job-hopping commitment avoiders? Survey research shows that millennials and Xers at the same point in their careers are equally likely to stay with their current employer for five or more years (22 percent v. 21.8 percent). It makes no sense for organizations to spend time and money changing HR policies when employees are just as likely to stick around today as they were 15 years ago.
Third, generations perpetuate stereotyping. Ask millennials if they are narcissistic job-hoppers and most of them will rightly be offended. Treat boomers like materialistic achievement seekers and see how it affects their work quality and commitment. We finally are starting to recognize that those within any specific group of people are varied individuals, and we should remember those same principles in this context too. We are (mostly) past it being acceptable to stereotype and discriminate against women, minorities, and the disabled. Why is it OK to do so to millennials or boomers?
The solutions are fairly straightforward, albeit challenging, to implement. To start, we need to focus on the why when talking about whether groups of people differ. The reasons why any generation should be different have only been generally discussed, and the theoretical mechanism that supposedly creates generations has not been fully fleshed out. We’ve jumped past testing that theory to acting on it. We need to go back to the science, figure out the why, and then use that knowledge to make better, more appropriate decisions.
Next, we need to quit using these nonsensical generations labels, because they don’t mean anything. The start and end years are somewhat arbitrary anyway. The original conceptualization of social generations started with a biological generational interval of about 20 years, which historians, sociologists and demographers (for one example, see Strauss and Howe, 1991) then retrofitted with various significant historical events that defined the period.
The problem with this is twofold. First, such events do not occur in nice, neat 20-year intervals. Second, not everyone agrees on what the key events were for each generation, so the start and end dates also move around depending on what people think they were. One review found that start and end dates for boomers, Xers, and millennials varied by as many as nine years, and often four to five, depending on the study and the researcher. As with the statistical problem, how can distinct generations be a thing if simply defining when they start and when they end varies so much from study to study?
In the end, the core scientific problem is that the pop press, consultants, and even some academics who are committed to generations don’t focus on the whys. They have a vested interest in selling the whats (Generation Me has reportedly sold more than 115,000 copies, and Google “generations consultants” and see how many firms are dedicated to promulgating these distinctions), but without the science behind them, any prescriptions are worthless or even harmful.
Generations and generational differences are intriguing and inherently appealing concepts. As such, the media will keep on reporting on them, academics will publish, pundits will talk, and consultants will sell to whoever is buying. But the science says that, despite their popularity, generations simply aren’t a thing. And until we recognize this, we will continue to waste time and resources while failing to understand how people really are, and are not, different.
