Associations between a one-shot delay discounting measure and age, income, education and real-world impulsive behavior

Abstract

There has been discussion over the extent to which delay discounting – as prototypically shown by a preference for a smaller-sooner sum of money over a larger-later sum – measures the same kind of impulsive preferences that drive non-financial behavior. To address this issue, a dataset was analyzed containing 42,863 participants’ responses to a single delay-discounting choice, along with self-report behaviors that can be considered as impulsive. Choice of a smaller-sooner sum was associated with several demographics: younger age, lower income, and lower education; and impulsive behaviors: earlier age of first sexual activity and recent relationship infidelity, smoking, and higher body mass index. These findings suggest that at least an aspect of delay discounting preference is associated with a general trait influencing other forms of impulsivity, and therefore that high delay discounting is another form of impulsive behavior.

Introduction

Given the choice, people generally prefer to receive pleasant things – such as money, food, or in the case of smokers, nicotine – sooner rather than later, and the attractiveness of a reward is reduced by increasing delay until its receipt. Often, people are willing to accept a smaller-sooner reward in lieu of a larger-later reward. Thus, most people would prefer $999 in a day’s time to $1000 in a year’s time. Some people would be willing to forgo much more, for example choosing $500 in a day over $1000 in a year. The more that people are willing to give up to reduce the delay to receipt of a reward, the higher their delay discounting.

From this description, it is clear that high levels of delay discounting are a form of financial impulsivity, a preference for immediate gratification over long-term financial benefits. Although choosing a smaller-sooner sum over a larger-later sum is often described as being an ‘impulsive’ choice, there is equivocal empirical evidence for (e.g., de Wit et al., 2007, Kirby et al., 1999) and against (e.g., McLeish and Oxoby, 2007, Mitchell, 1999, Reynolds et al., 2006) an association between discounting and trait-level impulsivity. This has led to discussion about the extent to which ‘impulsive’ choices in a delay discounting procedure generalize beyond the financial domain.

Many attempts to explain individual differences in delay discounting have focused on demographics, in particular, age. Green, Fry, and Myerson (1994) found that children discounted future rewards more than younger adults (aged around 20), who in turn discounted future rewards more than older adults (aged around 68; see also Green, Myerson, & Ostaszewski, 1999). Green, Myerson, Lichtman, Rosen, and Fry (1996) compared the discounting rate of 30-year-old participants with income-matched 70-year-old participants, and found no difference. Combining their findings with those from Green et al. (1994), they hypothesized that discount rate was high in young adults, declined to age 30, and remained relatively constant after that. A similar finding was reported by Whelan and McHugh (2009). Conversely, Read and Read (2004) found a curvilinear effect of age, with middle-aged participants (mean age = 44) discounting less than both younger (mean age = 25) and older (mean age = 75) participants. Harrison, Lau, and Williams (2002) found similar trends in their data, although the only significant age effect was higher discounting among retired people than those still working. Thus, discounting appears to be lower in middle age than in young adulthood, but there are conflicting accounts of how it varies between middle age and old age.

Other demographics, notably education and income, have been associated with discounting. De Wit et al. (2007) found higher education associated with lower discounting. Similarly, Jaroni, Wright, Lerman, and Epstein (2004) found a negative association between education and delay discounting in smokers (see also Harrison et al., 2002). De Wit et al. (2007) found that higher income was associated with lower discounting and lower non-planning impulsiveness. Similarly, Green et al., 1996, Harrison et al., 2002 found less discounting among high-income participants.

The effects of gender are less clear-cut, and for delay discounting there is evidence for women discounting more than men (e.g., Reynolds et al., 2006), no difference between men and women (e.g., Harrison et al., 2002) and men discounting more than women (e.g., Kirby & Marakovic, 1996). Thus, it appears that gender differences are particularly unstable.

Part of the validation of impulsivity and discounting measures has come from the fact that they have behavioral correlates that can be seen as reflecting impulsive decision making. Several areas of impulsive behavior have been investigated. These include obesity, substance use, and sexual behavior. Research in each area will now be briefly discussed.

In deciding whether to resist an opportunity for an extradyadic sexual encounter, one pits desire for immediate sexual gratification against the long-term benefits of an existing relationship. Similarly, during adolescence, there is a conflict between the immediate excitement and social acceptance of sexual experimentation and longer-term happiness from ‘saving oneself’ for a more satisfying significant romantic attachment, along with avoiding risks of sexually transmitted diseases, pregnancy, and, at least for women in many cultures, a negative reputation. As such, age of first sexual activity and infidelity rate might be expected to be correlated with other measures of impulsivity.

McAlister, Pachana, and Jackson (2005) found that propensity toward extradyadic kissing or sexual activity was positively associated with dysfunctional impulsivity. Furthermore, they found strong first-order correlations between dysfunctional impulsivity and age of first sexual encounter: higher impulsivity was associated with earlier age of first sexual encounter. Clift, Wilkins, and Davidson (1993) found that of patients attending a Genito-urinary medicine clinic, those who reported having sex without condoms scored higher on impulsivity than those who used condoms (see also Horvath & Zuckerman, 1993). McCoul and Haslam (2001) found similar results in heterosexual, but not homosexual, men.

In deciding whether or not to smoke cigarettes, drink alcohol in large quantities, or take illicit drugs, people make a choice between the short-term pleasure of the substance in question and longer-term health and financial benefits of abstinence. Unsurprisingly, then, there is a large body of research investigating the association between delay discounting, impulsivity and substance use (for a review, see Reynolds, 2006). The general pattern is similar for cigarette, alcohol and illicit drug use: regular users tend to discount future money more than non- or ex-users.

There is robust evidence that smokers discount future rewards more than non-smokers (e.g., Bickel et al., 1999, Mitchell, 1999, Reynolds et al., 2004). Similar, although more equivocal, results have been found for alcohol consumption. Some research has found a positive association between alcohol consumption and delay discounting (e.g., Mitchell et al., 2005, Vuchinich and Simpson, 1998), whereas other research has failed to find an effect (e.g., Kirby & Petry, 2004).

Finally, there is strong evidence that regular use of illicit drugs such as heroin and cocaine is associated with higher discounting of delayed rewards. For example Kirby et al. (1999) found higher discounting of money among opioid-dependent participants than controls. Madden, Petry, Badger, and Bickel (1997) found a similar result using hypothetical money, and a similar finding has been shown in cocaine-using individuals (Coffey, Gudleski, Saladin, & Brady, 2003).

Another form of impulsive behavior is choosing the pleasure of overeating now over longer-term health benefits of limiting one’s consumption. As such, one might expect body mass index (BMI) to be related to other measures of time preference including delay discounting. There is some evidence for this. Recently, Zhang and Rashad (2008) reported an association between BMI and time preference, measured by self-reported willpower (participants who reported having problems with willpower were categorized as having more immediate time preference). Borghans and Golsteyn (2006) found a marginally significant association between delay discounting rate and BMI in men, but not in women. They did, however, find a significant negative association between BMI and ability to manage finances and control expenditure, suggesting an association between gustatory and financial impulsivity. Conversely, Weller, Cook, Avsar, and Cox (2008) found higher levels of discounting in obese women relative to controls, but no such effect in men.

However, Nederkoorn, Smulders, Havermans, Roefs, and Jansen (2006) found no association between delay discounting and BMI, or between trait-level impulsiveness and BMI. Similarly, Epstein et al. (2003) found no association between BMI and delay discounting.